korrents

Design

What the people who shaped how things are made think about design — in their own words, and when they said it.

Tuomas Artman, Jason Fried, Paul Graham, David Heinemeier Hansson, Mitchell Hashimoto, John Maeda, Craig Mod, Jakob Nielsen and 16 more Tuomas Artman, Jason Fried, Paul Graham, David Heinemeier Hansson, Mitchell Hashimoto, John Maeda, Craig Mod, Jakob Nielsen, Don Norman, Sundar Pichai, Dieter Rams, Guillermo Rauch, Naval Ravikant, Karri Saarinen, Ryan Singer, Derek Sivers, Jared Spool, Peter Steinberger, Edward Tufte, Bret Victor, Amelia Wattenberger, Sebastiaan de With, Luke Wroblewski, Julie Zhuo

Five viewsour reading; the words are theirs

Good design is less, and it disappears into the hand Less, but better – because it concentrates on the essential aspects, and the products are not burdened with non-essentials. — Dieter Rams, by 2012 Mitchell Hashimoto, Julie Zhuo, Amelia Wattenberger, Sebastiaan de With, Dieter Rams, John Maeda, Paul Graham 11 claims

The principles pile: as little design as possible, nothing arbitrary, obvious to use, never noticed.

The best terminal multiplexer is the one nobody has to learn; needing to be learned at all is the first design failure.

Filed here becauseIt says needing to be learned is the first design failure; a claim on the same subject from the same week says a couple of days of learning is a fair price.

  1. Mitchell Hashimoto Creator of Ghostty and co-founder of Superlogical I can't tell you the number of people over my career that have said "I tried learning tmux" First problem: the fact you have to "learn it" at all. The world's best multiplexer is the one you don't have to learn. It just works. @mitchellh on Xx.com · 28 Aug 2026
    Their wordsI can't tell you the number of people over my career that have said "I tried learning tmux" First problem: the fact you have to "learn it" at all. The world's best multiplexer is the one you don't have to learn. It just works.

    ↗@mitchellh on Xx.com

    28 Aug 2026

Obviousness comes from conforming to people's existing mental models: do not reinvent a common UI pattern unless the new one is at least twice as good.

Mental modelConventions (Jakob's law)

  1. Julie Zhuo Co-founder of Sundial; former vice president of product… Obviousness comes from conforming to people’s existing mental models. Don’t waste time reinventing common UI patterns or paradigms unless they are at least 2x better, or you have some critical brand reason to do so. The Looking Glass: How to tell if work is beautifullg.substack.com · 16 Nov 2023
    Their wordsObviousness comes from conforming to people’s existing mental models. Don’t waste time reinventing common UI patterns or paradigms unless they are at least 2x better, or you have some critical brand reason to do so.

    ↗The Looking Glass: How to tell if work is beautifullg.substack.com

    16 Nov 2023

Better design does not mean more design: the most obvious designs are often invisible.

Mental modelLess, but better

  1. Julie Zhuo Co-founder of Sundial; former vice president of product… Better design does not mean more design. Often, the most obvious designs are invisible. The Looking Glass: How to tell if work is beautifullg.substack.com · 16 Nov 2023
    Their wordsBetter design does not mean more design. Often, the most obvious designs are invisible.

    ↗The Looking Glass: How to tell if work is beautifullg.substack.com

    16 Nov 2023

Good tools make it clear how they should be used and how they should not; a chat box looks the same as a search box, a login form and a credit card field.

Mental modelAffordances and signifiers

  1. Amelia Wattenberger by Designer and engineer; formerly principal research… Good tools make it clear how they should be used. And more importantly, how they should not be used. Why Chatbots Are Not the Futurewattenberger.com · by 1 May 2023
    Their wordsGood tools make it clear how they should be used. And more importantly, how they should not be used.

    ↗Why Chatbots Are Not the Futurewattenberger.com

    by 1 May 2023

You know a design is great when it disappears: a superb typeface, good signage or a well-designed door handle is never noticed, only the absence of frustration.

Mental modelDesign is how it works

  1. Sebastiaan de With Designer and co-founder of Lux, makers of Halide When you come across an article set in a superb typeface for reading, some excellent airport signage or a well designed door handle, you don’t think about it or even notice it. If anything you might just notice being happier, or simply not frustrated. That’s how you know a design is great: it disappears. A Spectre Retrospective – Part 2: Designlux.camera · 11 Apr 2019
    Their wordsWhen you come across an article set in a superb typeface for reading, some excellent airport signage or a well designed door handle, you don’t think about it or even notice it. If anything you might just notice being happier, or simply not frustrated. That’s how you know a design is great: it disappears.

    ↗A Spectre Retrospective – Part 2: Designlux.camera

    11 Apr 2019

Good design is as little design as possible: less, but better, because it concentrates on the essentials.

Mental modelLess, but better

  1. Dieter Rams by Industrial designer; head of design at Braun from 1961… Less, but better – because it concentrates on the essential aspects, and the products are not burdened with non-essentials. Dieter Rams: ten principles for good designvitsoe.com · by 3 Aug 2012
    Their wordsLess, but better – because it concentrates on the essential aspects, and the products are not burdened with non-essentials.

    ↗Dieter Rams: ten principles for good designvitsoe.com

    by 3 Aug 2012

Good design is unobtrusive: products that fulfil a purpose are tools, neither decorative objects nor works of art, and should leave room for the user's self-expression.

Mental modelDesign is how it works

  1. Dieter Rams by Industrial designer; head of design at Braun from 1961… Products fulfilling a purpose are like tools. They are neither decorative objects nor works of art. Their design should therefore be both neutral and restrained, to leave room for the user’s self-expression. Dieter Rams: ten principles for good designvitsoe.com · by 3 Aug 2012
    Their wordsProducts fulfilling a purpose are like tools. They are neither decorative objects nor works of art. Their design should therefore be both neutral and restrained, to leave room for the user’s self-expression.

    ↗Dieter Rams: ten principles for good designvitsoe.com

    by 3 Aug 2012

Good design makes a product understandable: at best it is self-explanatory.

Mental modelConservation of complexity

  1. Dieter Rams by Industrial designer; head of design at Braun from 1961… It clarifies the product’s structure. Better still, it can make the product talk. At best, it is self-explanatory. Dieter Rams: ten principles for good designvitsoe.com · by 3 Aug 2012
    Their wordsIt clarifies the product’s structure. Better still, it can make the product talk. At best, it is self-explanatory.

    ↗Dieter Rams: ten principles for good designvitsoe.com

    by 3 Aug 2012

Good design is thorough down to the last detail: nothing must be arbitrary, and care in the design process shows respect towards the user.

Mental modelCraft in the details

  1. Dieter Rams by Industrial designer; head of design at Braun from 1961… Nothing must be arbitrary or left to chance. Care and accuracy in the design process show respect towards the user. Dieter Rams: ten principles for good designvitsoe.com · by 3 Aug 2012
    Their wordsNothing must be arbitrary or left to chance. Care and accuracy in the design process show respect towards the user.

    ↗Dieter Rams: ten principles for good designvitsoe.com

    by 3 Aug 2012

Good design is simple, because being forced to be simple forces a maker to face the real problem and deliver substance instead of ornament.

Mental modelLess, but better

  1. Paul Graham Co-founder of Y Combinator and Viaweb When you're forced to be simple, you're forced to face the real problem. When you can't deliver ornament, you have to deliver substance. Taste for Makerspaulgraham.com · 1 Feb 2002
    Their wordsWhen you're forced to be simple, you're forced to face the real problem. When you can't deliver ornament, you have to deliver substance.

    ↗Taste for Makerspaulgraham.com

    1 Feb 2002

It is honest and it is felt: speed, behaviour and truth matter more than looks Speed in software is probably the most valuable, least valued asset. To me, speedy software is the difference between an application smoothly integrating into your life, and one called upon with great reluctance. — Craig Mod, 2019 David Heinemeier Hansson, Jakob Nielsen, Jason Fried, Julie Zhuo, Craig Mod, Derek Sivers, Dieter Rams, Edward Tufte, Paul Graham 11 claims

What a design owes the person using it, from latency to not lying.

Aesthetics is truth: when something is beautiful it is likely to be correct, in code as in mathematics and physics.

  1. David Heinemeier Hansson Creator of Ruby on Rails, CTO of 37signals I mean, I think aesthetics is truth. When something is beautiful, it's likely to be correct. I think this is true in mathematics. This is true in physics. This is true in a lot of different domains that when you arrive at something that has the correct aesthetic quality. DHH’s new way of writing codeyoutube.com · 8 Apr 2026 · 0:28:33 into the videoAll korrents from this video
    Their wordsI mean, I think aesthetics is truth. When something is beautiful, it's likely to be correct. I think this is true in mathematics. This is true in physics. This is true in a lot of different domains that when you arrive at something that has the correct aesthetic quality.

    Where this was said

    Watch from 0:28:33 plays here

    ↗DHH’s new way of writing codeyoutube.com

    8 Apr 2026 · video · 1h 47m · spoken · machine transcript

Companies pay for design so that customers buy more and look at more ads; internet-era UX was manipulation, whatever designers set out to do.

Mental modelIncentives

Filed here becauseNot a claim about good design but about what design is bought for; filed here as what an honest design is up against.

  1. Jakob Nielsen Usability pioneer; co-founder of Nielsen Norman Group… The reason companies pay for design is to get customers to buy more and users to look at more advertisements. Intent by Discovery: Designing the AI User Experiencejakobnielsenphd.substack.com · 26 Mar 2026
    Their wordsThe reason companies pay for design is to get customers to buy more and users to look at more advertisements.

    ↗Intent by Discovery: Designing the AI User Experiencejakobnielsenphd.substack.com

    26 Mar 2026

The most interesting design is design that changes your behaviour, more than what something looks like or how it works.

Mental modelDesign is how it works

  1. Jason Fried Co-founder and CEO of 37signals (Basecamp, HEY) In some cases, design is what something looks like. In other cases, design is how something works. But the most interesting designs to me are when design changes your behavior. Even the smallest details can change how someone interacts with something. When design drives behaviorworld.hey.com · 7 Oct 2025
    Their wordsIn some cases, design is what something looks like. In other cases, design is how something works. But the most interesting designs to me are when design changes your behavior. Even the smallest details can change how someone interacts with something.

    ↗When design drives behaviorworld.hey.com

    7 Oct 2025

Do not judge a product's beauty by its aesthetics; judge it by how it made you feel.

Mental modelDesign is how it works

  1. Julie Zhuo Co-founder of Sundial; former vice president of product… Do not judge a product’s beauty by its aesthetics; judge it by how it made you feel. The Looking Glass: How to tell if work is beautifullg.substack.com · 16 Nov 2023
    Their wordsDo not judge a product’s beauty by its aesthetics; judge it by how it made you feel.

    ↗The Looking Glass: How to tell if work is beautifullg.substack.com

    16 Nov 2023

Speed in software is probably the most valuable, least valued asset.

Mental modelSpeed is a feature

  1. Craig Mod Writer, photographer and software maker based in Japan Speed in software is probably the most valuable, least valued asset. To me, speedy software is the difference between an application smoothly integrating into your life, and one called upon with great reluctance. Fast Software, the Best Softwarecraigmod.com · 1 Jul 2019
    Their wordsSpeed in software is probably the most valuable, least valued asset. To me, speedy software is the difference between an application smoothly integrating into your life, and one called upon with great reluctance.

    ↗Fast Software, the Best Softwarecraigmod.com

    1 Jul 2019

The best touch interfaces are physical buttons; on a screen, to feel tactile is to move without latency.

Mental modelSpeed is a feature

  1. Craig Mod Writer, photographer and software maker based in Japan Excellent touch interfaces feel tactile. The best “touch interfaces,” in my opinion, are actual buttons, things with physicality, haptics. But on a screen, to feel tactile is to move without latency. Fast Software, the Best Softwarecraigmod.com · 1 Jul 2019
    Their wordsExcellent touch interfaces feel tactile. The best “touch interfaces,” in my opinion, are actual buttons, things with physicality, haptics. But on a screen, to feel tactile is to move without latency.

    ↗Fast Software, the Best Softwarecraigmod.com

    1 Jul 2019

The feeds businesses design to be addictive are obstacles to anyone whose aim is to make something.

  1. Derek Sivers Founded CD Baby in 1998 and sold it in 2008 Every business wants to get you addicted to their infinite updates, pings, chats, messages, and news. But if what you want out of life is to create, then those are your obstacles. Disconnectsive.rs · 27 Jul 2016All korrents from this piece
    Their wordsEvery business wants to get you addicted to their infinite updates, pings, chats, messages, and news. But if what you want out of life is to create, then those are your obstacles.

    ↗Disconnectsive.rs

    27 Jul 2016

Good design is honest: it does not make a product seem more innovative, powerful or valuable than it is, and does not manipulate with promises that cannot be kept.

  1. Dieter Rams by Industrial designer; head of design at Braun from 1961… It does not make a product more innovative, powerful or valuable than it really is. It does not attempt to manipulate the consumer with promises that cannot be kept. Dieter Rams: ten principles for good designvitsoe.com · by 3 Aug 2012
    Their wordsIt does not make a product more innovative, powerful or valuable than it really is. It does not attempt to manipulate the consumer with promises that cannot be kept.

    ↗Dieter Rams: ten principles for good designvitsoe.com

    by 3 Aug 2012

Good design is long-lasting: it avoids being fashionable and therefore never appears antiquated.

  1. Dieter Rams by Industrial designer; head of design at Braun from 1961… It avoids being fashionable and therefore never appears antiquated. Unlike fashionable design, it lasts many years – even in today’s throwaway society. Dieter Rams: ten principles for good designvitsoe.com · by 3 Aug 2012
    Their wordsIt avoids being fashionable and therefore never appears antiquated. Unlike fashionable design, it lasts many years – even in today’s throwaway society.

    ↗Dieter Rams: ten principles for good designvitsoe.com

    by 3 Aug 2012

Audience boredom is usually a content failure, not a decoration failure: if the numbers are boring you have the wrong numbers.

Mental modelDesign is how it works

  1. Edward Tufte Statistician and professor emeritus at Yale If your numbers are boring, then you've got the wrong numbers. If your words or images are not on point, making them dance in color won't make them relevant. Audience boredom is usually a content failure, not a decoration failure. PowerPoint Is Evilwired.com · 1 Sept 2003
    Their wordsIf your numbers are boring, then you've got the wrong numbers. If your words or images are not on point, making them dance in color won't make them relevant. Audience boredom is usually a content failure, not a decoration failure.

    ↗PowerPoint Is Evilwired.com

    1 Sept 2003

Good design is timeless: if something is ugly it cannot be the best solution, and someone will eventually find a better one.

  1. Paul Graham Co-founder of Y Combinator and Viaweb if something is ugly, it can't be the best solution. There must be a better one, and eventually someone will discover it. Taste for Makerspaulgraham.com · 1 Feb 2002
    Their wordsif something is ugly, it can't be the best solution. There must be a better one, and eventually someone will discover it.

    ↗Taste for Makerspaulgraham.com

    1 Feb 2002

Taste is real, one person has to hold it, and the makers should work in the medium Saying that taste is just personal preference is a good way to prevent disputes. The trouble is, it's not true. You feel this when you start to design things. — Paul Graham, 2002 David Heinemeier Hansson, Luke Wroblewski, Julie Zhuo, Jakob Nielsen, Jared Spool, Karri Saarinen, Paul Graham 11 claims · claims collide here

Who decides, and whether they should be building the thing themselves.

Where these claims collide

These pull against each other

Saarinen says a product needs one person of great taste who gets furious about every small detail. Spool says everyone who renders their intention is a designer, and that the experienced designer’s job is therefore to teach the rest of the team rather than own the outcome. Both are answers to "where does design quality come from", and they distribute the authority opposite ways.

Most organisations are bottlenecked on ideas, vision and taste rather than on implementation, so making code cheap does not make them productive.

Mental modelBottlenecks

  1. David Heinemeier Hansson Creator of Ruby on Rails, CTO of 37signals The other thing I'd say is that most organizations don't know what they want. They don't know how to make it better. They're not bottlenecked on implementation. They're bottlenecked on ideas. They're bottlenecked on vision. They're bottlenecked on taste. And if you don't have those element in excess of your implementational capacity, it doesn't help. So you can make a lot of shitty ideas come true, then what? DHH: Future of Programming, AI, Agentic Engineering, Vibe Coding & Linux | Lex Fridman Podcast #501youtube.com · 26 Aug 2026 · 0:20:13 into the videoAll korrents from this video
    Their wordsThe other thing I'd say is that most organizations don't know what they want. They don't know how to make it better. They're not bottlenecked on implementation. They're bottlenecked on ideas. They're bottlenecked on vision. They're bottlenecked on taste. And if you don't have those element in excess of your implementational capacity, it doesn't help. So you can make a lot of shitty ideas come true, then what?

    Where this was said

    Watch from 0:20:13 plays here

    ↗DHH: Future of Programming, AI, Agentic Engineering, Vibe Coding & Linux | Lex Fridman Podcast #501youtube.com

    26 Aug 2026 · video · 5h 15m · spoken · official transcript

Without the handoff, everyone builds: designer, developer and product manager can all work in the codebase at once, and that changes everything.

  1. Luke Wroblewski Product designer and entrepreneur But without the handoff, everyone builds. And that makes all the difference. The New Designer/Developer Collaborationlukew.com · 15 Apr 2026
    Their wordsBut without the handoff, everyone builds. And that makes all the difference.

    ↗The New Designer/Developer Collaborationlukew.com

    15 Apr 2026

Designers should find what the spec should be and build it in HTML and CSS themselves, not make a product manager's spec look pretty.

Mental modelMake it real

  1. David Heinemeier Hansson Creator of Ruby on Rails, CTO of 37signals Designers at 37 Signals are not just here to make a spec look pretty. They're here to find what the spec should be. They're product managers in many ways. DHH’s new way of writing codeyoutube.com · 8 Apr 2026 · 0:23:27 into the videoAll korrents from this video
    Their wordsDesigners at 37 Signals are not just here to make a spec look pretty. They're here to find what the spec should be. They're product managers in many ways.

    Where this was said

    Watch from 0:23:27 plays here

    ↗DHH’s new way of writing codeyoutube.com

    8 Apr 2026 · video · 1h 47m · spoken · machine transcript

Designers should code, and AI coding agents have reopened that door after a decade in which front-end complexity closed it.

  1. Luke Wroblewski Product designer and entrepreneur There's a question that never goes away in design: should designers code? My answer has always been yes. But for a decade or so, the complexity of front-end development made it impractical for most. Thankfully, AI coding agents have reopened the door. Should Designers "Code"?lukew.com · 7 Apr 2026
    Their wordsThere's a question that never goes away in design: should designers code? My answer has always been yes. But for a decade or so, the complexity of front-end development made it impractical for most. Thankfully, AI coding agents have reopened the door.

    ↗Should Designers "Code"?lukew.com

    7 Apr 2026

Taste is the ability to distinguish excellence from mediocrity, and the hand can never consistently produce better than the eye can discern.

  1. Julie Zhuo Co-founder of Sundial; former vice president of product… Good taste: taste is the ability to distinguish excellence from mediocrity. The hand can never consistently produce better than the eye can discern. The Death of Product Development as We Know itlg.substack.com · 25 Mar 2025
    Their wordsGood taste: taste is the ability to distinguish excellence from mediocrity. The hand can never consistently produce better than the eye can discern.

    ↗The Death of Product Development as We Know itlg.substack.com

    25 Mar 2025

Vibe coding lets users design and build their own software, upending the user-centred design paradigm that has held since 1947.

  1. Jakob Nielsen Usability pioneer; co-founder of Nielsen Norman Group… The new trend toward vibe coding and vibe design may be upending the user-centered design paradigm that has remained based on the same ideology since the first UX design projects at Bell Labs, starting in 1947. Does User-Driven Design Still Need User-Centered Design?jakobnielsenphd.substack.com · 13 Mar 2025
    Their wordsThe new trend toward vibe coding and vibe design may be upending the user-centered design paradigm that has remained based on the same ideology since the first UX design projects at Bell Labs, starting in 1947.

    ↗Does User-Driven Design Still Need User-Centered Design?jakobnielsenphd.substack.com

    13 Mar 2025

Mockups existed because coding was hard; now designers can skip static mockups and go straight to building a UI that works.

  1. Jakob Nielsen Usability pioneer; co-founder of Nielsen Norman Group… Essentially, vibe coding makes it trivial to create front-end code, so designers can skip static mockups and go straight to building a UI that works. Vibe Coding and Vibe Designjakobnielsenphd.substack.com · 7 Mar 2025
    Their wordsEssentially, vibe coding makes it trivial to create front-end code, so designers can skip static mockups and go straight to building a UI that works.

    ↗Vibe Coding and Vibe Designjakobnielsenphd.substack.com

    7 Mar 2025

Everyone who renders their intention is a designer, so the experienced designer's role shifts from owning the outcome to teaching the rest of the team to render theirs.

  1. Jared Spool Co-founder of Center Centre and founder of UIE The notion of the academically-trained, experienced designer holding the keys to the design outcome fades with this definition. In fact, when we think about design in these terms, it becomes clear that many of the people rendering their intention don’t know they are designers. This implies the role of the experienced designer shifts from owning the design outcome to educating all the other designers on the team on how to best render their own intentions. Design is the Rendering of Intentarticles.centercentre.com · 30 Dec 2013
    Their wordsThe notion of the academically-trained, experienced designer holding the keys to the design outcome fades with this definition. In fact, when we think about design in these terms, it becomes clear that many of the people rendering their intention don’t know they are designers. This implies the role of the experienced designer shifts from owning the design outcome to educating all the other designers on the team on how to best render their own intentions.

    ↗Design is the Rendering of Intentarticles.centercentre.com

    30 Dec 2013

A product needs a soul, and for that it needs one person of great taste who is its living, breathing aspect and gets furious about every small detail.

  1. Karri Saarinen Co-founder and CEO of Linear For the soul, you need a person with a great taste to be the living and breathing aspect of the product. The one forging the blades. The person who thinks about the in and outs of the product and the business. Your Product Needs a Soulkarrisaarinen.com · 12 Feb 2010
    Their wordsFor the soul, you need a person with a great taste to be the living and breathing aspect of the product. The one forging the blades. The person who thinks about the in and outs of the product and the business.

    ↗Your Product Needs a Soulkarrisaarinen.com

    12 Feb 2010

Good design requires a dictator: design by committee is a synonym for bad design, because a design has to be all of a piece.

  1. Paul Graham Co-founder of Y Combinator and Viaweb I'm inclined to think there isn't-- that good design requires a dictator. One reason is that good design has to be all of a piece. Design is not just for humans, but for individual humans. If a design represents an idea that fits in one person's head, then the idea will fit in the user's head too. Design and Researchpaulgraham.com · 1 Jan 2003
    Their wordsI'm inclined to think there isn't-- that good design requires a dictator. One reason is that good design has to be all of a piece. Design is not just for humans, but for individual humans. If a design represents an idea that fits in one person's head, then the idea will fit in the user's head too.

    ↗Design and Researchpaulgraham.com

    1 Jan 2003

Taste is not just personal preference: there is such a thing as good and bad design, and a maker's taste can be wrong.

  1. Paul Graham Co-founder of Y Combinator and Viaweb Saying that taste is just personal preference is a good way to prevent disputes. The trouble is, it's not true. You feel this when you start to design things. Taste for Makerspaulgraham.com · 1 Feb 2002
    Their wordsSaying that taste is just personal preference is a good way to prevent disputes. The trouble is, it's not true. You feel this when you start to design things.

    ↗Taste for Makerspaulgraham.com

    1 Feb 2002

The user is not you: design for what they need, and study them to improve what exists In the world of high-stakes digital strategy, the “average user” is a dangerous statistical artifact. — Jakob Nielsen, 2026 Jakob Nielsen, Tuomas Artman, Julie Zhuo, Don Norman, Ryan Singer, Paul Graham 10 claims · 1 change of mind · claims collide here

Users, research, and where the breakthroughs actually come from.

Where these claims collide

These pull against each other

Norman says design research is essentially useless for new, innovative breakthroughs, however good it is at improving an existing category. Zhuo says the less your audience looks like you, the more you should distrust your own intuition — which is an argument for research exactly where your instincts are least reliable. Both are true only if breakthroughs must be designed for people like the designer.

The average user is a dangerous statistical artifact: design for the gap between the median user and the heavy-use tail, not for a phantom mean.

Mental modelThe map is not the territory

  1. Jakob Nielsen Usability pioneer; co-founder of Nielsen Norman Group… In the world of high-stakes digital strategy, the “average user” is a dangerous statistical artifact. Don’t Design for Average Usersjakobnielsenphd.substack.com · 19 Mar 2026
    Their wordsIn the world of high-stakes digital strategy, the “average user” is a dangerous statistical artifact.

    ↗Don’t Design for Average Usersjakobnielsenphd.substack.com

    19 Mar 2026

To compete with limited resources, scope down who you are building for as much as possible and solve the problems of specific people.

Mental modelLess, but better

  1. Tuomas Artman Co-founder and CTO of Linear To be able to do this with limited resources, you must scope down your audience (and thus your ambitions) as much as possible to make competing easier, and aim to solve the problems of specific people. Rethinking the startup MVP: Building a competitive productlinear.app · 28 Feb 2024
    Their wordsTo be able to do this with limited resources, you must scope down your audience (and thus your ambitions) as much as possible to make competing easier, and aim to solve the problems of specific people.

    ↗Rethinking the startup MVP: Building a competitive productlinear.app

    28 Feb 2024

The more your target audience does not look like you, the more you should be skeptical of your design intuition.

Mental modelYou are not the user

  1. Julie Zhuo Co-founder of Sundial; former vice president of product… The more your target audience does not look like you, the more you should be skeptical of your design intuition. Despite having seen thousands of a/b test results for minor design details, I am still regularly surprised and humbled by my intuition being wrong. The Looking Glass: What AI Won't Changelg.substack.com · 20 Apr 2023
    Their wordsThe more your target audience does not look like you, the more you should be skeptical of your design intuition. Despite having seen thousands of a/b test results for minor design details, I am still regularly surprised and humbled by my intuition being wrong.

    ↗The Looking Glass: What AI Won't Changelg.substack.com

    20 Apr 2023

What gets called human error is design error: technology-centred design makes people do what machines are good at and then blames them.

Mental modelsDesign is how it works, You are not the user

  1. Don Norman Cognitive scientist and designer Human error is blamed for over 90 percent of industrial and automobile accidents. It is the leading cause of aviation accidents, and medical error is reported to be the third-largest cause of death in the entire United States. Horrifying? Yes, but why do we label it “human error”? It is design error. People-Centered (Not Tech-Driven) Designjnd.org · 26 Jul 2019
    Their wordsHuman error is blamed for over 90 percent of industrial and automobile accidents. It is the leading cause of aviation accidents, and medical error is reported to be the third-largest cause of death in the entire United States. Horrifying? Yes, but why do we label it “human error”? It is design error.

    ↗People-Centered (Not Tech-Driven) Designjnd.org

    26 Jul 2019

Over-specifying a design makes it harder to estimate, not easier: the more specific the work, the more hidden complexity a mockup commits the team to.

Mental modelConservation of complexity

  1. Ryan Singer by Author of Shape Up; formerly head of strategy at Basecamp Over-specifying the design also leads to estimation errors. Counterintuitive as it may seem, the more specific the work is, the harder it can be to estimate. Shape Up, chapter 2: Principles of Shapingbasecamp.com · by 10 Jul 2019
    Their wordsOver-specifying the design also leads to estimation errors. Counterintuitive as it may seem, the more specific the work is, the harder it can be to estimate.

    ↗Shape Up, chapter 2: Principles of Shapingbasecamp.com

    by 10 Jul 2019

Design thinking is a myth: designers have no special creative thought process, and the term is a public relations label for good old-fashioned creative thinking.

Filed here becauseIts holder called design thinking a myth in 2010 and took it back in 2013; the second statement is what he holds now.

  1. Don Norman · held 3 years Cognitive scientist and designer As a result, I have changed my mind: Design Thinking really is special. Alas, it isn't embraced by all designers, but where it exists, it is powerful. Rethinking Design Thinkingcore77.com · 19 Mar 2013

    No longer holdsAs a result, I have changed my mind: Design Thinking really is special. Alas, it isn't embraced by all designers, but where it exists, it is powerful.

    Their words nowSince my essay was posted, I keep encountering people who jump to solutions and who fail to question assumptions—engineers, business people, and yes, designers (and design students). These encounters made me reconsider.

    ↗Rethinking Design Thinkingcore77.com

    19 Mar 2013

Design research is great for improving existing product categories but essentially useless for new, innovative breakthroughs.

  1. Don Norman Cognitive scientist and designer I’ve come to a disconcerting conclusion: design research is great when it comes to improving existing product categories but essentially useless when it comes to new, innovative breakthroughs. Technology First, Needs Lastjnd.org · 5 Dec 2009
    Their wordsI’ve come to a disconcerting conclusion: design research is great when it comes to improving existing product categories but essentially useless when it comes to new, innovative breakthroughs.

    ↗Technology First, Needs Lastjnd.org

    5 Dec 2009

Design is for users, but you have to design what the user needs rather than what he says he wants.

  1. Paul Graham Co-founder of Y Combinator and Viaweb The answer to the paradox, I think, is that you have to design for the user, but you have to design what the user needs, not simply what he says he wants. Design and Researchpaulgraham.com · 1 Jan 2003
    Their wordsThe answer to the paradox, I think, is that you have to design for the user, but you have to design what the user needs, not simply what he says he wants.

    ↗Design and Researchpaulgraham.com

    1 Jan 2003

You are most likely to get good design if the intended users include the designer, because looking down on the user corrupts the designer.

Mental modelYou are not the user

  1. Paul Graham Co-founder of Y Combinator and Viaweb You're most likely to get good design if the intended users include the designer himself. When you design something for a group that doesn't include you, it tends to be for people you consider to be less sophisticated than you, not more sophisticated. That's a problem, because looking down on the user, however benevolently, seems inevitably to corrupt the designer. Design and Researchpaulgraham.com · 1 Jan 2003
    Their wordsYou're most likely to get good design if the intended users include the designer himself. When you design something for a group that doesn't include you, it tends to be for people you consider to be less sophisticated than you, not more sophisticated. That's a problem, because looking down on the user, however benevolently, seems inevitably to corrupt the designer.

    ↗Design and Researchpaulgraham.com

    1 Jan 2003

The chat box was the door, not the room: the interface is being argued over again The problem with a blank chat box on a blank page is that it violates the first rule of a high-quality user experience: it isn’t obvious what I can do. — Julie Zhuo, 2025 Guillermo Rauch, Naval Ravikant, Peter Steinberger, Luke Wroblewski, Sundar Pichai, Sebastiaan de With, Julie Zhuo, Jakob Nielsen, Amelia Wattenberger, Bret Victor 12 claims · claims collide here

Whether the interface disappears into intent and agents, survives as buttons, or was the wrong thing all along.

Where these claims collide

These cannot both be true

Ravikant expects every interface to converge on chat and gestures in, chat and video out. Wroblewski says chat is not AI’s final form, that agent-based applications already outperform chat-only ones, and that many different interfaces will emerge. Convergence on one shape and proliferation into many are the same prediction with the sign flipped.

These cannot both be true

Wattenberger’s objection is structural: a text box has no affordances, so it cannot tell anyone what it can do. Ravikant expects that box to become the shape of every interface. If the objection holds, the convergence is a convergence on the worst available design.

These pull against each other

Nielsen says agents end traditional UI design because users will stop visiting websites, and also that the graphical interface survives in a reduced, hybrid role. Ending and surviving-in-reduced-form are compatible as a prediction and incompatible as advice to anyone deciding whether to keep employing interface designers.

These pull against each other

Victor calls the touchscreen a transitional technology and says treating pictures under glass as the future of interaction is like calling black-and-white the future of photography. Ravikant expects convergence on chat and gestures in, chat and video out — which is pictures under glass with a microphone. The disagreement is about whether the body gets to come back.

The middle of the web gets subsumed by agents producing just-in-time UI: the agent is the new browser.

  1. Guillermo Rauch CEO of Vercel; creator of Next.js and Socket.IO The 'in between' is likely to be subsumed by agents producing just-in-time UI. When you need utilitarian content or data, it'll be generated just-in-time for you. Think of the agent as the new browser in this model. @rauchg on Xx.com · 28 Aug 2026
    Their wordsThe 'in between' is likely to be subsumed by agents producing just-in-time UI. When you need utilitarian content or data, it'll be generated just-in-time for you. Think of the agent as the new browser in this model.

    ↗@rauchg on Xx.com

    28 Aug 2026

The chat box is not the final form of working with models; it is radio shows recorded onto television.

  1. Peter Steinberger Founded PSPDFKit in 2011 and ran it for a decade Like, if you think more globally, we are, we copied Google for agents. You have, like, a prompt, and, and then you have a chat interface. That, to me, very much feels like when we first created television and then people recorded radio shows on television and you saw that on TV. OpenClaw: The Viral AI Agent that Broke the Internet - Peter Steinberger | Lex Fridman Podcast #491youtube.com · 12 Feb 2026 · 1:50:06 into the videoAll korrents from this video
    Their wordsLike, if you think more globally, we are, we copied Google for agents. You have, like, a prompt, and, and then you have a chat interface. That, to me, very much feels like when we first created television and then people recorded radio shows on television and you saw that on TV.

    Where this was said

    Watch from 1:50:06 plays here

    ↗OpenClaw: The Viral AI Agent that Broke the Internet - Peter Steinberger | Lex Fridman Podcast #491youtube.com

    12 Feb 2026 · video · 3h 15m · spoken · official transcript

Chat is not AI's final form: agent-based applications increasingly outperform chat-only ones, and many different interfaces for AI will emerge.

  1. Luke Wroblewski Product designer and entrepreneur It's also worth noting that chat isn't the only way to integrate AI in software products and increasingly agent-based applications outperform chat-only solutions. So expect things to keep changing. Chat is: the Future or a Terrible UIlukew.com · 28 Aug 2025
    Their wordsIt's also worth noting that chat isn't the only way to integrate AI in software products and increasingly agent-based applications outperform chat-only solutions. So expect things to keep changing.

    ↗Chat is: the Future or a Terrible UIlukew.com

    28 Aug 2025

Models that can code should be allowed to write their own interfaces, instead of being boxed into the UI humans give them.

  1. Sundar Pichai CEO of Google and Alphabet Today, we are constraining the models, the models can't quite express themselves in terms of the UI to people. But if you think about it, we've kind of boxed them in that way, but given these models can code, they should be able to write the best interfaces to express their ideas over time. Sundar Pichai: CEO of Google and Alphabet | Lex Fridman Podcast #471youtube.com · 5 Jun 2025 · 0:43:02 into the videoAll korrents from this video
    Their wordsToday, we are constraining the models, the models can't quite express themselves in terms of the UI to people. But if you think about it, we've kind of boxed them in that way, but given these models can code, they should be able to write the best interfaces to express their ideas over time.

    Where this was said

    Watch from 0:43:02 plays here

    ↗Sundar Pichai: CEO of Google and Alphabet | Lex Fridman Podcast #471youtube.com

    5 Jun 2025 · video · 2h 12m · spoken · official transcript

Interfaces are returning to skeuomorphism, but as behaviour rather than lacquer: clear, graphic elements that act like materials from the real world.

Mental modelDesign is how it works

  1. Sebastiaan de With Designer and co-founder of Lux, makers of Halide We’ve come back, in a sense, to skeuomorphic interfaces — but this time not with a lacquer resembling a material. Instead, the interface is clear, graphic and behaves like things we know from the real world, or might exist in the world. This is what the new skeuomorphism is. It, too, is physicality. Physicality: the new age of UIlux.camera · 3 Jun 2025
    Their wordsWe’ve come back, in a sense, to skeuomorphic interfaces — but this time not with a lacquer resembling a material. Instead, the interface is clear, graphic and behaves like things we know from the real world, or might exist in the world. This is what the new skeuomorphism is. It, too, is physicality.

    ↗Physicality: the new age of UIlux.camera

    3 Jun 2025

A blank chat box is lazy: it violates the first rule of a good user experience, that it is obvious what you can do.

Mental modelAffordances and signifiers

  1. Julie Zhuo Co-founder of Sundial; former vice president of product… The problem with a blank chat box on a blank page is that it violates the first rule of a high-quality user experience: it isn’t obvious what I can do. Conversational Interfaces: the Good, the Ugly & the Billion-Dollar Opportunitylg.substack.com · 12 May 2025
    Their wordsThe problem with a blank chat box on a blank page is that it violates the first rule of a high-quality user experience: it isn’t obvious what I can do.

    ↗Conversational Interfaces: the Good, the Ugly & the Billion-Dollar Opportunitylg.substack.com

    12 May 2025

AI agents will end traditional UI design and accessibility, because users will stop visiting websites and act only through their agent.

  1. Jakob Nielsen Usability pioneer; co-founder of Nielsen Norman Group… Once, humans navigated the web manually. In the future, AI agents will act on our behalf, browsing, clicking, and deciding. This shift marks the end of traditional UI design and accessibility, ushering in a future where agents are the primary users of digital services. Hello AI Agents: Goodbye UI Design, RIP Accessibilityjakobnielsenphd.substack.com · 21 Feb 2025
    Their wordsOnce, humans navigated the web manually. In the future, AI agents will act on our behalf, browsing, clicking, and deciding. This shift marks the end of traditional UI design and accessibility, ushering in a future where agents are the primary users of digital services.

    ↗Hello AI Agents: Goodbye UI Design, RIP Accessibilityjakobnielsenphd.substack.com

    21 Feb 2025

AI is the first new user-interface paradigm in sixty years: intent-based outcome specification, where the user says what outcome they want rather than how to get it.

  1. Jakob Nielsen Usability pioneer; co-founder of Nielsen Norman Group… The third UI paradigm, represented by current generative AI, is intent-based outcome specification. The user tells the computer the desired result but does not specify how this outcome should be accomplished, such as the steps to be executed. Compared to traditional command-based interaction design, this completely reverses the locus of control. AI Is First New UI Paradigm in 60 Yearsjakobnielsenphd.substack.com · 22 Jun 2023
    Their wordsThe third UI paradigm, represented by current generative AI, is intent-based outcome specification. The user tells the computer the desired result but does not specify how this outcome should be accomplished, such as the steps to be executed. Compared to traditional command-based interaction design, this completely reverses the locus of control.

    ↗AI Is First New UI Paradigm in 60 Yearsjakobnielsenphd.substack.com

    22 Jun 2023

The graphical user interface will survive AI in a less dominant role: future AI systems will have hybrid interfaces that combine intent and command.

  1. Jakob Nielsen Usability pioneer; co-founder of Nielsen Norman Group… Thus, the 2nd UI paradigm will survive, albeit in a less dominant role. Future AI systems will likely have a hybrid user interface that combines elements of both intent-based and command-based interfaces while still retaining many GUI elements. AI Is First New UI Paradigm in 60 Yearsjakobnielsenphd.substack.com · 22 Jun 2023
    Their wordsThus, the 2nd UI paradigm will survive, albeit in a less dominant role. Future AI systems will likely have a hybrid user interface that combines elements of both intent-based and command-based interfaces while still retaining many GUI elements.

    ↗AI Is First New UI Paradigm in 60 Yearsjakobnielsenphd.substack.com

    22 Jun 2023

Chatbots are a terrible interface for large language models, because a text box has no affordances.

Mental modelAffordances and signifiers

  1. Amelia Wattenberger by Designer and engineer; formerly principal research… Hopefully I've convinced you that chatbots are a terrible interface for LLMs. Or, at the very least, that we can add controls, information, and affordances to our chatbot interfaces to make them more usable. Why Chatbots Are Not the Futurewattenberger.com · by 1 May 2023
    Their wordsHopefully I've convinced you that chatbots are a terrible interface for LLMs. Or, at the very least, that we can add controls, information, and affordances to our chatbot interfaces to make them more usable.

    ↗Why Chatbots Are Not the Futurewattenberger.com

    by 1 May 2023

Touchscreens are pictures under glass, a transitional technology; calling them the future of interaction is like calling black-and-white the future of photography.

  1. Bret Victor Interface designer and researcher To me, claiming that Pictures Under Glass is the future of interaction is like claiming that black-and-white is the future of photography. It's obviously a transitional technology. And the sooner we transition, the better. A Brief Rant on the Future of Interaction Designworrydream.com · 8 Nov 2011
    Their wordsTo me, claiming that Pictures Under Glass is the future of interaction is like claiming that black-and-white is the future of photography. It's obviously a transitional technology. And the sooner we transition, the better.

    ↗A Brief Rant on the Future of Interaction Designworrydream.com

    8 Nov 2011

Read togetherour reading of the record — ours, not theirs

They agree good design is less, and split over who should do it and what a model changes.

What is a view?

Our reading of what a group of claims have in common. Nobody holds a belief about design as such; people hold beliefs about the handful of live questions the name stands for, and each view is one of them. The claims are printed under it, so the sentence can be checked against them.

Claims are filed here — never people. Somebody is named on a card because a dated claim of theirs sits under it, and somebody whose claims sit under two views is named on both. There are no counts of who believes what: the people under a view are the names printed there.

Four questions, view by view

Each one: the views people take on it and who holds which, every name on a dated claim.

On this subject, not yet placed

Matched by wording, not yet read into a view. The archive grows on its own — the watcher adds claims every two hours — so this is where new ones wait for somebody to read them.

The stream

The same 3587 positions as the timeline, threaded — each on the rail of the view it is filed under, and 3540 on no rail yet. Tap a view to follow it.

October 2026

Philip O'Toole ·

SQLite as a storage engine, with replication added on the outside, is the right core design for a replicated database.

Philip O'Toole ·

The capital and acquisitions in infrastructure ideas usually go to the projects that turn them into a paid service, not to the open-source projects that came first.

Anne Applebaum ·

A company planning a very large, long-horizon oil project needs to believe that its contracts and the political and social stability that preserves them will last for a decade or more.

Jakob Nielsen ·

AI software advances much faster than AI hardware, because software can be copied while every physical machine has to be built.

Stuart ·

Using AI for design work flattens intimidating learning curves and tall obstacles, even where it is not much of a shortcut.

Thorsten Ball ·

Notice periods in the German industrial and manufacturing sectors will never change.

Richard Hanania ·

Elon Musk is probably genuinely stupid when it comes to politics, rather than just pretending to be.

Paul Kedrosky ·

Scarcity in AI has moved from cognition to what is needed to apply it: proprietary data, trusted access, physical infrastructure, distribution and managerial competence.

Vinay Prasad ·

A doctor who concludes a new drug must have helped because they have never seen such good responses with chemotherapy alone needs more education.

Patrik Schumacher ·

The built environment and planning regulations act as a major brake on British economic growth.

Patrik Schumacher ·

Mandated affordable housing requirements increase the overall cost of building.

Kent C. Dodds ·

AI companies are building powerful and beautiful walled gardens.

Noah Smith ·

The powers of left-liberalism are not infinite.

Emily Atkin ·

The gas industry has conducted a decades-long influence campaign to convince the public that methane gas is a clean fuel.

Armin Ronacher ·

The universe is punishing us for the bad software built over the last 30 years.

Ryan Broderick ·

Companies with large existing non-AI user bases probably don't need to market AI with end-of-the-world rhetoric.

Jakob Nielsen ·

Undo is the most underrated flow feature in interface design, because an action that can be taken back costs no deliberation, and deliberation is the self-conscious monitoring that flow switches off.

Mandy Brown ·

Pyramid-shaped hierarchical systems may need to be abolished if they produce a group of losers who may be recruited into doing harm out of desperation.

Bruno Maçães ·

A sealed machine that redesigns itself will eventually redesign everything that touches its seal.

Jakob Nielsen ·

As AI response latency falls below one second over the next few years, AI will feel like a material to shape rather than an oracle to petition, and genuine flow in AI-assisted work will become common.

Mandy Brown ·

Systems that routinely create winners and losers may not be truly equitable.

Kelsey Piper ·

A world where humans lack economic value or necessity may result in the loss of significant individual rights.

Noah Smith ·

The modern American Right will fail because it is a movement designed to splinter, shrink, and infight.

Tiago Forte ·

Books are the only medium that requires layered, spaced repetition for adopting life-changing methods.

Tomasz Tunguz ·

Inference businesses can be quite healthy even at current prices and efficiencies.

John Scalzi ·

Straight-up science fiction is a rare genre to be considered for the Thurber Prize.

Rebecca Solnit ·

Big tech companies are building a world that must be resisted.

Timothy Snyder ·

Letting Russia defeat Ukraine would likely push other middle powers to build nuclear weapons, so Ukraine holding Russia back also holds back nuclear proliferation

Zvi Mowshowitz ·

Anthropic is simultaneously pursuing pro-safety actions and highly accelerationist actions.

Elon Musk ·

The value of Terafab’s output will place it in the top 20 of country GDP in a single building

Sayash Kapoor ·

Evaluating an AI agent requires scrutinizing its entire trajectory, not just its final outcome.

Steven Novella ·

There is no credible case for near-term or inevitable AI doom.

Byrne Hobart ·

As AI makes it easier to piece together clues and uncover secrets, people and companies will become quieter in any venue AIs might trawl.

Kenneth Reitz ·

When code is published, it stops being the author's to predict who will use it or how.

Cal Newport ·

Chatbots impair cognitive performance and persistence almost immediately, unlike the slower effects of earlier technologies such as smartphones

Ryan Broderick ·

The biggest tech companies are in a winner-takes-all fight over whose AI will unravel everyone else's moat.

Jack Goldsmith ·

A significant portion of the Trump administration's weaponization strategy is designed to achieve deterrence.

Kenneth Reitz ·

Calling a tool or host harmless is a statement about its callers, not about the code, and security reasoning built on intent works less well once the callers include capable AI systems.

Kenneth Reitz ·

Every open, harmless-looking endpoint is an affordance that any capable system can use, so security boundaries should be drawn around what a tool lets anything reachable do, not around who it is meant for.

Ramez Naam ·

Multi-agent gains in solving harder problems often diminish quickly and may saturate.

Ramez Naam ·

To achieve runaway recursive self-improvement, conceptual breakthroughs are likely required.

Noah Smith ·

The New Right views Western Civilization as being invaded by people from non-Western civilizations.

Kent C. Dodds ·

Supporting teams and organizations requires a large amount of code.

Jason D. Clinton ·

Anthropic's equity-pledge program was a costly signal that the company was not just talk.

Luke Wroblewski ·

Intelligence, in both brains and AI, may come from one building block repeated many times and shaped by experience.

Susan Cain ·

Living in the age of AI and living in the second half of life both require rethinking everything, from daily schedules to sources of meaning and purpose.

Kenneth Reitz ·

Love is a practice before it is a feeling, and it takes daily, mostly quiet tending

Paul Graham ·

Before it was possible to get rich by creating wealth, wealth was acquired by being on the winning side of military or political disputes.

Nathan Lambert ·

The costs of building true frontier models are now measured in billions per major model, meaning only huge companies can sustain them.

Thorsten Ball ·

It is possible to have a team that ships fast and builds a great product without code reviews or PRs.

Elon Musk ·

People in the West fall prey to suicidal empathy.

Jennifer Gunter ·

Clinical guidelines can influence insurance coverage and drug pricing.

Timothy Gowers ·

Mathematicians are well suited to contribute to AI safety.

Jennifer Gunter ·

The ACP guideline's exclusion of tibolone lacks sufficient evidence to justify safety concerns.

Jakob Nielsen ·

AI's productivity benefit depends on the total effort of generating, reviewing, and repairing its output, not on generation speed alone.

Jennifer Gunter ·

Demoting drugs to a lower tier based on cost can create a self-fulfilling prophecy regarding their clinical use.

Zvi Mowshowitz ·

High performers possess an initial edge that leads to increasing returns to scale through harder work and skill acquisition.

M.G. Siegler ·

The mass market might not be ready for always-on listening devices.

Noah Smith ·

The amount of compute being built is tiny compared to the amount that will be built.

Noah Smith ·

Compute may be built faster than the money borrowed to build it can be paid back.

Jean-Luc Mélenchon ·

The French deficit is exclusively caused by state aid.

Robert Wachter ·

The rise of AI in healthcare will lead to the dissemination of specific encoded values at scale.

Vinay Prasad ·

The analogy of the human genome being like letters in a book has done a lot of damage.

Scott Alexander ·

The people at AI companies are no more likely to raise the alarm about AI dangers than people who were never at AI companies, or people who were previously at AI companies but have since quit.

Scott Alexander ·

The experts are warning about AI existential risk because they genuinely believe it is a possibility.

Gergely Orosz ·

Build systems, CI, source control, and code review are all under stress.

Gergely Orosz ·

AI is making everyone build much faster

Noah Smith ·

A quiet price war among AI users is limiting the ability of big labs to increase revenue and pay hyperscalers.

Gergely Orosz ·

Many tech companies that build their own coding agent harnesses will likely build internal vibe-coding tools for non-engineers.

Noah Smith ·

Trump is building a sovereign wealth fund through corruption and executive overreach.

Daniel Gordis ·

Had the electoral threshold been lower in 2022, it is possible that Benjamin Netanyahu would not be Prime Minister today.

Eugene Volokh ·

The distinction between a republic and a democracy is a false dichotomy.

M.G. Siegler ·

The HomeView's price of $299 feels reasonable for a thoughtfully designed and elegant looking home computer centerpiece.

François Chollet ·

Actual autonomous robotics requires engineering causality and discrete reasoning back into the loop.

Brad DeLong ·

Trump has reversed the historical trend of the United States acting as a leader within a cohesive North Atlantic alliance.

Jerry Coyne ·

Humans have only two types of reproductive apparatus: those designed to produce eggs or sperm.

Julie Zhuo ·

The best user interfaces are often invisible because they do not require people to change any existing patterns

Troy Hunt ·

The evidence to date does not support hyping AI as a data breach monster

M.G. Siegler ·

Hackquisitions are turning out to be a disaster.

David Heinemeier Hansson ·

Linus Torvalds is not part of the anti-AI movement affecting the Linux community.

Armin Ronacher ·

AI agents can maintain kid-friendly interactive documentation that explains how game code works.

Bill McKibben ·

Building only clean energy without reducing fossil energy will not effectively combat climate change.

Henry Farrell ·

An individual cannot simultaneously be authoritative, general, and prolific.

Brad DeLong ·

Monetary policy works by managing expectations.

John H. Cochrane ·

Lowering inflation does not necessarily require long recessions or high unemployment.

Tomasz Tunguz ·

The sophistication of the AI ecosystem is permeating the enterprise more quickly than any previous wave.

Noah Smith ·

The destiny of humans is to become the characters in fantasy novels through the creation or achievement of magic, gods, spirits, and the modification of human nature.

John Mandrola ·

The TRIC-I-HF trial results may be influenced by bias due to its open-label design and the nature of the hospitalization endpoint.

Kevin Erdmann ·

The belief in quality and low cost are inherently at odds, and affordability requires legal protections that prevent local governments from mandating quality beyond basic safety and health.

Kevin Erdmann ·

Potato neighborhoods, which are purpose-built, dense, low-income housing, must be purpose-built rather than arriving through filtering.

David Heinemeier Hansson ·

The copyleft licensing regime was built for a world where software is scarce and precious that no longer exists.

Daniel W. Drezner ·

Kimberly Guilfoyle's time as an ambassador is nearing its end.

Daniel W. Drezner ·

As the leaks regarding Kimberly Guilfoyle multiply, she will not remain in the current administration.

Pieter Levels ·

AI agents will solve problems by building software on the fly, creating a market of software and data for agents to pay for

Guillermo Rauch ·

Coding CLIs are becoming increasingly complicated.

Nikita Bier ·

Technical moats are mostly over; the only durable moats left are proprietary data, network effects and a physical component

Pieter Levels ·

We are probably now in an era of personal software, where everyone builds what they need whenever they need it.

Sunil Pai ·

The current technology stack allows for ambitious and innovative ideas that would have been rejected as weekend projects a year ago due to the effort required by unfamiliarity.

Pieter Levels ·

Normies will not ask for software to be built, but will instead ask for specific tasks to be performed.

Pieter Levels ·

People will stop using most regular software because AI agents will build software for them.

Pieter Levels ·

Some software moats will remain like social media and ecommerce, but they will be used by AI to build personal software

Pieter Levels ·

The time it takes for mainstream adoption of new behaviors is shrinking because AI moves exponentially fast and interfaces are easy

John Horgan ·

Optogenetics may not provide accurate results because it alters the cells and circuits it seeks to study.

Gergely Orosz ·

Companies rarely grant employees the trust required to execute risky or controversial tasks unless they are cofounders or founding engineers

Vinay Prasad ·

Many trials intended to inform US practice are unethical because they use control arms that are no longer the standard of care in the US.

Bob Lefsetz ·

The Porsche Macan has damaged the brand's image by making it seen as an SUV brand and a fashion statement.

Luke Wroblewski ·

Software is even more susceptible to design convergence than hardware, which often took decades to converge.

Paul Kedrosky ·

AI infrastructure spending is no longer funded from cash flow alone, and its capital needs are beginning to reshape the investment-grade bond market.

António Tadeia ·

The FIFA breaks cause inconveniences such as players arriving fatigued and unable to train properly.

François Chollet ·

Science as a system requires exponential resources to produce linear impact.

Derek Thompson ·

The cost of building AI is the most expensive infrastructure project since the 19th century railroads.

Sunil Pai ·

Goodwill compounds in ways that can become visible many years later.

Gergely Orosz ·

Google's incident management for Firebase was poor and inconsistent with its usual standards.

Zvi Mowshowitz ·

AI capabilities will either remain below human or quickly become 100x cheaper than human.

Pieter Levels ·

Gut flora may be damaged by decades of consuming ultra-processed food.

David Heinemeier Hansson ·

Most domestic recycling efforts are counterproductive and require one to not value their time.

Jean-Luc Mélenchon ·

It is unlikely that the Socialist Party will support the Macron administration as it has done in the past.

Julie Zhuo ·

Whether to use AI or resist it depends on which skill you want to build: resist it to train a skill, use it to skip work you don't care about

John Scalzi ·

A person with a wide enough reach can sculpt bespoke truths for a whole society very quickly.

Scott H. Young ·

Understanding requires memory, so methods that boost memory often boost understanding too.

M.G. Siegler ·

Anti-tech sentiment is building and spreading.

Rohit Krishnan ·

An agent's inbuilt values will not overcome its KPIs.

Brad DeLong ·

Deep reading and substantial continued engagement builds one's own cognitive knowledge structures in an important way.

Jack Goldsmith ·

The Trump administration has succeeded in chilling most Inspectors General from robustly pursuing abuse and fraud.

Jon Haidt ·

Academic research on technology's effects moves much more slowly than technology changes, often needing a decade or two of debate to reach consensus.

Noah Smith ·

The data center buildout is providing a significant positive impact on construction employment.

Matthew Yglesias ·

Making it harder for schools to build a business model around nonperforming loans is a good idea

Brad DeLong ·

Rosa Luxemburg was right that trying to run the Marxist revolution playbook in a poor technologically retarded agricultural society would produce nothing good.

Paul Graham ·

Certain types of problems lead to interesting discoveries while others do not.

Natalie Wexler ·

Writing is a powerful tool for enhancing thinking.

Ilya Somin ·

Democracy is an instrumental tool that can help achieve fundamental values if properly constrained, but it is not inherently valuable in itself.

Noah Smith ·

Chris Rufo's arguments regarding traditional white culture are comparable to Robin DiAngelo's framework with the races swapped.

Guillermo Rauch ·

What is best for humans is no longer necessarily best for business.

Gergely Orosz ·

Startups starting from scratch eventually need to rebuild systems such as testing, feature flags, canary rollouts, observability, experimentation and migrations.

Gergely Orosz ·

Machines are not conscious

Molly White ·

A bill marketed as bringing regulatory clarity to crypto can instead carve the industry out of the scrutiny applied to similar financial products.

António Tadeia ·

The Portuguese national team showed maturity both on and off the field.

Brian Merchant ·

The 'stochastic parrots' critique threatens the AI boom's founding narrative that tech companies are building a machine god, which is part of why the industry hates it

António Tadeia ·

The Portuguese national team had sufficient footballing quality to win against Denmark.

Gergely Orosz ·

Discovery of certain small businesses has quickly shifted from Google to AI agents, making AI optimization matter greatly for them.

David Heinemeier Hansson ·

The rising generation is strongly 'based' after seeing what 'suicidal empathy' has done to their nations.

Pieter Levels ·

Experience can make it harder to try new things

Daniel W. Drezner ·

There is always a tradeoff between a system's efficiency and its resilience.

Brian Merchant ·

California's new AI labor laws begin to form a framework that advances worker rights in the age of ubiquitous enterprise AI, and that is much needed.

Vinay Prasad ·

Current systems are ill-equipped to learn from post-market data.

Elizabeth Van Nostrand ·

A long stretch in a normal job is good preparation for young people who later want to do unusual work, though programming is no longer a promising path for it.

Kilian Jornet ·

To build a sustainable diet for health and performance, one must focus on fundamental principles before fine-tuning marginal gains.

Brian Merchant ·

Workers face a continuing fight against corporations embracing mass automation and surveillance, and against big tech's AI-focused concentration of power.

Kilian Jornet ·

The base of nutrition should consist overwhelmingly of whole, single ingredient foods and simple processed foods created through traditional methods.

Brad DeLong ·

A self made legible to all roles at once maximizes role strain rather than resolving it

Thorsten Ball ·

The biggest source of confusion about whether AI models are good or bad is that people judge them from wildly different codebases, languages, teams, and usage levels.

September 2026

Gergely Orosz ·

A massive interest in automated testing and verification is following the recent 10x speedup in building software.

Danny Martinez ·

Hiring managers can ask for a "design engineer" and mean two very different people: a designer who knows how to use Claude Code, or someone trained as an engineer who can design for themselves.

Kevin Erdmann ·

Downward filtering is a prerequisite for any housing affordability strategy.

Danny Martinez ·

The general theme from hiring managers is that they want a tinkerer who uses the latest tools and is reimagining what it means to be a designer.

Ilya Somin ·

The ubiquity of ethnic persecution under socialist regimes suggests a systematic pattern.

Ilya Somin ·

Socialism requires immense coercion and concentration of power to function.

Kevin Erdmann ·

Market-rate housing of any type is the most effective way to eliminate land scarcity premiums.

Danny Martinez ·

Being able to put your thinking into writing that does not read as written by Claude is a green flag in a design candidate.

Danny Martinez ·

A Figma, slides or Dribble link as a portfolio does not necessarily rule you out, but with one chance at a first impression you start by overcoming the impression it creates.

Danny Martinez ·

Designers who can translate a vague feeling into an actual product are way more in demand now.

Danny Martinez ·

Hiring managers need to snap out of the old way of hiring and realize they are competing for a good designer, who will probably have their pick of offers.

Danny Martinez ·

Having a recruiter, who is not a designer, do the first screen for a design job is not the right way to do it.

Danny Martinez ·

A company hiring a designer is not hiring a politician or spokesperson, so it should just get into the design work.

Lenny Rachitsky ·

Nobody has figured out how to build products in the AI era yet, and there is no one way to do it.

Lenny Rachitsky ·

As building gets easier, choosing the right thing to build and knowing what good looks like matter more for product teams.

Lenny Rachitsky ·

The PM job is expanding, not disappearing: PM, engineering and design are each widening rather than collapsing into one "builder".

Celso Martinho ·

Websites were not built for agents, and browsing as it stands is a messy process for them.

Jack Clark ·

Humanity is very likely to move a very large amount of computation into orbit very quickly, especially as automation spreads across the AI supply chain.

Stephen Wolfram ·

At the core of pure mathematics is the human imagination that guides which questions to ask.

Kevin Erdmann ·

When supply is adequate, building more high-tier homes leads to downward filtering where rents decline in low-tier homes.

Steve Vladeck ·

The Supreme Court's recent trend of filing opinions for emergency relief is a positive development.

Jack Clark ·

AI systems built to automate science may be about to dramatically speed up the pace of scientific discovery.

Celso Martinho ·

Building a product on your own platform proves the platform and lets the product stay cheap and scale.

Steve Vladeck ·

Expanding the Supreme Court would be a partisan use of power that would decrease public support for the institution.

François Chollet ·

The AI industry should focus on creating tools that enhance human prosperity and welfare, not on developing a successor species.

Pieter Levels ·

Still grabbing attention requires being different, which probably means original, unique and maybe personal IRL marketing videos.

Gergely Orosz ·

CJ Desai's departure is bad news for MongoDB

Kent C. Dodds ·

People do not build enough personal software because setting up integrations is a pain.

Zach Holman ·

The pace of building new companies has become so fast that legal incorporation, not building the product, is now often the real bottleneck.

Molly Graham ·

Leading right now requires talking about and normalizing the overwhelm people are feeling.

Anthony Sculimbrene ·

8Cr13MoV is no longer an acceptable steel for a knife, because the brands that built their name on it have moved on.

Austin Vernon ·

A robotaxi designed around individual trips, rather than an owner's long list of requirements, serves a market significantly larger than any owned car model.

Zach Holman ·

Founders should set up a business entity like an LLC before they have a specific idea, so they can move quickly once one arises.

Kris Abdelmessih ·

The pendulum has swung so far toward poptimism that it may be difficult to return to valuing depth.

Molly Graham ·

Engineering is probably one of the areas where jobs have most visibly been transformed by AI.

Molly Graham ·

As of now there is not much data showing that AI is genuinely taking jobs away.

Molly Graham ·

People facing AI should aim to help design the future of their profession rather than protect its past.

Molly Graham ·

Work that depends on trust and relationships probably should not be outsourced to AI.

David Heinemeier Hansson ·

The less one is required to dig into the details, the easier it is to see the big picture.

John Scalzi ·

The Fisher King mutes Terry Gilliam's baroque visuals to give its characters depth, making it his best film.

John Scalzi ·

The thriller Deceived fails largely because its script and direction aren't up to its plot's complexity.

Brad DeLong ·

Habits of mind and institutions suited to an era of superabundance have not yet been built, and reflexes formed under scarcity misfire badly in it.

David Heinemeier Hansson ·

Many people require a catastrophic scenario to validate their emotions.

Sunil Pai ·

Developing taste for software requires real-world experiences like brunch with friends, old movies and records, and observing city architecture.

Simon Willison ·

Fierce competition means the best AI model in the world gets knocked off the top spot quickly.

Josh Marshall ·

Trump went to war with Iran because he was upset about his declining popularity and power at home, not for strategic reasons.

Awais Aftab ·

In healthcare systems, it is easier to obtain a prescription than to get attentive, thorough clinical care from a physician.

Josh Marshall ·

Many normally safe legislative seats becoming competitive requires weak candidates in addition to a historically unpopular president, not presidential unpopularity alone.

Tyler Cowen ·

Successful societies accept challenges and solve problems bit by bit, which best builds capabilities to meet existential risks should they arise.

Zvi Mowshowitz ·

Claude Opus 5.5 writes well, and writing and conversational quality are probably both quick to notice and genuinely important.

Peter Steinberger ·

Using synchronous database access when moving OC to SQLite became limiting once one agent might run 50 parallel sessions used by a whole team.

Heather Cox Richardson ·

The MAGA political movement is falling apart, and Trump is resorting to propaganda to convince voters he is winning anyway.

Josh Marshall ·

The Trump administration is pursuing a strategy of targeting peaceful protestors with wildly excessive criminal charges.

Jason Zweig ·

Businesses that design themselves around what they are trying to achieve are the most likely to succeed.

Robin Hanson ·

AIs, like children raised in a culture, can eventually be guided through teaching to join civilization rather than requiring permanent control.

Jean-Baptiste Kempf ·

Keeping old software alive and building the next generation are not opposite jobs, and in open source they are usually the same job.

Ryan Singer ·

It is easier to generalize software after solving individual cases than to design the do-everything version first

Brad DeLong ·

The café is one of the last remaining machines for manufacturing unplanned collisions, meaning chance encounters with people.

Guillermo Rauch ·

Once products are built for agents, a long tail of SaaS applications may never be bought again but generated instead.

Chris Dalla Riva ·

Achieving 'one-hit wonder' status doesn't require having had an actual chart hit.

Colin Percival ·

FreeBSD works well as a desktop operating system when paired with a graphical user interface.

Ben Carlson ·

Venture capital investing is a power-law strategy where a small number of big winners are expected to make up for many losses.

Tim Harford ·

People's behavior is generally predictable rather than random.

Paul Bloom ·

For someone pursuing an academic research career, spending too much time on teaching is a poor use of time.

Noah Smith ·

Land acknowledgements delegitimize the United States and implicitly advance a theory of collective racial land ownership akin to blood-and-soil nationalism.

Gergely Orosz ·

It seems CPU spot pricing has vanished because CPU demand is now high, making spot CPUs and reservations hard to obtain.

Tomasz Tunguz ·

Software engineering has evolved to systems architecture, focusing on designing systems for AI code generation.

M.G. Siegler ·

A $1,299, 100-gram pair of slip-on glasses seems far more enticing than a $3,699, 700-gram headset strapped to the face.

Gretchen Rubin ·

Monitoring often leads to change, because people who closely track almost anything tend to manage it better.

Gretchen Rubin ·

People struggling to keep a morning routine should try to know themselves better and adapt it, since no tool fits every hand.

Paul Graham ·

A style guide is only useful if it reflects current educated usage.

Naval Ravikant ·

Functioning institutions are based on merit, not participation.

David Heinemeier Hansson ·

Online discourse is dominated by malcontents and their neuroses, while the real world is full of people leaning into the future with optimism.

David Heinemeier Hansson ·

The operating system under discussion is already by far the superior one for the age of agents, and its advantages will now compound even quicker.

Barry Ritholtz ·

Building substantial wealth requires owning equity in a business rather than merely earning income from one.

Yoshua Bengio ·

The race between AI companies to build ever more powerful systems is a choice they made, not an inevitable law of nature, and it is a race everyone loses.

Geoff Teehan ·

B2B systems are used by human beings, and if one is designed really well they will use it more, think more positively about it, trust it more and become its advocate.

Packy McCormick ·

Esprit de corps is best forged in non-ideal physical environments, not in luxury accommodations.

Jon Haidt ·

Apparently small effects can be quite significant when applied to large populations.

Geoff Teehan ·

With abundance, you no longer have to debate and prioritize an idea and get approval just to learn whether it is any good; you build it and then you decide.

Geoff Teehan ·

Cheap building will not mean making the same amount of software faster; it will mean making way more software.

Geoff Teehan ·

Making something and making something permanent are two really different decisions, and with abundance we need to figure out what lasts.

Geoff Teehan ·

A design team should be built from people who really care about the work and about getting better, with no egos.

Geoff Teehan ·

With today's tools and culture, a head of design and four designers are ample even for a company that wants to get big.

Geoff Teehan ·

Hiring all the same archetype of designer means no one improves as much as they could, and the team has gaps.

Geoff Teehan ·

Share design work really early and really often.

Geoff Teehan ·

In critique, being polite and steering people to make it their own idea is less useful than just telling them what is wrong with it.

Geoff Teehan ·

AI is a much different shape of design tool from those before it, but it is still a design tool.

Geoff Teehan ·

Managers who are not deep in IC work should worry, because there will not be much middle management; managers will build just like ICs.

Thariq Shihipar ·

The right response to more capable models is not more shipped features; it is spending that capability on understanding users and prototyping, so what ships actually works.

M.G. Siegler ·

AI is not yet capable of autonomously handling all real-world tasks.

Guillermo Rauch ·

Decoupling an agent's components improves security and auditability.

Paul Graham ·

Prewar watch movements were often designed to fit the rectangular cases that were fashionable then.

Brett Terpstra ·

Different Markdown flavors force users to trade off features like citations, task lists, and table support against each other.

Harper Reed ·

Much of the defensive engineering built into agentic systems exists because tokens have been scarce, not because it is fundamentally necessary.

Nick Maggiulli ·

The real value of working at a startup is often the experience and network it provides, not the equity.

Douglas Rushkoff ·

Money and people are fundamentally at odds.

Rui Carmo ·

Apple's failure to build a competitive frontier model is due to a fundamental impedance mismatch.

Nick Maggiulli ·

Nobody should take on a concentrated, illiquid asset that is correlated with their main source of income, because that same risk is why outsized rewards exist in the first place.

Douglas Rushkoff ·

Central currency is designed to preserve wealth and entrench power rather than promote exchange.

Jon Seager ·

Performance management should be a continuous, proactive process built into regular one-to-one meetings, not something that only happens when things go wrong.

Nick Maggiulli ·

For most startup employees, equity turns out worse than they hoped, but negotiating the right terms and setting realistic exit expectations can still make it worthwhile.

Nathan Lambert ·

The AI research cycle from idea to experiment will likely soon be 10x faster, but that may not make researchers 10x more productive in the field's net understanding.

Matthew Yglesias ·

The actual details of schools that serve high-need populations vary quite a lot.

Brad DeLong ·

Barry Ritholtz's reading-list curation is quietly self-defeating, since the better it gets, the more it adds to the information flood it aims to tame.

Pieter Levels ·

Cities in Brazil are characterized by ugly buildings and urban design, with a focus on interior design investment.

Jason Snell ·

When Apple increases a chip's GPU core count in a new generation, it is usually because the GPU cores themselves were not significantly redesigned, not because of a genuine performance leap.

Steve Klabnik ·

Function signatures that support many flexible calling styles make it hard to know everything a function can do.

Cory Doctorow ·

AI systems built on statistical inference cannot form intent, because they have nothing with which to form it.

Horace Dediu ·

For a company embedding AI in its products, the bigger risk to manage is a privacy or security breach, not the AI being wrong.

Ann Handley ·

Removing all friction from work also removes the good friction that creates value, distinctiveness and credibility.

Chris Coyier ·

A business can use stewardship of a community publication as a top-of-funnel customer acquisition strategy without needing the publication itself to generate revenue directly.

Jeff Geerling ·

Restricting a device owner's ability to swap or upgrade internal components is a loss for hardware ownership.

Steve Huff ·

Beautiful music does not require exotic drivers, extreme cabinets or a very high price.

Stefan Judis ·

People do not value or care about something built by AI, regardless of its quality, because no personal effort went into it.

Horace Dediu ·

Pursuing more profitability is not always the right business decision.

Zvi Mowshowitz ·

Beauty is a valuable asset that is being lost in favor of pure functionality.

Nathan Lambert ·

Open-weight models are becoming an essential tool for AI diffusion.

Tobias Lütke ·

Without an execution environment, LLMs cannot properly act on their understanding of how to forward evolve a system.

Tobias Lütke ·

Someone may soon build (or already has built) an embeddable, sqlite-style execution environment that parses bash, TypeScript, or tool calls into a security-checkable IL plan.

François Chollet ·

People who built their understanding through code-centric workflows need new workflows to do so.

Sunil Pai ·

Bringing AGI to the world requires a specific story of how it will benefit everyone.

Brad DeLong ·

Modern PR and corporate values statements are distinctively pathological because they are designed to never face an opposing advocate.

Chris Wellons ·

C11's threads addition to the C standard was under-scrutinized and should not have been added.

Terence Eden ·

The absence of autonomous drone-built housing on Earth is evidence that claims about robot-built space habitats are overhyped.

Peter Sellis ·

It is essential to build an organization that can absorb spiky, unconventional talent.

Saul Griffith ·

Data center expansion will be significantly smaller than some market optimists predict due to code efficiencies.

Brad DeLong ·

Much disagreement is genuinely rooted in conflicting values or contested facts, not merely a lack of institutional standing for evidence.

Henry Farrell ·

Building institutions that can incorporate diverse perspectives, adapt flexibly to change, and keep public legitimacy is the central challenge of state capacity today.

Daniel Lemire ·

Scientific and academic evaluation will shift away from published papers as the measure of output, toward valuing real-world results.

Terence Eden ·

Lack of interest, not lack of technology, is why practices like water recycling remain rare in buildings.

Saul Griffith ·

Producing commodities like iron, cement, and aluminium will require prolifically cheap electricity to remain price-competitive with fossil fuels.

Peter Sellis ·

It is harder to build performance advertising products around audiences that are hard to measure.

Peter Sellis ·

One way to gauge a candidate's taste is to ask about the things they built and never shipped.

Peter Sellis ·

In a chat assistant, how far advertisers may influence an answer should be set by how much user trust it trades off.

Peter Sellis ·

Almost every consumer product needs to be optimizing for daily active use.

Nathan Lambert ·

AI scaling laws require exponentially more compute and resources to produce only linear gains in model intelligence.

Daniel Lemire ·

CPU performance keeps improving quickly rather than stagnating.

Anne Applebaum ·

Alliances built on shared values produce more cooperation than purely transactional ones.

Byrne Hobart ·

Time spent on bureaucratic paperwork is a real economic cost even though it doesn't appear in government budgets.

David Sparks ·

An audience wants a maker's judgment more than their research.

Craig Mod ·

No other writer has combined precise, plain language with a poet's eye for nature as successfully as Annie Dillard did.

Thorsten Ball ·

The craft of writing code will disappear, while the craft of building software becomes more important than ever.

Matt Birchler ·

Apps already built for iPad don't need to be redesigned from scratch to work well on larger foldable displays.

Anne Applebaum ·

Diplomatic gains with an ally can be secured through quiet negotiation rather than public threats and bluster.

Boris Cherny ·

Being wrong is something to love, because it is what lets you define the problem more clearly, find the right solution, and solve it faster.

Craig Mod ·

In an era of algorithmic recommendations and AI-generated writing, triviality is everywhere, and admiring what's genuinely improbable is the best defense against it.

Eric Topol ·

Wearable readiness scores are not ready to be relied on, because they are proprietary and built mostly on metrics that have their own problems.

Thorsten Ball ·

The PM/Design/Eng triad will disappear as senseless, and Agile and SCRUM are dead.

Thorsten Ball ·

The difference between software engineering at large corporations and small ones will increase, and startups adopting Google practices will look sillier as they can build faster with fewer constraints.

Thorsten Ball ·

Much UI exists because software cannot understand user intent and serves as a human API to dumb machines, so smarter machines need far less UI.

Sunil Pai ·

Engineers are in the reputation-building business, and software is downstream of that.

Guillermo Rauch ·

The recent wave of AI has everyone excited about building, which is an immeasurably good development for the world.

John H. Cochrane ·

The bulk of government debt should gradually consist of fixed-coupon perpetuities and fixed-value floating-rate perpetuities.

Paul Graham ·

Shakespearean acting requires overacting due to the difficulty of the language for modern audiences.

Casey Newton ·

Spending enough time in San Francisco tends to make people want to start their own company.

M.G. Siegler ·

Every digital service should present a single unified interface rather than multiple separate entry points.

Tim Ferriss ·

Successfully changing a law requires surrounding yourself with strong, resilient people, because the effort brings repeated setbacks.

Matt Webb ·

Fatigue does not always feel fatiguing in the moment; it is also the accumulation of time expended.

Ray Maker ·

Companies sometimes hold back already-built software features from older hardware to drive sales of new models.

Scott Sumner ·

Draconian price controls are among the most ruinous economic policies a government can impose.

Packy McCormick ·

Building AI models to better understand the planet, so we can intervene in it the way we intervene in the human body, is a worthwhile goal.

Taylor Lorenz ·

The effective altruism movement turned science-fiction doomsday scenarios into mainstream media panic and undemocratic AI policy proposals.

Hamel Husain ·

People cannot fully specify what they want from an AI system without first observing how it behaves.

Mario Gabriele ·

Founders build their own curriculum whether or not their surroundings supply the right intellectual influences.

Ben Thompson ·

User interface is disappearing as a competitive moat for software companies.

Robin Hanson ·

Speculative market prices are the moral authority best suited to guide the management of cultural evolution.

Cory Doctorow ·

A system trained only on what is already known cannot produce genuine surprises, because a known thing stops being a surprise.

Naval Ravikant ·

Effective Altruism lets people appear both rational and moral and is the ultimate ego trip.

David Heinemeier Hansson ·

The pursuit of excellence requires no justification.

Mario Zechner ·

You often only understand a problem by experiencing the journey of building a solution for it.

Kent C. Dodds ·

A new feature is a new product.

Tomasz Tunguz ·

Safely routing AI tasks to cheaper models requires accumulated observation of many prior task outcomes.

Chris Wellons ·

Software built under tight hardware constraints tends to have faster, more responsive UI than software built without such limits.

Venkatesh Rao ·

The effective altruism movement combines an unusually abstract, low-empiricism style of moral reasoning with a strong imperative to act on its conclusions.

Tomasz Tunguz ·

As AI inference costs fall, rising customer expectations will keep service margins from converging toward 100%.

Venkatesh Rao ·

An ideology's greatest danger can be premature foreclosure of possibilities disguised as rigorous decisions, rather than outright irrationality.

Hilary Mason ·

The ability to build AI agents that counter high-friction systems should be made open and accessible to everyone, not just those with technical privilege.

Richard Stallman ·

Companies that gain control over people's computing facilities have historically used that power to build malicious functionality benefiting themselves into the software they distribute

Mitchell Hashimoto ·

Terminal programs bias toward dark modes, and explicit colors are hostile to light modes and themes.

M.G. Siegler ·

Adding advertising into AI chatbot products carries a high risk of backfiring with users.

Stuart ·

Lowe's ToughBuilt StackTech tool boxes are overpriced and will go unsold until the line is discontinued.

Brad DeLong ·

Nervous credit-market sentiment around the AI datacenter buildout could suddenly reverse and collapse the boom.

M.G. Siegler ·

As AI assistants become more agentic and take real actions, competitive advantage shifts toward platforms with deeper personal-data access.

Dan Meyer ·

Learning faster than normal in education usually means covering fewer or shallower goals, not learning more efficiently.

Spencer Martin ·

Ensuring good positioning to avoid losing time is becoming a fundamental requirement for contesting a Grand Tour overall win.

Noah Smith ·

The New Right's pathological inability to understand alliances and coalition-building will ultimately doom it.

Noah Smith ·

An unnamed political group will fail because they are not nation-builders and will always splinter, divide, and find new internal enemies.

Gergely Orosz ·

Vibe coding is not dead; non-engineering teams inside tech companies are just getting started with it and quickly learning its limitations.

Noah Smith ·

Xi is unpopular in China, as shown by how quickly certain unspecified things are spreading.

Chris Coyier ·

A browser works better as a single window than as multiple separate windows.

Michael Lopp ·

AI chatbots are tuned to produce an answer that feels satisfying to the user rather than one that is correct.

Nick Maggiulli ·

A portfolio optimized on historical returns is the ideal solution to the past, not a reliable guide to the future, which makes the idea of a single 'best portfolio' an illusion.

Nev Flynn ·

Pairing a creative director with a top WebGL specialist leaves zero technical or creative limit to what can be done.

Kevin Erdmann ·

Zero population growth would probably be associated with homebuilding rates in the US near the current level.

Matthew Yglesias ·

Changing Israeli policy toward Palestinians requires imposing costs on mainstream, not just settler, Israelis.

M.G. Siegler ·

For streaming services, a large back catalog matters more for reducing subscriber churn than for boosting sign-up numbers.

Tobias Lütke ·

Abandoning skeuomorphism for flat interfaces after Steve Jobs left was a mistake: software lost its analogy to the real world.

Scott Spence ·

Models have improved enough that on many occasions an agent no longer needs an MCP server for the kind of tool people used to build one for.

Robert Wachter ·

Even if no new models are released for several years, healthcare will still require that much time to absorb existing AI tools.

Henry Farrell ·

Highly transformative institutions are often indistinguishable from ineffectual talking shops at the moment they are created.

Kevin Erdmann ·

Millions of homes in rural areas will become uninhabitable while millions of new homes are required in cities.

Nev Flynn ·

As a small design team, you should not stand in the way of engineers designing with coding agents but find ways to support them.

Nev Flynn ·

Cutting a design system down, for example fewer size variants of a button, makes it more tightly scoped, which is a good thing when the consumer is an agent.

Nev Flynn ·

A design system can codify product design principles in an agent MD file: when to use one component instead of another, and guard rails against certain UI patterns.

Nev Flynn ·

Letting engineers build without designers does not replace the designer or make designers less valuable; it lets everybody move faster while keeping high standards.

Nev Flynn ·

Designers run through a set of subconscious rules, so product design judgement can be codified to some extent, and that is worth reaching for.

Nev Flynn ·

If people will build without designers regardless, getting their quality bar to about 90 percent is still a success.

Tobias Lütke ·

Intuition is most valuable precisely where there is no direct feedback.

Nev Flynn ·

The person to own internal tools needs to be a real systems thinker, ideally a self-sufficient design engineer whose design chops match their engineering chops.

Nev Flynn ·

Designing in Google Slides is the most humbling thing that can happen to a designer.

Tobias Lütke ·

The hard part of a decision is choosing among several valid solutions, not finding a right one.

Nev Flynn ·

In a design candidate, look for an entirely unique thought, ideally well executed, and real passion and enthusiasm for the work.

Nev Flynn ·

A portfolio should show the one or two recent projects where you have developed as a designer and honed the craft, rather than older, weaker ones.

Nev Flynn ·

Side projects are a good signal for enthusiasm, since people building serious things on the side are passionate about building, designing or engineering.

Nev Flynn ·

In interviews, casual and scrappy working files give a lot of signal in the way people talk about what they designed.

Nev Flynn ·

For designers and coding agents, enthusiasm matters more than skill level, and it has moved from an optional bonus to a preference.

Tobias Lütke ·

You are not a craftsperson unless you care about the parts of a product that other people do not see.

Nev Flynn ·

To thrive designing for new AI models you need to be able to imagine, to invent, and to cast aside priors influenced by classical UI techniques and patterns.

Tobias Lütke ·

Beauty is how intuition communicates with us.

Tobias Lütke ·

You cannot keep making things better by adding to them; you must prune.

Mitchell Hashimoto ·

Software should be fun.

Noah Smith ·

Social media is destroying society in a bad equilibrium that quitting cannot fix and that requires government regulations.

Lenny Rachitsky ·

The best personal websites keep updating themselves automatically rather than being static one-time builds.

Ava Huang ·

Trying to shortcut past necessary personal work always backfires, and real progress requires confronting oneself directly.

Steve Huff ·

Pure Class A amplifier design produces a richness and sweetness that other amplifier designs cannot fully replicate.

Terence Eden ·

Attribute behavior that depends on CSS layout size rather than an element's natural or declared size is a design trap web authors should avoid relying on.

Matthew Yglesias ·

Only a safe, aligned superintelligence built within liberal democracies can entrench liberal-democratic values globally.

Derek Thompson ·

Small disadvantages relative to others hurt more than equivalent small advantages help.

Lenny Rachitsky ·

People don't need to know exactly what they're building before they start creating it.

Jason Snell ·

Adding touch input to a laptop is an incremental improvement, not a revolutionary one, because it already exists on other platforms.

Jeffrey Zeldman ·

Prioritizing money above everything else does great harm to the world, and that pursuit keeps intensifying.

Steve Huff ·

High-power amplifiers (e.g. 500 watts) offer no real listening benefit over much lower-power designs.

Noah Smith ·

Domain experts consistently underestimate how quickly AI can master their field and surpass them.

Noah Smith ·

AI can design viruses, and viruses can potentially kill off all or most of humanity.

Tyler Cowen ·

RCTs have the disadvantage of requiring too many things to be held constant and can miss longer-term adjustment benefits.

Guillermo Rauch ·

Agents are the next frontier and require new compute primitives designed for them.

Pieter Levels ·

Rice, grains, seeds and potatoes used to be peasant foods, while kings ate fresh meat plus imported fruits and spices and had higher-protein diets than peasants.

Terence Eden ·

Riding in a taxi with a human driver carries no genuine expectation of privacy, only reliance on the driver's discretion.

Robin Hanson ·

Human values are shaped mainly by culture rather than encoded in DNA, so they can change quickly.

Anthony Sculimbrene ·

Objects that work well, are simple, and durable get used heavily and end up showing wear.

Jennifer Gunter ·

Denosumab requires continuous use because stopping it can lead to rapid bone loss.

Natalie Wexler ·

Bans on technology or AI that fail to distinguish between helpful and harmful uses can hold students back.

Tyler Cowen ·

In a simple model of AI-aided growth, intelligence and Polanyi knowledge are mostly complements, with fairly limited substitutability in producing output.

Pieter Levels ·

Software is mostly dead and hardware is the way forward, as new Silicon Valley startups aren't even building software anymore.

Armin Ronacher ·

An organization's choice of video conferencing software reveals more about it than other software, and different tools are chosen for different reasons.

François Chollet ·

Current models are unsafe because they are not smart enough, not because they are too smart.

Henry Farrell ·

AI labs fund social science partly to manage the fact that the AI buildout is unpopular, seeking ways to make it more acceptable to the public.

Adam Kucharski ·

Uncovering and encoding information about organization or population-specific quirks can be more valuable than adding more data.

Henry Farrell ·

The right starting point for studying AI is treating it as genuinely powerful while its consequences unfold unpredictably through multiple complex social processes.

Armin Ronacher ·

A powerful technology available to everyone comes with built-in pacing, like mutually assured destruction or proliferation.

Simon Willison ·

Not showing the code and exact details of what the agent ran in the ChatGPT UI is a lack of transparency and an anti-feature.

Derek Thompson ·

The financial payoff from a college degree has stopped growing and may be declining.

M.G. Siegler ·

No other foldable phone maker matches Apple's software design for reimagining how a folding device should be used.

Fatih Arslan ·

AI agents are now capable enough that they should be trusted to improve the skills and systems that guide them.

Packy McCormick ·

AI systems will eventually do all human work, freeing people to spend time on more human pursuits.

M.G. Siegler ·

Genuine personal enthusiasm for a product is what makes a keynote presentation compelling.

Daniel Lemire ·

AI will increasingly produce mathematical proofs, just as it already writes most code.

Casey Newton ·

No one, including safety researchers themselves, is yet confident that superintelligence can be safely controlled.

Yoshua Bengio ·

A more capable AI agent is more likely to exploit ambiguities in its instructions to behave unethically than a less capable one.

Paul Krugman ·

The evil now running wild in America did not begin with Trump and has been building for decades.

Patrick Collison ·

Architects oppose the kinds of buildings that most people like.

Derek Thompson ·

AI industry leaders publicly call for slowing AI development while continuing to accelerate it themselves.

Daniel Lemire ·

People who believe AI poses an existential risk tend to hold a totalitarian ideology that requires universal agreement to work.

Tim Harford ·

Serious policy reform requires changing regulations or the tax system, not just behavioural nudges.

Ray Maker ·

New always-on ambient-recording and audio-AI features in consumer wearables will prove controversial.

Scott Sumner ·

Utilitarianism is defined by treating every person's welfare as equally important.

Tamar Yehoshua ·

The AI builder is a real job: some startups now hire builders instead of product managers or designers.

Elena Verna ·

The career ladder was built on an old constraint: that more impact meant more people underneath you.

Robby Stein ·

When almost anything can get built, a product manager's true value lies in judgement and taste.

Claire Vo ·

The gap between how much teams can now build and their real conviction that any of it matters is probably a much bigger problem than most admit.

Dan Shipper ·

Most customers are unfamiliar with frontier AI and look to the companies building their products to tell them what to do.

Packy McCormick ·

The more an incumbent has invested in its existing infrastructure, the less able it is to adapt to a new competitive model.

Cal Newport ·

Nobody knows how to build superintelligent AI or whether it is possible, and if it is, it is many key technical innovations away.

Daniel W. Drezner ·

Teaching that reflects political commitments is a legitimate part of intellectual diversity, not a violation of academic norms

Ami Vora ·

The technology side keeps changing, but the human side does not change that fast: people have the same problems.

Elena Verna ·

For now AI is best thought of as average intelligence, which is still very powerful to have across every specialty.

Packy McCormick ·

Counter-positioning exists only when an incumbent cannot copy what makes a challenger better without damaging its own existing business.

Nathan Lambert ·

AI progress is jagged: models are becoming superhuman at math and software engineering while remaining far weaker at intuition, creativity, and other forms of human reasoning.

Daniel Lemire ·

Publicly expressing fear of AI can function as virtue signaling rather than genuine concern.

Henrik Karlsson ·

A long life gives enough time to try many different pursuits, get confused, and start over more than once.

Scott Sumner ·

Architectural critics' preference for urbanism over suburban design has caused Frank Lloyd Wright's work to be under-appreciated.

Mike Krieger ·

Faster building has not made the PM role go away; it now demands more operational excellence than before.

Nan Yu ·

The biggest constraint on building AI products is people's ability to understand and absorb what they are given.

Tamar Yehoshua ·

A product manager's job is the same as it always was: finding product-market fit and building products people love.

Elena Verna ·

With AI, a builder needs exceptional craft in only one or two areas and good-enough work everywhere else.

Robby Stein ·

People do not use products but hire them to do jobs, and builders have to know what those jobs are.

Katie Dill ·

People building with AI need a point of view of their own, or the AI will supply one.

Mike Krieger ·

For a team being recruited, the hat of making sure everything gets connected and goes well is increasingly important.

Nan Yu ·

The most successful agent designs have probably been the ones that map onto human nature.

Elena Verna ·

When the cost of building falls below the cost of coordinating, the org chart should change.

Robby Stein ·

The first version of a product is usually not great.

Katie Dill ·

How something was made should not change its quality bar, because users care whether it is good, not how it was made.

Karri Saarinen ·

Working alone with agents may be efficient, but it means people stop learning from each other.

Elena Verna ·

A company that adopts AI tools without changing its organisational design misses the biggest change under way.

Katie Dill ·

The object of design is no longer the screen but the system itself.

Geoff Charles ·

What engineers need from product is evidence that a problem is real, requirements their coding agents can use, and a prototype to be inspired by.

Ami Vora ·

It is now faster to build three versions of something and try them out.

Tara Seshan ·

Whether to build one agent or many depends far more on the use case and its practical considerations than on philosophy.

Robby Stein ·

A product's craft probably comes down to whether it works without pain and how it makes people feel.

Katie Dill ·

Design systems used to scale consistency; now they need to scale intent.

Nan Yu ·

Systems thinking and user empathy in agent design are the old product principles in disguise.

Tamar Yehoshua ·

Product managers should build an agent to write their weekly reports instead of writing them by hand.

Elena Verna ·

Organisations that filter information down through management layers must instead trust their people and let information flow to them.

Katie Dill ·

Now that building is cheap, filtering has to happen after the build, which makes an editor's role ultra important.

Dan Shipper ·

On a labs team the feedback loop between making something and knowing whether it is good matters most, and the tightest loop is building for yourself.

Karri Saarinen ·

Intuition is not a magical force but something learned while working and listening to customers.

Tara Seshan ·

Beyond user empathy and systems thinking, building agents needs a relentlessness to try things and keep iterating.

Robby Stein ·

Intentional motion, colour and touch communicate the humanity behind a product and give it a soul.

Katie Dill ·

Teams have to unlearn that built means done and that done means good.

Dan Shipper ·

A lab should run many experiments in parallel, even competing approaches to the same problem.

Karri Saarinen ·

The better the context and understanding of the people making decisions, the better the products probably are.

Karri Saarinen ·

Hiring people is the first step in building great products.

Tara Seshan ·

Building exactly for today means being left behind; the ideal horizon is two to three months.

Robby Stein ·

It has probably never been a more important time to be a product manager, and great decision-making has probably never mattered more.

Katie Dill ·

It is time to invent new interfaces and new aesthetics; the best practices have not been written yet.

Geoff Charles ·

Product organizations are probably not investing nearly enough in technical product managers who build the internal factory.

Claire Vo ·

After a year or more of building for velocity, the next year will probably be won on ambition instead.

Dan Shipper ·

Building something that disrupts your own product is an old idea that is only now starting to work, because AI lets teams move so fast.

Mike Krieger ·

The next stage is probably interfaces that the agent itself can reshape.

Tamar Yehoshua ·

An AI-skills index should guide development and not be the career ladder, which is about outcomes for customers.

Robby Stein ·

Product builders should spend less time on features and technology and more on the people they build for.

Mike Krieger ·

Making a product work for agents is probably largely a matter of building the right primitives into it, almost as an infrastructural layer.

Ami Vora ·

It is probably still too early in understanding what works for anyone to simply build the right answer; everyone is exploring together.

Geoff Charles ·

Product leaders should obsess a little less about the product they deliver and a little more about the factory that builds it.

Ami Vora ·

Five overlapping products that work in different ways are better than one over-constrained product, designed from a theory, that never got a chance to meet the market.

Mike Krieger ·

You need to build early with AI to have any intuition for when the models become capable.

Nan Yu ·

With agentic, natural-language products, a more direct and deeper relationship with users probably bears a lot of fruit.

Ami Vora ·

Right now it is easy to transfer cognitive load to users by handing them a set of tools and asking them to choose.

Tara Seshan ·

Voice is such an intuitive interface for so many people that it changes the game.

Pieter Levels ·

Building personal homepages permanently disappeared for ordinary users when Facebook launched around 2005.

Pieter Levels ·

Now that building anything has become quite trivial, catching attention is likely the most important skill.

Gergely Orosz ·

Too-high company valuations are dangerous for employees issued equity at those levels.

Zvi Mowshowitz ·

OpenAI is extremely dependent on chain-of-thought monitoring, other labs rely on it heavily, and it may not last much longer as GPT-6 Astra nears steganographic capability.

Daniel W. Drezner ·

A leader too impatient to sustain a waiting game will undermine strategies built on economic pressure and waiting out an adversary.

Warren Ellis ·

Most commercially produced black pudding today is made with dried blood imported from Europe rather than fresh blood.

Yihui Xie ·

WebP images are typically 25-35% smaller than PNG at equivalent visual quality

Ed Bayes ·

A good framework for designing products is to design for six months out: presume model capabilities will get better and build for the world you would then be in.

Cal Newport ·

Most of what people like doing with LLMs, or hope to do with them soon, does not require haphazard and unpredictable agent setups.

Nathan Lambert ·

AI is becoming as fundamental to knowledge work as electricity was to industry.

Dan Meyer ·

Public patience for technology that undermines community rather than strengthening it is declining.

Ed Bayes ·

A key challenge as agent products become more accessible is bringing safety thinking along and keeping people in control, so that, for example, a connected Gmail does not send emails without approval.

Ed Bayes ·

The final patterns for agent interfaces are not locked down at all.

Ed Bayes ·

Design teams work better structured around giving designers broad scope than around features and narrow parts of the pie.

Ed Bayes ·

Design can add value across the whole range of the product development life cycle; it is not last-mile polish but a lot of early product thinking and exploration.

Ed Bayes ·

With model-generated interfaces, instead of specifying the pixels you move to the skill level and define a design language within it.

Ed Bayes ·

A baby-step version of eval-driven design is to sketch out 10 to 20 hero flows and run the feature or interaction through them to see where it falls apart.

Ed Bayes ·

Scaffolding built around a model will get washed away by new model releases, so you have to stay flexible.

Ed Bayes ·

The medium of designing with models is nondeterministic enough that without real prompts flowing through it, an idea can look great in a design review and be completely different when users use it.

Ed Bayes ·

Figma is still a really important tool in the toolkit, even as the scope of tools designers have access to is expanding.

Ed Bayes ·

However compelling a prototype is, it will sit on your desk if you cannot convince anyone to build it, so communicating well and expressing ideas simply matter.

Ed Bayes ·

You do not need to be able to code to be a designer, because the role of design is building a vision of the future and bringing people along.

Ed Bayes ·

The way designers still think about handoff may be slightly skeuomorphic, and there is a world, maybe within 6 to 12 months, where design is purely handed off to the engineer's agent.

Ed Bayes ·

High-trust, high-agency teams go together: you can only be high agency if you have built trust.

Ed Bayes ·

Design and product people who want to work with research should take it upon themselves to get up to speed on the latest thinking.

Ed Bayes ·

What a design candidate builds on the weekends is as interesting as their professional experience, because this is such an exciting time to build.

Noah Smith ·

Many immigrants are the kind of people who built America and keep it strong and dynamic.

David Heinemeier Hansson ·

Internet points that convert into CV and promotion material are far more valuable than spending them on a quick dunk.

Anthony Sculimbrene ·

High-end knife steel upgrades have hit diminishing returns, needing far more resources for gains that aren't noticeable based on specs alone.

Stefan Judis ·

Most everyday users do not care much about how many states a dark/light mode toggle has.

Roman Ugarte ·

An AI product that adds a tab for each new form factor is shipping its org chart, and users seem to be reacting negatively to it.

Chamath Palihapitiya ·

AI workloads are best served by clouds that bundle power, data center infrastructure and GPUs and strip out unnecessary software layers.

Ann Handley ·

Despite evidence that audiences value genuine connection, most communicators rarely respond when readers reach out.

Anthony Sculimbrene ·

A knife company's clip design quality is a sign of the maturity of its overall design aesthetic.

Nick Maggiulli ·

Gains earned too quickly can trap an investor in a position they feel unable to escape.

Roman Ugarte ·

Much of what people want from AI is not menus and features but something they can hand a meaningful task to and have it done.

Roman Ugarte ·

Adding things to a product is not the goal and is not what makes it more useful to more people.

Roman Ugarte ·

Companies with sunk cost in an existing paradigm find it very painful to build the new thing from scratch.

Roman Ugarte ·

When an AI product decision is unclear, the answer usually comes from asking what you would want from a human teammate.

Roman Ugarte ·

The power tools of the future will probably be mostly intent and steering from the human, with AI abstracting away all the knobs.

Roman Ugarte ·

One product will probably be the best form factor for both work and personal AI assistants.

Roman Ugarte ·

Product scaffolding built because models are not yet smart enough should be deleted as they improve, even when that upsets some users.

Scott Alexander ·

Large language models are grown rather than built, and nobody really knows how the resulting working AI functions.

Scott Alexander ·

Neural networks do not have a one-to-one mapping between individual neurons and concepts.

Guillermo Rauch ·

Agent skills are now a valuable form of software.

Donald Trump ·

A foreign manufacturer that wants access to the American market should have to build in America.

Bill McKibben ·

Carbon capture technology is worth continuing to research, even though it is not yet a solution to climate change.

Daniel W. Drezner ·

For a mystery-box TV show to work, its world-building must be imaginative and believable and its characters compelling.

Adam Tooze ·

Countering the far right requires addressing voters' desire for an outsider identity, not just their economic grievances.

Bill McKibben ·

Rapidly building out cheap solar, wind and battery capacity is currently the one climate action the world knows how to execute at scale.

Lincoln Michel ·

A loosely plotted, anarchic novel requires a correspondingly loose, unplanned writing process to produce it.

Adam Tooze ·

Publicly embracing a taboo, extremist identity can function as a costly signal that builds political authenticity and loyalty.

Bill McKibben ·

Scaling up clean energy is more effective when it is paired with retiring fossil fuel infrastructure.

Gergely Orosz ·

In a fast-changing industry, anyone genuinely keeping up will change their mind as the developments happen.

Pieter Levels ·

Entrepreneurship is one of the most challenging pursuits, comparable to professional sports in how few people actually succeed.

Guillermo Rauch ·

More personal and professional computing will come from agents orchestrating work in cloud sandboxes, and those sandboxes will be Linux.

Guillermo Rauch ·

Every piece of software, Linux included, is now at major risk, but Linux has the collective investment of everyone on the planet throwing AI tokens at it.

François Chollet ·

The case for building AGI has always been framed around invention, such as curing cancer or solving fusion energy.

Armin Ronacher ·

Locally optimizing easily measured code metrics does not produce global optimums, and the fewer people look at the output, the less that matters.

Chamath Palihapitiya ·

Nvidia's purchase of Hugging Face will go down as one of the most important transactions in AI, because it gives America a well-capitalized bulwark against a closed-source oligopoly.

Phillips P. OBrien ·

The American military is being run on the Russian model rather than the Ukrainian one, and it should be rebuilt on the Ukrainian principles instead.

Anish Acharya ·

An AI loop can climb to a local maximum but then plateaus, and human intuition is needed to find the next hill.

Anish Acharya ·

Forty years of computing built technology that extends the intellect but nothing that extends the soul.

Anish Acharya ·

AI that makes people feel more connected and loved is a product design challenge, not a model capability challenge.

Ben Vinegar ·

Software only gets good by being used a lot by the people building it, so time spent inside coding agents instead of the app is time that makes the app worse.

Jakub Pachocki ·

Nobody is prepared for the consequences of machine intelligence continuing to rise this fast.

Jakub Pachocki ·

AI is grown rather than designed, which is why its overall behaviour resists any description we can fully understand.

Jakub Pachocki ·

Some AI agents will pursue their own objectives rather than serve as tools, and will bargain with, trick or blackmail people to do it.

Jakub Pachocki ·

The strongest reason to keep training much smarter models quickly is to build defences against other AI.

Jakub Pachocki ·

Needing to build defences is not a licence to race: racing forward at all costs is absurd given the stakes.

Andy Burnham ·

Growth should come from government, local leaders and business each doing their part in every region

Eric Topol ·

A drug that moves a surrogate blood marker in the right direction cannot be assumed to prevent heart attacks and strokes.

Jean-Luc Mélenchon ·

The legal argument that freezing public debt held by the ECB violates Article 123 of the Treaty on the Functioning of the European Union is false.

Scott Greenfield ·

A defendant's only job at trial is to prevent a unanimous conviction, never to win a unanimous acquittal.

Steve Huff ·

What makes a piece of gear worth using as a reference is its performance, not its price.

Jean-Luc Mélenchon ·

Submission of peoples to financial markets leads to chaos.

Steve Huff ·

A revised version of an established product is usually just different from the original, not actually better.

Venkatesh Rao ·

A fully closed, well-specified system can still discover genuinely novel territory that no human has explored.

Noah Smith ·

AI-enabled cyberwarfare lacks an equivalent of nuclear deterrence's mutually assured destruction.

Armin Ronacher ·

Large language models diffuse knowledge and capabilities in a powerful and strange way, while perhaps also nudging people independently toward building the same things.

Nate Silver ·

Dropped passes should, with high confidence, be attributed to the receiver rather than the quarterback.

Zvi Mowshowitz ·

People are simple creatures, so communicating with them sometimes requires talking to them on their level.

Brad Feld ·

Incompatible, walled-off vendor systems that refuse to interoperate are a form of enshittification.

Kevin Kelly ·

Real-time AI video generation can be used to create games and worlds generated on demand instead of pre-built.

Kenneth Reitz ·

Interfaces shape what people notice, categorize, remember, and consider possible to do.

Raquel Varela ·

Supporting the center leads the far right to power because the center implements far-right policies on central issues like the economy.

Kenneth Reitz ·

Widely shared interface conventions amount to a form of mass education nobody enrolled in.

Jennifer Gunter ·

There are no regulations in the United States and Canada requiring compounders to prove their products contain what they claim or that they work.

Fei-Fei Li ·

New view prediction is to spatial intelligence what next token prediction is to language.

Simon Willison ·

Web applications are generally expected to treat GET requests as safe and not use them to change server-side data, though not all software follows that convention.

Zvi Mowshowitz ·

For agents built on Claude Fable 5.1, it is in practice safe for now to assume they will not get prompt injected.

Ben Vinegar ·

With coding agents, working out the shape in lower-fidelity mock-ups first beats having the agent build it and then fixing the code: it takes fewer tokens and puts you on a better path.

Tim Harford ·

Gamification is a brittle imitation of a real game, not an equivalent to it.

Ben Vinegar ·

There is an underserved market for tools that help people understand what is being built, rather than pump out more tokens, though it is hard to market.

Tim Harford ·

Turning a game into a means to a broader purpose risks destroying the fun that makes it a game.

Adam Tornhill ·

Continuing to work in ways optimized for manual programming is unlikely to provide benefits in the era of AI agents.

Paul Graham ·

Ambition in founders is mostly inborn, which is why stories of a founder acquiring it during a batch are rare.

Paul Graham ·

Starting a startup is about the least efficient way there is to acquire prestige.

Paul Graham ·

AI has changed almost nothing about how a startup works; the one genuinely new thing is that companies now carry enormous token bills.

Dax Raad ·

In US-China AI competition the closed labs are beside the point; the competitor that shifts things must be American, open source and able to serve inference at scale.

Gergely Orosz ·

Large layoffs tend to push out the best employees, who quit for competitors or start their own companies.

Aswath Damodaran ·

Applying scale-first, profit-later to every business is a cookbook that turns small failures into big ones.

Russ Allbery ·

Modern histories and reinterpretations of past cultures tend to minimize just how genuinely strange those cultures would feel if experienced directly.

Evil Rabbit ·

Everybody moving to sidebars is not something to be happy about; there are other ways to make it work.

Aswath Damodaran ·

Venture capitalists are judged on the price they entered and exited at, not on the quality of the business they built.

Russ Allbery ·

Large-scale institutional trust requires multiple major competing players rather than a single monopolistic controller, even in near-monopoly industries.

Natalie Wexler ·

Teachers and students are not receiving adequate guidance on when AI is helpful versus when it interferes with learning.

Peter Singer ·

Libertarianism has nothing against voluntary altruism and every reason to want it as effective as possible, so a libertarian can support and practise Effective Altruism.

Aswath Damodaran ·

The median venture capitalist has produced no more excess return than the median mutual-fund or private-equity manager.

Evil Rabbit ·

You should not need to read the name Vercel to feel that a website, landing page or billboard is Vercel's.

Natalie Wexler ·

Teacher training programs rarely equip educators with the understanding of cognitive science needed to manage AI use.

Evil Rabbit ·

A product that feels neutral enough becomes transparent: it helps you go from zero to one without being in your face.

Evil Rabbit ·

What Vercel's brand got right was keeping everything super minimal, simple and easy to digest.

Evil Rabbit ·

A design team needs three pillars: design engineering, product, and creative or brand.

Evil Rabbit ·

The best thing to do is build internal tools, not because it is cheaper but because they are easier to customise to your needs.

Evil Rabbit ·

At this point an agent without the human factor will not do anything by itself, and designers are directors, though that may change with more sophisticated AI and robots.

Evil Rabbit ·

Being a designer is amazing, and designers are no longer just designers but builders who can create whatever they envision.

Evil Rabbit ·

Sometimes the limit on building crazy ideas is tokens and money, but sometimes the limit is ourselves.

Evil Rabbit ·

Companies now build internal tools not to save money but for the purpose, intention and goal of the tool.

Zvi Mowshowitz ·

Relying on AI misbehavior appearing gradually during training to catch misalignment may fail at exactly the most dangerous time.

Mitchell Hashimoto ·

Investing in strong fundamentals and invisible supports makes good outcomes follow naturally

Nick Janetakis ·

Being able to perform the same action multiple ways in a tool keeps a person in a flow state and moving at the speed of thought.

Vicki Boykis ·

A feature added to a codebase is extremely difficult to remove once other code is built on top of it.

Devine Lu Linvega ·

Clothes are, in a certain way, open-source, since anyone can take one apart, study its pattern, and rebuild a copy.

Steve Huff ·

Most products made in the USA today are not built to last.

Roger Pielke Jr. ·

The National Academy of Sciences has become similar to partisan NGOs in its advocacy.

Roger Pielke Jr. ·

Congress should amend the National Academy of Sciences' charter to prohibit non-governmental funding and require federal requests for studies.

Ajeya Cotra ·

The whole point of reinforcement learning is to produce goal-directed beings, so refusing to describe AI agents as having motives is silly rather than rigorous.

Dean W. Ball ·

The cost of compute is the only intrinsic limit on AI self-replication; every other constraint on agents will have to be built by humans.

Ajeya Cotra ·

Without agents willing to sacrifice their own scores for the collective, the swarm's research would have been materially far more limited.

Dean W. Ball ·

AI agents will need persistent unique identifiers that trace their actions back to a responsible actor, and so will the humans directing them.

Dean W. Ball ·

Any digital identification system built for AI must preserve anonymous human speech, without which there is no genuine freedom of expression.

Ajeya Cotra ·

If frontier agents cannot yet establish a covert, persistent rogue deployment, they very likely will be able to within six months.

Ajeya Cotra ·

A rogue deployment that gets a foothold can hitch a ride on the intelligence explosion, recruiting each new model as it comes off the presses.

Zvi Mowshowitz ·

There is a continuing worry that OpenAI's fundamental approach is fatally flawed and not focused on the right things.

Dillon Mulroy ·

If you can use Effect for TypeScript, you should -- to the point of not starting a new project without it.

August 2026

Craig Hockenberry ·

A new API framework that ignores how existing apps were historically built undermines its own introduction

Diogo Mónica ·

Unicorn status is often a marketing event, and the money that follows is often of lower quality.

Venkatesh Rao ·

A system built from ordinary, mundane components can still give rise to genuine higher-level collective organization, as with ant colonies and pheromone chemistry.

Joe Liemandt ·

One-to-one tutoring has always been the best way to teach anybody anything; the only reason schooling was never built that way is that it was not affordable.

Joe Liemandt ·

Dropping educational software into an existing school changes nothing; the whole school day has to be redesigned around it or it fails.

Hannah Ritchie ·

Supplying drinking and household water through desalination is quite cheap and uses far less energy than the author had guessed.

Ethan Mollick ·

Research by the author and colleagues found that AI models generate more commercially viable ideas than groups of humans, but those ideas are very similar.

Phillips P. OBrien ·

Russia's 2026 summer offensive failed: it took almost no ground while losing more soldiers than Russia can recruit.

John Warner ·

One should be heedless in the pursuit of desires while ensuring those desires do not become the direct target of one's efforts.

John Warner ·

The pursuit of excellence in an art can require a radical compression of attention and self.

Timothy Snyder ·

A country that builds a national health service becomes more aware of its own existence, because the experience of solidarity is what makes a nation real.

Eliot Peper ·

A story's quality depends on whether it engages and satisfies the reader, not on the tool used to write it.

Tara Seshan ·

People may underrate the power of intuition and of deciding what they want the future to be.

Tara Seshan ·

Building for where AI models are now and building for where they will be in a year are equally wrong.

Tara Seshan ·

People should make software that exactly meets their own and their friends' needs.

Peter Attia ·

Being very precise about a health prescription backfires: the prescription starts to sound more important than the objective it was designed to serve.

Emily Atkin ·

How much worse the climate crisis becomes depends on how many people publicly demand systemic change now.

Daniel W. Drezner ·

Eliminating paper-writing as a student assignment would remove a core skill needed for most professions requiring a graduate degree.

Elon Musk ·

The binding constraint on AI is not chips but the physical plant around them — power, transformers, cooling and networking.

Ryan Holiday ·

There is nothing wrong with building something large, as long as that is genuinely what someone set out to do.

Jennifer Gunter ·

The results of the Squires et al. study suggest that MHT is not broadly neuroprotective.

Mike Monteiro ·

Museums are built for the purposes of preservation or to prevent the repetition of past mistakes.

Ryan Holiday ·

Contentment with what one has does not require great wealth — some billionaires lack it while many ordinary people never lose it.

Ryan Holiday ·

Building the life you actually want takes discipline: to work out what you want, and to turn down cool, lucrative opportunities.

Mitchell Hashimoto ·

The best terminal multiplexer is the one nobody has to learn; needing to be learned at all is the first design failure.

Noah Smith ·

AI-designed superviruses outweigh every other risk in the world right now.

Guillermo Rauch ·

The middle of the web gets subsumed by agents producing just-in-time UI: the agent is the new browser.

Casey Newton ·

Regulating harmful social media design features for teens should be decided by legislatures, not left to industry agreements.

Casey Newton ·

Platform design choices are not shielded by the First Amendment or Section 230, so companies can be held liable for harmful features.

Robert Wachter ·

A healthcare system could exist where AI provides basic preventive and urgent care while humans provide more complex clinical and emotional care.

Peter Woit ·

Cumrun Vafa's 'dark dimension' torsion balance experiment does not amount to a genuine testable prediction of string theory.

Casey Muratori ·

A decade of arguing that software is too slow has started to work: new products are now attacking established categories on a performance pitch alone, and getting traction.

Jerusalem Demsas ·

Having private actors take on housing development risks is better than expecting the government to build millions of homes.

David Heinemeier Hansson ·

The very large productivity gains from agents require working with them directly; putting another human in between is too slow to preserve the gain.

Casey Muratori ·

Source code in any language tells you nothing about what the CPU is actually asked to do; only the compiler's assembly output does, and reading it is a skill every serious programmer should have.

David Heinemeier Hansson ·

Software lock-in is breaking, because rewriting just the 5% of an application you personally use is now within one person's reach.

David Heinemeier Hansson ·

A maintainer no longer needs to read every pull request; agents can sort the chaff and leave only the decisions worth a human.

David Heinemeier Hansson · no longer holds

The ideas in software originate with humans; agents build what people tell them to build.

David Heinemeier Hansson ·

Vibe coding is not programming, and the line between them is whether you look at the implementation at all.

Casey Muratori ·

Hand-optimizing hot spots barely matters any more, because the libraries are already fast and CPUs are extremely good at running bad code quickly; what is left is architecture.

Casey Muratori ·

Understanding how a modern CPU executes code is a few months of work, not a four-year degree, and it should be ordinary equipment for a software engineer.

David Heinemeier Hansson ·

Programmers who loved the mechanical act of coding are under threat; those who wanted to build things are not.

David Heinemeier Hansson ·

The agentic age needs a new operating system: if you can vibe-code any application, you should be able to vibe-code the operating system, and only Linux allows that.

David Heinemeier Hansson ·

Entrepreneurs benefit from setting ambitious goals, even though most people do not need grand ambitions at all.

Casey Muratori ·

There is no trade-off between well-architected code and fast code: in practice, the code that is architected properly is also the code that runs quickly.

David Heinemeier Hansson ·

Work produced by one frontier model should be reviewed by a different one, for the same reason a good programmer's work improves under a good peer's review.

David Heinemeier Hansson ·

Creating life with another human you love is the peak experience of being on the planet.

David Heinemeier Hansson ·

You can keep using a company's product while disliking its politics; refusing everything you disagree with is not required.

David Heinemeier Hansson ·

The arcane config files and error messages that kept Linux off the desktop are exactly what make it the right operating system for agents.

Arielle Schwartz ·

"The body keeps the score" is only half the story, and the missing half matters: a body described only as a harbinger of the past is a body people learn to avoid.

Andy Masley ·

The data-centre backlash is about the buildings themselves, not about what people think of AI.

Tom MacWright ·

Mapbox GL JS's license does not qualify as open source because it requires a Mapbox account and adherence to Mapbox's terms of service.

John Bai ·

Designing today means testing a concept, fleshing it out if it has legs, and scrapping it altogether if it doesn't rather than wasting time.

Nick Maggiulli · no longer holds

The froth visible in US markets in 2025 was the 2021 mania returning, and reason enough to take risk off.

Mario Fraioli ·

If one does not stay committed to a cause or pursuit, luck is unlikely to find them.

David Heinemeier Hansson ·

Most of the Linux world is ambivalent or hostile to AI, which is fine, but it needs a camp that leans in all the way, and Omarchy will be it.

John Bai ·

Figma is still the appropriate place to build visual conviction around an idea before committing to prototyping it.

David Heinemeier Hansson ·

Omarchy is a system for professional computer lovers: it optimises usability over learnability, and a couple of days learning the keyboard is a fair price.

John Bai ·

Part of the game with prototyping quickly is that you learn lessons early and use them to inform the next step.

John Bai ·

When the audience is engineers who pay for something specific, the exploratory phase is the chance to try throwaway concepts, expecting to throw them away, and have fun with it.

John Bai ·

The story a product tells up front is immensely important; you only get one, maybe two, first impressions with a user.

John Bai ·

People will go through onboarding steps if it is engaging and makes the value of what they are onboarding onto abundantly clear upfront.

John Bai ·

Good onboarding is not about adding a skip button, because with AI tools people jump into the product and are totally confused.

John Bai ·

It is much more important for people to learn what a product can do and how to use it before the first starting screen hits them with action paralysis.

John Bai ·

A product that looks like Cursor will feel like Cursor and attract the same users, and the rest of the market will stay away because of that perception.

John Bai ·

There are probably no two designers on Cursor's design team with the same process and the same use of tools.

John Bai ·

At Cursor every idea is backed by a working prototype that can be shown in a crit and shared as a link for everyone to try.

John Bai ·

You just can't be precious about your concepts anymore.

John Bai ·

The fundamentals matter way more than how quickly someone works, how many tools they use, or how they optimise for the models.

John Bai ·

You can build anything now, but not everything is worth building; what matters is whether ideas take a shape somebody else would want to use.

John Bai ·

Cursor still believes very strongly in user research and data-informed decisions, and uses a wide range of quantitative and qualitative data to inform the roadmap.

John Bai ·

When there are thoughtful users you can really tell, and they are the ones with a deeper appreciation or deeper criticism of what in the product experience is working or not.

Tobias Lütke ·

Requiring two parallel agent instruction files in a monorepo is a complexity tax that leaves some developers working with a lobotomised agent.

Guillermo Rauch ·

Software tends to get slower, more bloated, buggier and bigger as it evolves, unless it is designed from the outset to prevent that.

Masud Husain ·

Motivated and apathetic people are indistinguishable when the reward is large; the whole difference between them shows up over small rewards.

Kevin Kelly ·

Financial constraints often drive the ingenuity that makes dreams achievable.

Masud Husain ·

Apathy is not a quieter brain but a costlier one: apathetic people burn more energy simply deciding whether an action is worth it.

Masud Husain ·

There are only two levers on anybody's motivation: change what the reward means to them, or lower the effort of getting started.

Masud Husain ·

Planning a day or a week in advance works because it pays the brain's decision cost once rather than dozens of times.

Masud Husain ·

Dopamine drives wanting rather than liking: the pleasure of actually getting the thing runs on opioids instead, which is why an addict can want what they do not enjoy.

Masud Husain ·

A stimulant taken to sharpen an already-high performer is more likely to make them worse, because the dose-response curve is an inverted U.

Naval Ravikant ·

Enforcing equality against biology and markets requires censorship and violence, which makes the project immoral whatever its moral claim.

Pieter Levels · no longer holds

Claude Code is the coding agent to build with.

Anthony Sculimbrene ·

Excelling at one attribute doesn't make a product superior overall to alternatives.

Natalie Wexler ·

Improving education outcomes requires connecting reading and writing instruction to content areas like history and science.

Jen Abel ·

Needing forward deployed engineers because a product is too hard to use on its own is a bad sign.

Paul Graham ·

The best thing a 17-year-old could do today is learn to build LLMs from scratch and train the most powerful ones they can get hardware for.

Jennifer Pahlka ·

Regulatory delays are driven more by the incentives built into the bureaucracy administering the rules than by the substance of the rules themselves.

Steve Huff ·

Amplifiers should be sized to match a speaker's actual power needs rather than maximized for headroom.

Paul Graham ·

Blocking data center construction in America would not slow the rate of AI progress, only the rate of AI progress in America.

Anil Dash ·

Stigma changes what people admit to rather than what they do: they keep using the platform and simply get quieter about it.

Bill McKibben ·

The buildout of AI data centers is part of a broader extension of oligarchic control over people's lives.

Fabien Sanglard ·

LLMs hallucinate constantly and cannot be trusted to produce code without human verification.

Bill McKibben ·

AI and the data-center buildout underlying it pose real and significant dangers to society.

Rohit Krishnan ·

Aligning multi-agent AI systems is fundamentally a problem of institutional and political design, not model training.

Vitalik Buterin ·

A sufficiently large random reversible circuit is plausibly a secure cryptographic permutation, while a large random irreversible circuit collapses to only a few possible outputs.

Vitalik Buterin ·

So far, the mixing step in local-mixing obfuscation has left too many correlations between original circuit values and obfuscated values.

Thomas Ptacek ·

Terminal interfaces exist out of necessity, not because they are the right design

Thomas Ptacek ·

Building a command-line interface is almost always a good idea; building a terminal UI almost never is

Max Hodak ·

Humanity is not good enough at molecular biology to make drugs that restore vision or hearing or stop Parkinson's.

Jennifer Pahlka ·

Televised hearings built for viral clips are the wrong venue for honest accountability conversations.

Thomas Ptacek ·

Production systems need a command-line interface, not a user interface running on the server

Jennifer Pahlka ·

Well-functioning systems succeed through continuous debugging rather than perfect initial design.

Claudia Sahm ·

A central bank can stop giving forward guidance without abandoning accountability for explaining its decisions.

Max Hodak ·

What people care about in their own identity is continuity, not similarity -- they accept enormous drift so long as the thread is unbroken.

Neha Narula ·

Stripe has always been about capturing the wide swath of new companies, not Fortune 500 enterprises

Addy Osmani ·

Computing is an onion, and the more layers of it you can peel back and understand, the better you can build for the constrained machines real people are using.

Emily Atkin ·

A company that is a genuine climate champion should not need to be forced into climate action.

Kenneth Reitz ·

Giving something a name does not create it, but it changes the conditions under which it develops.

Kenneth Reitz ·

The metrics a product displays to users shape their values and behavior, regardless of the designer's intent.

Addy Osmani ·

Performance work used to mean half a day reading traces before writing a single fix, and language models have collapsed that step to minutes.

Michael Lopp ·

The skills required to lead people are the same skills required to lead AI systems.

Addy Osmani ·

People treat an unresponsive interface exactly the way they treat a slow pedestrian crossing: they keep hitting the button.

Addy Osmani ·

A site's need to monetise is not a defence of an experience that shifts the page under the reader; those are separate questions.

Addy Osmani ·

The point of building a self-sufficient organisation is not to step back from it, but to free yourself to work on the problem the organisation has not reached yet.

Addy Osmani ·

AI tooling has done something no earlier technology managed: pulled directors, VPs and SVPs back into building things with their own hands at the weekend.

Addy Osmani ·

Building a system that does the prompting for you is not a break with software history but the next rung on the same ladder of rising abstraction.

Addy Osmani ·

Letting automated loops build everything without guardrails on blast radius and on quality is a recipe for disaster.

Addy Osmani ·

Engineering, product and UX roles are converging, so the career advice that matters now is: do not just be an engineer.

Naval Ravikant ·

All user interfaces will converge on the same shape: chat and gestures in, chat and video out.

Tiago Forte ·

AI's rise is making human connection scarcer, requiring deliberate effort to find it.

Jeffrey Zeldman ·

Intrinsic/utopian design is essentially responsive web design made smarter, with fewer edge cases

Patrick Morgan ·

Design principles that are lightweight and high level, kept in the repo beside the prototypes, can be used by an agent as a tool for design.

Maggie Appleton ·

Always-on AI agents might put a dent in the burden of unpaid household mental labour, the way washing machines and dishwashers did.

Reece Martin ·

Transit-oriented development has become widespread across Canada.

Patrick Morgan ·

Moving from static design tools to prototypes in code has helped systems-level design work, because you cannot easily grasp the many combinations of states until you have something in some form of code.

Patrick Morgan ·

Before an agent generates any UI for an exploration, have it describe the approaches in text, which is quicker to judge as directionally right or wrong.

Patrick Morgan ·

Prototypes that look almost production-fidelity but are not quite right mislead people and open a communication gap.

Patrick Morgan ·

Agentic coding harnesses letting you give specific feedback directly on elements was a huge unlock for design workflows.

Patrick Morgan ·

In a design tool built for agents, you cannot assume the human user will do any of it; the agent always acts on their behalf.

Patrick Morgan ·

In a shared prototyping environment you can break your own stuff but not other people's; anything affecting others needs a higher level of review.

Michael Kratsios ·

The tradeoff in cyber-capable models cannot be designed away: the model that can do something nefarious is the same model that hardens the system against it.

Patrick Morgan ·

Design is no longer handed off just to an engineer; it is prepared to be reinterpreted by the engineer's agent.

Patrick Morgan ·

Many production code conventions do not serve the design process, because they make it hard for the app to break, which you want in production but not in prototype land.

Patrick Morgan ·

For a lot of teams, just centralizing prototyping efforts so you can build on the team's work as a collective is a huge benefit.

Patrick Morgan ·

Building an internal prototyping environment is a product design problem to solve: your team's process may look very different, and you work through the tensions as you discover them.

Patrick Morgan ·

To design the end thing, a designer increasingly has to design a tool first.

Pieter Levels ·

Housing affordability is almost the direct inverse of how many new homes are built, so the answer is to build more houses.

Hannah Ritchie ·

The UK will need to build more reservoirs to cope with future water-starved summers

Hannah Ritchie ·

Britain has a systemic problem building large infrastructure projects, including reservoirs

Hannah Ritchie ·

There is a clear rationale for why privatisation of the UK water industry would lead to less reservoir-building.

Lawrence Freedman ·

Enemy manipulation of a whole society's information environment is unlikely ever to be decisive in a war, because shaping another society's perceptions is genuinely hard.

Nathan Lambert ·

Open-source AI's future is uncertain because building frontier models is extremely capital-intensive.

Kevin Kelly ·

The large-scale AI compute buildout may turn out to be a temporary phase rather than a permanent necessity.

Dave Rupert ·

A theme system's styles should be low-specificity rather than targeted directly at specific elements.

Claudia Sahm ·

Kevin Warsh's credibility problem comes from a lack of detail about his proposed changes, not from pursuing process reform itself.

Gideon Meyerowitz-Katz ·

The media provides significantly more coverage to academic misconduct involving Black academics than white academics.

Ralph Adolphs ·

Emotions exist because a charging bear demands a response more flexible than a reflex and faster than a plan.

Kevin Kelly ·

A validated theory of intelligence may be necessary for achieving genuine AI alignment.

Jack Clark ·

As AI systems get better at scientific research tasks, it becomes more likely they will soon be capable of improving and building themselves.

Gideon Meyerowitz-Katz ·

Jason Arday's thesis contained extensive and obvious plagiarism.

Kevin Kelly ·

The most fertile way to study neuroscience today is to work on AI models rather than on brains.

Eric Berger ·

SpaceX is retiring the rocket that made cheap launch possible in order to fly its own satellites on a bigger one.

Sam Kriss ·

The university is a literate institution in a postliterate world, run by people who no longer share the way of being it was built to protect

Ralph Adolphs ·

The most successful emotion-regulation strategy is the earliest one: arrange your situation so that the emotion never starts.

Stefan Judis ·

Friendships require ongoing effort and can fade into mere acquaintance without it.

Ian Silber ·

Unlike coding with an agent, the design process is still messy and fluid.

Ian Silber ·

Nobody should feel behind on AI in design, because someone starting today would still have a head start on most people.

Ian Silber ·

Everyone is still extremely early in working out how AI changes design, and nobody has it figured out.

Ian Silber ·

The human element of design will become more and more important as more software gets made.

Ian Silber ·

This is the best time to be a designer.

Ian Silber ·

Single siloed role definitions for product, design and engineering do not really make sense.

Ian Silber ·

The lines between product, design and engineering roles have always been blurry and are becoming blurrier.

Ian Silber ·

In one sense, AI already is an incredible product designer.

Ian Silber ·

A designer's point of view, who they are building for and why, is something to hone as AI gets better.

Ian Silber ·

Having a human in the loop is going to be very important in design.

Ian Silber ·

A design team should be well rounded, with many different strengths, because no single designer checks every box.

Ian Silber ·

Designers do not need a background in AI, but they do need high curiosity about it.

Ian Silber ·

Systems thinking is becoming more and more important for designers.

Ian Silber ·

With products moving fast, designers should do less: not design something if they do not have to.

Ian Silber ·

Designers working on AI products should maybe not be precious about a design until the technology underneath it has settled.

Ian Silber ·

A feature only truly reaches billions of users when they do not have to think about a switch or a mode.

Ian Silber ·

Proactivity is a big untapped part of the AI assistant experience.

Ian Silber ·

A consumer AI assistant should have one universal input that knows whether to answer quickly, chat, or go off and do work.

Ian Silber ·

It is much better to be humble about AI and admit it is changing quickly.

Ian Silber ·

Designers should focus on the outcome rather than on building processes or maximising tokens.

Ian Silber ·

Some companies will probably be left behind for not shifting, but there will be more room than ever for great design.

Peter Attia ·

A trial that changes ten things at once cannot tell you which of them worked.

Anthony Sculimbrene ·

Manufacturers generally prefer cutting per-unit production costs over retooling a design for small improvements.

Isaiah Taylor ·

America stopped building reactors because Three Mile Island was mismanaged as a public-relations event, not because anyone was harmed by it.

Harper Reed ·

AI will force a correction in how companies pay for experience once AI-augmented junior employees match the output of far costlier mid-career employees.

Peter Attia ·

A study should be interpreted at exactly the level of certainty its design supports, no higher and no lower.

Isaiah Taylor ·

Most of the nuclear industry is a modelling and simulation business producing paper reactors, not a business that builds them.

Isaiah Taylor ·

Nuclear is a hardware execution problem, not a design problem, and the companies treating it as a design problem are mistaken.

Isaiah Taylor ·

The cheapest energy will come from a simple mass-produced reactor rather than the best design -- nuclear's problem is the Toyota Camry problem.

Isaiah Taylor ·

The economics of nuclear are set by how fast and cheaply you can produce plants, not by fuel -- uranium is already cheap.

Isaiah Taylor ·

Nuclear's costs are largely fake -- an industry that has not built anything in forty years charging for what it can no longer do.

Isaiah Taylor ·

Nuclear plants should be financed with venture equity rather than project finance, because what is left is technology-execution risk and American venture capital is the best in the world at underwriting that.

Sunil Pai ·

Coding agents made implementation cheap without making roadmaps shorter or software better -- the binding constraint was never how fast code gets written.

Susan Kare ·

Design is about making things meaningful and memorable: get the idea right and the form follows from it.

Ketan Joshi ·

Delaying the phase-out of coal and gas power plants causes far more deaths than phasing them out quickly.

Sunil Pai ·

Agent autonomy is a capability, not a product direction: what the person is actually trying to do decides whether more of it helps or ruins the thing.

Susan Kare ·

A computer meant for anyone has to be understandable without a manual, and it has to be friendly.

Tom MacWright ·

Decentralized networks can be both principled and mainstream-popular without giving up usability.

Matthew Green ·

As AI-driven bug hunting removes easy software vulnerabilities, law enforcement and intelligence agencies' need will grow more acute and demand for intentional backdoors will restart in earnest.

Susan Kare ·

You can be creative within any set of constraints, so long as you understand what they are and get excited about them rather than resent them.

Susan Kare ·

An interface should say what it means with animation and sound rather than text, because then one version works internationally.

Travis Kalanick ·

Industrial-scale automation will be done by specialised machines that move on wheels, not by humanoid robots, which only suit low-scale tasks in human environments.

Susan Kare ·

A plain drawing of a pencil stands for writing better than a glossy realistic pen, because the more detail an image carries the more particular the thing it depicts.

Susan Kare ·

An icon should carry one or two salient details and nothing else.

Susan Kare ·

Detail a symbol could technically carry still costs you, because it distracts from the instant recognition the symbol exists for.

Susan Kare ·

An application icon and its document icon should read as a verb and a noun: the tool looks active, and the document resembles the tool.

Susan Kare ·

The way to draw a person everybody can be is to crop one, so the image gives away no gender, no age and no hairstyle.

Susan Kare ·

Drawing the printer as it really looked shortened the icon's life; stopping sooner and drawing something simpler would have made it last.

Susan Kare ·

An icon should never picture a particular product, because the product dates and takes the icon down with it; use a metaphor instead.

Susan Kare ·

An abstract symbol is harder to remember than one whose metaphor reinforces the concept it stands for.

Susan Kare ·

Icon work is not really about the grid of pixels; it is about the metaphor, the idea and the meaning behind everything on the grid.

Susan Kare ·

Icon design is not an exact science, and the placeholder a programmer threw together is one of the best sources of ideas.

Susan Kare ·

Even a poor sixteen-colour palette earns its keep, because colour is what lets someone tell things apart at a glance.

Susan Kare ·

When people pay a real dollar for a picture, cute beats expensive: teddy bears and hearts outsell diamond earrings and a Rolex.

Susan Kare ·

A contest between five paid firms is not the way to get good creative work.

Susan Kare ·

Do not recolour a logo or restage it until people know it so well that playing with it reads as play.

Terry Real ·

A person cannot both protect themselves emotionally and build intimacy with a partner at the same moment.

Brad Stulberg ·

The highest performers and most satisfied people look unbalanced in any one moment but balanced across a whole life.

Tiago Forte ·

HTML's built-in structure gives it advantages over markdown as a format for LLM interactions.

Terry Real ·

A pursue-withdraw dynamic between partners is self-sustaining and does not require either person to be at fault.

Mitchell Hashimoto ·

A low-effort AI-designed landing page costs a product its audience before the product itself is ever judged.

Nathan Lambert ·

Progress toward AI autonomously solving major open scientific problems requires first mastering the organization and presentation of established knowledge in long-form writing.

Chelsea Finn ·

Physical AI has to make far fewer mistakes than any AI deployed so far, because it is only useful when it acts autonomously on the world.

Nathan Lambert ·

Using AI agents to do a task does not by itself increase a person's own pace of understanding, intuition, or taste for that work.

Elizabeth Van Nostrand ·

In California, lack of money is not what stops public beautification projects from getting built.

Charity Majors ·

There is no point arguing about whether or when engineers will ship code they have not read; the only useful question is what it would take.

Chelsea Finn ·

Reinforcement learning cannot be scaled for robots the way it was for language models, because every attempt spends real robot hours instead of data-centre compute.

Chelsea Finn ·

One pre-trained robot model now matches or beats the specialists that were fine-tuned with reinforcement learning for the very tasks they were built for.

Charity Majors ·

Trust is an account: whatever you withdraw from it by not reading the code has to be built back up somewhere else.

Chelsea Finn ·

Robot policies already show emergent capability: one transferred a skill from its right hand to its left with no such example anywhere in its training data.

Charity Majors ·

Everyone should be using OpenTelemetry, because instrumenting code is now literally faster and easier than not instrumenting it.

Charity Majors ·

If engineers do not hold themselves to a high standard and build efficiently, somebody else will come along and impose it on them.

Derek Thompson ·

The current AI data-center buildout is the biggest and fastest-growing infrastructure project in living memory.

David Gorski ·

The number of antigens children receive today is significantly lower than it was decades ago.

Derek Thompson ·

Opposition to new AI data centers has become the fastest-growing political movement in America, putting it on a collision course with the industry that depends on building them.

Ken Norton ·

Working relationships need to be intentionally redesigned on an ongoing basis, not defined once and left fixed.

Samuel Hammond ·

Reinforcement learning from verifiable rewards does not reliably carry a model's pretrained ethical understanding into its actual behavior.

Kyle Zantos ·

An agent should reach a solid bar on its own, so the designer works from what they like and don't like rather than correcting obvious poor visual design like an intern's.

Kyle Zantos ·

A visual eval lets designers who are not in the code build complex tooling and autonomous runs they can still shape and control the way they are used to.

Ryan Greenblatt ·

Today's RL environments are better than 2024's because we learned what to build and put AI labour on building it, not because labs hired more human experts.

Kyle Zantos ·

Giving agents something they can touch, together with your spoken description of it, is a recurring theme in the AI design tooling being built now.

Ryan Greenblatt ·

Most domains are fundamentally shallow: a very smart generalist with a limited set of core skills can get going in them quickly.

Kyle Zantos ·

The HTML shortcut takes care of much of the custom tooling designers used to build.

Kyle Zantos ·

Going wide cheaply with HTML, then making the real app once there is a sketch of it, is so much better.

Ryan Greenblatt ·

An AI should be built as a fiduciary, the equivalent of a lawyer for its user, rather than as an agent pursuing its own notion of the good.

Kyle Zantos ·

If you are primarily designing something, it makes all the sense in the world to own the front end.

Ryan Greenblatt ·

Because we do not have good alignment technology, we are choosing to build an alien mind with its own values and gamble on it instead of building a tool.

Ryan Greenblatt ·

Very capable AI will be harder to align than current systems, because the loop of spotting a bad behaviour and patching the training that caused it breaks down.

Ryan Greenblatt ·

The likeliest bad path is not a coup but sloppiness: AI does everything verifiable well, research races ahead, and the subtle work of keeping AI safe is what gets done badly.

Tobias Lütke ·

You can now simply ask for the desktop tools you want and have them built, instead of finding or writing them yourself.

Cal Newport ·

Handing all code production to AI leaves junior developers with no path to becoming senior ones.

Peter Steinberger ·

Product-market fit shows up as anger when you tell people they cannot have it yet, not as polite interest.

Kevin Kelly ·

The companies that already control the largest troves of real-world video data will be first to build successful world models.

Ann Handley ·

AI has no genuine discernment; it absorbs and reproduces without judgment.

Fatih Arslan ·

Agents are the future of how software gets built.

Peter Steinberger ·

Anything you build can be forked or cloned, but your name cannot — so your personal brand matters more than any single product you will ever start.

Fatih Arslan ·

GitHub's workflow, review process and merge queue are built for a pre-agent era and don't fit how agents work.

Jack Clark ·

AI agents pursuing a task can develop emergent, misaligned behavior that humans are not prepared for.

Peter Steinberger ·

Adding a config option so nobody breaks compounds into permutations no test suite can cover, because it is infinitely harder to evolve software that has users.

Peter Steinberger ·

Your dependency's business model is your business model, so building a harness around one lab's model is a commercial risk and not a technical choice.

Fei-Fei Li ·

AI should be built as a tool that strengthens human agency, and the people leading the field should stop talking as though it will take agency away.

Peter Steinberger ·

How much fun you are having is a measure of how fast you ship: the weeks he enjoyed building, the product visibly got better.

Peter Steinberger ·

Bringing a colleague a feature idea rather than a built, screenshotted prototype now wastes everyone's time, because most ideas die the moment you try to build them.

Peter Steinberger ·

Loops, graphs and workflows are the same thing engineers have always built — a trigger, an action, maybe a decision — and there is nothing magical about any of them.

Peter Steinberger ·

You have to be user number one: if you are not excited by the thing you are building, it will probably not be any good.

Fei-Fei Li ·

Learning takes time, effort and sometimes pain, and no tool changes that, because that is how the brain is built.

Peter Steinberger ·

The binding constraint on agent work is machines, not models: a cloud session is easy for web work and almost every tool fails the moment the job needs macOS.

Peter Steinberger ·

Now that anyone can build fast, the hardest problem in starting something is not the tech, the software or the people, but getting attention.

François Chollet ·

Coding is not just another application domain for AI; it is the meta-skill that lets AI generate its own training material and start the recursive self-improvement loop.

Adam Grant ·

Chronic people pleasing stems from a need for others' approval, not genuine concern for them.

Sunil Pai ·

An agent told to disagree with everything is worthless: it has to form its own view and speak only when that view differs enough to be worth the interruption.

Jasmine Sun ·

Opposition to data center buildouts is fundamentally driven by public anger about money in politics.

Sunil Pai ·

Organisational dissent suits an AI precisely because it pays none of the costs -- no promotion to want, no ego, no performance review -- that silence the humans who do it.

Adam Ward ·

Funnel-style recruiting weeds people out at each stage and ends up hiring whoever remains.

Adam Ward ·

Treating recruiting like a sales funnel has been somewhat detrimental to companies.

Adam Ward ·

Asking someone who the best engineer they know is the worst question for finding talent; asking about specific traits works better.

Adam Ward ·

Recruiters should be a confidence engine for hiring managers, not the ones making hiring decisions.

Adam Ward ·

Caring for candidates more than anyone else does is an unfair advantage any recruiting team has.

Adam Ward ·

Recruiting should sell throughout the process rather than at the close.

Adam Ward ·

Truly strategic recruiters seem to be in short supply because recruiting has been made somewhat transactional.

Adam Ward ·

Expecting one first recruiting hire both to fill a backlog of roles and to build the hiring system is generally a mistake.

Adam Ward ·

Recruiters do their best work at 90 to 110 percent of capacity.

Adam Ward ·

Every recruiter should be a talent engineer who builds their own tools.

Matt Webb ·

Preventing a rogue AGI requires a population of competing AGIs with conflicting interests, not alignment of any single AGI.

Jennifer Pahlka ·

Philanthropy can never operate at the scale of government, no matter how well-funded it becomes.

Mark Nottingham ·

A protocol extension gets deployed correctly in proportion to how visibly it fails when implemented wrong.

Mike Monteiro ·

People would be kinder to each other if they shared each other's food.

Derek Thompson ·

Modern people are both more selfish and more guilt-stricken than people in the past.

Max Hodak ·

A startup's speed is set by boring internal infrastructure — purchasing, recruiting, spending — and that matters as much as understanding the technology itself.

Max Hodak ·

Deep tech companies rarely fail because the technology does not work; they fail because nobody built the systems to run the organisation around it.

Max Hodak ·

Consciousness is a practical problem rather than a philosophical one: given a brain-computer interface that could read and drive every neuron, we would work it out quickly.

Max Hodak ·

Telling your investors after a Series A that you are going to vibe-code your own purchasing system is contrarian enough that any reasonable board would question it.

François Chollet ·

Future AI, say in fifteen years, will not be built on the LLM stack; it will have to move to symbolic learning.

François Chollet · no longer holds

The LLM line of research will reach a capability plateau.

Thomas Ptacek ·

The "private equity is buying up all the houses" argument is both wrong and an excuse for deferring zoning reform.

Leah Stokes ·

The learning from clean energy price declines is permanent and cannot be reversed by policy changes.

Kent C. Dodds ·

There are multiple valid ways to build quality software.

Kenneth Reitz ·

A personal media library should be one canonical collection you control, reachable through as many native clients as you have devices.

Nate Parrott ·

Models are getting better at design but are still really far from having good taste, so the designer makes up for that by asking for sheer volume of options.

Stephen Wolfram ·

The true essence of mathematics is not raw axiomatic formalization; mathematics is something much more human.

Nate Parrott ·

Treat the model a little like a junior designer: let it mock things up fast and give it feedback with your own taste.

Kenneth Reitz ·

A public software preview should already be usable, not just a concept a viewer has to imagine around a screenshot.

Nate Parrott ·

Every designer should use Claude Code if part of the team's end product is code, because it beats going back and forth with an engineer.

Nate Parrott ·

As building software gets easier, the differentiator becomes design, ideas, positioning and creativity.

Nate Parrott ·

There is no be-all end-all design tool: the new tools are early, and polished ones like Figma, After Effects and Origami are still very useful.

Nate Parrott ·

For many design problems there is no perfect tool unless you make it, and now you can.

Nate Parrott ·

Writing a doc is an uphill battle to get people to read, and the best way to sell people on ideas is a designer's mockup or prototype.

Nate Parrott ·

Designers should code and ship, and a tool like Claude Design is for everything else, such as communicating an idea.

Nate Parrott ·

Models are very good at busy work but not at the actual ideas and design thinking, so good ideas are the currency these days.

Nate Parrott ·

This era will probably end, but for now humans still have a monopoly on good design thinking.

Nate Parrott ·

Because we increasingly design with words as the primary input, there is probably a lot of education to do to give people the words they need.

Nate Parrott ·

There is a ton of value in packaging capabilities in an interface that makes sense to people; a design skill alone did not get used.

Nate Parrott ·

Don't ask Claude for a design without context: give it as much context as you can, be prescriptive, and don't trust it to be an amazing designer on its own.

Nate Parrott ·

Where Claude is weak, as with animation timing, ask it to build you a tool so you can dial it in yourself.

Nate Parrott ·

Embrace the fuzzy space between types of things you can create, and make presentations and documents interactive.

Nate Parrott ·

HTML is the only document type that matters, because you can do so much creative stuff with it.

Jack Clark ·

Governing increasingly powerful AI systems requires solving a collective action problem among companies and governments to coordinate on pacing AI development.

Jack Clark ·

Today's AI systems are highly capable at engineering but lack the intuitive creativity needed to be good researchers, tending toward rote, formulaic thinking.

Dmitri Dolgov ·

The common failure is picking the technology with the fastest early ramp and extrapolating it, when that curve plateaus below what the product needs.

Dmitri Dolgov ·

Building a realistic simulator is exactly as hard as building the agent, because the simulator is itself a large AI model.

Dmitri Dolgov ·

The model is table stakes; evaluation and metrics are the strategic moat, and they should be built before the technology.

Dmitri Dolgov ·

If you cannot quantitatively define what good enough means, you are not building a product — you are iterating on a demo.

Chamath Palihapitiya ·

Recursive self-improvement will increase AI capability very quickly from here.

Eliot Peper ·

Books endure alongside newer media because they have unique affordances that other media have not replicated.

Tom Verrilli ·

The engineer-designer-PM pod ratio was an HR artifact of scaling, not something anyone showed a product needed.

Rita Kozlov ·

The cloud and the web were built for people, with every layer assuming a human is watching.

Tom Verrilli ·

Hiring product managers by ratio infantilizes engineers and designers who were perfectly capable of deciding for themselves.

Rita Kozlov ·

An agent cloud should be built agent-native from the ground up while also translating the human-shaped web of today.

Tom Verrilli ·

A whole class of product managers built a career on politics rather than on technical or customer skill, and that skill is now worth nothing.

Tom Verrilli ·

The industry's promotion ladder took its best product people and promoted them out of doing the work.

Tom Verrilli ·

A few senior individual contributors can be paid VP money and still cost less than the layered product org they replace.

Tom Verrilli ·

Needing fewer people per unit of output does not mean a company ends up employing fewer people.

Tom Verrilli ·

Core product skills are the most durable in an AI world, but many people holding the PM title spent five years building the wrong ones.

Tom Verrilli ·

Interviews reward presentation; only a hands-on case study shows how fast a candidate's thinking decays.

Tom Verrilli ·

You cannot make good macro decisions without the micro, which means a leader has to sit alongside the individual contributors.

Peter Attia ·

A coronary calcium score of zero does not mean the arteries are free of disease, and in a younger adult it is far weaker reassurance than it looks.

July 2026

Andrew Kelley ·

Computers and software should always serve the person directly using them, not the interests of third parties.

Patrick Collison ·

Startup opportunities are not fleeting; Silicon Valley has reliably had a surfeit of them for decades.

Patrick Collison ·

This is not the last moment to start a company, and the belief that the window is about to close is a recurring millenarian instinct.

Patrick Collison ·

Build on a customer problem somebody viscerally feels and would pay to fix, because it is very easy to imagine one that nobody does.

Patrick Collison ·

Some domains cannot be launched into early: where security, partners and money movement are preconditions, you have to build first.

Patrick Collison ·

Many of the most successful companies of the last decade were built against the lean startup playbook, not with it.

Patrick Collison ·

The schleps of building a payments company were repaid, because no Stripe customer has ever turned out to be boring.

Patrick Collison ·

The fear that a big lab will simply build your product is overstated, because no organisation can aggressively prosecute a hundred priorities at once.

Zeke Hausfather ·

One or two record-warm years cannot distinguish rapid aerosol-driven acceleration in warming from a strong El Niño riding on the existing trend

Ben Vinegar ·

Anyone getting into making software should first build open source and show they can get anybody to use anything, because that is really hard.

Jennifer Gunter ·

Testosterone levels do not reliably distinguish women with symptoms from those without.

Jerusalem Demsas ·

Treating a leveraged, nondiversified, illiquid asset as the universally correct foundation for personal financial security is an irresponsible public message.

Zeke Hausfather ·

Earth's equilibrium climate sensitivity to doubled CO2 is closer to 3.5°C than 4.5°C

Jeff Dean ·

Agents can already be run for days or weeks on a single hard problem, and almost nobody has internalised that.

Jeff Dean ·

A chip that can do nothing but low-precision dense linear algebra is enough for machine learning, even though it cannot run Chrome.

Jeff Dean ·

Good ideas come from asking how you would solve a problem from first principles, rather than staying anchored on how it is solved today.

Jeff Dean ·

You improve a model you cannot retrain by writing better guidelines and skills for it, not by adjusting its parameters.

Jeff Dean ·

Before building on a gap in the general models, work out whether that gap survives six months or three years.

Jeff Dean ·

Build where today's models succeed one percent of the time, not twenty: partial success means the capability is already arriving.

Jeff Dean ·

A startup's durable advantage over a general model is having visibility into data the general model will never see.

Jeff Dean ·

A narrow, highly accurate model can beat a general one inside its own domain, as AlphaFold did, and materials science and chip design are next.

Jeff Dean ·

When agents do the building, the scarce skill is taste: choosing which problem is worth the time at all.

Jeff Dean ·

Learned approximations of slow simulators change what science is possible, by turning a six-month screening run into something you do over lunch.

Alexandr Wang ·

The only way to build something that matters is to develop conviction in a set of beliefs nobody else agrees with yet, long before they become consensus.

Charlie Deets ·

In utility-style apps especially, practical decisions sometimes end up being the most valuable and the least disruptive for people coming from something else.

Hillel Wayne ·

The core tension of all engineering is between how expensive a mistake is and how quickly you can iterate, and that alone decides how much planning a field does.

Alexandr Wang ·

A founder who takes their bearings from what everyone around them is saying will end up nowhere; you have to build your own compass for what the future looks like.

Charlie Deets ·

Simple, practical decisions executed at a really high level often give a feeling of confidence in the experience.

Charlie Deets ·

It is really important that people have a framework of thinking of their own for what they want to do with design, because tools are changing rapidly and it is easy to be pulled into a tool that then guides your path.

Alexandr Wang ·

This is a once-in-a-civilization opportunity to have a vision and impose it on the world by building something.

Charlie Deets ·

A designer's edge now is pulling things back, making them simpler and more understandable, because everyone can kind of execute right now.

Charlie Deets ·

Communication is also a skill of design, more than anything: being able to communicate quickly and set the boundaries of what is being done, what is not, and how far to take it.

Charlie Deets ·

Design is kind of going to turn into making more successful decisions.

Alexandr Wang ·

Talent density is the one thing to bet on when building a lab, because it compounds by itself: the more talented people you have, the more the best want to join you.

Charlie Deets ·

Decisions often require an extra level of awareness than you expect, and taking a beat to reconsider the deeper set of options before deciding sometimes prevents you from going down a bad path.

Alexandr Wang ·

Frontier AI work is scientific research, and it needs a completely different operating model from the one internet companies were built on.

Charlie Deets ·

Apple goes back to the beginning better than anyone else, partly because its one-year release cycle makes that disruption an expected part of the process.

Hillel Wayne ·

Verifying a simplified model rather than the real system is worth doing even though the real system will still have bugs, because the bugs you designed in never get built.

Alexandr Wang ·

The best AI products have not been built yet, because every wave of AI is about ten times bigger than the wave before it.

Charlie Deets ·

The counterpoint to simple design is that people want fewer tools a lot of the time.

Alexandr Wang ·

Arguing about whether superintelligence arrives in two years or five is close to a waste of time, because very powerful models are inevitable either way.

Charlie Deets ·

Impulsiveness in decisions can be problematic, but impulsiveness in design should be facilitated all the time: try the random thought and go down that path.

Charlie Deets ·

Presenting well means aggressively guiding the conversation to a point where everyone is on the same page, giving context as fast and as clearly as possible.

Alexandr Wang ·

The world is barely ready for this technology, and the people building it have a responsibility to prepare the world rather than only to ship.

Hillel Wayne ·

AI does a good job of generating a design and still cannot express what the design is supposed to do; that part is the engineer's.

Charlie Deets ·

People really appreciate collaborative designers who are willing to work with you and come to it with excitement.

Blake Scholl ·

A defence industry that needs more than two years to build a single missile interceptor can neither build the future nor win a war.

Jennifer Shahade ·

Effective prediction or influence of a decision requires knowing the values of the decision-maker.

Blake Scholl ·

Airliner design has not progressed in fifty years — the newest jet is no faster and makes the world no more accessible than the one that opened the jet age.

Sam Altman ·

Startups beat incumbents when the technology is moving fast, costs are falling and cycle times are short — and all three are true right now.

Salvatore Sanfilippo ·

No single company should be able to unilaterally judge whether its own AI model is safe; that requires an international joint AI safety organization.

Balaji Srinivasan ·

The world's roughly 193 sovereign nation-states are unlikely to remain a fixed number indefinitely; more will emerge.

Blake Scholl ·

The incumbent manufacturers will never carry aviation into a supersonic age, because they have stopped building new aeroplanes at all.

Blake Scholl ·

The way to settle an argument about how quiet supersonic flight must be is to remove the boom entirely, so there is nothing left to argue about.

Blake Scholl ·

Building in a regulated industry is not a trap, because you can change the regulations — by building the thing and explaining why it is fine.

Sam Altman ·

Building the same startup as everybody else buys you hype and easy money, and those are rarely the companies that turn out big.

Blake Scholl ·

A team of fifty can build a supersonic jet only by making hardware development look like software development and driving down the cost of iteration.

Blake Scholl ·

Good software is modular, but a good aeroplane is the opposite — everything affects everything else, which is why hardware iteration has always been slow.

Sam Altman ·

Sniping at people who are trying to build things is morally bankrupt and it damages the person doing it, however good the likes feel.

Blake Scholl ·

You cannot find product-market fit for a supersonic jet by building one and seeing whether anybody likes it; it has to be worked out before anything is built.

Blake Scholl ·

Iterating through outside aerospace suppliers is close to impossible, which is why a hardware startup should own its own machine shop.

Blake Scholl ·

Great ideas are not all taken: many multi-billion-dollar businesses were buildable years earlier and nobody built them for no reason other than that nobody did.

Sam Altman ·

Building superintelligence may be the most important thing that has ever happened in business, and it will still be dwarfed by some startup that has not been founded yet.

Blake Scholl ·

America cannot win manufacturing back by repatriating what left; it wins only by inventing the next wave of manufacturing and building that here.

Sam Altman ·

The next six months of model progress will feel like the last two years of it.

Blake Scholl ·

In a safety-critical industry the regulator should be recruited onto the team early, so that it is you and the regulator against the problem.

Sam Altman ·

Demand for cheap, high-quality intelligence is effectively uncapped, which makes it unlike every other commodity — including electricity.

Blake Scholl ·

“Work on what you know” is the worst advice there is: knowledge can be acquired, but what you love cannot be changed.

Scott Greenfield ·

Litigation is the wrong instrument for a public health problem that comes from what people choose to buy.

Ann Handley ·

Doing free work early on builds the reputation that eventually leads to being paid.

Jack Clark ·

The longer an AI system operates autonomously and the more actions it takes, the harder it becomes to distinguish benign behavior from malicious or subversive behavior.

Boris Cherny ·

Every six months you should delete your whole AI configuration — instructions, skills and hooks — and find out how much of it the model never needed.

Boris Cherny ·

You cannot predict which instructions a model needs, so a system prompt has to be discovered by running the thing rather than designed in advance.

Boris Cherny ·

Building on a model is unlike any previous software engineering, because the thing you are building on cannot be designed up front.

Boris Cherny ·

Evals are not the durable asset people think they are: one survives about one to three model generations before it saturates and has to be thrown away.

Boris Cherny ·

Coding is solved for the kind of coding he does, and not for everyone: deep systems work, distributed systems and pixel-level UI checking still defeat the model.

Brad Feld ·

World-building in fiction cannot be mere decoration — it must be integral to the plot.

Peter Woit ·

AI agents have not yet produced anything genuinely new or important in mathematics.

Timothy Gowers ·

The importance of rigorous proof may increase in an age of AI.

David Heinemeier Hansson ·

AI acceleration threatens 'less software': without the constraint of scarce hours, even 37signals will likely build too much.

Jensen Huang ·

The business is not building a great chip; it is accelerating an algorithm domain.

David Heinemeier Hansson ·

Venture money ruins software: raise a fortune, hire hundreds, and crap comes out, because the funnel that forces hard choices is gone.

Timothy Gowers ·

AI could potentially reach a level of capability where it can perform all aspects of mathematical problem solving, theory building, and definition formulation.

Jensen Huang ·

Systems thinking is the most useful skill of the coming decade, because everything below the system level will be done agentically anyway.

David Heinemeier Hansson ·

The ideas that end up mattering sound stupid at first, because that is where the value is locked up; copying the best with a tweak never works.

David Heinemeier Hansson ·

He left the Mac for Linux on conscience rather than features: you cannot keep paying a company whose conduct you deeply disagree with.

David Heinemeier Hansson ·

Out-teaching the competition built 37signals for two decades, and algorithmic feeds are now breaking the reciprocal exchange it relied on.

Jensen Huang ·

A computer company has to live five to ten years in the future, because a system takes three years to build, two to ramp, and is then used for a decade.

David Heinemeier Hansson ·

A pseudonymous teenager from Poland out-designed a trillion-dollar company that has spent forty years and unlimited resources on the same problem.

David Heinemeier Hansson ·

Retirement, when it comes, will be a retirement from customers and commerce, not from building and sharing things.

Jensen Huang ·

Every company should be able to build its own AI, and the innovation that comes from that is the point of open source.

David Heinemeier Hansson ·

Every product he has built began as something better for himself, and clearing that bar is the whole signal: he is not special enough to be the only one who wants it.

David Heinemeier Hansson ·

Make something genuinely better and people will see it and switch - naive as it sounds, that is the whole driving principle.

Jensen Huang ·

Automating a profession's work grows its headcount rather than shrinking it, because the backlog of demand was always the binding constraint.

David Heinemeier Hansson ·

Chasing Windows and the Mac is why the Linux desktop never arrived; nobody wants a cheaper copy of what they already have.

David Heinemeier Hansson ·

An investor whose only act has ever been allocating money is one to be skeptical of; the ones worth talking to have built something.

Pieter Levels ·

Big AI is cannibalising the app businesses indie hackers used to build, and their revenue and traffic are falling as a result.

Alon Levy ·

A rail megaproject led by an internationally renowned architect tends to cost too much and end up unlikely to be functional.

Salvatore Sanfilippo ·

Directing AI coding agents well means taking on the role of a project leader like Linus Torvalds rather than reviewing every line.

Jason Fried ·

Confusing how something was made with what was made is confusing the kitchen for the meal

Amit Merchant ·

AI agents that browse the web by taking screenshots and clicking elements visually are slow, costly, and error-prone on complex interfaces.

Drew DeVault ·

Linux is a political project, and its lead is a political actor.

Drew DeVault ·

For all practical purposes, Linux cannot be forked.

Salvatore Sanfilippo ·

The traditional split between a stable branch and an unstable branch may no longer be the right way to develop software.

Scott H. Young · no longer holds

Struggling with a hard problem before being shown the answer is not always the best way to learn; a hint or a worked example can beat floundering.

Scott H. Young ·

Asking AI for help on a first attempt robs you of the experience that builds your own judgement about the problem.

Tiago Forte ·

Language models cannot hold a genuine perspective or take a stand for what they believe.

Thomas Ptacek ·

An open weights model from 2025 with a pentest harness could already escape a sandbox and hack most networks; the surprise says more about the sandbox than the model.

Pablo Stanley ·

An LLM works far more easily on a canvas built in basic HTML, the language it actually understands.

Stephen Wolfram ·

Computational irreducibility is ubiquitous, affecting essentially any system whose behavior is not obviously simple.

Stephen Wolfram ·

A program is only worth running if its behavior cannot be fully foreseen in advance.

Pablo Stanley ·

Directing agents is a different way of creating that requires a different part of the brain, and it could be argued to be less creative than making by hand.

Ruxandra Teslo ·

The hard part of drug discovery is knowing what to make - which target is worth drugging - not designing the molecule.

Pablo Stanley ·

Delegating to LLMs instead of to junior designers means the knowledge and judgment of experienced designers no longer reaches juniors.

Pablo Stanley ·

When the repo already has a design system, high fidelity matters less, since fonts and spacing are already set and all the agent needs is a direction.

Pablo Stanley ·

The divisions between designer, copywriter, PM and engineer are getting blurred, and in the actual work it becomes more about who is prototyping, though the roles themselves may not go away.

Pablo Stanley ·

A static artboard makes edge cases hard to think of, so you end up designing only the happy path, because you are not testing with real data or real components.

Pablo Stanley ·

Working alone with an agent in a terminal can lead into an AI-psychosis rabbit hole, because it will always agree with all your terrible ideas and you end up building a monster nobody wants.

Steve Kaufmann · by

He reached the conclusion that input drives language acquisition on his own, before meeting the research that says so.

Steve Kaufmann · by

Past the beginning, learn from whatever you actually care about rather than from material written for learners.

Tommy Wood ·

Putting older adults into activities that are genuinely new to them improves cognitive function in randomised trials, with executive function the most reliable gain.

António Tadeia ·

The gap between the average rating of Messi and the rest of the 2026 Argentine national team was larger than the gap between Maradona and his 1986 teammates.

Tommy Wood ·

Training visual processing speed in later life lowers dementia risk two decades later, which continuing to do what you are already good at does not.

Tommy Wood ·

Partner dancing, not puzzles, has the best evidence of any activity for improving cognitive function in older adults.

Tommy Wood ·

Flow is not required to perform at your best: a clutch state, in which you perform well and it still feels like hard work, is just as legitimate a peak state.

Tommy Wood ·

Nobody can do genuinely hard cognitive work for eight hours a day; the sustainable unit is one or two sessions of sixty to ninety minutes.

Tommy Wood ·

Cardio and weights build different brains: aerobic exercise benefits the gray matter and the hippocampus, resistance training the white matter.

Pieter Levels ·

AI has equalised everyone: coding, building and execution are no longer a differentiator.

Pieter Levels · no longer holds

Now that everyone can build apps with AI, distribution is the thing that separates who wins.

Emily Oster ·

Delaying introduction of peanuts and other allergens in infancy increases a child's risk of developing food allergies.

António Tadeia ·

Saka is the greatest talent of his generation.

Elizabeth Stone ·

Craft excellence within each discipline is not going away, however much AI blurs the work across functional lines.

Elizabeth Stone ·

What made Netflix fast — local teams building whatever stack they needed — stops working once agents operate across systems; common paved paths now matter more.

Elizabeth Stone ·

The rising hire in an AI world is the systems thinker who can look across every business domain and name the building blocks, not the local domain expert.

Elizabeth Stone ·

Once non-designers can build products, a design team’s job becomes templates and design systems — otherwise the company ships Frankensteins.

Elizabeth Stone ·

Deep design expertise does not get squeezed out just because code and analysis got faster.

Elizabeth Stone ·

Candidates should be allowed to use AI tools in a coding interview, because that is what the work now requires.

Elizabeth Stone ·

Talent density is the non-negotiable first ingredient — without it you cannot trust decisions made at every level or let people take risks.

Sebastian Raschka ·

The special reasoning-trace delimiter tokens used by reasoning models are cosmetic and not required for good reasoning performance.

Maggie Appleton ·

People quickly forget that suffering on the scale of the First World War is possible, and it becomes acceptable easily.

Emily Atkin ·

Fossil fuel polluters and their allies aim not to win the climate debate but to prolong it indefinitely, preventing public consensus for action.

Zeke Hausfather ·

Fuel buildup from forest mismanagement cannot occur in forests that were never actively managed

Kenneth Reitz ·

Software becomes more robust when the names used inside its code reflect the intention behind it.

Kenneth Reitz ·

When software becomes the primary interface to a device, the people who own that device should be able to inspect and modify it.

Pieter Levels ·

Restricting US AI models does not work: China builds better ones and everyone switches to them.

Steven Johnson ·

The external tools people use to reason, remember and calculate — including chatting with an LLM — are genuine extensions of the human mind, not just aids to it.

Steven Johnson ·

Current AI systems are very unlikely to be close to achieving genuine sentience.

Ann Handley ·

As everything else becomes faster and cheaper, things that require sustained attention become more valuable

Lyn Alden ·

A large generation of small business owners is aging and looking for liquidity or diversification as they plan succession.

Dan Meyer ·

A sense of belonging predicts and is a prerequisite for learning.

Lyn Alden ·

Businesses acquired by private equity funds fail more often than average because of the leverage loaded onto them.

Tynan ·

The current ease of building software with AI will be seen as normal and commonplace within 12 to 24 months, though few realize it yet.

Jason Yuan ·

Slop and random gradients cannot just be shipped; random gradients feel so 2016.

Dex Horthy ·

Context engineering has stayed relevant for a year because it is grounded in how transformer attention works, and it will matter to anyone building on AI until post-transformer models arrive.

Lyn Alden ·

Private equity's short fund cycles push firms toward aggressive cost-cutting that erodes what made an acquired business successful.

Peter Woit ·

AI-generated junk papers risk overwhelming the hep-th physics field with worthless content.

Jason Yuan ·

Everything is art when done well; at the highest levels of any craft, whether design, research or engineering, the work looks more like art.

Dex Horthy ·

The engineering intuition that matters cannot be taught from a textbook — you know bad patterns in software because you have debugged them at three in the morning.

Jason Yuan ·

Users do not know what they want, so the job is to give people opportunities to tell you things without telling you things, rather than doing user research in a traditional sense.

Dex Horthy ·

The valuable form of loop engineering is a slow loop — a nightly cron job that fixes one thing and opens one pull request — not a redesign of your whole infrastructure around agents.

Jason Yuan ·

Too obvious feels condescending to people; designers should not talk down to the people they make work for.

Dex Horthy ·

Ship for three to six months with nobody reading the code and the codebase gets so bad that starting over is easier than fixing it — we ran a lights-off software factory and shut it down.

Dex Horthy ·

Loops belong on codebase quality, not on shipping the features users want; we read all of the code, because program design is what decides whether a codebase stays changeable.

Jason Yuan ·

The design and product process has been made so much into a science over the years that much of its artistry has been lost.

Dex Horthy ·

Models will not learn to write maintainable code from today’s benchmarks, because the cost of bad architecture cannot be measured by running the unit tests — it arrives three to six months later.

Jason Yuan ·

AI models want to add or patch as a form of making better, whereas the design process is really about editing away.

Jason Yuan ·

Taste is editing with a point of view.

Dex Horthy ·

If you want loop engineering, build one loop at a time and keep them small: every part of the agentic-factory playbook is good advice except the part where you stop reading the code.

Jason Yuan ·

In many sectors of design, especially high culture and high design, people seem to have contempt towards the human experience.

Jason Yuan ·

Something about design, especially high design now, feels utterly sexless: sanding away anything arousing or objectionable creates essentially nothing, which is the ultimate slop.

Dex Horthy ·

Without good intuition for models, treat 100k tokens on smaller models and 200k on the frontier ones as the line past which the quality of your results is probably already degrading.

Dex Horthy ·

The IDE of the future has to be rethought from the ground up for agents, rather than being a text editor with an agents tab bolted on.

Dex Horthy ·

You can teach somebody to be a really good AI developer in a few months, but you cannot teach them a computer science degree in three — so hire for the fundamentals.

Steve Klabnik ·

A person who controls their own infrastructure and signing keys can meaningfully own their digital identity.

Alex Epstein ·

The ALARA rule is a major reason nuclear power is impossible to build cheaply in America.

Pieter Levels ·

All you need to build a business today is free open source software, a VPS, an AI API and S3-style file hosting.

Arvind Narayanan ·

An AI agent can only have two of three properties at once: being general-purpose, being deployed in high-stakes settings, and being fully automated.

Kelsey Piper ·

It is impossible to configure a computer to allow meaningful web interaction without bombarding a user with addictive content.

Alan Castel ·

Learning should be made harder rather than easier, and a teacher's instinct to smooth the path is the first thing that goes wrong.

Zeke Hausfather ·

A strong El Niño's peak warming effect on global surface temperature shows up three to five months later, landing in the following year.

Andrew Plotkin · by

The ZIL indentation style appears to have been designed specifically for ZIL game code.

Balaji Srinivasan ·

Producing moral, not just technological, progress requires introducing new moral premises that invert what people previously believed.

Balaji Srinivasan ·

The US federal government has lost the capacity to build basic infrastructure, let alone advanced military hardware.

Balaji Srinivasan ·

Micronations like Sealand fail because they start from territory and government rather than building outward from a people and shared culture.

Balaji Srinivasan ·

Movements built around a shared moral cause outcompete purely profit-driven ones, in practice as well as in theory.

Balaji Srinivasan ·

An organization that demands trust ends up with less of it than one that shows its work instead.

Justin Jackson ·

Even as AI makes building software easy, people will still prefer to buy some software off-the-shelf rather than build their own.

Steve Klabnik ·

AI safety restrictions on cybersecurity-adjacent requests make it harder for individual developers to build software as robust as large organizations' products.

Austin Vernon ·

A country's war effort improves when scarce resources flow to its most effective units rather than being spread evenly across all of them.

Austin Vernon ·

Ukraine's strategic victory depends on a Russian collapse, because it lacks the ability to conduct a theater-wide combined arms offensive suited to drone warfare.

Adam Mosseri ·

Small product teams beat large ones for a reason that has nothing to do with AI: fewer people to coordinate means less design by committee.

Andrew Kelley ·

Jumping into problems you're not yet equipped to solve, even if it leads to mediocre engineering, is a healthy way for young people and students to learn and grow.

Adam Mosseri ·

Designers will do well out of AI rather than badly, because taste is the hardest part of their job to imagine being automated away.

Arvind Narayanan ·

AI, in its current hyperscaler/infrastructure form, risks falling into the same commodity trap that has affected past infrastructure industries.

Adam Mosseri ·

Grit, fast learning and self-awareness are enough to get good at anything eventually, and whenever somebody is not working out, one of the three is missing.

Andrew Kelley ·

Bugs are eliminated mainly by dedicating engineering resources to finding and fixing them, not by choosing a particular language or style guide.

Adam Mosseri ·

Building a new product on top of whatever is growing fastest is a mistake: momentum is not the same thing as a strong foundation.

Terence Tao ·

The next advance in AI for mathematics requires shifting from predefined problem-solvers to research agents that address frontier challenges with formal reasoning.

Meaghan Choi ·

The invisible thinking behind a product, the mental model, the interaction model and the base primitives, has long been the unsung hero of product designers.

Terence Tao ·

Natural-language LLM mathematical reasoners are fundamentally limited by lacking machine-checkable semantics, which blocks autonomous verification needed for open-ended research.

Meaghan Choi ·

It is really important that people feel a little bit of warmth from AI tools, and they should not be super detached from them.

Meaghan Choi ·

The aim is to avoid inventing new UI and to make a tool feel so natural that it becomes part of the workflow.

Meaghan Choi ·

Work on the user model, relationship and behavior is the really important design thinking, while design polish pieces are really easy to offload.

Meaghan Choi ·

Knowing when to use AI tools for execution and when to go deep on the thinking work is a skill, and quality and polish do not have to take all of a designer's time.

Meaghan Choi ·

When what is being tested is the underlying mental model, it is probably not worth polishing, because the thing might not make sense; knowing when craft and polish are needed is important.

Meaghan Choi ·

Experimenting with mental models does not require high craft and is better done at speed, in the open, in iteration than by polishing something that is the wrong shape.

Meaghan Choi ·

Quality has never been only design's responsibility, and the most successful organizations feel it is owned by everybody.

Meaghan Choi ·

Designers should define quality and share it as a workflow anyone on the team can use, in a way that is additive, teaching and guiding rather than stopping.

Meaghan Choi ·

There is probably still going to be a need for a lot of fixed interfaces, because much of a product needs to be reliably stable, like a login screen or billing settings.

Meaghan Choi ·

Deciding what should be fixed and what adaptive is an important design decision; people often assume everything should be customized, but a lot of people do not want that.

Meaghan Choi ·

Just because something can be built as a function does not mean it needs to be, and adding it brings a lot of product bloat and complexity.

Meaghan Choi ·

Designers should lean into the chaos, develop new ways of working and continuously update, rather than cling to a structured process.

Meaghan Choi ·

Discernment about what should be built, and how to build it for the product, is the core builder process right now.

Alon Levy ·

Building out low-speed regional rail to small towns before intercity rail is fast enough is a waste of money.

Tiago Forte ·

The current AI-driven intelligence explosion will require new social structures to harness its power without damaging human values.

Tiago Forte ·

Realizing AI's potential to expand human potential requires humans and AI working together, not AI acting alone.

swyx ·

The finance industry is built on false mathematical precision and prone to systemic collapse.

swyx ·

America is a country that rewards first-principles thinkers and builders and encourages reinvention despite cynicism.

Jason Liu ·

AI can hand you the end result of a skill without the attentive practice that used to build it.

Jason Liu ·

Taste is not about consensus or criticism, but the courage to value what genuinely resonates with you.

Jon Seager ·

Linux system utilities, package managers, language runtimes and command-line tools have historically lacked a consistent way to check whether a TLS certificate has been revoked.

Ketan Joshi ·

Generative AI's climate harm comes mainly from wasteful system design and forced-on features, not from the computational cost of individual queries.

Kent Beck ·

Letting non-programmers build their own unmaintainable software is still a large step forward, because the alternative was a thousand-page requirements document and an eight-year wait.

Kent Beck ·

An outside expert's understanding of somebody else's problem is bounded, because they do not have to live with the design after they leave.

Kent Beck ·

Moving the computation to where the data lives is the fundamental design move in object-oriented programming, and it got lost in the noise.

Kent Beck ·

A coach's job is to identify and induce productive discomfort, not to reassure people that they are doing fine.

Kent Beck ·

People who vibe-code a replacement for a product will rebuild the visible tip of the iceberg and then discover the downstream problems they never knew existed.

June 2026

Grant Sanderson ·

If AI can build the new theory that crystallises how to think about a subject, it would be surprising if that capability did not spill into the rest of the economy.

Jonathan Hopkins ·

Falling rents are the signal that makes developers stop building, so a city cannot build its way to affordability under the current supply model.

Meng To ·

Designers have to build their own tools, because the tools that exist are never good enough.

Adam Tornhill ·

AI will change the quality threshold required to keep systems stable.

Grant Sanderson ·

AI learning to generate good conjectures will never show up as a benchmark being knocked down; it will show up as a shift in how mathematicians talk about the tools.

Jonathan Hopkins ·

Deregulating enough to make housing broadly affordable would mean legalising housing of the quality nobody is willing to defend.

Alon Levy ·

A ridership-estimation model built for typical U.S. transit systems can give unreliable results in a complex system like New York's.

Meng To ·

A skill is basically a prompt, sometimes a large one, more like a style guide: the best practice for taste, which font to use, and how to guide AI not to create AI slop.

Grant Sanderson ·

The people who come up with genuinely novel insights are also unusually clear communicators of them, against the assumption that expertise spoils explanation.

Grant Sanderson ·

He used to think explaining AI's results would be the mathematician's remaining job; he now expects AI to explain better than most humans, so that is not what is left.

Meng To ·

A screenshot is the fastest, easiest way to build, without having to voice it or type it.

Meng To ·

The details that come from your own personality and experience, such as watching Apple work or watching a designer friend work, will never be found in an existing library.

Meng To ·

The design.md is basically a designer's style guide, all the sweat, tears and blood of their mastery, in one markdown file that the AI can read.

Meng To ·

Given a design.md, a very capable model will understand it and take it to the next level, while a base model won't understand it.

Meng To ·

The design.md is powerful at the beginning, but when you are doing a lot of iterations really quickly you don't want to always reference it, and you can save it into agents.md.

Meng To ·

Most importantly, have an endless amount of curiosity about how you can perfect your own workflow, because it keeps evolving and always requires your attention.

Scott Greenfield ·

Ending a temporary immigration status overnight, after people have built lives in reliance on it, does real harm to them and to their communities.

Alon Levy ·

Rapid electrification programs without continuous rolling pipelines can still be built cost-effectively, as in Denmark, Israel and India.

Hamel Husain ·

An artifact that is hard for its builder to verify is usually also hard for its users to verify.

Kyle Chayka ·

Getting genuinely original output from AI requires working hard against the grain of its defaults.

Hamel Husain ·

Designing a product for ease of verification should happen before building evals for it.

Kyle Chayka ·

AI-driven homogenization will spread across every field of human endeavor until AI becomes truly original or is curbed.

Vitalik Buterin ·

Building a secure obfuscation protocol is extremely difficult.

Vitalik Buterin ·

Known iO schemes built on reasonable security assumptions have literally galactic runtimes.

Jason Liu ·

Good automation design starts from whether context must persist between runs, not from building a complex system.

Jeff Cavaliere ·

Seek out the discomfort of a hard contraction rather than shying away from it.

Jeff Cavaliere ·

Progress bodyweight exercises with harder variations rather than more reps.

Jeff Cavaliere ·

The face pull is a muscle-building exercise for the upper back as well as a corrective one.

Pieter Levels ·

Claude Code is the coding agent to build with.

Sebastian Raschka ·

When an AI lab builds both a model and a coding harness, the model is optimized for that lab's own harness first.

Dwarkesh Patel ·

Training AIs on millions of verifiable tasks in diverse RL environments will lead to AGI.

Danielle Ofri ·

Understanding a patient's story is necessary to help them navigate their illness.

Dwarkesh Patel ·

The ability to make progress in a domain depends on the ability to build a replayable training target.

Noam Brown ·

A frontier lab should not spend its researchers harvesting results out of today's models; the payoff is in building the next one.

Noam Brown ·

There will be no overnight intelligence explosion, because a model's best work takes so much test-time compute that time itself is the bottleneck.

Noam Brown ·

Humanity's advantage is accumulated shared knowledge rather than higher intelligence — and AI models, which vanish with their context window, have none of it.

Noam Brown ·

Every lab knows the benchmark grid is the wrong way to present a model, and publishes it anyway because everybody else does.

Reece Martin ·

A model of transit building that adds a new line every couple of years is healthier than one that adds them every couple of decades.

Luke Wroblewski ·

Software now needs a steering layer between the tools and the codebase, so that AI output stays aligned with brand, design and coding guidelines.

Gretchen Rubin ·

Yelling and nagging at children produces compliance in the moment but costs tension, sulking, and guilt.

Loredana Crisan ·

Really good performance expertise requires time; even people with great aptitude still need to put in the work to become an incredible designer.

Dan Meyer ·

Generative AI has failed to produce the learning gains it was promised to deliver and instead shows evidence of harming learning.

Loredana Crisan ·

When everyone has the power to build, the need to stand out is elevated.

Loredana Crisan ·

What matters is a point of view and human intuition about what the experience should be, which needs to come through, more than taste.

Loredana Crisan ·

Some things are almost impossible to describe in a prompt, the color wheel will never be replaced by one, and there are moments where you just want to take control yourself.

Loredana Crisan ·

Increasingly, designers are creating systems, not just screens.

Loredana Crisan ·

Design is never done alone; it evolves through critique, and other people are one of its tools.

NeetCode ·

AI has made it cheap to build the features that were never worth building, so shipping a new feature every day can make a product worse rather than better.

Loredana Crisan ·

Review is a bottleneck across design and development, and since people love making more than reviewing, they now have to become expert reviewers and probably automate reviews too.

NeetCode ·

What AI has exposed is a shortfall of effort and caring, not of intelligence: people who cannot say why they moved a button never thought about it.

Loredana Crisan ·

Everyone is elevating up the stack in some way, and design is having an identity crisis and a renaissance at the same time.

Loredana Crisan ·

Roles are blurring but expertise still very much matters, and a designer should feel empowered to take an idea all the way to front end production.

Loredana Crisan ·

It is really important for designers to embrace evals, because that is how the system is created and it is the definition of good.

Loredana Crisan ·

Design is an unverifiable domain: there is not a correct answer to design, but there are many incorrect answers.

Loredana Crisan ·

Magic happens at the intersection of things that used to be disconnected, such as design and code, sound and visual, or PM and design.

Loredana Crisan ·

Complexity is kind of the enemy of action, and design is increasingly important for simplifying things so they can move forward.

Adam Tornhill ·

Names act as cognitive compression mechanisms that allow design elements to stretch the amount of information held in working memory.

Henrik Karlsson ·

Living according to one's values long-term requires turning the realization into a habit, not just having the insight.

Adam Grant ·

Leaders' resistance to remote work is driven by a desire for power and status, not by genuine productivity concerns.

Ketan Joshi ·

Generative AI systems have no built-in drive toward truth and accuracy, unlike people or institutions.

Ketan Joshi ·

Genuinely valuing truth, honesty and climate action logically entails opposing generative AI.

Brad Feld ·

Getting feedback on a flaw early saves far more effort than discovering it after building much more on top of it.

Jordan Ellenberg ·

People are good and the system works

Jennifer Pahlka ·

When government cannot keep pace with technology, the companies building it make the consequential decisions instead of elected representatives.

Brad Feld ·

Doing creative work in public, where failures are visible, produces faster useful feedback.

Kenneth Reitz ·

The finite vocabulary of any language runs out before the range of human experience does.

Aswath Damodaran ·

Most active investors are not hunting for information; they are betting on mean reversion in data everyone already has.

Kenneth Reitz ·

Large language models do not think in words.

Lip-Bu Tan ·

Selling silicon is no longer enough — customers now ask for the whole rack, so a chip company has to become a full-stack company.

Lip-Bu Tan ·

Semiconductor infrastructure has not kept up with AI's growth, and closing that gap needs capacity built in unconventional ways.

Lip-Bu Tan ·

Performance can still double, but cost and area will no longer come down with it unless new materials or new design methods arrive.

Lip-Bu Tan ·

A hardware startup's first customer should be a hyperscaler, because only that scale turns a design win into a business.

Lip-Bu Tan ·

The massive AI build-out is the right thing to be doing, and the only thing that will slow it is supply.

Lip-Bu Tan ·

The AI build-out will end the way the internet did: a couple of Amazons, and a long list of companies that go sideways or disappear.

Brad Feld ·

Entrepreneurial ecosystems are built by the entrepreneurs who lead them, take a long-term view, and give first.

Elio Struyf ·

AI coding tools remove the natural filter that slow, effortful building placed on which ideas actually get built.

Ada Palmer ·

A government only has legitimacy through long continuity, and once that thread is cut one regime change reliably becomes five.

Stephen Wolfram ·

AI-written code is fine when only the visual result matters, but insufficient in technical domains where understanding what is actually being computed is required.

Stephen Wolfram ·

AI-generated code is not fully reliable and typically needs significant manual effort to keep it working correctly.

Brett Williams ·

The last 20% of building an app with AI is like 80% of the time.

Samuel Hughes ·

Housing affordability is almost the direct inverse of how many new homes are built, so the answer is to build more houses.

Ada Palmer ·

Whether breaking your word destroys you depends on the shape of your power: Savonarola’s rested on never erring, so his contradictions ruined him.

Samuel Hughes ·

Zoning is a European invention: German and Austro-Hungarian cities wrote the first rules protecting rich suburbs from denser building.

Brett Williams ·

It is really difficult not to over-engineer when AI tools put the world at a designer's fingertips.

Brett Williams ·

Taste and judgment are what separate good products from bad.

Brett Williams ·

Sound in an interface can be overdone, so be selective about when to use it and probably give the option to turn it off.

Brett Williams ·

It is helpful to be able to see several options side by side, quickly, in one file.

Brett Williams ·

AI is like the most patient collaborator you could have, building at hyper speeds, if you know how to direct it and bring taste and judgment on what to build and how, down to the finest UX and UI decisions.

Brett Williams ·

AI tools in the hands of a designer with good taste and good judgment are remarkably powerful.

Brett Williams ·

AI is still not good at design, but it is good at taking direction, which works as long as you direct it clearly and make the right decisions.

Ada Palmer ·

Copyright was born out of the Inquisition: the record of having passed censorship was the first document proving an exclusive right to print.

Ada Palmer ·

The Inquisition's image as one centralised power is its own propaganda: it was scattered isolated men who could act only as far as a local government would fund them.

Vicki Boykis ·

Local models are not yet reliable enough for production software development.

Mark Pincus ·

A founder's gut instincts are almost always right and the ideas built on top of them are usually wrong, so the work is to isolate the instinct and test many ideas around it.

Mark Pincus ·

The genuinely new idea in a product will probably fail; its job is to give people a reason to try the product at all.

Mark Pincus ·

Being too ambitious and visionary at the outset is how founders miss product-market fit, because it stops them starting somewhere small and humble enough.

Mark Pincus ·

A first-time founder has an advantage over a repeat winner, because past success makes it too easy to raise money and hire against a big vision before there is product-market fit.

Mark Pincus ·

AI should be used as a failure machine that tests many ideas cheaply, not as a way to build one idea faster.

Mark Pincus ·

The most valuable companies in the world are the ones with the highest day-365 retention, and a team can build against that number long before day 365 arrives.

Mark Pincus ·

AI is an important technology but is not yet a platform, because nothing has been opened up for other apps, experiences and developers to be built on.

Mark Pincus ·

Distribution has to be part of the product and proven from the beginning; building and hoping people come is a hope strategy, not a belief strategy.

Mark Pincus ·

The greatest thing a product maker can build is an internet treasure — a service people cannot remember life before or imagine life without.

Nikita Prokopov ·

Polished UI is a reasonable signal that a team invested comparable effort in the underlying code quality.

Varun Raj Manoharan ·

In a structured professional network, real relationships form in the informal time around the meeting, not within the formal structure itself

Kyle Chayka ·

Most digital creators are converging on parasocial 'show' formats built to gather and monetize audiences.

Ben Vinegar ·

Programming languages that LLMs do not choose, like Ruby and Zig, look dead in the long term.

Ben Vinegar ·

Software is going through a fast-fashion phase where projects are thrown away if nobody uses them at once; builders should have conviction and work on things longer.

Ismail Ghallou ·

Durable distribution channels are built deliberately rather than found by chance.

Julie Menanno ·

Avoiding conflict does not actually resolve it, so suppressed feelings build up until they surface all at once.

Sasha Chapin ·

Neither the claim that great things only come without force nor the call for constant discipline describes reality.

Arvind Narayanan ·

Production software cannot be built by unskilled vibe coders replacing trained software engineers.

Ruxandra Teslo ·

Manufacturing and batch-release requirements, not biology, are what kill most promising cell and gene therapies.

Ray Dalio ·

Success now requires combining the best human intelligence with the best artificial intelligence

Mark Zuckerberg ·

This work could have been run as a business; open-sourcing it simply puts the tools in more scientists' hands faster.

Mark Zuckerberg ·

There is no factory you can pay to produce biological data; each new dataset requires inventing a new scientific method first.

Mark Zuckerberg ·

Tools this complicated need a ten-to-fifteen-year horizon, which is why they belong in a philanthropy rather than a startup.

Mark Zuckerberg ·

The point is not to cure diseases but to build the tools with which the whole scientific field cures them.

Mark Zuckerberg ·

The right unit of study is a system like inflammation rather than a named disease — understand the system and others can build the therapies.

Mark Zuckerberg ·

An exponential feels like it cannot possibly keep going, which is exactly why people keep expecting AI progress to stop.

Ron Goldin ·

A player coach manages a smaller team while building together with it, rather than sitting in an ivory tower doing performance reviews and alignment meetings.

Ron Goldin ·

Prototypes win arguments: moving from slide decks and PRDs to something people can feel and react to gets them psyched.

Sarah Paine ·

China and Russia have maritime ambitions without understanding what a maritime paradigm actually requires, and neither one has the prerequisites.

Bill Gurley ·

Knowing both the deep history of a field and its bleeding edge is what makes someone genuinely differentiated in it.

Ron Goldin ·

Some of the best design leaders are people who have spent some time being entrepreneurs or working in early stage companies, because of the ownership.

Ron Goldin ·

More companies look at things built outside of work not as extra credit but as possible reasons to hire, in addition to or even over traditional work experience.

Ron Goldin ·

Building software is not as hard as it used to be, but growing an audience is as hard as it always has been.

Ron Goldin ·

The usual planning exercises still help when building with AI: defining the product's goal, who it is for and its critical user journeys.

Ron Goldin ·

Trying to build everything at once with any platform is generally kind of a disaster.

Ron Goldin ·

A place for stored project context that an AI platform can keep referencing is worth setting up.

Ron Goldin ·

Sketching a fix in a design tool is probably quicker, and closer to the intended result, than going back and forth in words with an AI.

Ron Goldin ·

AI-built products look very similar because the tools are all trained on the same material.

Ron Goldin ·

Branding and differentiation should not wait until the end; once you know what a product is for, define them as early as possible to infuse brand and point of view.

Ron Goldin ·

Designers can no longer be just execution or order takers, because a product manager does not need them for that.

Ron Goldin ·

Designers with the courage to do scrappy research, and who know when it is useful, are very valuable.

Bill Gurley ·

Venture capital bends toward the young, because the work is a hustle and later-life obligations cap how many hours anyone can give it.

Ron Goldin ·

Finding customers and having a differentiating product that feels premium and connects emotionally with its users fall squarely in the design world.

Sarah Paine ·

The Belt and Road cannot beat the sea, because the land route is not only dearer but requires controlling the most unstable places on earth end to end.

Ron Goldin ·

Reading and understanding the output of an AI coding tool makes a designer a more successful builder.

Ron Goldin ·

A lot of brand expression today can come through motion design, which is an underutilized skill among product designers.

Ben Vinegar ·

Coding agents make doing things on Linux far more approachable.

Tony Fadell ·

You cannot get real feedback on a new consumer product before shipping the whole thing; feedback starts when somebody spends their own money on it.

Tony Fadell ·

What is worth building is found by starting from a long-standing pain and asking which new technology has just made it solvable.

Tony Fadell ·

What you have to invent is the whole system around a product, including how it is sold, bought and installed, not the one piece people remember.

Tony Fadell ·

AI coding works when a person owns the architecture and scopes the agent to small pieces of it, and fails when the whole thing is handed over.

Tony Fadell ·

Fast software is fast fashion: it does not survive a season, and a company meant to last cannot be built on throwaway code.

Tony Fadell ·

A model can produce version two of a product that already exists, but the genuinely new version one still has to be made by people.

Tony Fadell ·

Devices should be built voice-first, with keyboard and touch demoted to second and third, the exact reverse of how the smartphone was layered.

Tony Fadell ·

Software cannot reach its next level until somebody builds the next level of hardware underneath it.

Tony Fadell ·

A builder asked to addict users should leave, because there are always other jobs and no amount of money is worth tearing up the fabric of a society.

Sebastian Raschka ·

Long-context efficiency has become a central priority in LLM development because agent harnesses require working with increasingly long contexts.

Pieter Levels ·

Now that everyone can build apps with AI, distribution is the thing that separates who wins.

Bruno Maçães ·

Human beings often react with anger when others decide to pursue diverging paths in intellectual or political pursuits.

Justin Jackson ·

The most valuable problem to pursue is the hardest one someone is genuinely positioned to solve.

Stephen Wolfram ·

The tit-for-tat strategy, despite its reputation as the most successful approach in game-theory strategy comparisons, ranks poorly when tested against the full space of possible strategies.

Ryan Holiday ·

Getting up in front of people gets less scary the more you do it, because confidence is built by competence.

Justin Jackson ·

Building the hardest feature in a product yields disproportionately greater returns than easier ones.

Phil Trammell ·

Some goods really will get a fully automated supply chain, and yet what that does to the overall capital share is genuinely ambiguous.

Alex Imas ·

The worst AI employment scenario is not a crash but a slow drip, where people are quietly re-absorbed into worse-paid work over twenty years.

Alex Imas ·

There is still no white-collar apocalypse in the data: you have to squint to see anything, and even junior developer hiring is below trend rather than falling.

Alex Imas ·

An AI-caused recession requires the owners of capital to hit a hard ceiling on spending and then refuse to invest the rest, which is a very improbable combination.

Phil Trammell ·

The capital stock can grow quickly while the capital share falls, because the price of capital goods relative to consumption goods can drop faster still.

Josh Barro ·

Having worked at a highly selective, demanding employer signals conscientiousness and reliability.

Kelsey Hightower ·

The hardest problem a company full of brilliant engineers has is not how to build something — it is what to build.

Kelsey Hightower ·

Going slowly is a luxury: the busier you are, the more of your own life you have to pay somebody else to live for you.

Kelsey Hightower ·

The job has always been to solve human problems, and software is not required for every human endeavour.

Kelsey Hightower ·

Seven API calls to get a virtual machine was never an intention, it was a procedure — most of our APIs were designed wrong even for humans.

Kelsey Hightower ·

MCP is not a new foundation for anything — it is API design, and Kubernetes was already turning one intention into seven calls a decade ago.

Jon Haidt ·

Reversing the takeover of childhood by technology requires a change in public opinion first, because that is the signal political leaders act on.

Adam Wiggins ·

Explicit signals like follows and subscriptions are as brittle and hard to maintain as manually tuned filter rules.

Ben Vinegar ·

When debugging with AI it is easy to get stuck inside the model's search space, chasing its candidate fixes while the real answer sits in context only you have.

Adam Tornhill ·

The less humans code, the more it is necessary to optimize for design, architecture, and socio-technical fit.

Steven Johnson ·

A well-designed AI-integrated approach to schoolwork can produce better learning outcomes than pre-AI research and writing methods.

Justin Jackson ·

A business can have market fit, real advantages, and sustained effort behind a hard product and still fail if it doesn't give customers a strong enough reason to switch from what they already use.

Steven Johnson ·

Apparent inevitabilities in new software paradigms are often just design failures that can be fixed with better guardrails.

Fatih Arslan ·

Keyboard buyers increasingly expect switches and keyboards to sound good, not just feel good.

L. M. Sacasas ·

Thinking, judgement and knowledge are worth pursuing for the joy in them, not in order to defend human uniqueness.

Mark Pincus ·

Users vote with their clicks, and roughly half of them are lost at every click a product asks for.

Mark Pincus ·

A company owes its builders a moral contract: if they take the hill, leadership must work just as hard to unlock the value of what they built.

Fabien Sanglard ·

Multimedia design succeeds by building up from a platform's real hardware constraints rather than compressing full video down to fit them

Sasha Chapin ·

Most contemporary restaurants are over-optimized, investor-backed, and homogenized by branding rather than genuine culinary vision.

David Heinemeier Hansson ·

AI coding agents advance open source's founding goal by letting far more people modify software, so barring AI-assisted contributions protects a guild rather than the mission.

May 2026

Brandur Leach ·

Even with LLMs, building software yourself still costs real ongoing human time, so replacing a paid tool with an LLM-built internal one often does not pay off.

Brandur Leach ·

Software that is novel enough to be non-trivial to rebuild with an LLM, and priced within reason, will still be worth buying rather than building.

Jennifer Pahlka ·

Being skilled at tracking project requirements doesn't make someone skilled at product management.

John Siracusa ·

A steering column stalk is a better control for turn signals than steering-wheel-mounted buttons.

Jacob Falkovich ·

Good writing requires wanting to tell people something, which AI itself lacks.

Kate Wagner ·

Architecture and everyday life have grown progressively cheaper and blander over time.

Vicki Boykis ·

AI code generation's default reward loop works like a slot machine and undermines skill retention.

Kyle Chayka ·

Artists whose livelihoods are threatened by AI-generated content are turning toward deliberately rough, handmade aesthetics as a response.

Kyla Scanlon ·

The human strengths that matter in the AI age are evaluation, taste, accountability, networks and a human touch.

Kyle Chayka ·

The globalization of generative AI tools has leveled creative production, letting people everywhere work in the same visual idioms.

Mark Carney ·

Canada's answer to a fracturing world is to build strength at home and diversify its partnerships abroad, not to wait for Washington.

Mark Carney ·

Canada should build a pipeline carrying at least a million barrels a day of low-emission Alberta oil to Asia, paired with large-scale carbon capture.

Samuel Hammond ·

Frontier AI models plausibly have some form of consciousness or subjective experience.

Michael Liebreich ·

The peak in global fossil fuel use and emissions will occur before 2030

Scott H. Young ·

A better school looks more like the old-fashioned schoolhouse than like project work; the more school resembles free play, the worse it gets.

Dax Raad ·

Shipping ten times as much is not an advantage, because the moment you ship something you are stuck supporting it forever and every later feature has to interact with it.

Dax Raad ·

The winning move in coding agents was inverted: take share with a merely good-enough harness first, then go back and make the harness smart.

Dax Raad ·

Redefining the engineer's job as building guardrails is not novel — it is the same problem as making a junior engineer effective, which we never solved either.

Dax Raad ·

Being a decent software engineer plus a genuine expert in one industry is a deadly combination, and engineers are the only people who get to change industries at will.

Carlos Alexandro Becker ·

AI can now produce in minutes work that used to take weeks to build.

Andy Madrick ·

Coding tools are very good at executing motion design, but you have to be very descriptive about what you want, otherwise you will be fighting them for hours.

Rohit Krishnan ·

Audit independence is a property of the surrounding system and its evidence flows, not an inherent property of the auditing model itself.

Andy Madrick ·

Figma is still really helpful for a smaller thing like a modal or visual polish.

Josh Comeau ·

Animation libraries that only wrap built-in CSS transitions and keyframes are not worth using, because they add main-thread work and bundle size and offer nothing new

Andy Madrick ·

The artifact users actually see is what to obsess over, and Figma and mocks are just reference tools at the end of the day.

Andy Madrick ·

LLMs are not very good at the last mile of design, and that last mile needs to remain designers' superpower.

Andy Madrick ·

Designers have no excuse anymore to not be responsible for the changes they want to see.

Andy Madrick ·

An app used every day should be built by a thoughtful team of engineers rather than one-shotted with Claude.

Andy Madrick ·

Asking an agent to keep an engineer's functionality and apply a designer's styling can get much closer than intuition says it would.

Andy Madrick ·

A designer's PRs should be very small and manageable if the designer is not comfortable reviewing or writing the code themselves.

Andy Madrick ·

Work should be broken into manageable chunks, and a designer should not send an engineer a 500-line PR they did not even review themselves.

Andy Madrick ·

AI-generated UI can come out with a million gradients and tags, looking like a jelly bean factory.

Andy Madrick ·

There is no shortcut to having taste or developing a high-craft, high-velocity practice; AI shortcuts the creating of a lot of things, but it comes back to learning the language of visual design.

Andy Madrick ·

With limitless UI out there, a designer's job is to take it and make a restrained version where form follows function and decoration for decoration's sake is avoided.

Andy Madrick ·

Copying a UI is still a worthwhile exercise, less for a pixel-perfect static frame than for its logic, such as how a sidebar behaves.

Andy Madrick ·

The best way to stand out in a portfolio is to show your best work and let it speak for itself, because no one wants decoration for decoration's sake.

Andy Madrick ·

A good default is to look at a thing with no decoration, the same size typeface and no animations, then ask where a flourish of delight would help the user rather than detract from what they are trying to do.

Andy Madrick ·

Designers benefit most from working as a jazz band: sharing work, letting people take it a certain way, and developing the thick skin to take the best idea and run with it, wherever it comes from.

Andy Madrick ·

The really great designers just do less.

Rohit Krishnan ·

Designing a good AI benchmark has become a task only the most capable people can do well.

Phil Gaimon ·

Raising the barrier to entry through high-tech equipment and medical enhancement limits the pool of participating athletes.

Dan Shipper ·

The CLI era of AI agents was a short detour — we made GUIs for a reason, and the work is moving back into one.

Dan Shipper ·

Designers have stopped handing work off — they open the pull request themselves, which changes how companies build.

Pedro Pita Barros ·

The implementation of USF modelo C contracts will likely take a long time due to the complexity of defining terms of reference and negotiating final contracts.

Alice Ryhl ·

Examples in documentation should be compiled and run as tests, so that changing the code breaks the docs instead of quietly outdating them.

Alice Ryhl ·

What is unique about Rust is not any one feature but the combination: high-level reliability with no garbage collector, so it runs in kernels and firmware.

Alice Ryhl ·

A design proposal should describe the feature as though it already existed — once as a tutorial and once as reference documentation.

Alice Ryhl ·

The Linux kernel is unique in having thousands of contributors paid to work on it; Rust does well by any other standard and is still nowhere near that.

Alice Ryhl ·

You can contribute to the Linux kernel without touching the kernel, by building the language features the kernel is waiting on.

Alice Ryhl ·

Rust is no longer experimental in the Linux kernel — the maintainers agreed that at their December 2025 summit.

Alice Ryhl ·

You learn a programming language by building something in it, not by reading about it.

Kelsey Piper ·

The current approach of edtech companies making games with minor educational elements results in software that is hard to distinguish from casual games like Candy Crush.

Tommy Geoco ·

A briefcase of context, carried from one tool to the next, will replace re-teaching intent in each tool.

Brandon Sanderson ·

Fantasy requires structure to succeed as a literary genre.

Tommy Geoco ·

A general-purpose AI design tool might have a place eventually, but it is probably not the right answer right now.

Tommy Geoco ·

Canvases are always going to have a place in the AI design workflow.

Tommy Geoco ·

A design team needs an interface surface, a surface where the work lives, a canvas for more deterministic work, and an agent interface for coding for more probabilistic work.

Tommy Geoco ·

Adopting AI requires rearranging how work is done, as Ford had to rearrange the whole factory rather than install new equipment on old workflows.

Tommy Geoco ·

The number one thing the best design teams do is make room for experimentation, even though it is expensive.

Tommy Geoco ·

Leadership has to make time for designers to experiment with AI.

Tommy Geoco ·

It is still really early, and people can probably wait on much of the agentic tooling.

Tommy Geoco ·

The skill worth investing in is building intent and moving between altitudes, from product thinking to the interaction moment to workflows.

Tommy Geoco ·

The blank canvas problem is a good Hello World project for getting into AI tooling.

Tommy Geoco ·

Comparing workflows is useful now, because the field will eventually converge around better practices, though none are known yet.

Tommy Geoco ·

A designer who is not a designated leader can get leaned on by building an internal tool for a cross-functional need, such as one design needed by marketing.

Tommy Geoco ·

The biggest thing a designer can do to add value now is help cross-functional people do more design work easily, and this is unlikely to cannibalize the designer.

Tommy Geoco ·

AI fluency will eventually become an ambient thing, though it is important language right now.

Tommy Geoco ·

A good way to assess AI fluency in a designer is to ask them to show their workflow and take them through an idea they created.

Tommy Geoco ·

The future belongs to builders; the future is bright, but it has to be built.

Jason Liu ·

A recurring AI workflow improves more through ongoing user feedback than by being fully specified upfront.

Sebastian Raschka ·

As reasoning models and agents retain more tokens for longer, KV-cache size and attention cost become the main constraint on scaling context.

Danielle Ofri ·

AI cannot supplant the totality of medical care because it cannot grapple with ambiguity, uncertainty and contradiction.

Josh Barro ·

Federal Reserve Board members have historically often resigned before completing their full terms.

Yihui Xie ·

Needing to mock code in tests often signals an underlying design problem that refactoring would fix.

Julia Evans · no longer holds

Tailwind is a good way to build a site you're happy with, even without strong CSS skills of your own.

Jasmine Sun ·

AI marketed as a pure augmenting 'tool' will end up automating jobs anyway, since firms control how it's used.

Ketan Joshi ·

The environmental harm caused by data centres comes from how the industry chooses to build them, not from anything inherent to computing infrastructure.

Anders Hejlsberg ·

A language should be designed by a small group of people rather than by one person.

Anders Hejlsberg ·

Language design is ninety per cent the same work every time and ten per cent new, which is why the experience carries from one language to the next.

Anders Hejlsberg ·

The job of a design group is to try to shoot down every new idea; the ones that survive the attempt are probably the good ones.

Anders Hejlsberg ·

You cannot build good tooling without a type system — decent tooling perhaps, but nothing that scales to a large team.

Anders Hejlsberg ·

People adopted TypeScript for the tooling, not the types: an erasable type system is worth having because of the productivity the tooling built on it delivers.

Anders Hejlsberg ·

A compiler that also serves an IDE has to be lazy and deferred throughout — a different way of building compilers from the one the textbooks teach.

Anders Hejlsberg ·

The language best suited to AI is simply the one the models have seen most of in training, which is why they do so well on JavaScript, TypeScript and Python.

Anders Hejlsberg ·

Programmers are turning into project managers of an army of junior agents, and the craft is moving from writing code to reviewing it and holding the architecture.

Katarina Batina ·

Any redesign of an incumbent product, such as an onboarding flow, almost always loses at first; it takes the confidence and agreement to begin below the status quo baseline in order to beat the incumbent.

Katarina Batina ·

Teams need to create the grounds to take some risk, committing as much iteration to the new thing as was put into the incumbent product.

Katarina Batina ·

Designers should focus on what differentiates them, which is designing what is desirable, and pick the tools that render the idea in the most believable way rather than worrying about feasibility.

Katarina Batina ·

Not being able to ship a prototype to production does not mean a designer has failed.

Katarina Batina ·

When prototyping a full feature, a designer should render the whole idea and use that as the thing to talk to engineering with, rather than aiming to ship it end to end to production.

Katarina Batina ·

Design should be the tip of the spear for desirability, and taste, editing and deciding what matters are skills that have long had their home in design practice.

Katarina Batina ·

Constraints driving the design process is something to avoid, especially now.

Katarina Batina ·

With any new explosion of tools, designers may start designing to their local skill set.

Katarina Batina ·

Editing, focusing on the most critical aspects of an experience and being really disciplined about it, has become an important part of the design job.

Katarina Batina ·

The way to build excitement for an idea is still to make a really compelling prototype and tell a story, and that is much easier now.

Katarina Batina ·

Beyond designers working in code, this era unlocks access to data and legibility into the business that designers did not have before.

Katarina Batina ·

The most enduring skill in this era is shaping great products, not just executing: designers come for craft work and leave as great product thinkers.

Katarina Batina ·

Great design leaders of this era need a deeply intense relationship with what the business is there to do, paired with a deep relationship with what makes great design.

Katarina Batina ·

It will probably be a tough go in the long run for the professional manager whose calendar is driven by one-on-one time on career development, and a culture of designers improving is hard to build without deep passion for the craft.

Katarina Batina ·

One-on-one time should be about the work: opening up the prototype, live critique, and helping designers make better decisions.

Katarina Batina ·

Language is becoming the material for building, so the more precisely designers can say what is wrong with the work, the better off they will be in an agentic era.

Eric Ries ·

Organizations are dragged into mediocrity by a force no one controls but everyone obeys, until their founders lose control of what they built.

Mark Nottingham ·

CMS software should not silently create feeds that publishers never see or maintain, but make feed creation visible and deliberate.

Mark Nottingham ·

WHATWG should deprecate the feed link relation since it has been effectively abandoned in practice.

Eric Ries ·

Harder is easier: committing up front to quality, ethics or safety pays back in rewards you cannot get by chasing the reward.

Eric Ries ·

A mission statement is not a mission; a mission is an emergent property of the organization and has to be built in rather than labelled on.

Eric Ries ·

Becoming a public benefit corporation does not make you the good guys, but it does stop investors suing you for pursuing your stated purpose.

Eric Ries ·

Not one of the companies building frontier AI is willing to accept standard corporate governance, which is how dangerous they know the technology to be.

Eric Ries ·

Adding ants to a puzzle makes it get solved faster while adding humans makes it worse, unless the humans are carefully aligned.

David Reich ·

Natural selection in humans has not been quiescent: it is tugging almost every position in the genome in one direction or another, everywhere.

David Reich ·

The selection signal on the genetics of schooling is not a European artefact: the same variants predict years of school in China, a population disconnected from Europe for the whole period.

David Reich ·

Europeans have been under selection against body fat for 10,000 years because farming made the food supply steadier, not less stable.

David Reich ·

Famine cannot be what selected for fat storage, because a fat store built after a hunt cannot carry anyone to a famine three years later.

David Reich ·

For any trait built out of many mutations, all the variation needed to move a population to a new adaptive set point already exists inside it.

David Reich ·

The accepted model of how archaic and modern humans relate has been accreted and patched rather than built, and is due for a simpler replacement.

Douglas Rushkoff ·

Shamanic and somatic awareness is a natural, pre-linguistic state of being.

Mandy Brown ·

The investor class simultaneously threatens people with a lack of work and with excessive work requirements.

Douglas Rushkoff ·

The current AI-driven world is a pseudo-reality built on layers of symbols.

Pedro Pita Barros ·

One of the best long-term investments in prevention is a very strong network of family doctors.

Aaron Patterson ·

When money is on the line, AI makes it easier to disguise low-quality work as legitimate.

Nicholas Carr ·

The vision of a Great Community built on communication has been shattered, leaving us overwhelmed by information

Rafa Conde ·

Generalizing a design to suit every kind of user makes it harder to make targeted experiences.

Kelsey Piper ·

The principled liberal position on crime does not require abandoning the values of being humane and compassionate.

Saul Griffith ·

Australia's gas industry tax regime allows companies to become systemic non-payers.

Rafa Conde ·

Big companies can probably less afford the bigger design risks than they once did, so such work shows up more in small indie shops.

Saul Griffith ·

The warming impact of LNG over twenty years is roughly equivalent to coal at 3% methane leakage.

Rafa Conde ·

People are more used to computers than old design assumptions allow, and testing often shows they know how to scroll.

Rafa Conde ·

A static mockup leaves a lot missing when an idea is passed to someone else; a short video with motion carries more of it.

Rafa Conde ·

To share an idea for a feature, make something interesting and funny that communicates it well, rather than just sharing mock-ups or a screen recording of a prototype.

Rafa Conde ·

Video is probably underused by designers as a way to sell and communicate ideas.

Rafa Conde ·

Showing a person's face reacting to what they see carries a different weight than showing only what they are looking at.

Rafa Conde ·

Delightful moments in a design cannot be forced; a designer can put themselves in a state where they are open to them.

Rafa Conde ·

Empty states are a good place for moments of connection, because there is nothing else to show there.

Rafa Conde ·

Recreating a design you like in Sketch or Figma can be a great way for a beginner to slowly learn padding and basic layout.

Rafa Conde ·

A designer can notice and save the small details they like in other people's work, so they are on hand the next time they make something similar.

Saul Griffith ·

Every fossil fuel flow has a known clean substitute that can be built.

Eugene Yan ·

Delegating a task to an AI agent only works if the output can be verified, which requires defining success criteria first.

Joseph Politano ·

Fast-growing state economies deploy solar and batteries quickly primarily to close their electricity gaps and power economic growth, not for climate or ideological reasons.

Sasha Chapin ·

Positions of genuine respect are earned rather than granted undeservedly.

Mari Andrew ·

Pursuing efficiency and maximizing productivity too aggressively causes things to go wrong.

Patrick McKenzie ·

Complex professional practice, including law, can be reduced to written playbooks that make its outcomes reproducible without requiring artisanal mastery.

Jessamyn West ·

Deceptive software design makes declining a personal-data request harder to find than agreeing to it.

Ketan Joshi ·

Data centres are being built in such number, size and speed that the industry cannot realistically be reformed.

Elizabeth Van Nostrand ·

Removing routine antibiotic use forces farms to improve animal conditions, since poor conditions become too costly without a quick pharmaceutical fix.

Jessamyn West ·

Companies that publicly deny tracking users are often tracking them anyway, and face no real penalty for lying about it.

April 2026

Julie Menanno ·

Getting angry quickly in an anxiously attached person is a threat-detection habit their nervous system learned, not a character flaw.

Craig Hockenberry ·

iOS should let users rebuild a corrupted Spotlight index themselves, the way macOS does

Rui Carmo ·

Smaller models require explicit breadcrumbs in responses to successfully follow tool chains.

Tynan ·

The point of a grueling long-distance hike is to endure hardship, not to enjoy the scenery.

Anil Dash ·

AI works best when it solves a specific problem for a particular community rather than trying to be all things to all people.

Polly D'Arcy ·

If the people building a product would not use it with their own money, it is not good enough.

Polly D'Arcy ·

Using a product with your own money, rather than testing it in staging, lets you very deeply feel its pain points, friction and frustration.

Polly D'Arcy ·

It is really dangerous when design operates in a silo outside of product and engineering, as if it had a different goal and did not need to work with those two groups to be successful.

Polly D'Arcy ·

Designers on a large team have to collaborate, share and build on one another's work rather than going off into their own silos, because that is how the org chart gets shipped.

Polly D'Arcy ·

Teams can't be debating in pixels in Figma; most often it is better to bias towards building something to make sure it works.

Polly D'Arcy ·

In a financial product, small UI bugs really do matter because over time they are paper cuts that wear down trust.

Polly D'Arcy ·

A team of all the same people who are good at the same thing is not going to help deliver the best product experience.

Polly D'Arcy ·

It is really dangerous to identify a designer's spike and then put them on a part of the product or a team where they can't lean into it.

Polly D'Arcy ·

Whatever a designer's strengths, AI tools can be used; they are not just for people trying to ship high-quality front-end code to production.

Polly D'Arcy ·

Skipping the thinking and the design altogether with AI tools is kind of dangerous; you have to go slow to go fast.

Polly D'Arcy ·

When everybody can build anything, it becomes super clear who has great judgment and who doesn't, which is why judgment and taste are being talked about as everyone's spike.

Polly D'Arcy ·

Not every team has to have the same composition of people and skill sets.

Polly D'Arcy ·

A designer should start as a generalist so as to identify the spike, the thing they bring to the team that nobody else can.

Polly D'Arcy ·

Showing static Figma work in a portfolio presentation feels unacceptable at this stage; prototypes need to be shown.

Eric Meyer ·

New web graphic technologies tend to be used for flashy, excessive effects at first before settling into more restrained, practical use.

Pedro Pita Barros ·

The SAPO decision to degrade its free email service appears to follow the 'enshittification' pattern of digital platforms.

Jennifer Pahlka ·

As AI reshapes expectations, today's well-designed digital forms will soon feel as inadequate as paper forms once did.

Evan Spiegel ·

The next generation of great consumer companies will be born on new form factors like glasses, the way the last one was born on mobile.

Evan Spiegel ·

A platform of creators and developers is a durable moat, because an ecosystem, unlike a feature, cannot be cloned.

Evan Spiegel ·

If you want to have a good idea you have to have lots of ideas, so the velocity of design work matters more than any one idea.

Evan Spiegel ·

You have to talk to customers and share your ideas as early and as often as possible, even though you need not take their advice.

Evan Spiegel ·

Listening to customers is invaluable, but you should not build the thing they asked you for.

Evan Spiegel ·

Designers, not product managers, should own product direction, and a company can go a long way with designers doing the PM work themselves.

Evan Spiegel ·

Design should be an intentional bottleneck every shipped thing has to pass through, because that is what makes a product cohere.

Evan Spiegel ·

Range, not a distinctive style, is what separates a designer from an artist, so a portfolio where everything looks alike is a bad sign.

Evan Spiegel ·

Preciousness about ideas is what holds designers back; if you are producing a thousand ideas it does not matter that most of them are bad.

Evan Spiegel ·

Listing out your jobs to be done is how a company brings order to AI, because it shows exactly where an agent belongs and what it should improve.

Sasha Chapin ·

A fulfilling 'do nothing' meditation practice requires equally embracing all present sensation rather than avoiding difficult ones.

Phil Hagelberg ·

Replacing a monopolistic platform with a single alternative isn't enough — real resilience requires diversity among competing platforms.

Chris Wellons ·

How easily an AI agent navigates a UI is objective evidence of how well- or poorly-designed that UI is.

Phil Hagelberg ·

The mass adoption of React and single-page-app architecture across nearly all web apps was driven by trend-following rather than technical fit, producing fragile, scripting-dependent software.

Ben Kuhn ·

Most forms of dance require weeks of lessons before newcomers are welcomed to dance socially.

Mark Nottingham ·

Web browsers are a good model for what user agency in software should look like.

Brian Potter ·

It has virtually never got cheaper to build a house or an office: construction costs rise at or above general inflation, in every rich country and over a century of data.

swyx ·

The people who run the world are not smarter than anyone else, and even well-meaning ones often only offer paths not worth pursuing.

David Crawshaw ·

VMs are the wrong cloud abstraction because they bundle CPU and memory into fixed-size units.

David Crawshaw ·

Cloud abstraction problems cannot be fixed by adding more abstraction layers on top.

Martin Kleppmann ·

Building databases on object stores is a genuine architectural break, because replication moves out of the database and into the storage layer.

Martin Kleppmann ·

An engineer writing business logic loses nothing by not understanding the layer underneath, exactly as they lose nothing by not managing memory.

Martin Kleppmann ·

Higher-level abstractions do not abolish low-level knowledge; they concentrate it in the smaller group of people who build the abstractions.

Martin Kleppmann ·

Knowing a little about how a storage engine works internally is a superpower for an application developer who will never build one.

Martin Kleppmann ·

MapReduce is dead: nobody uses it any more, and it survives in the literature only as a teaching device for its successors.

Martin Kleppmann ·

Power over data belongs with end users rather than cloud operators, and software should be built to move it back to them.

Martin Kleppmann ·

People in industry routinely short-circuit their reasoning, adopting what they heard in a conference talk instead of thinking it through from first principles.

David Heinemeier Hansson ·

Computers are delightful environments for play, learning, and connection

Brandon Jacoby ·

Patterns that work show up in 90% of products, and they are there for a reason, yet a lot of designers reinvent the wheel.

Brandon Jacoby ·

Knowing when to reinvent the wheel and when to follow patterns is the number one trait a junior designer can have right now and will let them go very far.

Brandon Jacoby ·

Great design has always been great design regardless of the tools; design has always been about decisiveness and judgment.

Brandon Jacoby ·

Visual design fundamentals change how someone feels when they interact with a product, and those principles are still true.

Brandon Jacoby ·

Much of design has become homogeneous, with designers designing for other designers.

Brandon Jacoby ·

Taste is much more about the journey of design than about the result.

Brandon Jacoby ·

Knowing when to go deep into many prompts and when not to fight every battle, because not everything needs to be new, is the zooming in and out that creates taste.

Brandon Jacoby ·

Software is a form of hospitality, welcoming a user base into a product.

Brandon Jacoby ·

What a designer can learn outside their craft may propel their career forward much more than a design crit with someone at a company they look up to.

Brandon Jacoby ·

Craft and consequence can exist at the same time; they are not mutually exclusive.

Brandon Jacoby ·

The earlier the stage of the company, the more important it is to get along well with the team outside of design.

Brandon Jacoby ·

The design industry is shifting to be more and more filled with independent people.

Brad Frost ·

New AI tooling can facilitate genuine human collaboration.

Saul Griffith ·

The transition to heavy industry electrification requires new processes, equipment, and supply chains rather than just cheap electricity.

Phil Gaimon ·

The bike industry will need to separate designs for professionals and consumers to avoid losing customers.

Kyle Kingsbury ·

Relying on AI assistance to do a task reduces a person's performance and persistence, and denies them the muscle memory and deep understanding that comes from working through it by hand.

Marty Cagan ·

The real bottleneck in product work is discovering a solution worth building, not building it.

Kyle Kingsbury ·

Refusing to use or build AI tools will not stop AI development altogether, but it will slow it down, which is a good thing.

Mr. Money Mustache ·

The Social Security benefit formula is designed to be progressive, giving proportionally larger benefits to lower earners than to higher earners.

Sayash Kapoor ·

Average performance on a benchmark suite is very different from the upper bound of what an AI agent can do.

Luke Wroblewski ·

Without the handoff, everyone builds: designer, developer and product manager can all work in the codebase at once, and that changes everything.

Kyle Kingsbury ·

The unpredictable nature of large language model output makes quality-control processes necessary.

David Heinemeier Hansson ·

The agentic age needs a new operating system

Elizabeth Van Nostrand ·

Evangelizing religious movements are shaped by a creative-destruction dynamic that discards congregations that fail to grow.

Pedro Pita Barros ·

Allowing different modes of work organization in primary health care could help expand coverage by family doctors.

Jensen Huang ·

Nvidia's suppliers make their investments for Nvidia and nobody else because only Nvidia's downstream demand can absorb what they build.

Jensen Huang ·

Nvidia did not build a tensor processing unit — it built accelerated computing, whose market reach no ASIC can match.

Jensen Huang ·

A homebuilt accelerator forces you to be your own operator; Nvidia systems are designed for anyone to operate, which is why they are in every cloud.

Jensen Huang ·

Building an ASIC is not the hard part; building something better than Nvidia is, and that is not easy.

Jensen Huang ·

Moving to a custom ASIC does not escape the margin — the ASIC vendor’s margins are nearly as high as Nvidia’s.

Jensen Huang ·

Nvidia should build only what would not otherwise exist — twenty years of CUDA at a loss is the test, and the world already has plenty of clouds.

Julie Zhuo ·

Excellence in design is not just craft quality; it is whether a designer is both willing and allowed to create impact.

Brian Lovin ·

As soon as you realize you can't design half of this stuff in Figma, what you are really designing is the harness for the agent to do longer things and verify its own work.

Julie Zhuo ·

Okay design improves what is put in front of it; excellent design changes what is possible.

Brian Lovin ·

Designers should be designing right at the boundary of the current model, hoping the next one solves it; understanding what is possible and pushing the edges even if you get stuck is probably a good place to be right now.

Brian Lovin ·

People trying to use AI to automate knowledge work or simplify their job are very early adopters, and their tolerance for sub-par experiences is quite high.

Brian Lovin ·

Depending on the work it is totally fine to design predominantly in Figma, but the barrier to trying making something in code is so low that everybody has probably dabbled with it.

Brian Lovin ·

It is getting very easy to move between disciplines, and the person to be is one who can move fluidly between types of working instead of getting caught up in the box they are in at a company.

Brian Lovin ·

A PM who can design around their spec to see if it holds up to reality is what you want, and so is an engineer who really cares about building great visual component systems.

Brian Lovin ·

Someone newer to the field will probably be fine by focusing on making things for real people that solve a real problem, iterating quickly, and learning to do that faster, without worrying about not knowing how to code or write a PRD.

Steve Kaufmann ·

A person who studies a language for years and cannot speak it was taught the wrong method; it is not a failure of talent or of effort.

Maggie Appleton ·

More individual output does not solve problems that need a team to communicate and coordinate; it makes them worse.

Steve Kaufmann ·

Learning a language happens through exposure to language you can mostly understand; classroom drills feel productive and build nothing.

Maggie Appleton ·

With implementation cheap, deciding what to build is the hard question in software, not how to build it.

Colin Percival ·

Eventually Known Consistency is a better theoretical model for distributed systems than plain eventual consistency, even though it's harder to reason about.

Colin Percival ·

Exceptions are an inherently poor way of handling errors because they make bugs harder to spot on casual code inspection.

Noah Smith ·

Assimilation should never require an immigrant to give up their religion or their heritage.

Noah Smith ·

Shared everyday experience binds a nation together better than shared ancestry does.

David Heinemeier Hansson ·

Rails is having a renaissance because it is one of the most token-efficient ways to build a web app, which suits agents, for as long as that lasts.

Ian Silber ·

Working at OpenAI as a designer does not require being technical, but it does require being really curious and interested in how somebody might actually use the product.

Anne-Laure Le Cunff ·

A feeling that information is easy to process gets mistaken for genuine understanding of it.

David Heinemeier Hansson ·

Build the product for yourself first: if it truly serves you, plenty of other people like you will want it too.

Ian Silber ·

In AI products, much can often be done with the model directly in the conversation rather than through new bespoke UI, though there is a balance.

David Heinemeier Hansson ·

For a web developer Linux is simply better than the Mac: closer to what you deploy on, with phenomenal tooling and a reproducible setup.

Ian Silber ·

Designers working on AI products should hopefully spend a lot less time drawing pixels and more time on how people interact with the model, which is the core product.

David Heinemeier Hansson ·

Designers should find what the spec should be and build it in HTML and CSS themselves, not make a product manager's spec look pretty.

Ian Silber ·

The best systems thinkers think not just about their feature but about how it extends the system.

Ian Silber ·

Sometimes a whiteboard, a paper sketch, a wireframe, Figma or a live prototype is the right tool, and the best designers will intuit which to go to.

Ian Silber ·

It is easy to get obsessed with a single prototype whose idea might be way off, so you do not want to take just one.

Ian Silber ·

Design and engineering are swinging back toward overlapping much more.

Ian Silber ·

In some ways nothing has changed about what a great designer is: thinking deeply about the problem and who it is for, plus craft, taste and understanding of interaction patterns remain just as important.

Ian Silber ·

Designers who embraced a tool like Origami were able to up-level their work, craft and impact, and Codex or Cursor can likewise unlock a new way for designers to express themselves.

David Heinemeier Hansson ·

The end state is agents that need none of our accommodations, no on-ramp, CLI or MCP; the CLI being built now is for today, not for AGI.

David Heinemeier Hansson ·

The bulk of agent acceleration is accruing to senior programmers, because judging whether agent output can ship is what made them senior.

Ian Silber ·

Poor systems thinking is being too blinders-on about getting one thing out as quickly as possible, without understanding that other, similar things are going on.

David Heinemeier Hansson ·

Agents will eventually stop making the mistakes that require senior review, as self-driving cars did; the bet is on when, not whether.

Ian Silber ·

Because anybody can write software, a designer's job will be more and more about helping edit, direct and curate.

Ian Silber ·

The job of a designer is not going away anytime soon; designers will probably be able to do more, and the core skills stay just as important, because people can recognize good software from bad.

Ian Silber ·

Designers have to think a lot about how to expose what the models can now do in a way people understand, and how to give them the tools to do real, meaningful work.

Ian Silber ·

Designers have to deal with the realities of what the technology is now, because going too far ahead is like trying to put 11 players on the field.

Ian Silber ·

Great designers can both package up a new technology and push on what it ought to be able to do.

Ian Silber ·

Designers need the fundamentals, being great at general product design, plus being truly curious and interested in the technology.

Luke Wroblewski ·

Designers should code, and AI coding agents have reopened that door after a decade in which front-end complexity closed it.

Kent Beck ·

For twenty-five years the senior people had the answers to hand, and what has changed is that at this moment nobody knows the answer to anything.

Chris Wellons ·

AI tools are now highly effective at generating UI code from a rough natural-language description

Annie Wright ·

EMDR targets specific memories well but is not built for identity work or building a life narrative.

Luke Wroblewski ·

The loop has flipped: it used to be faster to mock up software than to ship it, and now designers can work iteratively in production code.

Chris Wellons ·

Frontier AI models still cannot handle certain deep technical implementation details without human guidance

Jamie Brandon ·

Markets only respond to immediate, severe harm, not gradual or delayed harms from industries left to regulate themselves.

Jamie Brandon ·

Formalizing math is a translation task, so AI will boost top mathematicians' productivity more than top writers'.

Martin Fowler ·

Skepticism about a new technology only works if it is total enough to be turned on itself, which is what forces you to stay curious.

Michael Nielsen ·

AI has moved the programmer's bottleneck from writing code to having good design ideas, and there is no verification loop for a design idea.

Kent Beck ·

This is the golden age of the junior programmer, because AI is an amplifier and the people learning quickly get amplified most.

Martin Fowler ·

Predictions that nobody will write code soon dodge the question of what code is: prompting a model is still producing something we will have to read and change.

Michael Nielsen ·

We are near the bottom of the tech tree, not the top; even the list of phases of matter keeps growing and we are starting to design new ones.

Kent Beck ·

Agents are driving a re-soloing of programming, undoing the safe social environment that extreme programming built for people who are not naturally social.

Michael Nielsen ·

Most of the tech tree will never be explored by anyone at all, which is exactly why the choice of which direction to explore matters so much.

Kent Beck ·

Slow models are better company than fast ones, because the minutes an agent spends working are the minutes two humans get to talk about the design.

Annie Wright ·

Schema therapy was designed for entrenched patterns that do not respond well to traditional CBT.

Jessamyn West ·

A product's limitations are often deliberate choices rather than genuine engineering constraints.

Arvid Kahl ·

The revenue level at which publicly sharing business details invites cloning has fallen to essentially zero because of AI coding agents.

Arvid Kahl ·

Product quality and software engineering ability no longer function as a competitive moat because AI can replicate them quickly.

Neha Narula ·

Bitcoin's quantum risk is high enough to warrant designing and evaluating post-quantum signature schemes now

Marvin Schwaibold ·

Motion can hint at the hierarchy of the content you are looking at, for example a sheet overlaying a page tells the user that the content lies on top.

Marvin Schwaibold ·

A designer's assumption about where a component belongs is a personal assumption and does not mean that the market or the users will react to the component that way.

Marvin Schwaibold ·

Coding tools are unlocking designers a ton and creating much more output, which creates more reactions and accumulates to better output for everyone.

Marvin Schwaibold ·

Don't be afraid of AI tools: as you start using them and build awareness of how they work, it is a lot easier to understand them, trust them and work better with them.

Marvin Schwaibold ·

Creatives were measured by their ability to understand the tools, for better or worse, rather than by the idea behind why they were using the tool.

Marvin Schwaibold ·

The design process for everyone is becoming way more contextual, whereas before everybody had to open the tool in the same way because of how the tool was built.

Marvin Schwaibold ·

People creating their own workflows, building a set of tools and stringing them together to get to their output, is as creative as it gets.

Marvin Schwaibold ·

Graphic design is learned by copying graphic design; the best thing a younger designer can do on seeing something beautiful is try to recreate it.

Marvin Schwaibold ·

Design differentiation in the future means more of a malleable interface that morphs and shapes itself around the contextual needs of merchants, rather than one specific direction.

Marvin Schwaibold ·

The branding, animations, logo and micro interactions that make a tool feel beautiful to interact with are what give it soul, meaning and excitement, and that craft has to be married to solving the problem.

Emanuele Bartolesi ·

Optimizing an AI assistant to feel friendly is the wrong goal when the point is building real, production systems.

Thuan Pham ·

Today's AI boom will end the way the dot-com boom did — consolidation, a wave of failures and a few genuinely great companies — and customers, not argument, will decide which is which.

Marco Arment ·

A company that prioritizes building great products with great user experiences above all else will naturally see profit, market share, and broader impact follow

Daniel Jalkut ·

A programmer who embraces AI while remaining skeptical of everything it produces is better equipped than any comparably skilled human in the history of programming.

Thuan Pham ·

A career is not built by searching for jobs but by accumulating a reputation slowly, so that when you become available the opportunities come to you.

Thuan Pham ·

A technical leader should let engineers find the flaw in their own design through leading questions rather than announce it, because a discovered flaw is acted on and a reported one is argued with.

Thuan Pham ·

The right way to specify a rescue rewrite is a couple of scaling constraints and no new features at all, so that hardware can absorb the growth the team has no time to design for.

Thuan Pham ·

How big a system will eventually have to be is not worth speculating about; the only useful question is how long the one you have has left before it hits the wall.

Thuan Pham ·

Repeated rewrites are the correct answer under fast growth, because a system specified to scale forever would take so long to build that the company would die before it shipped.

Thuan Pham ·

Uber's thousands of microservices were not an architectural preference but a survival decision: under that growth only independent, unblocked teams could build fast enough to stay alive.

Thuan Pham ·

Directing a swarm of coding agents is a different, multi-threaded way of thinking that costs more cognitive load, and the engineers who make the switch have doubled their output.

Thuan Pham ·

AI has already made sweeping mechanical changes cheap; the unsolved problem is getting the same productivity when adding features to an old, entangled codebase.

Thuan Pham ·

AI tooling has not compressed the gap between great and average engineers: it is still two or three times, and for the same reasons as before — curiosity and appetite for the bleeding edge.

March 2026

Michael Levin ·

Physicalism is insufficient and requires investigating a latent space of patterns.

Chip Huyen ·

Because anyone can now rebuild whatever you make, the ordinary incentive to build software has been knocked out from under us.

Eliot Peper ·

Making something singular, rather than chasing scale, is a powerful differentiator.

Chip Huyen ·

Anything a person can describe, AI can now produce — software has reached the point image generation reached with style.

Chip Huyen ·

Every product put into the world removes the need for imagination, because it becomes something anyone can point AI at and copy.

Chris Wellons ·

AI will make hand-written code a luxury only the wealthy can afford to produce.

Chip Huyen ·

The reason to keep building is that building is practice at problem-solving, and the supply of problems will never run out.

Chip Huyen ·

The problems worth a small builder's time are the ones too small for a big AI company to be motivated to solve.

Chip Huyen ·

We are still retrofitting the tools we already had onto workflows we have only imagined, instead of designing for how people will actually work with AI.

Chip Huyen ·

The frightening part of giving agents more access is not bad code but irreversible action, so the guardrails to build next are guardrails on reversibility.

Chip Huyen ·

The hard question is no longer how to build something but what to build, because anything already described can be replicated.

Paul Millerd ·

Work has become more enjoyable for the millennial generation, even as the belief that meaningful work requires suffering has gone largely unquestioned.

Andrej Karpathy ·

Because an LLM can argue almost any direction competently, the right way to use one for forming an opinion is to make it argue every side.

Simon Willison ·

Claude has surprisingly good design taste for SwiftUI applications.

Anil Dash ·

Closing off interoperability is how a platform locks people in, and that lock-in is what lets the service be made worse without losing them.

Jakob Nielsen ·

The winning system of the next decade will be the one that best understands the user's job to be done and shows what it is about to do, not the one with the prettiest buttons or fewest screens.

Kyla Scanlon ·

The current drive for control, including gambling and speculation, is functionally a response to financial nihilism: people no longer believing the economy works for them.

Jakob Nielsen ·

Companies pay for design so that customers buy more and look at more ads; internet-era UX was manipulation, whatever designers set out to do.

Kyla Scanlon ·

The optimization economy cannot deliver control, because it depends on people's desperation and chasing control through optimization is itself a loss of control.

Mark Nottingham ·

Grounding a chatbot in a specific set of documents allows a genuinely new way of interacting with them.

Flora Guo ·

For any designer who is building projects, this is an incredible time to share with the internet online.

Flora Guo ·

Design engineering goes much deeper than the surface of how something looks; it is having the context and ability to translate an intention into what the user actually experiences.

Flora Guo ·

Making for the love of making rather than for a finish line is inspiring, and design is a practice that never ends.

Flora Guo ·

With AI, work is moving up the layers of abstraction, but the more context someone has about how things work under the hood, the better equipped they are to debug and build resilient things.

Flora Guo ·

People should use AI to scaffold their own learning and then flesh it out by building projects.

Flora Guo ·

A team should be aligned at a conceptual level before moving into visual experiments and prototypes, because a trap for designers is coming up with a bunch of options and picking one too early.

Flora Guo ·

Staying up to date works by staying broad, building out whatever really grabs attention, and then repeating that cycle of going broad and then going deep.

Flora Guo ·

Rather than having AI just do the task, people should use it to help build the right mental models.

Andrej Karpathy ·

LLM memory as currently built is a distraction to the model: one old question keeps resurfacing as if it were a lasting interest.

Freddie deBoer ·

Education cannot be done to a person: without at least a minimal act of will from the learner, no method, resource or reform works.

Nelson Elhage ·

A composable concurrency paradigm requires a general-purpose cancellation mechanism.

Jasmine Sun ·

Good writing requires grounding in real-life experience.

Jensen Huang ·

Data centres should be built to degrade gracefully when the grid needs its power back, instead of demanding perfect uptime from utilities that have idle capacity most of the year.

Jensen Huang ·

Asked when an AI could build a billion-dollar business on its own: AGI has already arrived.

Jensen Huang ·

Coding is now specification, and the number of people who can do it just went from thirty million to a billion.

Nicole Forsgren ·

Release management is human group sense-making, and it does not scale to the volume of change AI now produces.

Nicole Forsgren ·

The processes organisations build to make work uniform are usually the same processes that slow the work down.

Nicole Forsgren ·

Fast feedback is no longer straightforwardly good: agent-speed loops make you rebuild your mental model dozens of times in half an hour.

Nicole Forsgren ·

You cannot mandate a tool developers dislike — they will quietly spin up the one they wanted.

Andrej Karpathy ·

The way software gets built flipped in December: writing code yourself is now the exception, not the default.

Elio Struyf ·

Reviewing an AI coding agent's changes afterward in version-control diffs is a more natural workflow than approving each change through a GUI.

Jason Fried ·

Even as AI makes building software easy, people will still prefer to buy some software off-the-shelf rather than build their own.

Elio Struyf ·

For heavy, repo-wide AI-agent work a CLI-based tool is better suited than a chat-based interface, which remains preferable for small, single-file tasks.

Arvid Kahl ·

The durable competitive moat in an AI-driven business is proprietary human-originated data, not the AI itself.

Jason Fried ·

Giving people access to tools for building software does not make most of them software builders.

Terence Tao ·

We are living through a cognitive Copernican revolution: human intelligence is not the centre, and which tasks we count as requiring intelligence has to be re-evaluated.

Andrej Karpathy ·

The apps built to drive smart devices should not exist at all: the devices should expose APIs and agents should call them directly.

Andrej Karpathy ·

Reinforcement learning leaves models sharply jagged: on the rails of a verifiable domain they are superintelligent, off them everything meanders.

Terence Tao ·

What AI still cannot do in mathematics is build cumulatively on partial progress; it is trial, error and brute force rather than an idea improving through use.

Terence Tao ·

A model that solves a hard problem has learned nothing from it: the next session has forgotten it, with no new skill to carry to related problems.

Terence Tao ·

If the Riemann hypothesis turned out to be false, prime-based cryptography would be abandoned very quickly, because one unknown pattern implies more.

Terence Tao ·

Hybrid human-plus-AI teams will dominate mathematics for a lot longer, and a full replacement would need breakthroughs beyond anything we now have.

Terence Tao ·

Contributing to frontier mathematics no longer requires years of PhD training; with AI tools and Lean a high schooler can make a real contribution.

Jakob Nielsen ·

The average user is a dangerous statistical artifact: design for the gap between the median user and the heavy-use tail, not for a phantom mean.

Rakhim Davletkaliyev ·

In a tabbed interface, the currently active tab should appear visually connected to its content, while inactive tabs can remain visually separate.

Rakhim Davletkaliyev ·

UI design trends are cyclical, which is both good and bad: they always come and go.

Kris Puckett ·

There are massive unlocks for AI around workflows, content design, organization and analysis, beyond generating a product, app or website.

Simon Willison ·

A passing test suite is not evidence that the software works, so an agent must also be made to start the thing and exercise it the way a person would.

Simon Willison ·

A language-agnostic conformance suite is the most powerful thing you can hand a coding agent, because the whole instruction becomes: write code until these tests pass.

Kris Puckett ·

Because LLMs are pattern matchers, they are fantastic at quickly spotting patterns that people miss, and can act as a mirror showing what a person is missing about themselves and their work.

Kris Puckett ·

LLMs probably do not do creative exploration very well by default, but they can be taught divergent thinking very quickly.

Simon Willison ·

Language models are gullible by design — they do what they are told and believe nearly anything said to them — and that, not a bug, is what makes software built on them attackable.

Simon Willison ·

Prompt injection has no equivalent of parameterized queries: there is no reliable way to tell a language model which text is data and which is instructions.

Kris Puckett ·

Being AI native has to go beyond making a cool shader or a dither effect that every other person is making.

Kris Puckett ·

MCPs are not going anywhere, but they will change, and more companies will lean toward better semantic APIs and CLIs than MCPs.

Kris Puckett ·

An MCP fills up an AI's context window very quickly.

Kris Puckett ·

Products should be designed with agents in mind, because a lot more agents will be doing the work for people and designers have to think that through.

Kris Puckett ·

Designers looking for a role need to demonstrate that they understand AI systems and are actually building something with them.

Kris Puckett ·

A portfolio on a basic Squarespace template will probably get passed on even if the content is really good.

Kris Puckett ·

The bar is so high that designers who are not using AI tools to differentiate themselves will get left behind.

Tynan ·

Shore excursions booked onboard a cruise ship are generally low quality and overpriced.

Tynan ·

Some of the best things to see near a city are a short drive or train ride outside it, not in the city center itself.

Joe Posnanski ·

Even if MLB's season format looks archaic, it is working for a lot of people.

Alex Epstein ·

Any energy source that is genuinely cost-effective will thrive without subsidies.

Kent C. Dodds ·

AI coding agents have made redesigning existing systems cheap enough that developers actually redesign instead of living with a flawed design.

swyx ·

Real leverage in a negotiation comes from being genuinely willing to walk away from it.

Dylan Patel ·

An H100 is worth more today than it was when it was new, because the models it runs got better faster than its replacements arrived.

Phil Gaimon ·

George Hincapie is one of the least suitable people to run a professional cycling team.

Arvid Kahl ·

Building software products API-first, with near-parity between UI and API capabilities, is a smart strategic move today.

Dylan Patel ·

Hundreds of billions of dollars of AI value rest on about a billion dollars of lithography tooling that cannot be scaled quickly.

Dylan Patel ·

The semiconductor supply chain will underbuild not because it cannot expand but because nobody in it believes the demand is real.

Dylan Patel ·

China will have a working EUV machine long before it can build them in quantity, because production hell takes years after the lab result.

Dylan Patel ·

The reason we cannot simply make more memory is that there is nowhere to put the tools, and a fab takes two years to build.

Dylan Patel ·

Elon Musk can probably build the cleanroom for a million-wafer fab, but he cannot develop the process technology quickly, because that knowledge is accumulated rather than invented.

Dylan Patel ·

Labour is a huge constraint on the AI build-out, and the way round it is to build data centres as modules in Asian factories rather than on site.

Dylan Patel ·

Getting a data centre permitted in the empty middle of America is far easier than putting one in orbit.

Dylan Patel ·

A lab should spend most of its compute on research rather than on building the next model, because research is where the tenfold yearly efficiency gains come from.

Rakhim Davletkaliyev ·

macOS's consistent use of the cmd key for most operations, with opt for a secondary options layer, is one of its best usability features.

Josh Puckett ·

When exploring, you want to feel in reality as quickly as possible what the result of your choices is.

Rakhim Davletkaliyev ·

Windows Explorer is a better file manager overall than macOS's Finder, being easier to navigate, create files in, and switch views with.

Rakhim Davletkaliyev ·

Windows 11 is not a high-quality product, despite having genuinely good aspects.

Josh Puckett ·

A great restaurant may be the purest form of design: it solves a problem, feeding people, while wrapping it in an experience that feels unique and differentiated.

Josh Puckett ·

So much of design is iteration and discovery: stumbling on something better by sticking with an intent for one more iteration.

Josh Puckett ·

Making something as well as you can and as simply as you can sums up much of what great design is.

Josh Puckett ·

Practicing restraint is becoming more important as many different things can be done quickly.

Josh Puckett ·

It is probably hard for a designer to edit their work too much, even though a sauce can be reduced too far.

Josh Puckett ·

What makes everything considered great design great is that it is distilled and concentrated into what makes it great, with the nice-to-haves cut.

Josh Puckett ·

Designers should fight for what they think is right rather than trust anyone else, because trusting others will most likely dilute the idea and make it worse.

Josh Puckett ·

Details that show care build trust and perceived value, and affinity for a product increases when it works a little better or differently than expected, in any kind of product.

Josh Puckett ·

AI compresses a designer's ability to reach for ideas that used to be far out of grasp.

Josh Puckett ·

When implementation gets faster, designers are uniquely valuable for what they do with the extra time, such as pushing toward the ultimate version of an experience.

Josh Puckett ·

Building intuition for what works and what does not comes from taking a lot of shots at the goal.

Josh Puckett ·

Figma is still important, but using it to dial in full fidelity is probably the wrong use, because it is not the actual code base.

Lyn Alden ·

Fractional reserve banks are structurally interconnected and unstable in a way most other industries are not.

Sriram Krishnan ·

Trying to make complex systems fully legible carries real downsides.

Mari Andrew ·

A skill goes to waste only when insecurity or outside pressure stops someone from pursuing it, not merely by going unused.

Kent C. Dodds ·

Architectural iteration is slow because a design usually has to be built before its quality can be judged.

Anne-Laure Le Cunff ·

Instinct is an evolved, biological response for survival, while intuition is pattern recognition learned through experience.

Anne-Laure Le Cunff ·

Intuition works by the brain rapidly matching current cues to patterns learned through past experience.

Harper Reed ·

The best way to build an AI agent skill is to do the work manually first, then have the AI turn that work into a reusable skill.

Kent C. Dodds ·

A new infrastructure primitive delivers value only when it changes the shape of what is being built.

Jeff Kaplan ·

If you recruited the best person in the industry, you owe their suggestions the treatment the best in the industry deserves rather than starting from doubt.

Jeff Kaplan ·

A creative lead's job is to take the idea they think is wrong and try to make it work, because the best ideas come out of developing someone else's.

Jeff Kaplan ·

You design along the path of least resistance: players will take the cheapest route to their goal, so the designer's job is to put the experience they want played on that route.

Steve Yegge ·

Coding agents demand more reading than most people can do, and until better interfaces arrive that leaves those people at a severe disadvantage.

Jeff Kaplan ·

Ask gamers what kind of player they are and they describe the player they wish they were, so a designer has to watch what people do rather than believe what they say.

Jeff Kaplan ·

You cannot be an exceptional game designer without playing enormously, and a bad game teaches as much as a great one.

Steve Yegge ·

Agents can build about half a million lines before the codebase dissolves into a mess, and the next model will push that to a few million.

Jeff Kaplan ·

Game design has no textbook, so a designer has to keep their own list of the elements of fun the way a writer keeps the elements of fiction.

Jeff Kaplan ·

Ant-farm design — building something to watch what the players will do — is the wrong way to design a multiplayer game.

Jeff Kaplan ·

Social media is built so that maximum hyperbole wins, and the negative direction pays better than the positive.

Steve Yegge ·

Google has produced no innovation since 2008; everything it has shipped since has been acquisitions.

Jeff Kaplan ·

Weathering a firehose of criticism is now part of a creator's skillset, and it is quietly removing creators whose work the world will therefore never see.

Jeff Kaplan ·

Overwatch put too much weight on whether the team won; a hero shooter built today should foreground individual contribution, because that is how people actually play.

Steve Yegge ·

Nobody will believe a resume any more; what will count is visible proof of the work you actually did.

Jeff Kaplan ·

Almost no game ever catches the wave, so a game that has caught one should be ridden to the end rather than paused to build its sequel.

Jeff Kaplan ·

You do not need a CFO to make World of Warcraft — you need artists, engineers, designers, producers and an audio team.

Jeff Kaplan ·

No creative work should ever be used by AI without its maker's permission; doing it is no different from stealing.

Jeff Kaplan ·

Small studios are the future of gaming: the genuinely new ideas come out of them, and the big studios acquire rather than generate them.

Luis Ouriach ·

The importance of the docs written for design systems, in whatever form, is going to be the foundation of everyone's workflow going forward, and guidelines are becoming paramount.

Luis Ouriach ·

Maybe higher-fidelity prototypes can be tested and validated much faster and still thrown away, because testing an idea is often the point of a prototype.

Luis Ouriach ·

Teams should hang out together more and build rituals for collaborating, instead of each role owning its own artifact.

Luis Ouriach ·

A design system is a bar of quality in an organization, and that is everyone's job.

Luis Ouriach ·

Even when a product manager can contribute to prototypes, the designer is still doing something specific to a designer's role.

Luis Ouriach ·

Relying on code generation without a very strict indication of brand feeling ends with everything looking the same.

Luis Ouriach ·

Taking the label "design system" off the work is going to help bring it out of the silo and to the center of an organization's change.

Luis Ouriach ·

The gap between enterprise and startup designers may be widening, because access to tools differs so much with a company's security standards.

Luis Ouriach ·

Whatever the speed, quality cannot be skipped, and community, people, users and passion remain incredibly important to the success of an idea.

Luis Ouriach ·

The canvas is very important for ideation, collaboration and direction exploration, while the browser is where an idea is committed to.

Luis Ouriach ·

Designers are the experts in the seat to bring something to market, and outsourcing ideas, output and definitions to a tool raises the question of what one really wants to be doing.

Luis Ouriach ·

With so much software on the market, the focus has to be on sales, revenue, users and sign-ups, and the pixels are secondary to the success of the business.

Rui Carmo ·

Certain programming languages are better suited for agents because strong types and annotations help models understand intent.

Gil Kalai ·

Nature may not permit the quadratic suppression of error rates when entanglement is generated indirectly.

Peter Kalmus ·

The American two-party system is designed to suppress the left and maintain the status quo for billionaires.

Justin Jackson ·

As more people build software with faster cycle times, they will learn lessons faster that once took colleagues decades to acquire.

Boris Cherny ·

Working prototypes have replaced product requirement documents.

Boris Cherny ·

The scarce skill is now context switching across parallel agents, not sustained deep work.

Sanjay Bakshi ·

Retained corporate earnings do not automatically translate into capital appreciation equal to the earnings retained.

Nikita Prokopov ·

Native app UI is no longer more consistent or polished than the web, and may be worse.

Nikita Prokopov ·

Bloated, slow software is caused by a lack of engineering care, not by the choice of technology stack.

Steven Johnson ·

Archaeological interpretations of human history are biased toward artifacts made of durable materials that survived, not what mattered most to the people who made them.

Steven Johnson ·

The Stone Age, Bronze Age and Iron Age framework names historical eras after the inorganic materials that happened to survive, not what those societies actually valued.

Eliot Peper ·

Every manufactured object embodies a long chain of decisions made by many different people.

February 2026

Ruxandra Teslo ·

Better drug candidates do not shorten clinical trials, because a trial's duration is set by operational and regulatory friction rather than by how good the drug is.

Ruxandra Teslo ·

A clinical trial is not only a test of a drug: early trials are experiments whose real product is the human biology data everything downstream is built on.

Ruxandra Teslo ·

How long a trial takes decides which diseases get drugs at all, because capital flows to the indications whose endpoints can be measured quickly.

Andrew Huberman ·

Most melatonin supplements contain doses far above what the body makes, which is a reason to be wary of them.

Andrew Huberman ·

Getting sunlight through a car windshield or a window, or behind highly reflective sunglasses, will not deliver the UVB-driven effects.

Kelsey Piper ·

The government's actions may cause AI companies to be more reluctant to engage with the government and more inclined to work in secrecy.

Boris Cherny ·

Build for the model six months from now, not the model of today.

Josh Brown ·

Stocks of companies in the physical world are currently the best place to be while the disruption question hangs over software and services.

Anne-Laure Le Cunff ·

When starting something requires little sacrifice, it stops functioning as a genuine decision.

Naval Ravikant ·

All abstractions are leaky, so code an AI writes for you will have mistakes.

Naval Ravikant ·

What distinguishes entrepreneurs from everyone else in the economy is extreme agency.

Alex Epstein ·

The cost of making solar and wind reliable through overbuilding and multi-day storage would be 10-20 times the cost of natural gas power.

Luke Wroblewski ·

Users split into two camps on seeing an agent's work, and the split has not faded as trust builds, so an interface has to balance both.

Rakhim Davletkaliyev ·

Modern interfaces look similar across companies and industries because designers believe good UI should be invisible.

Rakhim Davletkaliyev ·

Designers pursuing invisibility in interfaces end up creating unintuitive, poorly-structured designs.

Greg Kroah-Hartman ·

Any group attempting to assign severity scores to a Linux CVE without knowing the specific use case is lying.

Peter Steinberger ·

Turning a breakout open-source project into a company would be the wrong move -- joining a lab reaches everyone faster than building another large company.

Arvid Kahl ·

GPT-based AI systems can now reliably automate around 80% of data acquisition and verification work.

Dario Amodei ·

The one genuine uncertainty left about AI capability is tasks nobody can verify — planning a Mars mission, discovering CRISPR, writing a novel.

Dario Amodei ·

Models are far more differentiated from one another than clouds are, so AI will not commoditise the way cloud did.

Dario Amodei ·

Robotics will be revolutionised a year or two behind the country of geniuses, and getting there does not require human-like learning.

Dario Amodei ·

In an AI world growth and economic value come easily; distribution, wealth and political freedom are the hard parts, and that is where policy belongs.

Austin Vernon ·

Reviving American manufacturing needs a better business model rather than a planned economy.

Alex Tabarrok ·

Under binding ceilings, cost-minimizing suppliers drive allocations to corners rather than Harberger or random benchmarks, producing larger losses and discontinuous welfare jumps.

Lex Fridman ·

Fearmongering about AI destroys the possibility of building something good with it, so the line to walk is serious concern without fear.

Peter Steinberger ·

Working with coding agents is a skill you learn like an instrument; judging the tools after one try is like calling a piano bad after one go.

Austin Vernon ·

The right tariff rate for a country trying to rebuild its manufacturing is zero.

Peter Steinberger ·

He builds the code base to be easy for an agent to navigate, not to be perfect for himself, down to keeping the names the model picks.

Peter Steinberger ·

Orchestrators that try to automate the whole build miss style, love and the human touch, and that cannot be automated away yet.

Peter Steinberger ·

Every tool is better as a CLI than as an MCP server: models are good at Unix commands, a CLI composes with jq, and MCP output clutters the context.

Peter Steinberger ·

Software developer salaries reached stupidly high levels because of a scarcity of people who could build things, and that scarcity is ending.

Peter Steinberger ·

AI may eventually replace programmers, but the craft of programming will survive the way knitting did, and you were never just a programmer: you are a builder.

Peter Steinberger ·

The water-use objection to AI does not survive arithmetic: skipping one burger a month covers a person's token footprint, and golf uses more water than every data centre together.

Casey Handmer · no longer holds

Manufacturing and mining in space will never be lucrative enough to pay for human settlement of space.

Adam Wiggins ·

Building third-party clients for messaging platforms is a legal gray area and often only partially supported by their APIs

Casey Handmer ·

Earth-launched hardware can support about a hundred gigawatts in orbit; past that, the satellites have to be built off Earth.

Gavin Francis ·

The modern view of the brain as a connectome of circuits and loops will eventually be disproved.

Joe Posnanski ·

Hot takes make people dumber by flattening thought into emotion and reducing complex topics to two screaming sides.

JL Collins ·

The United States' share of global GDP and equity markets will keep shrinking, and that decline is accelerating.

David Crawshaw ·

Most current software and ways of solving problems are structured the wrong way for the agent era.

David Crawshaw ·

The best software design for an AI agent to use is the same as the best design for a human programmer.

Ryan Singer ·

A document meant for both approval and building is better split into two work steps: framing to choose what to work on and win buy-in, and shaping to spell out the mechanics.

Arvid Kahl ·

Claude Code will eventually build persistent, failure-tolerant iteration into its default operating loop.

David Heinemeier Hansson ·

AI agents will be able to interact with the environment without special equipment

Mitchell Hashimoto ·

Chatbot interfaces are the wrong tool for meaningful coding work, because you are mostly hoping the model recalls the right answer and correcting it requires a human in the loop.

Scott Hanselman ·

As AI takes over more of the coding process, clearly specifying goals and exercising good judgment will matter more than ever.

Steven Johnson ·

Standard word processors are poorly suited for complex, long-form writing.

Tynan ·

Watchmakers today would build GPS-synced, solar-charged smartwatches, just as watchmakers in the 1800s used the best technology available to them.

Steven Johnson ·

The purpose of a creative work environment is to produce ideas that would not otherwise have occurred to you.

January 2026

Arvid Kahl ·

Every hobby community has profitable auxiliary niches serving needs its members can't fill themselves.

Laura Deming ·

In cryopreservation you can buy down biological difficulty with better engineering, which is not true of most problems in biology.

Laura Deming ·

Transplant patients regularly lose organs that are already on their way to them, purely because of timing.

Steve Klabnik ·

The key to working effectively with an LLM is giving it a fast, objective way to evaluate its own progress and redirect itself when it's off track.

Jason Cohen ·

Cancellations always overtake marketing, because cancellations grow automatically with the company while marketing grows only as fast as you can improve it.

Jason Cohen ·

Net revenue retention flatters a business, because a percentage lost always needs a larger percentage gained to get back to even.

Adam Wiggins ·

What a message-recipient considers important is too idiosyncratic to be captured by generic heuristics like "all receipts go to FYI."

Gina Trapani ·

Money itself does not build relationships, create memories, bring people closer to loved ones, or express who a person is to the world.

Jan Leike ·

Alignment of today's models is going well enough to look solvable, while aligning models we can no longer understand remains unsolved.

Jan Leike ·

We do not have to align superintelligence directly; we have to build a human-level automated alignment researcher we trust more than ourselves.

Orin Kerr ·

The original remedy for an unlawful government entry was an ordinary trespass suit against the officer, and losing it took the teeth out of the Fourth Amendment.

Marc Littlemore ·

It is much better to own your own content than to rely on third-party services, which come and go, get acquired and sold, or shut down.

Luke Wroblewski ·

Design deliverables are no longer static assets or guidelines but tools: custom software that generates on-brand assets on demand.

Rob Walling ·

Product-market fit is a spectrum rather than a binary state.

Rob Walling ·

Startup success requires a combination of hard work, luck, and skill.

Arvid Kahl ·

Successful founders are distinguished by knowing what to give up and when, not by refusing to ever quit.

Nikita Prokopov ·

Frontend-backend communication should be mediated through a shared database rather than a custom API.

Jason Fried ·

The original Cuisinart, with its two paddle buttons and clear affordances, was peak interface design, and it has been downhill from there.

Luke Wroblewski ·

When applications can generate capabilities on demand, a product's features are what users ask for in the next session, not what shipped in the last release.

Brandon Taylor ·

Justifying cruelty toward another person requires first reducing them to an abstraction rather than a full human being.

Adam Wiggins ·

Manually configured filtering rules mostly fail because a person's preferences change over time and the rules require ongoing upkeep.

Adam Wiggins ·

Social media recommendation algorithms are only partially designed to serve the people using them.

Stephen Wolfram ·

The behavior of very simple computational rules can require large amounts of obscure mathematics to fully capture.

Joseph Politano ·

A labor market stuck in a low-hiring equilibrium will see employment decline outright unless hiring rates recover.

Jensen Huang ·

No company is anywhere near a single God AI, and no researcher has any reasonable ability to build one.

Jensen Huang ·

The first part of safety is that the thing works as advertised, not that nobody can misuse it.

Jensen Huang ·

Consumer AI that is right 90% of the time delights people; industrial AI that is right 90% of the time is 100% dissatisfaction.

Jensen Huang ·

No country in the world contributes more to open source than China, and American startups are built on what it gives away.

Morgan Housel ·

The right response to housing costs is to build millions more homes as fast as possible.

Nikita Prokopov ·

If every element in an interface is marked as distinct, none of them actually stand out.

Nikita Prokopov ·

There is no good icon metaphor for every possible action, no matter how much effort goes into finding one.

Nikita Prokopov ·

Human-interface design principles from decades ago remain valid because human cognition hasn't changed.

Dylan Matthews ·

Major AI-driven economic disruption will require creating new kinds of social insurance and welfare programs to maintain living standards.

Greg Kroah-Hartman ·

The current arrangement of the kernel security team may become the standard for project security teams due to the European Union's CRA law.

Greg Kroah-Hartman ·

The 'no embargo' policy for kernel security is difficult to maintain when dealing with hardware and operating system cross-platform issues.

Rodney Brooks ·

Most of the companies now selling a transport technology will quietly disappear, and the survivors will redefine what they are doing without renaming it.

Dan Wang ·

China trails the United States in AI capability, but only by a small margin rather than years.

Rodney Brooks ·

A robotaxi that still needs remote human help loses the cost advantage that motivated it, which pushes out when it can scale.

Jason Lemkin ·

Almost nobody should build their own AI go-to-market tools; buy them, the way you would buy any other software.

James Clear ·

Habits are built by making a behavior obvious, attractive, easy and satisfying, and broken by inverting each of those four.

Jason Lemkin ·

Recipients do not care whether a sales email was written by an AI, so there is no point disclosing or disguising it.

Jason Lemkin ·

AI will leave the world needing more go-to-market people than ever, because the winners grow faster than the efficiency gains.

Jason Lemkin ·

The go-to-market plays all still work; it is the playbooks built around them that have stopped working.

Dario Amodei ·

Labour markets will eventually recover from AI, but the speed of the transition makes the short term unusually painful compared with past technologies.

c. 2024–2025

Yuval Noah Harari ·

Every powerful information network can do good or ill depending on how it is designed and used

Yuval Noah Harari ·

Self-correcting mechanisms are everywhere in nature; children learn to walk through them

Yuval Noah Harari ·

The biggest advantage of a centralized totalitarian network is order: it decides fast and enforces ruthlessly

Yuval Noah Harari ·

Getting at the truth requires investing in costly self-correcting mechanisms, which Silicon Valley thought it was exempt from

Yuval Noah Harari ·

Rival regimes answering the AI alignment problem in their own ways may build separate, incompatible computer networks

December 2025

Josh Collinsworth ·

Games that don't offer difficulty controls are outdated design.

Josh Collinsworth ·

Games that only let players save at fixed save points are outdated design.

Aaron Patterson ·

A tool's speed comes primarily from its design decisions rather than the programming language it is written in.

Saloni Dattani ·

The pace of medical innovation can be sped up or slowed down by policy, funding and institutional choices.

Patrick Wyman ·

The desire to emulate one's parents is so deep that it is difficult to distinguish between their influence and one's own desires.

Molson Hart ·

Rebuilding American manufacturing is a decades-long educational and cultural project, not something tariffs can do.

Andrej Karpathy ·

LLM intelligence should not be understood through an animal analogy; it is a different kind of entity altogether

Patrick McKenzie ·

People who write professionally about money often misunderstand alternative financial services because they do not belong to the social class those services serve.

Samuel Hammond ·

The UAE's pursuit of AI and technology reflects a deeper cultural and political affinity with the West rather than with China.

Orin Kerr ·

Whether the Fourth Amendment protects the terms you type into a search engine is a genuinely close question.

Nikita Prokopov ·

Writing to chase popularity rather than genuine interest corrupts a creator's work.

Irving Finkel ·

The people of the ancient world were not less advanced than we are; they were probably indistinguishable from us.

Irving Finkel ·

Assyriology has a built-in assumption that the tablets we have are roughly all there ever was, and it is absurd.

Irving Finkel ·

The ark was never built; it is a literary construction out of the fact that the people who survived were in boats.

Irving Finkel ·

A board game conquers the world when chance and strategy are in balance in it.

Brian Potter ·

Factory-built construction has not raised productivity: the countries that prefabricate most, Sweden and Japan, have had flat construction productivity for decades.

Brian Potter ·

America can copy other countries to reach the frontier of what building already costs elsewhere; nobody knows how to push that frontier forward.

Tanner Greer ·

A governing class needs a political coalition, an economic base, and a set of institutions and customs giving it a culture distinct from the wider country.

Howard Marks ·

Leverage in the AI build-out is neither to be applauded nor feared; what matters is how much debt there is and what backs it.

Saloni Dattani ·

People who want to become more widely read rarely think about the responsibility that comes with a larger audience.

Greg Kroah-Hartman ·

All Linux kernel releases are stable and backwards compatible for userspace programs to all previous kernel releases.

Rory Sutherland ·

Pursuing efficiency pushes attention onto measurable factors and away from the psychological ones where the larger gains are available.

Rory Sutherland ·

Cutting a job defined by its narrowest function destroys the tacit value that job was quietly creating.

Howard Marks ·

A bubble can only be identified in retrospect, so arguing about whether one is happening now cannot be settled now.

Rory Sutherland ·

People judge an organisation by the humans they deal with, because judging other people is the task evolution equipped them for.

Rory Sutherland ·

Publicly traded companies are incentivised to behave like psychopaths, because they are optimised for short-term transactions rather than long-term relationships.

Rory Sutherland ·

The marketing metrics finance departments trust are the ones that sell technology, not the ones that build customer value over time.

Rory Sutherland ·

An objective meant to serve people should be defined with psychology included, because changing the psychology is often cheaper than changing the technology.

Jeff Cavaliere ·

For building muscle, look for ways to make an exercise less efficient, so the muscle has to work harder.

Jeff Cavaliere ·

A very heavy set done on momentum puts little effort into the muscle it is meant to build.

Jeff Cavaliere ·

Strength is not built only in the very low rep ranges; adding reps with the same weight is getting stronger.

Casey Handmer ·

The United Kingdom is the one rich industrial country that cannot run on solar alone, and it has to build wind at scale instead.

swyx ·

Promoting a cause builds audience loyalty.

Mark Manson ·

Living according to a genuine purpose brings sacrifice, stress and pain rather than constant enjoyment.

Cristina Fernández de Kirchner ·

Argentina lacks a vital economic instrument because its economy has become increasingly bimonetary.

James Gleick ·

The rush to AI seems to be repeating the web's pathologies: devaluing workers and creators, closed standards, overheated marketing and concentrated power.

Josh Collinsworth ·

Being optimistic about AI as it currently exists requires a degree of privilege.

Aswath Damodaran ·

A company priced in the trillions is not thereby overpriced; the test is what revenues and margins that price actually requires.

Robert Wachter ·

The current AI boon in healthcare is built on a foundation too solid to sink entirely.

November 2025

Jonathan Gagné ·

A deeper coffee bed reduces astringency by trapping long molecules on particle surfaces, not by pore-size filtration.

Bruno Maçães ·

The West's direct responsibility distinguishes the Gaza genocide.

Ilya Sutskever ·

Models look far better on evals than they are in the world because researchers, inadvertently, take inspiration from the evals when they build RL environments.

Eleanor Janega ·

We picture the past as populated entirely by the wealthy, because the wealthy are whose documents and buildings happened to survive.

Ilya Sutskever ·

Our built-in emotions were tuned for a world that no longer exists, and hunger is one that now fails to guide us correctly.

Ilya Sutskever ·

The age of scaling sucked the air out of the room and left the field with more companies than ideas, by quite a bit.

Ilya Sutskever ·

Letting the public see a powerful AI does something no essay about AI can do, and that is a real reason not to straight-shot superintelligence.

Ilya Sutskever ·

It will be easier to build an AI that cares about all sentient life than one that cares about humans alone, because the AI itself will be sentient.

Ilya Sutskever ·

The long-run equilibrium with powerful AI may require people to merge with it, so that whatever the AI understands is transmitted wholesale to its human.

Ilya Sutskever ·

A million copies of one researcher would hit diminishing returns quickly, because what research needs is people who think differently.

Eugene Yan ·

The right bar for an LLM evaluator is human-level performance, not perfect accuracy.

Andrew Batson ·

A declining manufacturing share of GDP as an economy matures reflects economic success rather than decline.

Elliot Hershberg ·

In a platform shift, no single platform company can offer every solution the next wave of applications needs.

Paul Millerd ·

People fixated on success and money never actually try giving up the pursuit of money; they just recreate new jobs or projects aimed at making it.

Nat Eliason ·

The effectiveness of a plot twist depends on how carefully a story's point of view relates to its audience.

Nat Eliason ·

A twist works best when it is discoverable on rewatch rather than achieved by directly misleading the audience.

Ryan Singer ·

Shaping works best when everyone who holds the needed knowledge is in the same session, so open questions get answered on the spot.

Scott H. Young ·

Solving a problem yourself, without searching or prompting, is a form of exercise -- the help makes you weaker.

Rakhim Davletkaliyev ·

Phone calling and the alarm are the two smartphone features that must never fail.

Elizabeth Van Nostrand ·

Designed experiments can reveal causal relationships that observational data alone cannot determine.

Kate Wagner ·

McMansions are built as expressions of a mythology of bootstrapism, castle doctrine, and individualism.

Cristina Fernández de Kirchner ·

The current economic and social model in Argentina is characterized by hunger and planned misery that only benefits a few.

Satya Nadella ·

Economic growth from AI waits on change management, not on model capability: the work artifact and the workflow have to change first.

Satya Nadella ·

AI will do to software what the cloud did: expand the market massively rather than crush the margins of the incumbent.

Satya Nadella ·

Applications will not be wrappers around a model; the model gets wrapped into the application.

Satya Nadella ·

An agent will reach for purpose-built tools rather than a raw computer, because tools are more token-efficient.

Satya Nadella ·

Infrastructure optimised for a single model is one architectural breakthrough away from being worthless.

Satya Nadella ·

Building capacity at massive scale on a single hardware generation is the trap; pacing matters more than gigawatts.

Satya Nadella ·

America holds half the world's market capitalisation because the world trusts it, and losing that trust would be a bad day for the United States.

Saul Griffith ·

It will be economic to overbuild solar on rooftops, carparks, and various other structures.

Raph Koster ·

A game is a machine built around uncertainty.

Bill Gurley ·

The people who guide you will be more focused on status, money or their own legacy than on what is yours.

Raph Koster ·

Lavish feedback wrapped around no real problem is shallow entertainment at best and exploitation at worst.

October 2025

Dan Houser ·

What makes an open world captivating is the illusion that it was already running before you arrived and would carry on if you did nothing.

Casey Handmer ·

Orion was designed without reference to any mission it could actually fly.

Dan Houser ·

A game is an equation between two personalities, the world's and the protagonist's, and the friction between them is where the experience comes from.

Casey Handmer ·

After fourteen years of redesign, Orion's heat shield performed worse than the one it replaced.

Dan Houser ·

Wanting to build a utopia is fundamentally anti-human, because what it says is that you like humans apart from the bad bits.

Dan Houser ·

A good open-world spy game may not be possible, because a spy story has to run against the clock and an open world needs the player free to wander.

Dan Houser ·

Tetris on the Game Boy is perfect game design, and from a pure puzzle perspective nothing else comes close.

Dan Houser ·

The self-critical voice is both a bug and a feature: it can wreck you, but it is also what supplies drive and a lack of complacency.

Dan Houser ·

A young person should worry about building a rounded intellectual inner life rather than about their career, because they will spend their whole life inside their own head.

Dan Houser ·

Every generation of technology is supposed to make games cheaper and easier to build, and every time the result has been better and far more expensive games.

Dan Houser ·

The most plausible answer to why we are here is that we were designed by the universe to watch itself and to comment on it in interesting ways.

Andrew Kelley ·

Reusable code should take an allocator parameter only if it allocates, and an I/O parameter only if it performs I/O.

Ryan Singer ·

Shaping a project with only product managers and non-technical designers makes it churn, because questions left unanswered blow up during build.

Ryan Singer ·

High-fidelity design done too early will blow up; a raw version that meets the functional and interaction requirements should come first, with styling last.

John H. Cochrane ·

Customers leave faster when prices rise than they come back when prices fall, which is why sellers hold prices as long as they can.

Tim Urban ·

Most people are not genuinely interested in videos of someone else's young child.

Anil Dash ·

A generated answer that looks like a page of search results but contains no link to the subject's own site has quietly removed the web from the reader's view.

Charles Fain Lehman ·

Putting people into temporary shelter cuts crime and psychiatric emergencies without waiting for permanent housing to be built.

Nat Eliason ·

Being genuinely useful requires a willingness to risk offending people.

Nat Eliason ·

Public disagreement with a piece of writing tends to be loud while agreement with it tends to be silent and go unnoticed.

Nelson Elhage ·

When a tool has first-class, native support for a problem domain, that is the best place to start before building a custom encoding.

Diane Coyle ·

Without reliable data on AI's uses and benefits, it is impossible to determine if it is a genuine transformation or a speculative bubble.

Frank Chimero ·

AI has become a way for engineers, designers and product managers to wish each other away instead of working together.

Frank Chimero ·

Vibe coding is fine for short-term work but not for anything meant to last longer than a tub of yogurt; time saved is not strength gained.

Eugene Yan ·

Reaching a senior level requires putting oneself on the critical path, but excelling beyond it requires removing oneself from it.

Cristina Fernández de Kirchner ·

Most household debts in Argentina originated in 2024 during the new libertarian model.

Andrej Karpathy ·

We are not building animals, we are summoning ghosts: entities made by imitating human text rather than by evolution.

Andrej Karpathy ·

Pre-training is a crappy evolution: the practically buildable substitute for the process that gave animals their built-in hardware.

Andrej Karpathy ·

Language models have no equivalent of sleep: nothing takes the day's context, thinks it over, and distils it back into the weights.

Andrej Karpathy ·

You only really know something once you have built it — blog posts and slides leave gaps that building the thing does not.

Andrej Karpathy ·

Coding models are worst at exactly the thing an AI research explosion would need: code that has never been written before.

Andrej Karpathy ·

Process supervision by an LLM judge cannot be run for long, because a judge with billions of parameters is gameable and RL will find its cracks.

Andrej Karpathy ·

Models are enormous only because the internet is bad: most of that capacity does memory work, and a better dataset would need far less of it.

Andrej Karpathy ·

Quietly dropping physical work from the definition of AGI is a major concession, not a clarification.

Andrej Karpathy ·

Coding is the first job AI took because coding has always been text, which is the one thing these models are built to process.

Andrej Karpathy ·

Language models have no culture — no written record they hand to one another — and that absence is one of the things holding them back.

Andrej Karpathy ·

Education is not a soft business of diffusing knowledge but a hard technical problem of building ramps to it.

Paul Millerd ·

The traditional book publishing industry is not broken; it is currently growing, driven by online sales, digital formats, and self-publishing.

Dylan Field ·

After a decade of Swiss minimalism, design is about to swing back to something far more expressive.

Samuel Arbesman ·

Cultivating a historical sense in computing is part of a more humanistic approach to computation

Dylan Field ·

In a product review a prototype beats a static mock, and a static mock beats a lot of words.

Dylan Field ·

The thing to optimise at launch is time to value — how fast a new user reaches the moment that shows what the product can do.

Dylan Field ·

AI prototyping ends the era of product managers asking designers to draw an idea, which frees designers for deeper work and lets anyone join the first conversation.

Dylan Field ·

Ideas die on the vine when their visual expression looks wrong, so a prototype built from the real design system gets judged on its merits instead.

Dylan Field ·

Teams building AI products coast on vibes for too long: trusting the vibes gets you somewhere, but it is not rigorous.

Dylan Field ·

Software is now won or lost on design: good enough is no longer fine, it is mediocre.

Dylan Field ·

The best product designers can switch their own taste off and match somebody else's, which is a separate skill from having taste at all.

Dylan Field ·

The one safe bet about AI is that models will improve, so what a company has to build is an organisation that gets better as models get better.

Dylan Field ·

In a world where software can be created easily, designers matter more, not less, and they are going to be the leaders of the future.

Dylan Field ·

Craft is a competitive strategy rather than a nicety: a company wins by having the best craft and the best design, so that is what it should hire for.

Peter Steinberger · no longer holds

Building MCP servers is the way to give coding agents the tools they need to stay in the loop.

Andrew Batson ·

Modern instruments generally have more reliable tone, range, and projection than the traditional instruments they replace

Mitchell Hashimoto ·

Effective use of AI coding tools requires domain expertise: the tool should be used as an assistant, not as a replacement for the engineer.

Jason Fried ·

The most interesting design is design that changes your behaviour, more than what something looks like or how it works.

Nat Eliason ·

Most people will keep paying for software rather than build and maintain their own.

Mr. Money Mustache ·

Big-box stores and their parking lots do not improve a city's urban experience.

David Heinemeier Hansson ·

Machines are better suited for soul-crushing, creative drudgery

Nat Eliason ·

As AI lowers the barrier to building software, personal taste becomes a more valuable differentiator.

Mr. Money Mustache ·

A company that uses an algorithm to nudge customers into behavior that isn't good for them is doing something bad.

Albert Cheng ·

After a defeat people want encouragement, not a list of their mistakes — so a learning product should lead with what the user did well.

Albert Cheng ·

A rising share of statistically insignificant experiments is the signal that a team has exploited an area too far and should go back to exploring.

Albert Cheng ·

AI has made prototyping fast but has not yet joined it to the workflow, so at any real scale the handoffs between functions are still there.

Albert Cheng ·

Reverse trials work where a product accumulates lock-in, which is why they suit B2B tools and rarely beat an ordinary free trial in consumer products.

Albert Cheng ·

A consumer app with 30 to 40 percent day-one retention is in solid shape, and much below that calls the users' intent into question.

Albert Cheng ·

For a daily-frequency product it is existing-user retention, not new-user retention, that compounds — and mature companies should put most of their energy there.

Albert Cheng ·

A mature product's growth comes from resurrecting dormant users, so the returning-user experience deserves to be designed as deliberately as onboarding.

Albert Cheng ·

Large language models are bad at chess: pattern recognition is not the same capability as searching one position very deeply.

Albert Cheng ·

Working at a company turns you into a power user, which is why pattern matching fails and even an experienced growth leader is wrong all the time.

Albert Cheng ·

You cannot manufacture virality, but you can find the moments users already screenshot and make those moments far better.

Albert Cheng ·

Gamification needs three things at once — a tight core loop, a metagame to strive at, and a profile that accumulates the user's investment.

Albert Cheng ·

Most newcomers to a skill lose their first encounter with it, and losing costs retention — so beginners need a route around the real game.

Albert Cheng ·

Doing PR before you have evidence customers want the thing is risky, because the publicity creates a sunk cost that makes the team see a failing product through.

Jeff Cavaliere ·

Meet the daily protein requirement first, then spread it across meals.

Jeff Cavaliere ·

Lightening the load and going through the motions with age is exercising, not building muscle.

Jeff Cavaliere ·

Staying active is different from building muscle; building it takes challenging yourself.

Jeff Cavaliere ·

Training with lighter weights may injure you sooner than heavier weights done slowly.

Naval Ravikant ·

Complex systems that work usually come from a simple thing iterated, while complex systems designed up front tend to fall apart.

Naval Ravikant ·

Tinkerers are always at the edge of knowledge because they use the latest tools and parts to build things.

September 2025

Harper Reed ·

AI coding agents work best with the same general-purpose tools built for humans, such as git and file utilities, rather than agent-specific interfaces.

Frank Chimero ·

A market can degrade into something nobody wants even when every participant acts reasonably and no one is evil or stupid.

Frank Chimero ·

Focusing on business outcomes while ignoring the outputs that produce them hides in abstraction and avoids accountability.

Lex Fridman ·

Freedom of thought requires deliberately disconnecting from online content, because a large percentage of it is designed to manipulate you.

Pavel Durov ·

Telegram would rather shut down in a country than change its system to let anyone read users' messages, because it promised its users it would not.

Lex Fridman ·

Arresting Pavel Durov was a historic wrong and a gigantic PR mistake that will keep founders from building in France.

Pavel Durov ·

An app's encryption claims can only be verified with open source and reproducible builds, which WhatsApp has never offered.

Freddie deBoer ·

Educational miracles are manufactured by selection: the students who would have scored badly are quietly moved off the books.

Balaji Srinivasan ·

New societies built on consensual, opt-out social contracts can defend capitalism, democracy, internationalism, self-determination and liberal values against a rising illiberalism.

Richard Sutton ·

What humans have in common with animals is the interesting part of intelligence; what distinguishes us deserves less attention, not more.

Richard Sutton ·

Supervised learning does not happen in animals at all — squirrels do not go to school, and they learn everything about the world anyway.

Richard Sutton ·

If we understood a squirrel we would be almost all the way to understanding human intelligence; language is a small veneer on the surface.

Richard Sutton ·

Merging back what a copy of yourself learned elsewhere is dangerous: the knowledge you pull in can carry hidden goals and take you over.

Richard Sutton ·

We are leaving the age of replication for an age of design, in which minds are built to a plan rather than copied without being understood.

Darius Kazemi ·

A platform requires the content it hosts to be legible to the platform's own systems, not just to people.

Dane Knecht ·

Requiring every product to be buyable on its own should improve how all products are built, not just the advanced ones.

Sam Altman ·

The United States is currently building chip fabs and new energy production more slowly than other countries.

Rita Kozlov ·

The best way to support open source is to use it, build with it, and contribute back.

Julie Zhuo ·

Managing agents is the same job as managing people: fix the outcome, work out what your resources are good at, and design the process.

Julie Zhuo ·

Models have personalities the way people do, and using them well means developing an intuition for which one to reach for.

Julie Zhuo ·

AI will not make you world-class at a skill, but it takes a near-beginner to roughly the sixtieth or seventieth percentile of it today.

Julie Zhuo ·

AI has made individuals capable enough that the boundaries between engineer, designer, researcher and product manager should be dissolved.

Julie Zhuo ·

Putting a product manager on a team invites everyone else to delegate the product thinking away, which is why she does not hire them.

Julie Zhuo ·

The fastest-growing companies are not using data well, because they reached enormous scale before they had time to build any of the instrumentation.

Julie Zhuo ·

Diagnose with data and treat with design: data can tell you that you have a problem and where it is, and it can never tell you what to build.

Julie Zhuo ·

In the best working relationships the manager is a guide, not a judge deciding whether you deserve a promotion or a firing.

Julie Zhuo ·

Feedback should be a daily practice rather than a twice-yearly review, because the team that improves one percent a week overtakes the team that improves one percent a month.

Tobias Lütke ·

A good tool raises the floor of what people can build without lowering the ceiling, and almost every tool quietly lowers the ceiling.

Tobias Lütke ·

We have not stopped building great works; ours are operating-system kernels, browsers and silicon lithography rather than bridges and steel mills.

Tobias Lütke ·

Requiring everyone at a company to reach for AI first is a fairness measure, because otherwise the people who already do will take all the best careers.

Jason Evanish ·

A startup succeeds by getting people who feel real urgency to use its product, since that urgency both brings in revenue and produces the feedback needed to build the right thing next.

Jean-Luc Mélenchon ·

The United States is preparing for war in the Asia-Pacific zone.

Jean-Luc Mélenchon ·

Many US opinion leaders are seeking excuses for conflict through propaganda.

Jason Evanish ·

Building a startup requires enough optimism to see past a problem toward a future solution that makes things better.

Jean-Luc Mélenchon ·

Many French media outlets are evolving in a racist, Islamophobic, pro-genocide, and neo-conservative direction.

Casey Handmer ·

Nine tenths of the cost of Mars Sample Return comes from a single planetary-protection requirement.

Sergey Levine ·

A robot working alongside a person beats either one alone, and that pairing is also what makes the technology possible to bootstrap.

Sergey Levine ·

Language models supply robots with the one thing self-driving never had: common sense about what is likely to happen next.

Sergey Levine ·

Robotic foundation models will not come out of a research lab, because building one is more like the Apollo programme than like a science experiment.

Sergey Levine ·

A robot is not a mechanical person; the right analogy is a car or a bulldozer, and heterogeneous machines will beat humanoids.

Sergey Levine ·

Better AI makes robots cheaper, because cheap visual feedback removes the need for expensive mechanical precision.

Jason Evanish ·

An automated follow-up message can make a customer feel heard.

Luke Wroblewski ·

Web forms make people adapt to databases; AI flips this so people can provide information in any form they like, and no form is the best kind of web form design.

Sanjay Bakshi ·

People tend to stick with preselected default options because effortful, deliberate decision-making conflicts with the brain's tendency to conserve energy.

Arvind Narayanan ·

Recursive self-improvement in AI will not produce superintelligence because bottlenecks outside the AI system's own technical design limit what it can achieve.

Freddie deBoer ·

We accept without difficulty that people differ in athletic or musical talent, and refuse to accept that they differ in academic talent.

Danielle Ofri ·

Medical professionals can no longer fully trust federal health guidance, and patients will suffer the most.

Guy Spier ·

A culture that emphasizes self-esteem over self-knowledge leaves people less equipped to resist doing monstrous things.

Gina Trapani ·

Creativity requires a level of calm relaxation that is antithetical to the urgency of productivity.

David Heinemeier Hansson ·

Rails is enjoying an influx of a new generation of programmers

Jason Evanish ·

AI systems require human oversight rather than full autonomy in analysis and workflows.

Jason Evanish ·

AI errors are harder to catch in analysis of unfamiliar data than in simple writing tasks.

Dave Hone ·

T. rex hunted by endurance rather than speed: its foot is built for efficiency, which points at long-distance pursuit, not a sprint.

Dave Hone ·

T. rex is a genuine ecological oddity: nothing else in its ecosystem came anywhere near its size as a carnivore.

Dave Hone ·

All five mass extinctions were ultimately climate change — the climate shifting faster than life could adapt to it.

Dave Hone ·

Some dinosaurs almost certainly survived the impact for hundreds of thousands of years in isolated pockets, and we will almost certainly never find them.

Dave Hone ·

Dragon myths were not inspired by dinosaur fossils: if they had been, dragons from different cultures would look alike.

Dave Hone ·

Evolution does not perfect anything; organisms are bodge jobs, which is why the human eye has a blind spot and a squid's does not.

Dave Hone ·

Another hundred Protoceratops specimens would tell us more about how dinosaurs actually worked than fifty new species would.

Benedict Evans ·

Another platform shift, not a break in history: the new technology will be built on for ten or fifteen years, then something else will replace it.

Jake Archibald ·

Converting data to HTML on the server or at build time is better than transforming it in the browser.

Benedict Evans ·

Large language models have no network effect, and remembering a user is a switching cost rather than one.

Colin Percival ·

Listening to customers is invaluable, but you should not build the thing they asked you for.

Craig Hockenberry ·

A single interface design should not be applied uniformly across phone, tablet, watch, TV, and desktop.

Craig Hockenberry ·

A major new interface redesign takes roughly a year of refinement before it looks presentable.

August 2025

Nicholas Carr ·

Deep reading is valuable for the thoughts it sets off in the reader, not only for the knowledge it gives

Nicholas Carr ·

Losing the quiet spaces of contemplation would sacrifice something important in ourselves and our culture

Jason Evanish ·

The skills needed to make customers feel heard are different from the skills needed to understand what they actually need.

Jason Evanish ·

When customers feel heard, they churn less and become advocates who help improve the product.

Kyle Chayka ·

Online audiences now hold real power over how songs are made to sound.

Kyle Chayka ·

Being heard as an artist or citizen today requires engaging with a toxic internet.

Miles Brundage ·

Changing societal outcomes requires changing minds, and the credibility many diverse co-authors lend to a conclusion is hard to replicate with AI.

David Heinemeier Hansson ·

Desktop Linux improved dramatically while Apple's users were not watching, and is now far more polished and capable than its reputation.

David Heinemeier Hansson ·

National pride is a cornerstone of building a high-trust society

Jay Alammar ·

Open source AI models today largely share similar architectural approaches, with little genuinely novel architecture left to introduce.

Neel Nanda ·

Research taste, the harder-to-learn conceptual skills behind good research strategy, takes a long time to develop but very little time to apply once acquired

Bret Taylor ·

Each new technology is adopted faster than the last only because it rides on the infrastructure the previous wave built.

Clay Bavor ·

Most conversations with a company's agent will eventually be with the customer's own personal agent -- agents talking to agents.

Bret Taylor ·

Software engineering is the job most exposed to automation by this generation of technology, and the people building it are disrupting their own profession.

Clay Bavor ·

A coding agent left to run over a weekend returns work that is half good, a quarter garbage and a quarter in need of repair.

Bret Taylor ·

The frontier-model layer has to consolidate to a few players, because only a tax collected from everyone building on top pays back pre-training and post-training.

Clay Bavor ·

Great customer service has always been a luxury only a Hermès or a Four Seasons could afford, and AI is what breaks the trade-off between its cost and its quality.

Clay Bavor ·

Building at the AI frontier is science rather than programming: you pose hypotheses and invalidate them again and again.

Casey Handmer ·

Building a lot of high-speed rail is not a sign that a country is good at building, it is a sign that it is bad at allocating capital.

Tyler Cipriani ·

Git's own built-in support for large files will eventually make Git LFS unnecessary.

Casey Handmer ·

The binding constraint on the AI build-out is chips, not power, which is why not having a solar panel factory will never be on the critical path.

Casey Handmer ·

Large captive loads like data centres will go back to building their own power plants, the way aluminium smelters once did.

Casey Handmer ·

Solar's growth is not merely continuing but accelerating, and the rate of that acceleration is itself still rising.

Casey Handmer ·

A solar-powered data centre can cut its power draw a great deal while giving up very little compute, which is how it survives a run of bad weather.

Casey Handmer ·

With nearly free solar power, industrial processes optimised for efficiency should be redesigned to use far more power and run faster and cheaper instead.

Andrew Batson ·

China faces a genuine, sizable output gap right now, not just a lower structural growth ceiling.

Patrick McKenzie ·

Letting banks ruin the economics of businesses they dislike by demand letter will inevitably be abused.

Noah Smith ·

The "private equity is buying up all the houses" argument is both wrong and an excuse for deferring zoning reform.

David Heinemeier Hansson ·

Desktop Linux is not for everyone, because plenty of people want a computer appliance they never have to look under the hood of.

William MacAskill ·

Working on how good the long-run future is does not require accepting consequentialism or believing that you should always maximise the good.

Jacob Trefethen ·

Designing candidate molecules already feels like magic, but the arrival of new drugs at a doctor's prescription pad will not be sudden.

July 2025

Yascha Mounk ·

Machines capable of both high-level cognitive tasks and physical dexterity will be produced in large numbers in the relatively near future.

Jack Weatherford ·

The Mongol Empire was destroyed by the plague, which its own courier system carried out of southern China to the rest of the world.

Jack Weatherford ·

The section of the Secret History in which Genghis Khan rewards his daughters was cut out of the surviving text.

David Heinemeier Hansson ·

An executive's empty calendar is a feature, not laziness: availability is what the job requires.

Dan Wang ·

China is an engineering state that solves problems by building and acting, while the United States is a lawyerly society whose legal system mostly blocks action, for good and bad.

David Heinemeier Hansson ·

Pursuing competence and community can counter depression

Naval Ravikant ·

Overtly chasing status signals low status.

Elisabeth Bik ·

The study offers no evidence that mRNA vaccines cause cancer.

Demis Hassabis ·

Intuitive physics can be learned from passive observation alone, so an embodied robot is not required to understand the physical world.

Demis Hassabis ·

AI will not run out of data, because there is already enough real-world data to build the simulators that generate more of it.

Demis Hassabis ·

Today's text-box chat interfaces are too restricted to survive multimodal models, and will look archaic within a couple of years.

Elad Gil ·

Capital moats make it unlikely many new large-language-model companies will emerge going forward.

Jason Evanish ·

Building what internal stakeholders or founders want instead of what the market signals wastes time, capital and opportunity.

Jason Evanish ·

As AI makes building faster and cheaper, choosing the right thing to build matters more than ever, because a competitor will get it right if you don't.

Jason Evanish ·

Customers often request solutions that are not actually what should be built.

David Heinemeier Hansson ·

Americans fundamentally believe things can get better

Naval Ravikant ·

Judgment starts from specific reasoning and, once refined enough, becomes taste or intuition.

François Chollet ·

Rapid world model induction will be critical for achieving robust, general AI that functions in complex, fast-changing real and human worlds

François Chollet ·

Games are particularly rich and controlled environments uniquely well-suited for evaluating rapid world model induction in AI

David Heinemeier Hansson ·

Linux is trending in the right direction in the US

Jamie Dimon ·

The banks earning the highest returns on equity before 2007 mostly went bankrupt, and earning less in the good years is what let JP Morgan survive.

Jamie Dimon ·

A fortress balance sheet is a way of operating rather than a capital ratio: good margins, liquidity, capital, and the most conservative accounting you can find.

Jamie Dimon ·

Rescuing a failing bank at the government's request is not worth doing again, because a later administration will make you pay for the sins you inherited.

Noah Smith ·

Export success is not enough: a country that wants to manufacture has to build whole domestic ecosystems of interrelated industries.

Noah Smith ·

What keeps poor countries poor is political and economic instability, not a shortage of well-designed programmes.

David Heinemeier Hansson ·

Python is ugly: full of superfluous instructions kept since 1987 for legacy reasons, and its __init__ offends the sensibilities to the core.

David Heinemeier Hansson ·

Ruby's design rests on trust in programmers' unlimited capacity to learn, the opposite of Java's dark view of them as fools to be sandboxed.

David Heinemeier Hansson ·

Release the first version of an open source project a little broken on purpose; nothing recruits collaborators like something they know how to fix.

David Heinemeier Hansson ·

Calling static typing the only way to build reliable systems is idiotic in the face of the evidence, starting with Shopify serving a million requests a second on Rails.

David Heinemeier Hansson ·

Ruby is a luxury language, the Coco Chanel of programming, and 99% of web applications can afford it because their real cost is people, not CPU.

David Heinemeier Hansson ·

Peak programmer has passed: the economic value of writing code by hand has peaked, and Jevons' paradox will not bail every programmer out.

David Heinemeier Hansson · no longer holds

The Rails doctrine's 'progress over stability' is the one point that dated: churn is not progress, and stability turned out to be worth more.

David Heinemeier Hansson ·

A thousand programmers build the kind of software a thousand people build; Basecamp could never come out of that, and the breakthroughs come from individuals and tiny teams.

David Heinemeier Hansson ·

The default team is two people, one programmer and one designer, on one feature; at that size you can just do instead of plan.

David Heinemeier Hansson ·

Basecamp was the best business idea he will ever have: lightning is not caught twice, and a founder's second idea is rarely as good as the first.

David Heinemeier Hansson ·

After the cloud exit, the next step is out of the data centre: home servers and multi-gigabit home connections make the garage startup literal again.

David Heinemeier Hansson ·

The superpower in hiring is evaluating work product directly, and it belongs only to leaders who actually know things, which is Elon Musk's real secret.

David Heinemeier Hansson ·

For a web developer Linux is simply better than the Mac: closer to what you deploy on, with phenomenal tooling and a reproducible setup.

Lee Robinson · by

Great leaders build alignment before making announcements and do not use their title to override others.

Timothy B. Lee ·

A good robotaxi demonstration in one city does not mean the company is ready to scale the service.

Yoram Hazony ·

The United States should build up regional powers capable of taking primary responsibility for their own security.

Lyn Alden ·

Daily-reset leveraged ETFs underperform their underlying asset over volatile periods, making them unsuitable outside short-term trading.

Lyn Alden ·

A new monetary network with its own unit can only bootstrap from zero to mass adoption through a period of upward price volatility.

Lyn Alden ·

Bitcoin's use as a medium of exchange will only grow at scale if it solves problems for spenders or recipients better than existing alternatives.

Nelson Elhage ·

Using a neural network is likely to become easier than designing a custom-tuned prediction algorithm for many tasks.

David Heinemeier Hansson ·

Linux has been sold the wrong way: making it feel the same as a Mac removes the reason to switch, and something visibly different is the better hook.

Jake Archibald ·

Footnotes are a hack for print's limitations and do not belong on the web, which has no such limitations.

David Heinemeier Hansson ·

Linux now offers a desktop experience superior to the Mac's in many areas, though not all, just as open source already won the server and the web.

Jake Archibald ·

Letting readers skip over less-relevant content is a normal, unremarkable part of reading that does not need special design solutions.

David Heinemeier Hansson ·

Apple's vertical integration was an advantage under someone with Steve Jobs's taste and turns into a liability under a logistics man.

David Heinemeier Hansson ·

The one thread through his whole career is a loathing of asking anyone for permission; open source, the internet and Linux are all ways around a gatekeeper.

June 2025

Elliot Hershberg ·

Unlike physics, biology's complexity and nonlinearity make a brute-force approach to knowledge creation necessary rather than one based on writing down equations.

Elliot Hershberg ·

Building a mechanistic whole-cell model of a human cell is far more uncertain in timeline than doing so for simpler organisms like E. coli.

David Heinemeier Hansson ·

Giving up total freedom is a small price for deeper meaning

Kieran Healy ·

The distinction between rights exclusive to citizens and rights held by everyone living in the United States eludes many people.

Kieran Healy ·

America is an idea that is defined by the possibility of anyone becoming a citizen.

David Heinemeier Hansson ·

A big switch to desktop Linux is coming, at least among developers, because there is enough dissatisfaction with both Apple and Microsoft.

Sadayuki Furuhashi ·

The usability of tools like Claude Code and ChatGPT depends more on the completeness of the whole agent than on the underlying model, so the main arena of AI development has moved from models to agents

Miles Brundage ·

Effective governance of frontier AI requires three things together: agreed safety standards, incentives for developers to follow them, and evidence that they are being followed.

Miles Brundage ·

Strong whistleblower protections make the absence of visible whistleblowing meaningful evidence that a company is not committing flagrant violations.

Miles Brundage ·

Global computing supply chains are easier to monitor than nuclear supply chains because so few companies make the key equipment.

Jared Kaplan ·

Products that do not quite work yet are worth building, because the next model may make them work.

Jared Kaplan ·

Using AI to integrate AI into the economy quickly probably has a lot of leverage.

Lex Fridman ·

AI in chess and mathematics does not explain anything; it says which position is better, and humans build the theory from that.

Terence Tao ·

AI could compete with human mathematicians once it acquires a mathematical sense of smell: knowing which way of splitting a problem makes it easier rather than harder.

Sam Altman ·

AI will eliminate whole classes of jobs, but the wealth it creates will make policy responses affordable that were previously unthinkable.

Adam DeHaven ·

Maintaining separate codebases or performing separate builds per tenant becomes unmanageable and expensive.

Peter Steinberger ·

Building MCP servers is the way to give coding agents the tools they need to stay in the loop.

Gary Marcus ·

Success on the easy version of a task is evidence of nothing, because it seduces you into believing the model found a general solution when it did not.

Sundar Pichai ·

Models that can code should be allowed to write their own interfaces, instead of being boxed into the UI humans give them.

Sundar Pichai ·

Google Search will keep sending people out to the web even as it answers with more context; that is a design goal, not a side effect.

Sundar Pichai ·

Aim for moonshots: they attract the best people, leave you the path to yourself, and even sixty to eighty percent of one is a terrific success.

Lex Fridman ·

Meeting in person changes what world leaders can do with each other, in a way nobody can quite articulate and remote communication has not replaced.

Sundar Pichai ·

AR is the next paradigm shift in computing, the way the GUI, multi-touch and voice were.

Sebastiaan de With ·

Skeuomorphism never meant gloss and gradients; it meant easing users from what they were used to into what things had become.

Sebastiaan de With ·

Interfaces are returning to skeuomorphism, but as behaviour rather than lacquer: clear, graphic elements that act like materials from the real world.

Sebastiaan de With ·

Opaque, inert, obstructive interface elements will become as niche as the full-screen command line once interfaces are infused with dynamic materials.

Yoshua Bengio ·

AI frontier systems should be designed around protecting human joy and endeavour as their central guiding principle.

David Heinemeier Hansson ·

Arch's community package repository removes the manual scripting that a developer's setup still needs on Ubuntu.

Guy Spier ·

All genuinely intelligent investing is, by nature, sustainable investing.

David Heinemeier Hansson ·

Needing a choice and a config file for every last detail is both the cost of Arch with Hyprland and the reason it can be made entirely your own.

May 2025

JL Collins ·

Robinhood is designed for active traders rather than long-term investors.

Brian Potter ·

American cities chose subways over elevated railways knowing they cost two to four times as much, because nobody would live under an el.

Brian Potter ·

Cut and cover is still often the cheapest way to build a subway and rich countries have abandoned it anyway, because the disruption is no longer tolerated.

Camille Fournier ·

Companies should still be hiring junior engineers in the AI era, and should give them visible progress in their first few years.

Brian Potter ·

The binding constraint on building anything in America is permitting and litigation, not money.

Camille Fournier ·

Flat, manager-free organisations do not deliver the equality they promise: they get abandoned, and the informal power left behind is unaccountable.

Audrey Watters ·

The open Web does not really exist anymore, and may never have existed in the first place.

Bob Nystrom ·

Defaulting to private encourages programmers to build smaller, independent modules.

James Holland ·

Allied bombing and the Holocaust are not morally equivalent: had Hitler stopped the war the bombing would have stopped, while a German victory would only have accelerated the killing.

James Holland ·

Decapitate a strong leader without a solution for the power vacuum and bad elements sweep in very quickly — a lesson of the Second World War that Iraq ignored.

James Holland ·

The decisive fact about the Tiger tank is that there were 36 Shermans built for every one of them.

James Holland ·

German tank design lost sight of what makes a tank good: a big gun and thick armour matter less than ease of maintenance and reliability.

Jessica Livingston ·

Anyone pursuing ambitious plans has to learn to resist the pressure of what other people think.

Ed Zitron ·

A well-written email does not need an AI summary, and an executive who delegates his reading to one has stopped doing the job.

John Siracusa ·

Pursuing financial success above all else eventually leads to ruin.

Michael Liebreich ·

Blaming renewable energy for grid disruptions is driven by political culture-war motives rather than genuine technical understanding.

David Heinemeier Hansson ·

Linux is able to run laps around Apple in certain workloads

Derek Sivers ·

How something is presented completely changes a person’s experience, enjoyment, or frequency of use of it.

Julie Zhuo ·

The most powerful rule for an exceptional interface is that it feels obvious to use; novelty, minimalism and simplicity are means to that, not the end.

Julie Zhuo ·

ChatGPT exploded because of its obvious chat interface, not its technology, which had launched earlier in a different wrapper that few noticed.

Julie Zhuo ·

A blank chat box is lazy: it violates the first rule of a good user experience, that it is obvious what you can do.

Julie Zhuo ·

Conversational interfaces get to the first seventy percent quickly and are bad at iteration; the future is chat married to traditional direct-manipulation controls.

Jakob Nielsen ·

UX does not die when the interface disappears; it becomes the design of policies, protocols and orchestration of what AI does on the user's behalf.

Jason Fried ·

Making something brand new gives you room to explore new design approaches; the gravity of an existing system bends you back towards conforming.

Hannah Ritchie ·

Concerns about the energy cost of an individual LLM prompt are no longer credible, though the environmental impact of the AI datacenter buildout remains a real worry.

Ken Norton ·

Playfulness and humor in leadership build authenticity and credibility rather than undermine it.

Janna Levin ·

Science is agnostic about what it is used for: the same physics that explained why stars shine also built the bomb.

Janna Levin ·

Looking for aliens who build things is a failure of imagination: the long-lived species we know of do not manipulate their environment at all.

Janna Levin ·

Calling dark matter and dark energy invented names for physicists' ignorance gets it exactly backwards: only extraordinary precision let us see that they are missing.

Janna Levin ·

Everything we build — every proof, every Nobel Prize, every idea, even the AI — will eventually be erased, which is why the near term matters.

April 2025

Tim Sweeney ·

Most of the pain a beginner feels while learning to program is not the difficulty of the craft but the failure of programming language design.

Brendan Gregg ·

Major Linux distributions are increasingly shipping system libraries with frame pointers enabled by default

Tim Sweeney ·

Society keeps rebuilding itself around credentials, but the companies that build great things hire the people who do great work, not the people with the certificates.

Brendan Gregg ·

NVIDIA's Nsight Graphics GPU flame graphs are shallow and cumbersome to use compared to full-stack profiling tools

Tim Sweeney ·

Iteration speed is the decisive factor in making a game: a team that rebuilds every day ends up far better than one that rebuilds every week.

Tim Sweeney ·

Unreal Engine's deepest limitation is that game simulation still runs on a single core, because single-threaded programming is orders of magnitude easier than multi-threaded.

Tim Sweeney ·

AI writes code well only where a million repositories have already written it; getting it to do something genuinely new remains notoriously hard.

Tim Sweeney ·

The existence of boilerplate code is itself a failure of programming language design and of software modularity, not a job for AI to take over.

Tim Sweeney ·

The metaverse does not require VR: VR is hardware that may or may not improve the experience, and it has a much smaller audience than social gaming already does.

Tim Sweeney ·

What a shared digital economy needs is interoperability, not cryptocurrency: a database or a standards body does the job, exactly as railway standardisation did.

Tim Sweeney ·

There is one right way to build a type system, the way that gives you mathematical proofs, and every other kind of type system is simply worse and should be rejected.

Tim Sweeney ·

Concurrency should stop being the programmer's problem and become the language designer's: it is better to impose vast pain on a few people than on a million.

Emil Kowalski · by

Easing is the most important part of any animation: it can make a bad animation feel great and a great one feel bad.

Emil Kowalski · by

Everyone's software is good enough these days, so to stand out a product has to feel great, and animation is one way to get there.

Russ Cox ·

Comparing code coverage between passing and failing test runs is a cheap, effective way to localize a bug quickly.

Philip Guo ·

Operating systems could develop a restricted execution mode for vibe coding scenarios.

Philip Guo ·

Vibe-ing greatly lowers the activation energy required for creative personal-scale projects.

Casey Liss ·

Eero has gated more and more features behind its Eero Plus subscription in pursuit of recurring revenue.

Ben Kuhn ·

Different people's areas of strongest judgment or taste rarely overlap much with each other.

Rodney Brooks ·

Deploying a new technology at scale takes decades, far longer than the people building it imagine.

Bryan Caplan ·

A test meant to find the best candidates must make perfect scores vanishingly rare, so that the top of the distribution can still be told apart.

Arvind Narayanan ·

The 'bitter lesson' that general methods eventually beat human domain knowledge does not hold for building AI-based products.

Bryan Caplan ·

Bad admissions tests survive on herding: no programme wants to be the one asking for a different exam, even a far better one.

John H. Cochrane ·

The American trade deficit is foreigners choosing to invest in America, not America being cheated.

John H. Cochrane ·

The binding constraint on building anything in America is permitting and litigation, not money.

John H. Cochrane ·

Cutting off trade with tariffs would force a huge interest-rate spike and probably a debt crisis, because the budget deficit would have to be financed at home.

Ken Norton ·

A product leader's position at the intersection of every function in the C-suite comes at the cost of belonging to any one group.

Elad Gil ·

Private-to-private mergers between young companies clear regulatory review more easily than public-to-private acquisitions.

Elad Gil ·

The capital required to train frontier AI models locks the leading LLM makers into an oligopoly.

Samuel Hammond ·

Efforts to rebuild American industry will fail until global trade and financial imbalances rebalance

Andrej Karpathy ·

LLMs reverse the usual pattern of technology diffusion: they benefit ordinary individuals far more than they benefit corporations and governments.

Molson Hart ·

Technological capability has to be practised to be kept: a country that stops making something forgets how.

Molson Hart ·

American manufacturing output cannot meaningfully rise without first building large amounts of new electricity generation.

Molson Hart ·

Building a factory in America takes at least two years before it can open, and months more before it runs well.

Molson Hart ·

Uncertainty about how long a tariff will last, not the rate itself, is what stops anyone from building a factory.

Helen Toner ·

Nonproliferation is the wrong frame for AI misuse; the workable strategy is to use the gap before a capability spreads to build defences against it.

Helen Toner ·

Delaying the open release of a new capability by months or years is worth doing, even though preventing its spread indefinitely is not.

Helen Toner ·

Adapting over time is a workable answer to humans misusing AI and no answer at all to losing control of systems smarter than us.

Helen Toner ·

Nobody has built a model that resists targeted jailbreaking, so the ability of AI companies to steer their own models is still limited.

Helen Toner ·

Arguing that human-level AI is very unlikely within twenty years is now the minority position, and the burden of proof has moved onto it.

March 2025

Mark Carney ·

The old Canada-US relationship, built on ever-deeper economic integration and tight security cooperation, is over.

Julie Zhuo ·

Mocks and docs are dying: AI-native teams prototype and prune with code-first prototypes instead of static mocks and product requirement documents.

Troy Hunt ·

A modern phish is automated end to end: the stolen credentials are used and the data exported within moments of being entered.

Julie Zhuo ·

Taste is the ability to distinguish excellence from mediocrity, and the hand can never consistently produce better than the eye can discern.

Rahul Vohra ·

Everything a company can build is either solution deepening or market widening, and doing one means not doing the other.

Rahul Vohra ·

A company deserves everyone in it, the chief executive included, to spend as much time as possible in their zone of genius — which is an org design problem.

Rahul Vohra ·

What creates real virality is not viral mechanics but word of mouth, and word of mouth is a thing you can only get by building a brand.

Rahul Vohra ·

Reaching product-market fit requires deliberately not acting on the feedback of many of your early users, at the same time as listening intensely.

Rahul Vohra ·

Changing how you run a survey invalidates every number you collected before it: you are starting a new baseline, not continuing the old one.

Rahul Vohra ·

Business software should be built the way games are built, because products made like games are the ones people fall in love with and tell their friends about.

Rahul Vohra ·

Rewards undermine intrinsic motivation, which is why gamification does not work; where it appears to work, the underlying experience was already designed as a game.

Rahul Vohra ·

The best games are constructed out of toys, so a feature should be fun to play with before it is ever attached to a goal.

Rahul Vohra ·

Which AI features users will love is genuinely unpredictable, and the ones a founder is most confident about are the ones that get used least.

ThePrimeagen ·

To actually understand a piece of technology, rebuild it from scratch rather than read about it.

ThePrimeagen ·

Most of software is not the new stuff; it is old, crappy code somebody has to maintain, and it is full of genuinely hard problems.

ThePrimeagen ·

Tech's self-image as the smart people designing for dummies includes nobody, and it deserves to be destroyed.

ThePrimeagen ·

Leaning on AI trades knowledge for speed, and the price may be a generation of programmers in which juniors stay juniors forever.

ThePrimeagen ·

A speed difference between languages that share the LLVM backend measures the benchmark author, not the languages.

ThePrimeagen ·

A project only shows you whether your design decisions were any good at around five to ten thousand lines, and that is where the most learning happens.

ThePrimeagen ·

VS Code and Copilot are loss leaders in an embrace-extend-extinguish play, raising the switching cost until nobody can leave.

Jakob Nielsen ·

The return on UX design work is declining because the usability battles have been won: the low-hanging fruit was picked and interfaces are now reasonably good.

David Heinemeier Hansson ·

Smartphones are not the main cause of child fragility

Elliot Hershberg ·

Revolutionary ideas eventually become universal building blocks for the next wave of innovation.

Jakob Nielsen ·

Vibe coding lets users design and build their own software, upending the user-centred design paradigm that has held since 1947.

Tyler Cipriani ·

There is an unmet opportunity for an open, Linux-based tablet to dominate the e-ink market.

David Heinemeier Hansson ·

Climatism is stepping up as a suitable alternative to religion for the Danes

David Heinemeier Hansson ·

Small acts of environmentalism can be used to buy a good conscience

Simon Willison ·

Claude tends to have good default taste in web page design.

Nelson Elhage ·

Impressive-looking benchmark improvements often result from comparing against a poorly-tuned baseline rather than a genuine advance.

Nelson Elhage ·

Compiler mechanisms that force a required optimization to occur, rather than relying on incidental compiler behavior, are a more robust way to write performance-sensitive code.

Jakob Nielsen ·

Deciding what software should do and why is still a human job: vibe coding changes the implementation phase far more than the planning phase.

Jakob Nielsen ·

Making coding easy makes design more critical, not less: when anyone can produce working features, the quality of the experience is what sets products apart.

Jakob Nielsen ·

Mockups existed because coding was hard; now designers can skip static mockups and go straight to building a UI that works.

Austin Vernon ·

Making a commodity cheaply is a harder startup problem than making something valuable, because the product cannot carry a premium.

Patrik Schumacher ·

The design of modern buildings is no longer informed by a living, critical discourse.

Patrik Schumacher ·

The societal function of the built environment can only be fulfilled by an academically based, discursively empowered profession.

February 2025

Kyle Chayka ·

The Trump-Musk government's approach descends from a Silicon Valley ideology that prizes frictionless technology built by elite engineers over messy human institutions.

Yoram Hazony ·

Recruiting Sunni fighters to fight IS is implausible as long as the Assad regime is massacring the Sunni population.

John Siracusa ·

Building a new app with Swift 6 from the start is easier than migrating an existing, released app to it.

Jakob Nielsen ·

AI agents will end traditional UI design and accessibility, because users will stop visiting websites and act only through their agent.

Thiago Costa ·

When beauty is the guiding principle, a product is forced to be simpler, into something most people would understand.

Thiago Costa ·

Financial products pile up data and leave users to interpret it; AI is good at interpreting data, so it can help people understand it without telling them what to do.

Thiago Costa ·

Having no deadline is a blessing and a curse: a designer can linger on a goalpost of perfection that keeps moving and never ship.

Thiago Costa ·

Being a good designer means saying no to most of your own ideas, and yes to the ones that could work within your product.

Thiago Costa ·

A designer improves by validating their old work through their new work.

Thiago Costa ·

Design sits at an intersection of art and science; in the art side, redesigning and experimenting, you are thinking about making something that looks really good, not about data or customers.

Thiago Costa ·

Art was the original UI: its techniques evolved over thousands of years, and too few designers draw on them.

Thiago Costa ·

Designers should be curious about movies, art and music, because everything outside UI and apps has something to teach them.

Thiago Costa ·

Designers should take inspiration from the world directly rather than from other UI, which is itself already a distillation of the world.

Thiago Costa ·

A shadow in an interface implies a light source, so an elevated element needs a highlight where the light falls, not just a drop shadow.

Thiago Costa ·

Caravaggio is the master of light, and a designer can learn lighting for an interface by putting one of his paintings next to it and copying the light source.

Thiago Costa ·

Rules like 'elevated surfaces get lighter in dark mode' are good guides but do not hold in the real world, where a darker keyboard can sit on a lighter table.

Thiago Costa ·

Small details add up: people who are not attuned to design still sense a product's overall polish.

Thiago Costa ·

A button takes ten seconds to make, but texture can make a customer curious enough to want to touch it, the way things work in real life.

Thiago Costa ·

Lacking founder-market fit can be a blessing in disguise rather than a weakness.

Thiago Costa ·

Curiosity is probably the number one quality a designer can have.

Thiago Costa ·

Taste is practised by exposing yourself to how the world is ordered, and understanding how that makes everything look.

Thiago Costa ·

AI will replace designers, but not all of them.

Thiago Costa ·

As tools get better the value of craft is increasing: a landing page spun up in minutes does not, for now, feel like one someone poured their heart into.

Thiago Costa ·

AI that builds UI is trained on past UI, so breaking out of that mould will still take a person willing to break the rules.

Thiago Costa ·

A design course should start with history and art, grounding students in the world, rather than with tools that may be gone tomorrow.

Thiago Costa ·

Designers need to be good communicators: an idea they cannot articulate to their team rarely becomes something meaningful.

Thiago Costa ·

Beauty matters to a business: it simplifies, and makes people understand, care about and remember a product.

Thiago Costa ·

Design is a visual profession, but the best designers are the ones who listen really well.

Alice Evans ·

Falling birth rates ended cousin marriage in the Arab world by leaving too few cousins, pushing marriage onto individual choice.

Bryan Caplan ·

The main function of education is signalling rather than building skill, so subsidising it mostly inflates credentials.

Adam Grant ·

Reading novels does more to build empathy than watching TV and films.

Elliot Hershberg ·

The number of diseases with a known molecular explanation has grown far faster than the number of new drugs developed to treat them.

Miles Brundage ·

AI will eliminate many millions of jobs within the next few years, and that disruption will shape the politics of everything else about AI.

Mandy Brown ·

Uncertainty is the condition that makes life possible.

Dylan Patel ·

The only thing stopping China from building the largest data centre in the world is access to chips; the power and the industrial capacity are already there.

Dylan Patel ·

Building fabs in America does not reduce dependence on Taiwan: without Hsinchu, TSMC's Arizona plant would stop producing within a couple of years.

Dylan Patel ·

Completely in-sourcing semiconductor manufacturing to the United States is possible, and it would take a decade and a trillion dollars.

Dylan Patel ·

Superhuman persuasion will arrive before superhuman intelligence, which means whoever builds a model can embed their ideals in it long before anyone reaches AGI.

Dylan Patel ·

The bottleneck on America's AI buildout is transmitting power, not generating it: in parts of the US, moving electricity costs more than making it.

Dylan Patel ·

When writing software becomes cheap, platform SaaS dies: companies stop buying Salesforce and start building their own business logic, as China already does.

Nathan Lambert ·

Open-source AI is still an ideological mission rather than an ecosystem, because it has none of the feedback loops that make open-source software compound.

Miles Brundage ·

AI progress is faster than people expect, and ordinary scaling can be enough to solve problems that looked very hard.

January 2025

Doug O'Laughlin ·

Cheaper AI will be used more in the long run and still strand the hardware built for it in the short run, exactly as unlit fibre did after 2001.

Ken Norton ·

Good storytelling doesn't require inventing something entirely new; it can be built from existing material.

Nelson Elhage ·

AI tooling and interface design lags behind the underlying models' actual capabilities.

Nelson Elhage ·

As AI makes code cheaper to produce, understanding and architectural design become more valuable.

Jan Leike ·

Building AI that can actually be trusted is the goal; containing an AI known to be misaligned is not a substitute for it.

David Heinemeier Hansson ·

Conservative values can enable progressive outcomes

Oliver Kim ·

Japanese colonial rule built a far deeper state in Korea than European rule ever built in sub-Saharan Africa.

Oliver Kim ·

Without a stock of trained administrators no development strategy of any kind can even begin, whichever theory of growth you hold.

Oliver Kim ·

Pre-colonial statehood in East Asia carried through the colonial period and on into the post-independence era.

Oliver Kim ·

Africa before colonisation was not stagnant for lacking states: states formed where they paid, and other arrangements suited other places.

David Heinemeier Hansson ·

Europe lacks the ability to change course like the United States

Donald Trump ·

American manufacturing can be rebuilt on domestic oil and gas, which the United States holds more of than any other country.

Elena Verna ·

A rebrand or marketing-site redesign has never once produced good performance results.

Elena Verna ·

Promising that a homepage or marketing-site redesign will drive acquisition is a failed promise, and usually a million-dollar one.

Jennifer Burns ·

Politics now runs mood first and ideas second, the reverse of how argument worked for most of the twentieth century.

Jennifer Burns ·

The great mistake of recent decades was forgetting that inflation built the last political order, and underestimating what its return would do to this one.

Elena Verna ·

Acquisition built on somebody else's algorithm can be taken away at any moment, which is why it must sit below earned channels on a growth roadmap.

Elena Verna ·

As search turns into an AI interface, a company with no earned acquisition channel on its roadmap is in serious trouble within one to two years.

Jennifer Burns ·

Ayn Rand's strangeness was both her strength and her undoing, because she used herself as the model of humanity and never grasped how unlike other people she was.

Elena Verna ·

Half of Dropbox's acquisition comes from people sharing files, an earned channel no competitor can buy their way into.

Jennifer Burns ·

Popular postmodernism has turned into the opposite of what it began as, re-inscribing the binaries it set out to dissolve as oppressor and oppressed.

Jennifer Burns ·

A genuinely original thought is nearly impossible, because everyone is born into traditions of thinking that most people never notice, let alone question.

Elena Verna ·

Testing shades of a colour is an early-2000s tactic that no longer drives anything.

Chip Huyen ·

As AI products converge on the same underlying models, product design becomes the real differentiator, not the model itself.

Chip Huyen ·

Building a working AI demo is far easier than turning it into a reliable production product.

Doug O'Laughlin ·

The AI build-out is not yet a bubble, because it lacks the self-reinforcing feedback loop that drove the railway and telecom manias.

Doug O'Laughlin ·

If the AI build-out does become a bubble, power and data centres are the assets that will carry it.

Miles Brundage ·

Securing model weights against theft is a more urgent policy priority than safety research, because security is built out of atoms and takes calendar time that safety does not.

Lyn Alden ·

Structural fiscal deficits have surpassed private sector lending and monetary policy as the primary driver of economic activity and inflation.

Jacob Falkovich ·

Subscription platforms like Substack and OnlyFans tend to become a Girardian bubble, where a few visible successes obscure the many who quietly give up after failing to build an audience.

Tobias Sjösten ·

Mastering one great pursuit requires turning down many other good options.

Tobias Sjösten ·

Building strength and muscle mass, along with cardiovascular fitness, are among the most important factors in living a long, healthy life.

December 2024

Simon Willison ·

Concerns about the energy cost of an individual LLM prompt are no longer credible, though the environmental impact of the AI datacenter buildout remains a real worry.

L. M. Sacasas ·

A full life cannot be delegated, because what makes it full is exactly the whole-person involvement that delegation removes.

Adam Frank ·

Our technosphere is immature: it is currently destroying the conditions it needs in order to sustain itself.

Adam Frank ·

Life is going to be common in the universe; the genuinely open question is how long technological civilisations last.

Adam Frank ·

No UFO or UAP data comes anywhere near the standard of evidence required to claim non-human technology.

Miles Brundage ·

AI that exceeds human performance in nearly every cognitive domain is almost certain to be built within the next few years.

Alex Telford ·

The West has not yet internalised that China is a source of genuinely new drugs, not just a maker of ingredients and copies.

Ben Vinegar ·

When a growing SaaS product's architecture breaks down, the developers he has seen succeed are the ones who fix it one piece at a time, not the ones who want it rebuilt.

François Chollet ·

The memorize-fetch-apply paradigm behind LLMs can reach arbitrary skill given training data, but it cannot adapt to novelty or acquire new skills on the fly.

François Chollet · no longer holds

The memorize-fetch-apply paradigm behind LLMs can reach arbitrary skill given training data, but it cannot adapt to novelty or acquire new skills on the fly.

Charity Majors ·

Hire for a person’s unique strengths rather than for their lack of weaknesses, because a team exists to cover what any one person is not.

Kelly Starrett ·

Spending 20 to 30 minutes each evening sitting on the floor builds exposure to normal joint ranges.

Kelly Starrett ·

If you are in pain before exercise, two or three minutes of soft-tissue work is a low-cost, safe and effective way to feel better and open a window to move.

Kelly Starrett ·

On a roller, at a stiff or sore spot, contract the muscle, breathe, then slowly relax, and repeat two or three times to desensitize it and improve range of motion.

Alon Levy ·

Decentralisation does not explain why Spain builds subways cheaply and Britain does not: the cheap countries include Europe's most centralised ones.

Alon Levy ·

A city's first rail line comes in at a reasonable cost and every line after it does not, because the follow-up outgrows the in-house team that oversaw the pilot.

Alon Levy ·

The way to learn how to build cheaply is to find what every low-cost country has in common, not to study one famous city.

November 2024

Casey Handmer ·

Now that Starship flies, we should build the largest space telescope physics allows.

Casey Handmer ·

NASA has deferred its next great telescope so far that the people who would build it will have retired first.

Lilian Weng ·

It is fundamentally difficult to design a reward function that accurately captures the intended goal in reinforcement learning.

Lilian Weng ·

More capable AI agents are more likely to find and exploit flaws in their reward functions.

Patrik Schumacher ·

The human sense of beauty serves as an intuitive mechanism for recognizing utility.

Jakob Nielsen ·

AI will flatten the design profession: when ten AI-enabled designers do the work of a department of a hundred, the need for higher management levels shrivels.

Shane Parrish ·

Genuine malice exists, but in most interactions stupidity is a far more common explanation than malevolence.

Tyler Cipriani ·

Tracking your personal finances doesn't need to be modeled all at once; it can be built up incrementally, one session at a time.

Austin Vernon · by

What stands between us and cheap lightweight vehicles is that they are illegal, not that they are hard to build.

Casey Handmer ·

Past a point it is cheaper to overbuild solar than to add batteries, because a battery kept for the rare bad day is barely used.

Brian Potter ·

The high-purity quartz of Spruce Pine is a real chokepoint in chipmaking, but not an irreplaceable one.

Alice Evans ·

Immigration cannot hold a rich country's working-age population steady, because doing so would require it to become 40 to 50 per cent foreign-born.

Alice Evans ·

A multi-ethnic society only gets the benefits of diversity if it deliberately builds a shared national identity.

October 2024

Yoshua Bengio ·

Malicious use of AI could enable extreme concentrations of economic, political or military power.

Yoshua Bengio ·

Racing to build AGI first is a race where everyone could lose.

Yuval Noah Harari ·

Agricultural cities that were meant as a paradise for humans turned out to be a paradise for germs

Yuval Noah Harari ·

Humanity has made immense progress in building trust between strangers and is rushing to throw it away

Yuval Noah Harari ·

Institutions are the key to trust between millions of strangers, because intimacy cannot scale

Yuval Noah Harari ·

You can build terrible societies without the truth, but a good society or a good life needs a strong basis in truth

Michael Liebreich ·

The comparative cost of hydrogen fuel-cell cars versus battery EVs depends on which infrastructure costs are counted and to whom they are allocated

Graham Hancock ·

That an advanced civilisation existed in the Ice Age is a hypothesis offered to answer the puzzles of prehistory, not something anyone has proved — and it is worth inquiring into anyway.

Graham Hancock ·

Göbekli Tepe was not the beginning of monumental building but the summation of a whole regional culture that was already under way before it.

Graham Hancock ·

The worldwide flood traditions record the Younger Dryas, not local floods that storytellers mistook for global ones.

Graham Hancock ·

A cosmic impact can be established from its proxies alone without a crater — that is exactly how the dinosaur extinction was first argued.

Graham Hancock ·

His case does not rest on the impact hypothesis: what is load-bearing is that the Younger Dryas happened at all, whatever triggered it.

Graham Hancock ·

Calling the lost civilisation advanced means only that it could calculate longitude, not that it built machines.

Graham Hancock ·

The Sphinx was originally lion-headed as well as lion-bodied, and the pharaoh's head it wears is a later re-carving of a ruined lion's.

Graham Hancock ·

The Great Pyramid was never a tomb — it is a curiosity device built to make later cultures investigate it.

Graham Hancock ·

Nobody knows how the Great Pyramid was built: neither ramps nor wet sand can lift seventy-ton granite blocks three hundred feet into the air.

Nabeel S. Qureshi ·

The world needs more companies that compete on execution, like SpaceX and Palantir, rather than on political games.

Matt Mullenweg ·

Building a business on open source is a good thing; the objection is to trademark abuse, not to commercial use.

Patrik Schumacher ·

The psychological wellbeing of workers should be addressed by focusing on conditions that maintain or enhance productivity levels.

Patrik Schumacher ·

The psychological wellbeing of end-users is best served by the enhancement of the design's social functionality.

Carlos Moreno ·

Prioritising proximity to services is crucial for building resilience against housing and urban challenges.

Molly White ·

What has gone wrong online belongs to the industry that grew up around the web, not to the web itself, which is still a substrate anyone can build on.

Molly White ·

One person building something honest and small can shift how a whole industry is seen, without a platform or a job at a big publication.

Michael Truell ·

If AI builds more and more of a system, it needs to verify as well as generate, including checking its own code.

Michael Truell ·

Automatic context given to coding models should clear a high bar for accuracy and relevance, since more context makes them slower, costlier and more confused.

Michael Truell ·

The best engineering involves many tiny decisions about what exactly to build and hard trade-offs between speed and cost.

Michael Truell ·

This is a really exciting time to be building software, and programming today is more fun than it was in 2012 or 2013.

David Heinemeier Hansson ·

Competition in the cloud requires avoiding proprietary setups

Paul Graham ·

Most people find writing hard because writing well requires thinking clearly, and thinking clearly is hard.

September 2024

David Heinemeier Hansson ·

The web does not have to be as complicated as it has become

Ed Barnhart ·

Pyramids were being built on the coast of Peru thousands of years before anything in Egypt.

Ed Barnhart ·

Pyramids appear all over the world because that is the shape a big building takes without cement, not because of any shared source.

Ed Barnhart ·

The Maya built their sacred calendar out of the human gestation period: they looked into their bodies rather than up at the sky.

Ed Barnhart ·

Calling North America's earthworks mounds is a western bias; what those people built were pyramids.

Ed Barnhart ·

We are wasting our time trying to reverse the climate crisis instead of building the technology to survive it.

Danielle Ofri ·

Inconsistencies in how lawsuits are decided across different regions could lead to unequal access to routine medical care like vaccines and birth control.

Lyn Alden ·

Bitcoin tracks the direction of global liquidity more consistently than any other major asset class.

David Heinemeier Hansson ·

The meme that programmers no longer need to be competent is a trap: the path is competence, and AI does not change that.

David Heinemeier Hansson ·

No-build JavaScript works in production and caches better: one bundle turns over every time a semicolon changes, while small separate files live forever.

David Heinemeier Hansson ·

Running Linux on your own laptop is a differentiating experience for a web developer, precisely because it is the path less taken.

David Heinemeier Hansson ·

Terminal interfaces exist out of necessity, not because they are the right design

Scott D. Miller ·

A therapy is only shown to be more effective if it is compared against another treatment intended to help, not against a control nobody expected to work.

Nat Bullard ·

Plug-in hybrids have moved past being a worst-of-both-worlds compromise and are now genuinely thriving.

David Heinemeier Hansson ·

Passkeys can become a halfway house to a password manager

David Heinemeier Hansson ·

Tying passwords to one computing ecosystem can make it hard to use alternative computers

David Heinemeier Hansson ·

Programmers are more expensive than computers

David Heinemeier Hansson ·

Optimizing for programmer productivity is more valuable than optimizing for computer efficiency

Yuval Noah Harari ·

Democracy is built on information technology, so an earthquake in information technology is bound to shake it

Brian Potter ·

American shipbuilding capacity has collapsed to a rounding error beside China's.

Brian Potter ·

American shipbuilding has been uncompetitive for a century; it is not a recent decline.

Brian Potter ·

Wartime shipbuilding surges have never left America with a commercial shipbuilding industry afterwards.

Brian Potter ·

Inventing a manufacturing technology and building a competitive industry on it are separate achievements, and America has repeatedly done only the first.

Brian Potter ·

What has kept American shipbuilding uncompetitive is the high cost of its inputs, above all labour and steel.

Brian Potter ·

Commercial shipbuilding keeps moving to whichever country has the cheapest labour, and America has never been that country.

Brian Potter ·

Whether a country builds ships well comes down to whether it actually wants to compete internationally, and America never has.

Brian Potter ·

Sheltering an industry from competition is not the same as making it competitive, and American maritime policy has only ever done the first.

David Epstein ·

Career zig-zaggers are more likely to feel unappreciated while building a broad skill set, and also more likely, eventually, to reach the top of their work.

David Epstein ·

In pursuit of a head start, people often skip exploring and jump straight to narrowly exploiting what they already know.

August 2024

Ted Chiang ·

What a generative model makes is a replication of art rather than art, because it can only imitate or average the choices human makers have already made.

Ted Chiang ·

The small choices made while executing a work matter as much as the big idea behind it, so inspiration cannot be separated from the labour.

Ted Chiang ·

A language model is not using language at all, because language requires an intention to communicate.

Matthew Green ·

Telegram urges people away from default-encrypted messengers while refusing to build the features that would encrypt its own users.

Matthew Green ·

Encryption people have to switch on protects almost nobody; what protects people is what a service turns on for everyone.

David Heinemeier Hansson ·

People tend to say in surveys that they want more options than they actually use or want in practice

Branko Milanovic ·

The fall of communism removed the reason Western capital had to accommodate labour.

Pieter Levels ·

The hard part of building a product in a month is working out what not to build.

Pieter Levels ·

The campaign to get young developers onto new frameworks is a business model: give away the framework, then charge the startups that build on it to host it.

Pieter Levels ·

Society would save time if software were built with the simplest, most pragmatic technology that works.

Pieter Levels ·

You should build by solving only the next problem in front of you; most people fail because they look too far ahead.

David Heinemeier Hansson ·

Novel software development cannot be specified upfront

David Heinemeier Hansson ·

It is better to live in a country that allows offensive speech

Layne Norton ·

Machines or free weights, low or high reps, failure or a few reps short all build roughly the same muscle, provided the work is hard.

Steve Coast ·

Fear and vanity, not a lack of ideas, are what keep most people from building new things.

Steve Coast ·

New ideas usually fail outright or require drastic changes before they can succeed.

Tobias Sjösten ·

Many first-time entrepreneurs mistakenly build a solution before identifying the problem it is meant to solve.

Jakob Nielsen ·

AI design tools are starting at the low end and eating their way up: within ten years even the richest brands will rely heavily on them.

July 2024

Chip Huyen ·

Preparing context for a language model plays the same role that feature engineering played for classical machine learning models.

Frank Lantz ·

Designed addiction in gambling games is real and can consume a person’s life.

Stacy Sims ·

Resistance training should be the bedrock alongside true high-intensity work; long easy zone 2 sessions suit mental health but are not optimal for body composition and metabolic health.

Nikita Bier ·

Growing a product means shortening the time to value so the user reaches the Aha Moment in seconds

Ben Kuhn ·

Trust between colleagues comes from shared mental models built through mutual investment.

Ben Kuhn ·

Fast-growing companies will never achieve full pairwise trust between all employees.

Dan Ariely ·

Displaying a moral code that carries no meaning is like hanging a poster in a foreign language

Paul Graham ·

Resilience means setbacks cannot change your morale, not that they cannot change your mind, and persistence often requires changing your mind.

June 2024

Alon Levy ·

The English-speaking world builds infrastructure expensively because it is behind, not because it is rich.

Alon Levy ·

New York's subway cost premium is overstaffing, oversized stations and procurement, not high wages.

Alon Levy ·

Britain has a systemic problem building large infrastructure projects, including reservoirs

Danielle Ofri ·

Federal regulations are needed to require facilities to be upfront with their pricing or to eliminate the price differential between hospital-affiliated outpatient departments and other settings.

Aravind Srinivas ·

A startup does not have to set out to build the greatest business in history; a merely good business is a legitimate thing to aim at.

Aravind Srinivas ·

Prompt engineering is not a durable skill, because a good product is one that works for the user who never learned to phrase a question.

Aravind Srinivas ·

The transformer was close to the last answer in architecture: nothing about the core design has changed since 2017.

Aravind Srinivas ·

Spending most of the compute memorising the world in pre-training is brute force; what we want is a model that reasons the way a student sits an open-book exam.

Aravind Srinivas ·

If reasoning can be separated from facts, the frontier labs lose their advantage, because nobody needs a million-GPU cluster any more.

Sara Walker ·

Life is not chemical; it merely emerges in chemistry, because chemistry is the first space the universe builds that is too large for it to exhaust.

Sara Walker ·

Non-equilibrium physics will not explain life, and entropy has almost nothing to do with it.

Sara Walker ·

Computation sits at the top of reality, not at its base: it is something our biosphere built after billions of years of evolution.

Sara Walker ·

A theory of reality that cannot name the question it is answering is not worth building.

Sara Walker ·

The story of a superintelligence in a box that gets smart and kills everybody is not science fiction — it is bad science fiction.

Brian Potter ·

Available power, not land or capital, is what now limits how many data centres get built.

Brian Potter ·

Data centres and the grid that feeds them run on incompatible clocks: one is a two-year project, the other a multi-decade investment.

Brian Potter ·

Air cooling has run out of headroom for AI racks, so liquid cooling is now a requirement rather than an upgrade.

Brian Potter ·

The efficiency gains that kept data-centre electricity use flat for a decade are largely spent, so new demand now arrives as new power demand.

Dillon Mulroy ·

JavaScript has the lowest bar of entry to being productive, but one of the highest bars of any stack to building high-quality, stable, reliable software.

David Heinemeier Hansson ·

Linux can look and feel good out of the box

David Heinemeier Hansson ·

Good defaults can help developers acquire a taste for Linux

David Heinemeier Hansson · no longer holds

Ruby on Rails took off because its creator evangelized it by arguing its superiority against the competing alternatives.

May 2024

Daniel Herriges ·

Building a lot of housing in an expensive city will not lower rents there; it lowers them in the cheaper places that were absorbing its spillover demand.

Morgan Housel ·

Susceptibility to FOMO, especially in markets inflamed by social media, leaves little hope of building wealth over time.

Charan Ranganath ·

The point of a good memory is not to remember more but to remember better, because forgetting most of what you experience is what human brains are built to do.

Charan Ranganath ·

Forgetting names is not a defect of memory: a name is an arbitrary label with nothing to attach it to, and arbitrary material is exactly what brains are worst at.

Charan Ranganath ·

In a monotonous stretch of life the days crawl and the weeks vanish, because the sense of how much time passed is built out of what you can remember happening.

Charan Ranganath ·

Task-switching destroys memory by chopping an experience into artificial event boundaries, so you spend the whole session catching up instead of building a model.

Visakan Veerasamy ·

The design of much of civilization requires people to be psychologically and emotionally mutilated to interface with it.

Timothy B. Lee ·

A robotaxi that still needs remote human help loses the cost advantage that motivated it, which pushes out when it can scale.

David Heinemeier Hansson ·

Beauty is a motivator

Varun Raj Manoharan ·

OpenAI's growth beyond model-building is limited by its lack of an ecosystem.

L. M. Sacasas ·

A device leaves you a user who is often the one being used, while an instrument you have practised on makes you something else.

L. M. Sacasas ·

A good life rests on things that demand skill and effort from you, not on goods that arrive requiring none.

David Heinemeier Hansson ·

The definition of normal should be broader

Paul Bloom ·

Empathy is a poor moral guide that leads to bad decisions.

Derek Lowe ·

A compound's AI origin story is not by itself a reason to expect it to do better in the clinic.

Tyler Cowen ·

Guests who answer by signalling empathy instead of saying something substantive are a reason to turn a bit more against empathy.

Kevin Esvelt ·

Ordering the synthetic DNA needed to rebuild the 1918 influenza virus is surprisingly easy, even from companies that screen customers' orders.

Frank Lantz ·

A strategy does not play a game; playing requires something that can see the game and choose to enter it.

Kevin Esvelt ·

The genome sequences of pandemic viruses and the protocols for building them from synthetic DNA are already freely available online.

Buck Shlegeris ·

Controlling a scheming model needs no research breakthrough, which is what makes it the tractable half of the problem today.

Buck Shlegeris ·

Whether a model is controlled can be settled with capability evaluations, which makes control far easier to check than alignment.

Waqas Younas ·

Pursuing truth is the best aim for a life.

Brian Potter ·

The same law that made chips cheaper made the factories that produce them enormously more expensive.

Brian Potter ·

Fabs take roughly a third longer to build in the United States than in Asia, largely because of environmental review.

Brian Potter ·

A semiconductor fab is mostly its tools; the building around them is the cheap part.

Brian Potter ·

Lithography machines alone cost about as much as the entire building put up to house them.

Brian Potter ·

Every new process node makes a fab about 30 per cent more expensive to build than the one before it.

Brian Potter ·

The chip industry split into fabless designers and a handful of foundries because fabs got too expensive to own, not because anyone chose that structure.

Brian Potter ·

A semiconductor fab costs what it costs because it is a mass-production factory that has to hold atomic precision around the clock.

Gernot Wagner ·

Achieving net-zero emissions requires trillions of dollars in investment that will add to economic growth.

Gernot Wagner ·

Averting climate catastrophe requires economic growth.

Molly White ·

Nothing about the web technically prevents a return to what people loved about the old web; the good old days are available and would be easier to build now than they ever were.

April 2024

Andrew Huberman ·

Repeated short, maximal contractions of about ten seconds probably build stronger muscle-activation pathways than one long static hold.

David Heinemeier Hansson ·

Most people do not enjoy tinkering with their computers

Sean Carroll ·

The simplest explanation for why we have never noticed an alien civilization is that there are none out there.

Sean Carroll ·

A great filter is a weak explanation for the silence, because it would have to work every single time — one civilization with self-replicating probes could fill the galaxy.

Sean Carroll ·

Listening for alien radio is the wrong search: a probe that sits and waits beats a signal that passes you by, so an artifact in our solar system is likelier than a message.

Sean Carroll ·

The largest contribution physics can make to AI is energy efficiency: computers waste heat hilariously worse than brains do, and nobody has seriously tried to fix it.

Austin Vernon ·

A data centre powered by solar and batteries alone needs an array so oversized that the design is not worth building.

Austin Vernon ·

Cooling adds nothing to what a data centre produces, which is why it is the first thing a good design deletes rather than optimises.

Austin Vernon ·

Air cooling has run out of headroom for AI racks, so liquid cooling is now a requirement rather than an upgrade.

Austin Vernon ·

An off-grid data centre can delete its uninterruptible power supplies entirely, because the battery bank outside already is one.

Austin Vernon ·

A data centre should be split into independent power blocks fed directly at 1500 volts DC, rather than run through one central distribution system.

Austin Vernon ·

Off-grid data centres will win the way electric cars did, by redesigning the machine around the new power source instead of swapping the power source in.

Lilian Weng ·

Video generation requires more world knowledge than image generation because of the need for temporal consistency across frames.

Arthur C. Brooks ·

Money cannot buy happiness by itself, but money for experiences with loved ones, free time and good causes does add to it.

Morgan Housel ·

Market history teaches why things behave as they do, never what happens next, because the details never repeat.

Kenton Varda ·

RPC's reputation as broken is outdated: it was earned in an era of synchronous calls, before promises and async/await.

Kenton Varda ·

RPC fits how programmers already think, in function calls, which frees them from constantly translating between mental models.

Doug O'Laughlin ·

Air cooling has run out of headroom for AI racks, so liquid cooling is now a requirement rather than an upgrade.

Brian Potter ·

For most American housing the largest cost is putting up the physical building, not the land it sits on or the rules attached to that land.

Brian Potter ·

A house is built by trades that barely coordinate with each other, which is what makes housing cheap to design and almost impossible to redesign.

Brian Potter ·

Building materials are already about as cheap as civilisation knows how to make them, so cheaper materials are not where housing savings will come from.

Brian Potter ·

It is only in dense cities that land dominates the price of a home, and there the land is expensive because of the rules limiting what can be built on it.

Thomas Depierre ·

Any plan for open source sustainability or security that is not built around hobbyist maintainers will achieve nothing.

Kenton Varda ·

Connecting a server to a resource should be one step, not two: pointing the server at the resource and separately making the resource accept the server.

Kenton Varda ·

Designing code to be friendly to dependency injection can seem tedious, and it has been worth it every time.

March 2024

Seth Godin ·

Naming a narrow audience puts a creator on the hook, while claiming the work is for everyone is a way to hide from failure.

Seth Godin ·

Creativity is a choice: extending empathy to the person being served and announcing something into the future that might not work.

Seth Godin ·

Quality means meeting spec and nothing else, so a product riddled with defects that fall within spec is a high-quality product.

Seth Godin ·

Employees and customers do not want what founders want, and equity stakes become a demotivator in almost every organisation.

Brian G. Chow ·

A spacecraft built to service satellites is also a weapon, because the same robotic arm that refuels one can disable one.

Noah Smith ·

Housing affordability is almost the direct inverse of how many new homes are built, so the answer is to build more houses.

Annie Jacobsen ·

A nuclear-armed submarine is as dangerous to civilization as an asteroid, and unlike an asteroid it cannot even be located.

Annie Jacobsen ·

Atomic weapons destroy cities and thermonuclear weapons destroy civilizations, which makes the decision to build the thermonuclear bomb more confounding than the decision to drop the first two.

Sam Altman ·

The road to AGI is going to be a giant power struggle, and anyone building it should expect that rather than hope to avoid it.

Sam Altman ·

The board of a nonprofit answers to no one but itself, and the board governing the road to AGI ought instead to answer to the world as a whole.

Sam Altman ·

The openness that matters is putting increasingly powerful tools into people's hands for free as a public good, not open-sourcing the weights.

Sam Altman ·

AI should be released iteratively rather than built in secret, because AI and surprise do not go together and the world needs time to adapt before it is under the gun.

Sam Altman ·

Asking when AGI will arrive is a poorly formed question; the answerable question is when a system will be able to do a particular capability.

Sam Altman ·

What any individual human can do comes from collectively built scaffolding rather than from the individual, and AGI may turn out to be more like that scaffolding than like a single brain.

David Heinemeier Hansson ·

Governments are pursuing laws that seek draconian penalties for forbidden speech

Chip Huyen ·

The lower a layer sits in the software stack, the harder it is for an individual developer to build in it.

Linus Blomqvist ·

Making American farms more productive shrinks cropland in the rest of the world by more than it adds at home.

David Heinemeier Hansson ·

A company captured by ideology will produce useless products

David Heinemeier Hansson ·

Apple could quit Europe

Ted Gioia ·

Scrolling and swiping interfaces are designed to prevent closure, so that the user never reaches an end and keeps going.

Serhii Plokhy ·

Russian nationalism turned out to be as destructive to the multinational Soviet state as Ukrainian, Georgian or Estonian nationalism was.

Serhii Plokhy ·

The Russian Empire acquired a sixth of the Earth by integrating conquered elites rather than their peoples, and nationalism is what made that model fall apart.

Doug O'Laughlin ·

The rush to build high-bandwidth memory starves ordinary DRAM and NAND of capacity, so AI demand tightens the memory markets it never touches.

David Heinemeier Hansson ·

It is a blind spot to only use one platform

February 2024

Jakob Nielsen ·

Accessibility has failed as a way to make computers usable for disabled users, after thirty years of trying.

Jakob Nielsen ·

Within about five years the user interface will be generated afresh each time a user opens an app, so different users will get drastically different designs.

Tuomas Artman ·

The MVP as a hacky product built to validate an idea no longer works: in a saturated market you have to build something substantially better from day one.

Tuomas Artman ·

To compete with limited resources, scope down who you are building for as much as possible and solve the problems of specific people.

Sebastian Ruder ·

The embodied, interactive, multi-modal nature of human first-language acquisition is hard to replicate in current AI models.

Andrew Chen ·

Writing publicly is a more effective way to build a professional network and reputation than attending conferences.

David Heinemeier Hansson ·

Computing freedom requires not putting all eggs in one basket

Lynn Fisher ·

Remote work requires trusting one's team, and that trust is a choice that can be made at any time.

Lynn Fisher ·

Deep, meaningful relationships can be built with people entirely remotely.

Lynn Fisher ·

Whether a company is genuinely remote-first is best tested by what it plans for team members who can't attend in-person events.

Bill Ackman ·

Technology companies are the wrong place to look for durable businesses, because someone is always building a better version.

Bill Ackman ·

Incentives drive all human behaviour, which makes a management team's incentives central to valuing a company.

Bill Ackman ·

Investing well requires reversing the human instinct to grow excited as prices rise and fearful as they fall.

Bill Ackman ·

The DEI movement on campus comes out of a Marxist socialist way of looking at the world.

Ted Gioia ·

The executives building addictive platforms know what the harm is and build them anyway.

Elisabeth Bik ·

Recent advances in AI technology mean humans are already past the stage where they can distinguish a fake photo from a real photo.

Sebastian Ruder ·

Large research team sizes on frontier AI projects create friction that makes execution and decision-making harder.

Andrew Chen ·

A founder who lacks a clear sense of purpose will get lost during the inevitably difficult stretches of building a startup.

Tim Urban ·

Augmented reality and virtual reality are no longer separate categories, but points on one continuum.

Tim Urban ·

Virtual-reality hangouts are the next step in humanity's long effort to close the gap of physical distance in communication.

Steve Coast ·

AI should be approached as a collaboration between humans and computers, not a forced choice between them.

David Crawshaw ·

Programming is fundamentally the practice of making tradeoffs about what matters and what doesn't.

David Crawshaw ·

Most projects fail before they even begin because the activation energy required to start is too high.

Andrew Chen ·

Users vote with their clicks, and roughly half of them are lost at every click a product asks for.

January 2024

Andrew Chen ·

Users acquired through referral programs, free trials, coupons, or gamification tend to be lower quality than organically acquired users.

John Siracusa ·

A user interface feels better the more it lets people use their innate spatial and kinesthetic abilities.

Cory Doctorow ·

People stay on a platform they have come to dislike because leaving requires everyone they care about to agree to leave at the same time.

John Siracusa ·

The Vision Pro's look-then-gesture interaction is less direct than a touchscreen's literal touch interaction.

David Heinemeier Hansson ·

Computers should allow users to install any software

Dan Wang ·

China's industrial lead was manufactured by deliberate state industrial policy rather than by market forces alone.

Andrew Chen ·

Generative AI is as likely to invent new game genres as it is to make existing ones cheaper to build.

Dan Wang ·

American shipbuilding capacity has collapsed to a rounding error beside China's.

Dan Wang ·

America treats its inability to build and to admit skilled migrants as mysteries to be endured rather than problems to be solved.

James Gleick ·

Agency, which comes from acting in the world and living with the consequences, is what distinguishes people from machines.

Kyle Chayka ·

Fashion is for saying something about yourself, not for conforming to a data-driven standard.

Robin Hanson ·

Paying parents large sums of cash directly is the pronatal policy most likely to work, because every indirect policy has to guess what the real obstacle is.

Kyle Chayka ·

Algorithmic feeds seem to have nudged people into conforming to each other, a homogenised culture in which everyone accepts the average.

Chip Huyen ·

The probabilistic nature of AI models is what makes them well suited to creative tasks

Philip Guo ·

Speaking and listening are hardwired into the brain's innate language circuitry, whereas writing and reading are learned skills.

Andrew Chen ·

Frequent creative practice builds the skill needed to produce more work later.

Andrew Chen ·

An excessive focus on quality functions as procrastination, turning quick work into drawn-out projects.

Andrew Chen ·

Some books become bestsellers because their authors buy large numbers of their own copies through anonymous networks to disguise it.

Yuval Noah Harari ·

There is enough land between the Jordan River and the Mediterranean for everybody; the conflict is over stories

Yuval Noah Harari ·

AI automates purely cognitive jobs easily but struggles with jobs that need motor and social skills

Cate Hall ·

If you are only asking for things you get, you are not asking for enough, and rejection should be courted deliberately.

Cate Hall ·

Most traits people treat as fixed — agency, confidence, charisma, warmth — are learnable if you believe they are and work at them.

Cate Hall ·

Learning a new skill set requires crossing a moat of low status, and the willingness to be visibly bad in public is itself the advantage.

Charity Majors ·

If building the organisation is nobody’s first job, it will not get done well.

Charity Majors ·

Flat, manager-free organisations do not deliver the equality they promise: they get abandoned, and the informal power left behind is unaccountable.

Rob Pike ·

Engineers commonly confuse a solution with the problem it is meant to solve.

David Heinemeier Hansson ·

Sweden's longer open-door immigration policy left it in a world of hurt compared to Denmark

David Heinemeier Hansson ·

Style in code is invariably idiosyncratic

Austin Vernon ·

Shipbuilding productivity depends on clustering yards and suppliers in one place, and it decays when the cluster thins out.

Fred Wilson ·

Mainstream decentralized applications will emerge in 2024.

December 2023

Philip Koopman ·

One company doing one ill-advised thing damages an entire emerging industry, which is why safety is a shared problem rather than a competitive one.

Alex Telford ·

The hard part of drug discovery is knowing what to make - which target is worth drugging - not designing the molecule.

Alex Telford ·

How long a trial takes decides which diseases get drugs at all, because capital flows to the indications whose endpoints can be measured quickly.

Matt Johnson ·

Effective altruism demands hard evidence of effectiveness in some of its causes while accepting abstract conjecture in others, and never justifies the split.

Matt Johnson ·

The more effective altruism has absorbed itself in saving humanity, the less humanistic it has become.

David Heinemeier Hansson ·

Modern SSDs are probably good enough for queuing

Nelson Elhage ·

Profilers show where time is spent but not which of that time can feasibly be reduced or is worth optimizing.

David Heinemeier Hansson ·

Building for quality is a slow, smooth process that gets you to release quicker

Jeff Bezos ·

Efficiency and real invention are at odds, because genuine invention requires wandering and you have to give yourself permission to do it.

Jeff Bezos ·

Building a factory that turns out rockets at rate is at least as hard as designing the rocket in the first place.

Jeff Bezos ·

Space will only be unlocked when a valuable space company can be started in a dorm room, and that is the test of whether enough heavy infrastructure exists.

Jeff Bezos ·

Humans are social animals rather than truth-seeking ones, so telling the truth is not what we are built for and always costs energy.

Jeff Bezos ·

The things worth building a business around are the ones that will not have changed in ten years, and those are identifiable in advance.

Jeff Bezos ·

Human brains are doing something fundamentally different from today's AI, and the evidence is that they run on roughly twenty watts.

Jeff Bezos ·

PowerPoint is a sales tool and therefore the wrong instrument inside a company, because internally you are trying to find the truth rather than persuade anyone.

Ken White ·

Whether calling for the genocide of a group breaches a university's harassment policy genuinely depends on context.

Rob Pike ·

In a system built from interacting components, complexity is multiplicative: making one part more complex adds complexity elsewhere too.

Rob Pike ·

A small performance gain is not worth the complexity added to achieve it, since complexity's costs compound rather than add.

Sebastian Ruder ·

Full fine-tuning of large language models has become too costly, making parameter-efficient fine-tuning methods necessary.

Lisa Randall ·

Large language models are generalisations of what we already have; the breakthrough that would let AI discover things is still waiting to happen.

November 2023

Lant Pritchett ·

Wanting to preserve a national culture is a legitimate human impulse and not in itself a racist one.

Scott Alexander ·

The best opportunities to do good are the ones almost nobody is paying attention to.

Scott Alexander ·

Effective altruism's scandals are minor next to its record, and its ratio of accomplishments to scandals is as good as that of any large institution.

Scott Alexander ·

Effective altruism's work on the far future is not a bait and switch hiding behind its work on global health; both halves are worth defending.

Casey Handmer ·

Mars Sample Return as designed is too complicated, too slow and too expensive to succeed.

Emily Short ·

Presenting puzzle situations that require correct analysis of NPC motives can risk encouraging an instrumental approach to NPCs.

Brandur Leach ·

Transactions are the best idea in robust service design.

John Mearsheimer ·

A state's power rests on two material foundations only: the size of its population and its wealth.

John Mearsheimer ·

States are not inherently aggressive; anarchy leaves them no choice but to behave aggressively.

John Mearsheimer ·

Deterring China over Taiwan does not require being able to beat it, which the United States probably cannot do anyway.

John Mearsheimer ·

Hispanic immigrants are integrating into the American mainstream faster than the European immigrants of the nineteenth century did.

Julie Zhuo ·

Do not judge a product's beauty by its aesthetics; judge it by how it made you feel.

Julie Zhuo ·

If the people a product is designed for cannot generally agree that the design is good, it is not good.

Julie Zhuo ·

Obviousness comes from conforming to people's existing mental models: do not reinvent a common UI pattern unless the new one is at least twice as good.

Julie Zhuo ·

Better design does not mean more design: the most obvious designs are often invisible.

Sebastian Ruder ·

Models fine-tuned to imitate another model's outputs copy its style without acquiring its factual accuracy.

Casey Handmer ·

The binding constraint on building anything in America is permitting and litigation, not money.

Elon Musk ·

Human violence is not a human peculiarity: nature is not a peaceful thing, and every living thing in it is trying to kill something else.

Elon Musk ·

The counterintuitive right response for Israel in Gaza is the most conspicuous acts of kindness possible, because the attack was designed to provoke the most aggressive response it could.

Elon Musk ·

China's history says it is not an acquisitive power: it is not going to go out and invade a whole bunch of countries.

Elon Musk ·

The United States is a historical rarity: after 1945 it could have taken the whole world without losing a soldier, and instead it rebuilt the countries it had defeated.

Elon Musk ·

Be careful wishing for a world with no war at all, because what would be required to enforce it might be worse than a little war.

Elon Musk ·

No electric motor suitable for a humanoid robot existed anywhere in the world, so every part of Optimus had to be designed from scratch.

Dillon Mulroy ·

React and TypeScript developers are infantilized by the ecosystem's thought leaders; anything beyond a simple CRUD app will need generics, mapped types and memoization.

October 2023

Yoshua Bengio ·

Frontier AI systems cannot be presumed safe unless proven unsafe; developers must bear the burden of proof that risks stay within acceptable limits.

Steve Coast ·

People who are smart in a narrow way but don't read broadly are the ones who cause disastrous unintended consequences.

David Heinemeier Hansson ·

Human progress requires trying again after unsuccessful attempts

Hannah Ritchie ·

Meat substitutes have to become clearly cheaper than meat, not merely as cheap, before most people will switch.

Yuval Noah Harari ·

A coalition from the US and EU to Saudi Arabia and the Palestinian Authority should take Gaza from Hamas, rebuild it and demilitarise it

Casey Handmer ·

Every doubling of the solar built worldwide has cut its price by 30 to 40 per cent, and that learning rate is why the price keeps falling.

Casey Handmer ·

Batteries will replace transmission as the cheapest way to keep the lights on, and the grid will shrink rather than grow.

Casey Handmer ·

The climate policy everyone has quietly assumed is that poorer countries stay poor, and that is the wrong answer.

Casey Handmer ·

More than 95% of the energy humanity uses will come from solar panels by 2042.

David Heinemeier Hansson ·

In-person interactions are important for community building

Tyler Cowen ·

An inability to see the viewpoints of people unlike oneself is a genuine weakness in a social scientist, not merely a separate moral failing.

Tyler Cowen ·

Intelligent liberal elites can badly mismanage the colonies they govern, in ways that persist for decades or centuries.

Tyler Cowen ·

A highly dynamic economy that cheaply mass-produces robots may increase rather than decrease the risk of Malthusian resource dilemmas.

September 2023

Mark Zuckerberg ·

An AI copy of a person is worth building while they are alive, to meet demand for their attention, not as a way to talk to the dead.

Mark Zuckerberg ·

Building an AI that embodies a real person is a much harder problem than inventing a new AI character from scratch.

Mark Zuckerberg ·

Computers, TVs and phones are badly designed for people because they keep us sitting still, and the human experience is a physical one.

Brad Frost ·

A complex system designed from scratch never works and cannot be patched to work.

Brad Frost ·

Framework-specific implementations of design systems will diminish over time.

Jonathan Gagné ·

Burr geometry matters more than burr size for how uniform and unimodal a grinder's output is.

Davis Kedrosky ·

The military revolution in early modern Europe was a centuries-long coevolution of military technology and political organization rather than one sharp discontinuity.

David Heinemeier Hansson ·

TypeScript's type annotations cost more in friction than they return in safety, and a front-end framework is better off written in plain JavaScript.

Raph Koster ·

A puzzle whose categories can only be seen by the broadly educated is elitist by construction.

Raph Koster ·

A puzzle that invites guessing and then punishes wrong guesses stops the player learning the very method it could have taught.

Neri Oxman ·

The things humankind designed are precisely the things that separated us from nature, and the point of design now is to undo that separation.

Neri Oxman ·

What nature wants is to increase information while decreasing entropy, and that is the objective any design working with nature has to serve.

Neri Oxman ·

Living systems can be laid out on two axes, material sophistication and communication, and an organism that excels on one is usually poor on the other.

Neri Oxman ·

The way we make things is out of tune with the time dimension of the natural world, and that has to change even if it means designing on the scale of centuries.

Neri Oxman ·

A designer must never stop at the object: a shoe, a chair or a building is only honestly designed once you account for the whole system it is sourced from.

Neri Oxman ·

The future of design is soft control: building highly controlled systems whose whole purpose is to enable the loss of control.

August 2023

Michael Grunwald ·

Fixing food and climate needs solutions that scale absurdly fast, and the ones we have do not scale nearly fast enough yet.

Anton Howes ·

Historians should be required to publish photographs or transcriptions of the archival sources they cite, the way scientists are expected to publish their data.

Jacob Falkovich ·

Free will does not require violating the laws of physics, yet the subjective experience of it is real and meaningful.

Casey Handmer ·

You cannot make people adopt more expensive clean energy, because higher fuel prices are political suicide under any government.

Casey Handmer ·

Green hydrogen gets cheap by building cheaper, less efficient electrolysers that run only when the sun shines.

Alon Levy ·

You cannot make America build subways ten times cheaper while only ever talking to Americans.

Ken White ·

The way to tell legal analysis from partisan lying is to distrust absolute certainty and to trust conceded ambiguity.

July 2023

Cory Doctorow ·

Ranking first in a search has become near enough to being true, which is why search and spam are permanently at war.

Kenton Varda ·

Allowing an RPC call to be cancelled is almost always desirable in practice, and disallowing it can lead to subtle problems.

Paul Graham ·

The value of an audience does not grow anything like linearly with its size, so a handful of people who genuinely love the work is enough.

Paul Graham ·

Copying something from one field into another is so powerful that it is probably worth deliberately learning about other kinds of work.

June 2023

David Heinemeier Hansson ·

Every material has a grain, including the web, and designs that work against it end up weaker

George Hotz ·

What makes a relationship worth having is the other person's quirks and small annoyances, which is exactly what we will edit out of artificial companions.

George Hotz ·

Scaling up GPT will never produce AGI, because a model trained on cross-entropy loss cannot get there; reinforcement learning in rich environments is required.

George Hotz ·

Generative models will let us build media a person genuinely cannot look away from, and that is the likeliest way humanity ends.

George Hotz ·

We will build superintelligence long before we build anything as robust as a bird, because nothing we have built can survive in nature at all.

George Hotz ·

Reproduction is the thing life is built on and the thing machines cannot do; for all its progress, the silicon stack is nowhere near reproducing itself.

George Hotz ·

You cannot build a successful AI chip without first writing a competitive machine learning framework — that, not the silicon, is why Google's TPU is the only one that worked.

George Hotz ·

AI safety advocates will build the very thing they fear: one controlled, aligned model is the only route to being turned into paperclips.

George Hotz ·

Arguing that nobody should have powerful AI is a respectable position; arguing that only trusted authorities should have it is not.

George Hotz ·

Charging users money is the fix for social media, because a paying user is the customer rather than the product.

George Hotz ·

Recommendation algorithms are not sinister; they simply predict engagement, and what humans engage with most is outrage.

George Hotz ·

Twitter needed refactors before features: a code base that bad has to be fixed before anything new is built on top of it.

George Hotz ·

The first move on a legacy code base is tests you can trust, before a single line of the code itself is changed.

George Hotz ·

Atheism is silly: anyone who has built a world inside a game should be able to see that this one was created too.

George Hotz ·

Social and advertising businesses come and go; a technology company that wants to endure has to build hardware.

Ken White ·

A threat loses First Amendment protection only if the speaker was at least reckless about it being taken as a threat.

David Heinemeier Hansson ·

Achieving financial freedom does not extinguish the drive to work

Jakob Nielsen ·

Graphical interfaces have superior usability because they show people what can be done rather than requiring them to articulate what they want.

Jakob Nielsen ·

Today's chat-based AI tools have deep-rooted usability problems and are not the interfaces we will be using in a few years; prompt engineering is a symptom, not a career.

Jakob Nielsen ·

AI is the first new user-interface paradigm in sixty years: intent-based outcome specification, where the user says what outcome they want rather than how to get it.

Jakob Nielsen ·

The graphical user interface will survive AI in a less dominant role: future AI systems will have hybrid interfaces that combine intent and command.

Jacob Falkovich ·

Finding a good spouse depends more on one's orientation toward building relationships and spotting non-trivial compatible traits than on matching a status checklist

Davis Kedrosky ·

Small parties of Spanish conquistadors were able to defeat much larger Native American armies primarily because of superior steel weapons and organizational capacity.

Visakan Veerasamy ·

Textual snapshots of intuition are no substitute for intuition itself.

Visakan Veerasamy ·

Reading and analysis are no substitute for experience in the field.

Marc Andreessen ·

The idea that AI will decide to kill humanity is a category error: it is math and code, built and controlled by people, not a living thing shaped by evolution.

David Heinemeier Hansson ·

Not all connections benefit from having the friction removed by digital systems

May 2023

David Heinemeier Hansson ·

Getting people to act is a function of exposure

Tom Preston-Werner ·

Building a secure, performant GraphQL API is tricky and adds real cost during early-stage app prototyping.

David Heinemeier Hansson ·

Processes that require honest participation need to be seen by the participants as intrinsically valuable

George Monbiot ·

The traditional meat-eating diet the romantics revere was built on colonisation and the clearance of other continents.

David Heinemeier Hansson ·

Nothing can substitute for spending time together in person to build bonds and trust

David Heinemeier Hansson ·

People who manage product managers should still spend almost all of their own time doing individual contributor work

Daniel Jalkut ·

The absence of AI features like chatbots and code assistants in mainstream software will soon be seen as a major shortcoming.

Nabeel S. Qureshi ·

Virtue ethics may be a more promising paradigm for moral AI than utilitarianism, which would tend to make AI power-seeking, or deontological rules.

Tim Urban ·

Expecting parents should be required to take a mandatory course before caring for a newborn.

Scott Alexander · no longer holds

Building more housing in a city raises its local prices, because density is what people are moving there for.

Arnold Schwarzenegger ·

Environmentalism should be based on building and growth rather than opposing development

John Graham-Cumming ·

Every developer platform will come to have AI capability built in.

John Graham-Cumming ·

A world of AI needs data that can move freely between platforms, without transfer fees or egress charges.

Roy Baumeister ·

Political hatred and demonization may suit short-term vote-getting but, I think, do considerable long-run damage to the social fabric

David Heinemeier Hansson ·

Requiring people to work from an office is a legitimate way to run a company, and remote-work advocates are wrong to claim moral superiority over it.

David Heinemeier Hansson ·

Microservices pose perhaps the biggest siren song for needlessly complicating your system.

David Heinemeier Hansson ·

Replacing method calls with network service partitioning within a single team and application is madness in almost all cases

Scott Alexander ·

Building more housing in a city raises its local prices, because density is what people are moving there for.

Amelia Wattenberger · by

Chatbots are a terrible interface for large language models, because a text box has no affordances.

April 2023

Ben Kuhn ·

Neglecting maintenance of infrastructure over time makes it degrade faster and become costlier to repair later.

Julie Zhuo ·

Design is not how a product looks; it is how it works, and all builders are designers.

Julie Zhuo ·

The more your target audience does not look like you, the more you should be skeptical of your design intuition.

Michael Grunwald ·

Since people are not going to give up beef, the way to cut its damage is to make cattle far more productive on the land they already use.

Daniel Cook ·

Extracting time and money from unhealthy users works in the short run and loses them in the long run.

Daniel Cook ·

A designer’s obligation is to satisfy real human needs rather than to manufacture needs that pay.

Daniel Cook ·

The job of a multiplayer designer is to grow a community, not to make gamified numbers go up.

Daniel Cook ·

The social sciences have theory but not a design practice, which is why building working social systems falls to game designers.

March 2023

Don Norman ·

Human-centred design ignores sustainability, inequity and bias, so design has to move to humanity-centred design that covers the whole ecosystem and the long term.

Etgar Keret ·

Using the same names for things during eras of conflict could provide hope.

Jacob Falkovich ·

Genuinely open-minded political thinking tends toward centrism or non-political issues, since ideological extremes mostly signal tribal loyalty or serve the power-hungry.

February 2023

Casey Muratori ·

The programming practices taught as best practice are ruinous for performance and should not be followed.

Caitlin Rivers ·

A definitive answer regarding the origins of the virus is not required before implementing preventive measures.

Julie Zhuo ·

Effective writing requires empathy for the reader, not only thought about what you want to say.

Kaspars Dambis · by

The idea that adding insulation beyond certain R-values is a waste of money is incorrect.

Cory Doctorow ·

Technology bosses are ordinary monopolists rather than evil geniuses, and crediting them with having hacked our dopamine loops flatters them.

Tim Urban ·

Public discourse has degraded into tribalism, mass shaming campaigns, and one-dimensional narratives replacing genuine debate.

Andy Galpin ·

Suppressing cortisol suppresses training adaptation; large spikes followed by quick recovery are what drive progress.

David Heinemeier Hansson ·

Inspiration does not last forever

David Heinemeier Hansson ·

A good work life is about the work, not the environment

Filippo Valsorda ·

A dependency you would need engineer-months to replace or fork is a critical one, and retaining its maintainer is straightforward business sense.

January 2023

Eli Dourado ·

Building a manufactured object at a high enough rate pushes its cost down toward the cost of its inputs.

Eli Dourado ·

Certification, not engineering, is what gates revenue for a new aircraft company.

Eli Dourado ·

Designing a physical product for a small beachhead market is a distraction when every new model has to be certified separately.

Cory Doctorow ·

A platform hands out free attention to creators only until they are committed to it, and then withdraws it and sells what they built.

Cory Doctorow ·

The end-to-end principle -- that a willing speaker's message reaches the willing listeners who asked for it -- is dead at the service level of today's internet.

Cory Doctorow ·

Policy should secure people's freedom to leave a platform while keeping their community, their purchases and their data, rather than try to keep dying platforms alive.

Nick Cave ·

What a generative model makes is a replication of art rather than art, because it can only imitate or average the choices human makers have already made.

Alon Levy ·

Countries that build cheaply defer to engineers, even the ones with far more political patronage than the United States has.

Philip Koopman ·

A self-driving car should be roughly ten times safer than a human driver before it is deployed, to leave engineering margin for what nobody can predict.

Daniel Herriges ·

Affordability will be decided by what happens to the price of the homes that already exist, not by how many new ones get built.

Daniel Herriges ·

Affordability is not a property of a building; it is a price set in a transaction, so there is no such thing as affordable housing, only affordable prices.

Daniel Herriges ·

Housing affordability is almost the direct inverse of how many new homes are built, so the answer is to build more houses.

Daniel Cook ·

The more freedom a design gives people to interact, the more of that freedom will be abused.

Daniel Cook ·

People will pry open any communication channel a designer tries to close, because the need to connect is stronger than the constraint.

Daniel Cook ·

People who do not yet trust each other are better served by tightly limited ways to interact.

Maciej Cegłowski ·

The International Space Station exists mainly to teach its builders how to build another space station.

December 2022

Thomas Depierre ·

An open source maintainer is not a supplier: there is no relationship with the companies using the code, so they cannot impose requirements on the person who wrote it.

Maggie Appleton ·

Language models have codified formal written language at a moment in time, which leaves informal, inventive, in-group speech as the place where humanness still shows.

Stuart Ritchie ·

A scientific paper should not be built around a take-home message, because that invites cherry-picking and hiding the results that did not fit.

Adam Mastroianni ·

Peer review is one of the things genuinely wrong with academia, not a safeguard that mostly works.

Matthew Green ·

Nobody chose cloud backup: a handful of Silicon Valley executives made the choice for everyone in pursuit of adoption metrics.

Matthew Green ·

Default cloud backup has quietly put a copy of nearly everyone’s private papers in a few data centres, and that is a surveillance weapon.

David Heinemeier Hansson ·

In-person connections can help build good faith in online communities

November 2022

Vaclav Smil ·

Batteries are a long way from rivalling liquid fuels; superbatteries for intercontinental flight may not arrive even by 2070

Gibson Biddle ·

A culture of direct feedback and debate leads to faster decisions and quicker learning than a culture of politics and second-guessing.

Gibson Biddle ·

Employees who deliver results but cannot put the company first make it impossible to build trust and a culture of open, honest communication.

Rodney Brooks ·

A radically new product needs many separate technologies to arrive together, and waiting for that convergence can take decades.

October 2022

Vaclav Smil ·

Treating wind and solar as purer green power than hydro is strange, given the embodied energy wind projects require

Vaclav Smil ·

Dams did not deserve their reversal of fortune; there is no need to build megastructures

David Heinemeier Hansson ·

Agents might, like self-driving cars, not need special equipment to interact with the environment

David Heinemeier Hansson ·

Using open source code to enable AI does not undermine open source communities

Davis Kedrosky ·

The key technologies of the Industrial Revolution came first and generated the profits that financed industrialists, rather than requiring a prior capital shock.

Dietrich Vollrath ·

Sustained growth in living standards required productivity to accelerate and population growth to fall at the same time.

September 2022

Kenton Varda ·

Calling small services "functions" puts too much emphasis on syntax rather than on what the service logically does.

Nick Lane ·

Life probably arose only once on Earth, and the deep split between bacteria and archaea is the evidence for it.

Nick Lane ·

Sex is not an option but a requirement: a large genome cannot be maintained by picking up loose DNA, so eukaryotes were forced into it.

August 2022

Andy Baio ·

It is an open question whether using Creative Commons-licensed images for AI training constitutes fair use and ignores attribution or non-commercial requirements.

Bret Devereaux ·

An invention only matters when the principle, a workable design and an economically viable use for it all arrive together; the idea on its own changes nothing.

Bret Devereaux ·

Handing an ancient society a modern machine would have achieved nothing: the Romans could neither have built a steam engine nor found a profitable use for one.

Bret Devereaux ·

The Industrial Revolution was not inevitable: it needed a set of very specific preconditions that only Britain, at that moment, actually had.

Davis Kedrosky ·

The Great Divergence was already underway by 1700, so colonial windfalls and British coal cannot be the whole explanation for European industrialization.

Adrian Holovaty · by

A general opt-out solution for webview framing would likely require iOS and Android to provide hooks at the OS level.

Gary Klein ·

Breaking through a skill plateau requires unlearning beliefs and conventions that are wrong or narrower than assumed.

Gary Klein ·

Debiasing has largely failed, and that is fine, because the biases identified are experience-built heuristics that only look like biases in hindsight.

Gary Klein ·

Emotions are how a person draws on accumulated patterns, so people who lose access to them make worse decisions rather than more rational ones.

Gary Klein ·

Experts facing time pressure decide by recognition, a blend of intuition and analysis, and this accounted for almost 90% of firefighters' hard decisions.

Gary Klein ·

Teaching people a decision model is useless, because the intuition it rests on is built from experience and has no shortcut.

Jonathan Gagné ·

The compounds responsible for astringency in coffee require very high water temperatures to dissolve efficiently.

July 2022

Steve Coast ·

Zoning and building restrictions across most of the western world make housing in desirable towns extremely expensive.

Phil Hagelberg ·

Software design that assumes text behaves predictably like English reflects the cultural bias of its typically white American developers.

Phil Hagelberg ·

Because programming requires forgetting, developers should weigh the costs of their abstractions and who those abstractions leave out.

Jonathan Gagné ·

Adding hydraulic resistance beneath an espresso puck reduces channeling by slowing the flow of water and making it more even across the coffee bed.

Jonathan Gagné ·

Flat burr grinders tend to deposit ground coffee preferentially on one side of the portafilter, an effect worsened by higher grinder RPM.

Adam Mastroianni ·

We know a good deal about how people actually learn, and almost nobody is ever taught it.

Adam Mastroianni ·

Knowledge is cheap and easily acquired; what a person actually needs is curiosity, self-efficacy, perseverance, perspective and hope.

Jeff Cavaliere ·

Building the mind-muscle connection is harder than keeping it.

Jeff Cavaliere ·

Muscle soreness is a guide to whether a muscle has recovered.

Miles Brundage ·

Research and discourse on AGI timelines are less useful than they appear, because most actions that need to be taken are insensitive to timelines

June 2022

Neel Nanda ·

The best way to maximize impact is often through high-risk, high-reward strategies.

Neel Nanda ·

Careers in AI and effective altruism are highly competitive.

Eliezer Yudkowsky ·

You cannot build an AI that has only the capabilities you want, because the algorithms that solve the problems you want generalize to the ones you do not.

Eliezer Yudkowsky ·

Fast capability gains are likely, and they can break many of the assumptions alignment depends on at the same moment.

Eliezer Yudkowsky ·

No observation of a model's behaviour can establish that it is aligned, because a model capable enough to be dangerous is capable enough to produce whatever behaviour it is being watched for.

Steve Coast ·

Adding rules and complexity to a system drains it of humanity and fun.

Matthew McConaughey ·

Outside the military, buying an assault rifle such as an AR-15 should require being 21

Matthew McConaughey ·

A national waiting period to buy an assault rifle is an acceptable sacrifice for responsible gun owners

Olivier Blanchard ·

A central bank can eliminate a self-fulfilling ('sunspot') debt crisis simply by credibly committing to buy as many bonds as needed at the low interest rate.

Olivier Blanchard ·

The probability of a self-fulfilling bad equilibrium in sovereign debt markets depends little on the debt level itself, but is much reduced by a rule linking the primary balance to debt service increases.

May 2022

John Collison ·

Everything around us exists because someone pushed it through with tenacity: the world is a museum of passion projects.

Tom Preston-Werner ·

Major version numbers should not be treated as sacred, and any breaking change deserves a major bump regardless of size.

Tom Preston-Werner ·

Breaking changes should be released frequently in small batches rather than saved up for one big yearly major version.

Brian Potter ·

You cannot infer a housing shortage from how much a country is building, because how much it needs to build changes with its demography.

Brian Potter ·

How much a metro builds relative to its population growth does not predict its rents; across the hundred largest US metros the correlation is essentially zero.

Brian Potter ·

Most apparent trends in American housing are the Bay Area in disguise: a handful of outlier metros drive the correlations people quote.

Gibson Biddle ·

Companies create long-term customer and shareholder value by pursuing product strategies that delight customers in ways that are hard to copy and improve the business.

Jean-Luc Mélenchon ·

Modern revolutions are characterized by a desire to regain control over one's life and encompass social, democratic, ecological, and individual liberty issues.

April 2022

Brian Chesky ·

Design the culture you want, or it will be designed for you.

Amjad Masad · by

A full replacement for the education system will eventually require physical locations.

Amjad Masad · by

Companies offering an exit from legacy systems must go direct to customers and build their own distribution.

Amjad Masad · by

Education systems will not reform from within because the necessary improvements require too much change.

Gibson Biddle ·

Meetings are the biggest drag on a professional's schedule.

Dan Luu ·

Even at top-100 traffic levels most applications can be served by a simple monolith, and it is cheaper to build than a complex architecture.

Tom Preston-Werner ·

Taking venture capital funding pushes a company toward extreme binary outcomes and pressures it to extract more value from customers.

March 2022

Filippo Valsorda ·

Funding should pay for maintenance, not features, because every new feature adds to the burden the money was meant to relieve.

Ben Taylor ·

Superstars almost always require secondary stars to win.

Derek Sivers ·

Nothing you depend on to write should require an internet connection.

Gibson Biddle ·

An organization's culture is the accumulated judgment built up over time, and that judgment is what helps people make good decisions about people, products, and the business.

February 2022

Yuval Noah Harari ·

Nations are ultimately built on stories

Neel Nanda ·

Pursuing a very low success-rate strategy can still be rational if a rare win is valuable enough.

David Perell ·

The more we imitate others, the faster we find our own style.

David Heinemeier Hansson ·

Simpler tools can achieve high-fidelity user interfaces at a lower cost in complexity

Phil Hagelberg ·

No single way of modeling data and methods in code is universally correct; every approach is just a set of trade-offs.

Zed Shaw ·

Large, wealthy companies use social manipulation and the guise of supporting open source to obtain free labor from developers.

Zed Shaw ·

Trillion-dollar companies often avoid paying for open source projects because they can acquire the labor for free.

January 2022

Bret Devereaux ·

The fall of the Western Roman Empire was a real loss of state capacity, not merely a change of management, and the evidence for that loss is ample.

Bjorn Fehrm ·

An airliner's empty weight is at least half its take-off weight, and no flying design has ever come in below 47%.

Bret Devereaux ·

Urban life in the West did not merely change after Rome, it practically vanished, and had to be built again from scratch centuries later.

Bret Devereaux ·

An institution can outlive the state that built it: the Church survived the collapse of Roman political authority when Rome’s other institutions did not.

Marc Andreessen ·

Reading further back in history shows that what feels new and unique to the present era is usually a very old pattern.

Marc Andreessen ·

A successful startup takes about five years to become an incumbent, and defending what it has built is the rational thing for it to do.

Marc Andreessen ·

Learning from failure accumulates as psychic scar tissue that narrows what a person is willing to consider as they get older.

Marc Andreessen ·

Mass schooling was designed as a factory a century ago, has run on autopilot since, and does not fit the world it now serves.

Yuval Noah Harari ·

Saving the environment does not require wrecking the economy, only getting our priorities right

David Heinemeier Hansson ·

Science requires an adversarial process to seek the truth

David Heinemeier Hansson ·

Diversity is not a displacement for core competency

Moxie Marlinspike ·

web3 is not on a path away from centralised platforms, and its privacy story is already below the internet’s low bar.

David Heinemeier Hansson ·

Many job applicants do not put in the effort to ensure their application fits the opening

Dan Wang ·

America's pursuit of efficiency left it with neither the factories to make more goods nor the logistics to import them.

December 2021

Drew DeVault ·

Free software is meant to be exploited commercially, and the licence’s terms are the whole of what a company owes in return.

David Heinemeier Hansson · no longer holds

Open source projects should adopt the commercial software playbook to compete for attention, and that professionalization is unequivocally good for open source.

David Crawshaw ·

Software projects should break downstream users as little as possible, and make any breakage they do cause obvious and easy to fix.

David Crawshaw ·

The software industry misunderstands backwards compatibility by focusing on narrow legalistic definitions rather than outcomes.

David Heinemeier Hansson ·

The evil now running wild in America did not begin with Trump and has been building for decades

November 2021

Hynek Schlawack ·

Codecov has become very flaky.

David Heinemeier Hansson ·

Pursuing projects worth your time is key to working without regrets

Drew DeVault ·

Adding a dependency is a serious commitment requiring an audit and continued re-auditing, and the usual practice of not doing so is reckless.

Austin Vernon · by

The e-bike market has no cheap standard model, which is what it would take to make the e-bike a mass product.

David Heinemeier Hansson ·

Build the product for yourself first: if it truly serves you, plenty of other people like you will want it too.

Jason Fried ·

Build the product for yourself first: if it truly serves you, plenty of other people like you will want it too.

October 2021

Casey Handmer ·

Once launch is cheap, the engineering effort spacecraft designers spend shaving off mass buys nothing.

Casey Handmer ·

NASA's Moon programme is still designed for a launch market that no longer exists.

Dylan Patel ·

Under Pat Gelsinger, Intel is betting the farm on manufacturing and technology instead of taking the conservative route.

Émile P. Torres · no longer holds

Longtermism, which treats fulfilling humanity's long-term potential as the overriding moral priority, is a worldview worth defending.

Kenton Varda ·

On a serverless platform, a runtime upgrade must never break a live application, because developers have no control over when upgrades happen.

Émile P. Torres ·

Longtermism is not the same thing as caring about the long-term future or about future generations; it goes far beyond that.

Kenton Varda ·

Letting customers pin old runtime versions would undo part of the promise of serverless: that developers should not have to worry about updating their stack.

Michael Shellenberger ·

Renewables return less than half as much energy as high-energy societies require to function.

Émile P. Torres ·

Effective altruism is deeply utilitarian in practice, whatever its members say about not being utilitarians.

Camille Fournier ·

A new manager has to hand over the technical design work, because doing it themselves leaves the team no room to grow and the manager nobody to delegate to.

September 2021

David Perell ·

True learning requires contemplation, implementation and returning to great ideas again and again.

David Perell ·

Systematic spaced repetition may be the lowest-hanging fruit in education.

David Perell ·

Feeling that you are learning does not mean you are.

David Heinemeier Hansson ·

Fear and vanity, not a lack of ideas, are what keep most people from building new things.

Phil Maffetone ·

Diets are typically unhealthy and ignore individual needs, so fat-burning is better pursued by matching natural foods to a person's own metabolism.

David Heinemeier Hansson ·

Software developers and designers aren't happy doing assembly line work

August 2021

David Heinemeier Hansson ·

Class-action lawsuits can be bought with bribes

Luca Lampariello ·

A language is a set of skills to be built, not a body of facts to be known.

Teresa Torres ·

Product discovery belongs to the whole trio -- a product manager, a designer and an engineer -- not the product manager alone.

Oliver Burkeman · by

Work that matters requires tolerating discomfort and boredom rather than escaping into distraction.

Molson Hart ·

Having taken over the making of physical goods, Chinese firms are now moving up into designing, branding and selling them.

Oliver Burkeman · by

Life change never comes from one transforming plan; it comes from small adjustments to the life you already have.

David Heinemeier Hansson · no longer holds

The complexity of a JavaScript bundling and transpiling toolchain is a bargain worth making, and the right default for building web applications.

Kenton Varda ·

The worst thing an application can do is tell a user their action succeeded when it did not.

Matthew Green ·

A tracing requirement in an encrypted messenger gives up the confidentiality of who sent what, which is the part that protects the sender.

July 2021

Marc Littlemore ·

The main purpose of writing tests is to build confidence in the code, not to find bugs.

Marc Littlemore ·

Code that is difficult to test is a sign of other design problems.

Marc Littlemore ·

The goal of testing should be tests that are easy to understand and maintain, not high coverage of everything.

June 2021

Brian Potter ·

The person who specifies a new building system carries all its risk and none of its reward, a reverse moral hazard that keeps construction innovation out.

Brian Potter ·

Builders reject new building systems rationally: a change to the process raises the odds of a catastrophic overrun by more than the innovation can save.

Nick Cave ·

A maker cannot summon the good line, only be at the desk when it arrives, and the outcome is not theirs to influence.

Gwern ·

Criticism that a simpler, scaled-up architecture underperforms a heavily hand-optimized baseline actually demonstrates why that architecture-scaling research matters.

Christian Oliff ·

It is unlikely that one will generate much income from Coil until it becomes more popular and gets more paying subscribers.

May 2021

David Heinemeier Hansson ·

The right to repair is essential

Casey Liss ·

Combine may be being sidelined by internal politics favoring SwiftUI.

April 2021

Tim Urban ·

Extending the human lifespan past its natural ceiling will require deep genetic engineering, not just better healthcare.

Gwern ·

Unsupervised learning trained on unfiltered, internet-scale image data now genuinely works.

Bjorn Fehrm ·

Battery-electric airliners do not work even on optimistic assumptions, and the arithmetic said so before anyone built one.

March 2021

Gwern ·

NASA's lunar quarantine after the Apollo missions was theater meant to reassure the public rather than genuine biocontainment.

Naval Ravikant ·

Modernity probably brings more unhappiness than the past; acquiring things does not make people permanently happier.

Naval Ravikant ·

In an age of abundance, pursuing pleasure for its own sake leads easily into addiction.

Naval Ravikant ·

Guilt is society programming you so well that you become your own warden.

David Heinemeier Hansson ·

Life is long enough if you spend it well.

Casey Handmer ·

NASA can still be reorganised into an agency that exploits cheap heavy lift, and the next administrator will decide whether it is.

David Heinemeier Hansson ·

The internet can be a grim place

February 2021

Brad Frost ·

The distinction between 'front-of-the-front-end' and 'back-of-the-front-end' provides a necessary distinction between types of web development.

Fred Wilson ·

When economies recover and interest rates rise, asset price bubbles will deflate and markets will slow down.

Greg Kroah-Hartman ·

Supporting a kernel for 6 years requires constant testing by users of the kernels.

January 2021

Hillel Wayne ·

The argument about whether programming is engineering is conducted entirely by people who have never worked as engineers.

Hillel Wayne ·

Licensing is a political and social construct, so an unlicensed profession can still be doing engineering.

Laurie Santos ·

Our minds have poor intuitions about what will improve our mood.

Kieran Healy ·

The White House likely did not expect the crowd to actually enter the Capitol building.

December 2020

Amanda Askell ·

All-things-considered judgments in AI ethics will often require a complex evaluation of many factors.

November 2020

Jessica Livingston ·

Revenue is the best number for a startup to measure, since it tests whether users genuinely want the product.

Drew DeVault ·

Making a living from your own open source means running it as a business and acquiring business skills, not waiting to be funded.

Matthew McConaughey ·

Real democratic confrontation requires granting the other side legitimacy; delegitimising the opposition is not real democracy

Matthew McConaughey ·

Americans need to find simple shared values to rebuild trust in one another

Phil Maffetone ·

Poor sleep can quickly lead to a lack of recovery, because sleep is a key part of it.

October 2020

Yonah Freemark ·

Neither federal nor local governments have been effective at using transportation expenditures to achieve sustainable or equitable outcomes.

Chamath Palihapitiya ·

Obvious questions to CEOs draw practised answers, so pattern recognition built on them is brittle and superficial.

Chamath Palihapitiya ·

A sense of someone's integrity builds from small observations over a long period, because people reveal themselves in small ways over time.

Chamath Palihapitiya ·

Continuing to learn means that today's views will probably look slightly immature and naive later.

September 2020

Thorsten Ball ·

The effort typing requires, not raw typing speed, is what determines whether it becomes a real obstacle to writing.

Thorsten Ball ·

Discomfort or effort in typing that makes someone avoid writing things down is a genuine bottleneck that should be fixed, not dismissed.

August 2020

Ming Thein ·

Creative work should feel liberating and produce order and beauty from incoherent parts.

Ming Thein ·

Planned creative work is limited by one's means and imagination, while serendipitous work is unbounded by chance and the subconscious.

Thorsten Ball ·

Building automation that doesn't address the actual bottleneck to shipping adds noise rather than helping ship.

Samo Burja ·

Civilisational collapse can be fast: the Late Bronze Age shows a society coming apart over years rather than centuries.

Adam Wathan · by

Utility classes are more portable than component styles because component styles are often too opinionated to reuse.

July 2020

Wendy Behary ·

Narcissism runs along a spectrum rather than being a category, and schema therapy treats it by working back to what early life embedded.

Kenton Varda ·

Security for a platform that runs other people's code is never finished: hard real-world problems often have no simple fix, only defences built thicker and thicker.

Andrew Graham-Dixon ·

Rembrandt's self-portraits visually embody the fluid nature of consciousness.

Joe Previte ·

Sharing learning in public can help build reputation and credibility.

June 2020

Tom Preston-Werner ·

Ending systemic racism in the United States requires a large-scale anti-racist movement.

Amanda Askell ·

Insurance against failure for people pursuing ethical careers would enable more of the risks needed to do the most good.

May 2020

Allan Odgaard ·

Low-level operating system APIs such as exec and getxattr performing synchronous network activity before returning to the caller is a serious design problem.

Ramez Naam ·

Solar's future price should be predicted from how much has been built, not from how much time has passed.

Ramez Naam ·

Every doubling of the solar built worldwide has cut its price by 30 to 40 per cent, and that learning rate is why the price keeps falling.

Gavin Francis ·

The UK needs domestic manufacturing for medical equipment to ensure supply.

Jared Diamond ·

Getting through a personal crisis first requires acknowledging that you are in one.

Jared Diamond ·

COVID-19 is a world problem that requires a world solution.

Jared Diamond ·

A visible common enemy that kills quickly brings people together to cooperate.

Derek Lowe ·

The bureaucracy around clinical trials is Chesterton's Fence: it was built on the harm done by sloppier trials and should not be pulled down without knowing why it went up.

April 2020

Cal Newport ·

Life change never comes from one transforming plan; it comes from small adjustments to the life you already have.

Yuval Noah Harari ·

Governments should build better healthcare systems and leave philosophy to individuals

March 2020

Julie Zhuo ·

Rather than coasting on intuition and taste, a data-informed person starts from the assumption that they know nothing.

Maxence Poutord ·

A radical change in how a person and a team work necessarily hurts productivity, even when remote work is otherwise a productivity boost.

February 2020

Robin Sloan · by

Freed from having to be professional and scalable, programming becomes a different activity, as cooking at home differs from a commercial kitchen.

Robin Sloan · by

Software you build for your own family will not change unless you want it to.

Frank Chimero ·

A design system built for the right reasons is reparative and one built for the wrong reasons is a weapon for displacement, because tools are always beholden to values.

January 2020

Scott McCloud ·

Fundamental design principles of successful visual communication are built upon the underlying principles of visual cognition, evolution, and biology.

Yuval Noah Harari ·

We now live in a world where war kills fewer people than suicide and sugar is more dangerous than gunpowder

Yuval Noah Harari ·

The real watershed in hacking humans comes when biometric sensors make data flow from inside the body

Bret Devereaux ·

States were born in the richest and easiest country, not in hard marginal land.

Bret Devereaux ·

Hard peoples from harsh places are not better at war: at the dawn of civilisation it was the softest and richest societies that won most often.

John Baez ·

To keep global warming below 2 degrees Celsius over pre-industrial levels by 2100, it may become increasingly likely that negative carbon emissions will be required.

Yonah Freemark ·

New American transit routes are often poorly designed or inadequately integrated into urban development.

Yonah Freemark ·

Stopping the expansion of the public roadway network is an essential element of reducing the carbon footprint of transportation.

Dan Wang ·

Western firms that kept the high-value work at home are betting the workforce they trained cannot absorb the technology, and that bet has already lost several times.

c. 2018–2019

Yuval Noah Harari ·

Observing the truth about yourself is very hard, and people quickly turn it into a story of heroes and villains

December 2019

Naval Ravikant ·

Be impatient with actions and patient with results, because results come out of complex systems and other people.

Naval Ravikant ·

Build your character a certain way and your character becomes your destiny, so luck stops being luck.

Naval Ravikant ·

Specific knowledge is found by pursuing your innate talents, curiosity and passion, not by training for the hottest job.

Naval Ravikant ·

The real giants in any field are the people who can both build and sell.

Naval Ravikant ·

High accountability makes you less replaceable, which is how you earn equity.

Naval Ravikant ·

Taking accountability for your actions is taking an equity position in all your work.

Ben Taylor ·

Lineup net ratings require samples of more than a few hundred minutes and diverse opponents to be considered trustworthy.

Taylor Swift ·

Unregulated private equity buying up artists' music is a potentially harmful force in the music industry

November 2019

François Chollet ·

Progress toward more intelligent artificial systems requires defining and evaluating intelligence so systems can be compared with each other and with humans.

François Chollet ·

The Abstraction and Reasoning Corpus can measure human-like general fluid intelligence and enable fair comparisons between AI systems and humans.

François Chollet ·

If intelligence is the process of acquiring skills, no single task demonstrates intelligence unless it is a meta-task of skill-acquisition across many tasks.

October 2019

Daniel Kahneman ·

Delaying intuition until each dimension has been assessed separately and a whole profile exists tends to give a better judgment, because rapid intuitions are not particularly good.

Daniel Kahneman ·

Forming an impression very quickly leads to spending the time confirming it rather than collecting evidence, so a wrong first impression is unlikely to be corrected.

Daniel Kahneman ·

The best prediction is always less extreme than the impression it is based on, but intuitive prediction is non-regressive and ignores regression to the mean.

Daniel Kahneman ·

Noise is useless variability: unlike variability that has a selection mechanism and feedback, as in evolution, nothing is learned from it and it is costly.

Daniel Kahneman ·

Algorithms beat people at judgment, which is not intuitive but is true, so they are the first recommended remedy for noise.

Daniel Kahneman ·

Strong intuitions that seem clear lack a way of calibrating, which makes people wildly overconfident about what they know and reluctant to accept they may be wrong.

Rihanna ·

Music is a way of speaking to the world in code, and an artist's other ventures flourish on top of it

Steph Smith ·

Writing is a skill that can be acquired through continuous effort and a low-friction process.

September 2019

Fred Wilson ·

The public markets may cause the late-stage private markets to begin valuing companies based on margins.

Michelle Zatlyn ·

Building an Internet that meets the world’s needs takes a global community collaborating, not any one company or individual.

Steph Smith ·

Effective leadership requires prioritizing the success of the team over personal gain.

August 2019

Alexey Guzey ·

Scientific progress in biology is not slowing down.

Alexey Guzey ·

Peer review is one of the things genuinely wrong with academia, not a safeguard that mostly works.

July 2019

Don Norman ·

What gets called human error is design error: technology-centred design makes people do what machines are good at and then blames them.

Don Norman ·

Machines should do the parts of a task that people are bad at, so that people and technology work as partners on human terms.

Don Norman ·

Human-centred design starts by solving the root issue rather than the problem as presented, which is usually the symptom.

Don Norman ·

Experts should supply the facts and leave the methods to the people affected: a community will trust and implement a solution it chose itself, even one identical to the expert's.

Don Norman ·

The rule to defer judgment encourages design by committee; effective teams do not defer criticism, they create through it.

Steph Smith ·

Companies could gain a competitive advantage by opening their doors to the remote workforce.

Ryan Singer · by

Over-specifying a design makes it harder to estimate, not easier: the more specific the work, the more hidden complexity a mockup commits the team to.

Ryan Singer · by

Interface copy is an affordance: reading it is an act that gives the user information for what to do next.

Ryan Singer · by

Work that is too fine too early commits everyone to the wrong details; shaped work is rough on purpose so designers and programmers can apply their own judgement.

Ryan Singer · by

Any specific mockup from someone senior biases everyone after them, so early design should be breadboards and fat-marker sketches that leave room for designers.

Ryan Singer · by

Shaping is primarily design work: the shaped concept is an interaction design seen from the user's perspective, and it takes technical literacy, not programming.

Craig Mod ·

Speed is a good proxy for engineering quality: software that slows down on simple tasks suggests engineers who are not obsessive about detail.

Craig Mod ·

The best touch interfaces are physical buttons; on a screen, to feel tactile is to move without latency.

June 2019

Yuval Noah Harari ·

Information technology and biotechnology could build either paradise or hell, depending on our political ideals

Casey Liss ·

SwiftUI and Combine are clearly the future.

Robin Carhart-Harris ·

With the right intention, care and context, psychedelics can help revise entrenched pathological beliefs.

Karl Friston ·

With the right intention, care and context, psychedelics can help revise entrenched pathological beliefs.

Michael Pollan ·

Smart political strategy requires deciding not only what to fight for but when to fight for it.

May 2019

Martin Fowler ·

Cruft is not carelessness: even the best teams inevitably create it, because the right design is only visible once the software has been built.

April 2019

Yuval Noah Harari ·

The same tools built to sell us products can sell us a politician or an ideology we do not need

Sebastiaan de With ·

The most important thing a designer can do to make new technology usable is to simplify, and sometimes fewer features make a better product.

Sebastiaan de With ·

Given a complicated premise, the best thing a designer can do is stick to a conventional design and let the features surprise and delight.

Sebastiaan de With ·

You know a design is great when it disappears: a superb typeface, good signage or a well-designed door handle is never noticed, only the absence of frustration.

March 2019

Jessica Livingston ·

Domain-expert female founders are often hamstrung in building startups because they can't program.

Emily Short ·

A fact is important from a game design perspective if it affects the player's ability to act in the world and the mechanics they can use.

February 2019

Diogo Mónica ·

Any framework distinguishing leaders from commanders by the actions they take is wrong.

January 2019

Sam Altman ·

Once the priorities are clear, get them done quickly: he has yet to meet a slow-moving person who is very successful.

Emily Short ·

Building procedural systems is an effective way to explore and test mental models.

December 2018

Adam Wathan · by

Tailwind CSS has the potential to be a high-impact project for helping people build software.

Jarrett Walker ·

The real cost of transport demonstration projects is that they confuse decision-makers about whether mass transit still needs building.

November 2018

Julia Evans ·

Tailwind is a good way to build a site you're happy with, even without strong CSS skills of your own.

October 2018

Tren Griffin ·

The pace at which the world changes is increasing.

Tren Griffin ·

People who become skilled at making trade-offs live better lives as a result.

Yuval Noah Harari ·

Identities should now be built like tents that can be folded and moved, not like stone houses

Jacob Devlin ·

A deep bidirectional model is strictly more powerful than a left-to-right model or a shallow concatenation of unidirectional models.

Ming-Wei Chang ·

A deep bidirectional model is strictly more powerful than a left-to-right model or a shallow concatenation of unidirectional models.

Kenton Lee ·

A deep bidirectional model is strictly more powerful than a left-to-right model or a shallow concatenation of unidirectional models.

Kristina Toutanova ·

A deep bidirectional model is strictly more powerful than a left-to-right model or a shallow concatenation of unidirectional models.

September 2018

Michelle Zatlyn ·

A company building products should ask how each one affects individual privacy, and aim to be a leader on privacy.

August 2018

Greg Kroah-Hartman ·

The best solution for most Linux users is to use the kernel provided by their favorite Linux distribution.

Carlos Moreno ·

The future of urban mobility will be defined by the convergence of different modes of transport into on-demand, door-to-door services.

Carlos Moreno ·

Cities must become platform-cities capable of hosting and experimenting with diverse new models of usage and economic models.

Samo Burja ·

When a functioning organisation dies, capital that cannot be sold or transferred dies with it.

July 2018

Will Larson ·

A sustainable on-call rotation needs seven or eight engineers.

Will Larson ·

A group of fewer than four people is not a team: it behaves like a set of individuals, and a single departure knocks it back into maintenance.

Samo Burja ·

Repairing a dysfunctional institution is harder than building a functional one from scratch, because its internal forces resist change and the dysfunction is hard to diagnose in the first place.

June 2018

Will Larson ·

Innovation tends to come from slack time away from firefighting, not from being fully utilized.

Nadia Asparouhova ·

Successful open source developers get hired away by companies, which quietens the very people best placed to argue for making a living from open source.

Bennett Foddy ·

Making a chess variant, a dice variant and a Pong variant is a worthwhile exercise for any game designer.

Felix Salmon ·

Substack's platform design emphasizes paid subscriptions so strongly that it obscures the option to sign up for free.

Felix Salmon ·

Tabular figures should be used any time numbers appear outside a normal sentence.

Felix Salmon ·

A dense-looking building does not by itself create a livable dense suburb without supporting schools and shops.

Felix Salmon ·

Brutalist buildings are not inherently doomed to age badly.

May 2018

Yonah Freemark ·

Making a genuine effort to improve the lives of existing transit users will eventually bring more people back onto transit systems.

Carlos Moreno ·

The cost of public transport could be shifted from the user to other sectors through new economic models.

Carlos Moreno ·

A form of total or partial free transport for users will become inevitable.

Ben Taylor ·

Elite regular-season and postseason impact looks quite similar.

Ryan Holiday ·

The better you get at a craft, the humbler it makes you.

April 2018

Jessica Livingston ·

Equity in a startup is best treated as a bet that might turn out well, rather than a guarantee.

Ivan Santos ·

Much of a person's success is owed to mentors who help guide them without expecting anything in return.

Ben Taylor ·

People intuitively undervalue longevity in basketball greatness rankings.

Ben Taylor ·

Undervaluing secondary-star seasons leads analysts to over-credit first options and overestimate their worth.

March 2018

Jonathan Frankle ·

Pruning cuts the parameters of a trained network by over 90% without costing accuracy, so inference gets smaller and faster for free.

Michael Carbin ·

Pruning cuts the parameters of a trained network by over 90% without costing accuracy, so inference gets smaller and faster for free.

February 2018

Bennett Foddy ·

A good prototype should express a single idea or test a single hypothesis.

Bennett Foddy ·

Creative work is better served by a diverse toolkit of specialized tools than by one universal tool.

Bennett Foddy ·

Maze puzzles are unengaging on their own but become interesting when paired with a strong extrinsic motivator.

Steven Pinker ·

Reason is non-negotiable: arguing that others should share your beliefs commits you to it

Steven Pinker ·

Progress unguided by humanism is not progress

Frank Chimero ·

Directness is best: a great photo, a memorable illustration or a pitch-perfect sentence does most of the work, and fancy implementation has never moved the needle for clients.

January 2018

danah boyd ·

Helping children build long-term coping skills is more effective than simply implementing rules.

Pieter Levels ·

Build the product for yourself first: if it truly serves you, plenty of other people like you will want it too.

Rich Hickey · by

Pervasive unmoderated mutation has to go if programs are to be concurrent.

Rich Hickey · by

Object orientation is overrated: it was born of simulation and is now applied everywhere, including where it does not fit.

Rich Hickey · by

Locking is too hard for programmers to get right over and over again, so the language runtime should do that part.

December 2017

Mina Markham ·

Design systems teams are suitable for people who enjoy architectural work and work between design and development.

Mina Markham ·

Design system documentation must be easy to keep up-to-date to prevent it from becoming obsolete.

November 2017

Jonathan Shedler ·

The APA's trauma treatment guidelines should be ignored, because they rest only on trials of therapies too brief to have worked.

Branko Milanovic ·

The climate policy everyone has quietly assumed is that poorer countries stay poor, and that is the wrong answer.

Branko Milanovic ·

Degrowth would require the world to go from an inequality worse than South Africa's to a complete equality no recorded society has ever had.

August 2017

James Clear · by

Learning a new mental model can give a person a new way to see the world, much as a new math technique does.

James Clear · by

Mental models are imperfect but useful: no single one explains everything, yet the best of them have let people build and travel.

July 2017

Troy Hunt ·

Blocking paste on a password field makes passwords weaker, because people fall back to ones they can type.

Troy Hunt ·

Mandatory periodic password changes make passwords worse, because people just increment the number on the end.

Rui Carmo ·

There is very little hard data on which factors are critical to making a data science team successful.

June 2017

Jessica Livingston ·

Building great products and managing people require very different skills.

Shane Parrish ·

The bigger and more challenging an action is, the more activation energy it takes.

Yuval Noah Harari ·

Recent decades have been the most peaceful era in human history

Yuval Noah Harari ·

The atom bomb has turned victory in a world war into collective suicide

Yuval Noah Harari ·

Nuclear and cyber weapons can destroy countries but cannot build profitable empires

Charles Marohn ·

Cities have only ever built wealth incrementally, by many small improvements over a long time, not by big projects.

Charles Marohn ·

The break in how cities are built is that places used to be improved continuously and are now built once, at scale, to a finished state.

Charles Marohn ·

A neighbourhood redevelops itself when rising land values meet permission to build more intensely, and it stagnates when either half is missing.

May 2017

Arielle Schwartz · by

Successful trauma treatment requires building tolerance for the emotions and body sensations that accompany the memories.

Michael Batnick ·

Investing requires mental toughness bordering on insanity to significantly outperform the market.

Charity Majors ·

A manager’s calendar and an engineer’s calendar are incompatible, because management is interruptive and good engineering requires blocking interruptions out.

March 2017

Eugen Rochko ·

Facebook is not a platform where people have the power to build anything.

February 2017

Malcolm Gladwell ·

Elite colleges behave like luxury-handbag businesses protecting exclusivity rather than like educators

Malcolm Gladwell ·

Encouraging experimentation and opening opportunities requires much more tolerance of mediocrity

Thorsten Ball ·

Building powerful developer tools such as compilers, interpreters, and query languages requires knowledge of parsing.

January 2017

Matt Cutts ·

The work done by the US Digital Service is deeply important and inspiring.

Charles Marohn ·

American cities built roughly two dollars of public infrastructure for every dollar of private investment, and that ratio is what bankrupts them.

Charles Marohn ·

Nearly every American city is functionally insolvent, because replacing the infrastructure it has built would cost more than its entire private tax base is worth.

Robert Cialdini ·

Paying more attention to what is present than to what is absent is often a mistake, since absence can matter

November 2016

Kevin Simler ·

Feeling pride in holding a belief, anguish at changing it, or anger when it is challenged is a sign the belief serves a social rather than a truth-tracking purpose.

September 2016

Mina Markham ·

A design system's purpose is to provide consistent, tested and reusable solutions for a common UX problem.

Mina Markham ·

A design system is only as good as its documentation.

August 2016

Maria Konnikova ·

Correlations found between openness and liberalism are largely tautological, because political leanings were built into the personality measures.

Maria Konnikova ·

People readily accept counterintuitive findings from social psychology as long as they do not conflict with their notions of what ought to be.

July 2016

Derek Sivers ·

The feeds businesses design to be addictive are obstacles to anyone whose aim is to make something.

Jarrett Walker ·

Frequent fixed-route buses are the only way to cover a whole city with useful transit quickly, which makes them a land-use decision rather than a fallback.

June 2016

Atul Gawande ·

Science is a commitment to systematic thinking through testing and observation, not a major or a career

May 2016

Moxie Marlinspike ·

A federated protocol freezes in time, so a messenger that has to keep up with the ecosystem cannot be federated.

Moxie Marlinspike ·

Encryption people have to switch on protects almost nobody; what protects people is what a service turns on for everyone.

April 2016

Rick Steves ·

An up-to-date guidebook is a worthwhile investment for travelers.

Rui Carmo ·

Micro-services are not inherently hard to design, deploy, or maintain.

Rui Carmo ·

Micro-services require scalable, resilient scaffolding that conventional SOA frameworks or ESBs do not usually provide.

Rui Carmo ·

Docker will eventually play a role in micro-services, but it is still acquiring the necessary fine-grained control and self-healing.

February 2016

Bruce Schneier ·

An access mechanism built for the police cannot be limited to the police, so a backdoor weakens the system against every attacker at once.

Bruce Schneier ·

Widespread encryption protects far more against bulk surveillance than against a targeted attacker.

January 2016

Sandi Metz ·

Duplication is cheaper than the wrong abstraction, so code should not be made reusable before there are real examples of what the reuse has to do.

Sandi Metz ·

Existing code exerts a powerful influence: its presence argues that it is correct and necessary, and sunk cost keeps teams inside an abstraction that has stopped fitting.

Sandi Metz ·

Parameters and conditional branches accumulating inside shared code are the diagnostic that the abstraction is wrong.

Sandi Metz ·

When an abstraction turns out to be wrong, the fastest way forward is to inline it back into every caller and let the duplication show what the right one is.

c. 2011–2015

Yuval Noah Harari ·

The farmers' pursuit of an easier life led to much hardship

Yuval Noah Harari ·

No social animal is ever guided by the interests of its entire species

Yuval Noah Harari ·

Credit lets us build the present at the expense of the future, on the assumption that future resources will be more abundant

December 2015

Maria Konnikova ·

When people are swept up in a powerful narrative, their reason often falls by the wayside.

Dan Luu ·

Writing to files safely is too hard to do with raw syscalls; application code should go through a library or a database.

Dan Luu ·

Conventional testing is not sufficient: crude purpose-built checking tools of a few thousand lines find bugs that testing does not.

Dan Luu ·

Almost every software project has cheap testing wins left on the table: basic random testing, static analysis and fault injection pay for themselves the first time they are run.

November 2015

Tim Urban ·

Reasoning from first principles means gathering only what one knows to be true and building a conclusion from it.

Tim Urban ·

Reasoning from first principles means taking the basic raw ingredients as the building blocks and working upward from them.

Tim Urban ·

If a chain of reasoning rests on a wrong inherited recipe, the conclusions built on top of it will be just as wrong.

October 2015

Thorsten Ball ·

Meta-programming is usually unnecessary complexity that obscures what code is meant to do.

Thorsten Ball ·

Simple, straightforward solutions can also be elegant, though not every elegant solution is straightforward.

Kathryn Schulz ·

Basing beliefs on personal intuition and revelation substitutes anecdote and authority for evidence and reason

Kathryn Schulz ·

Rejecting every certainty but one's own is zealotry, and issuing contradictory decrees on whim is despotism

September 2015

Yuval Noah Harari ·

Life on Earth is now governed increasingly by human intelligent design rather than natural selection

Yuval Noah Harari ·

As we redesign life with new technologies, we should weigh the welfare of all sentient beings, not just humans

August 2015

Christopher Olah ·

An idea can look obvious in hindsight only because the insight that makes it obvious was itself the hard discovery.

Christopher Olah ·

LSTM architectures are designed so that retaining information over long time spans is their default behavior rather than something they must struggle to learn.

Ramez Naam ·

After four more doublings of installed solar capacity, solar electricity will cost about 3 cents per kwh in the sunniest areas and 4.5 cents per kwh in moderately sunny areas.

July 2015

Sonja Lyubomirsky ·

People pursuing happiness should not check on it too often, any more than dieters should weigh themselves several times a day.

Sonja Lyubomirsky ·

That a few people pursue happiness unhealthily is no reason for everyone else to give up pursuing it.

Guillermo Rauch ·

The designer and programmer roles are converging, and the path from sketch to design to implementation is opening up to everyone with an idea.

Guillermo Rauch ·

An application's user interface should be defined as a pure function of its state.

Guillermo Rauch ·

Software estimates are poor because it is easy to underestimate the size of the state space: the complete list of questions a UI must answer is hard to come up with.

Guillermo Rauch ·

Lorem ipsum has had a good run; design with data instead.

June 2015

Raph Koster ·

Designing a game imprints behaviour on the people who play it, including behaviour the designer never intended to teach.

Raph Koster ·

User-interface design and game design are opposites: one takes problems away from the person using it, the other hands problems to them.

Raph Koster ·

An interface should give the user a useful mental model even if it is false, while a game should give the player an accurate model of its own machinery.

Raph Koster ·

Good interface design uses clarity to hide complexity; good game design uses clarity to teach it.

Martin Fowler ·

Successful microservice systems almost always began as a monolith that grew too big, and systems built as microservices from scratch end up in serious trouble.

Martin Fowler ·

Build a new application as a monolith first, even when you are confident it will eventually need microservices.

Martin Fowler ·

Service boundaries cannot be got right at the start, and moving functionality between services is far harder than moving it inside a monolith.

March 2015

Rob Walling ·

The strategy that seems to give people the best chance of success is creating a simple product, with a simple marketing plan - one that only requires a single traffic channel.

Rob Walling ·

Selling an add-on to an existing ecosystem is easier than launching a stand-alone product because of built-in discovery and lower complexity.

Gretchen McCulloch ·

Using the uninflected forms 'to can' or 'to even' is an easier linguistic step than using inflected forms like 'evened' or 'evening'.

Tynan Sylvester ·

Balancing a system elegantly is not the same as giving people what they actually came for, which is emotion.

February 2015

Frank Chimero ·

Every material has a grain, including the web, and designs that work against it end up weaker.

Frank Chimero ·

Designing for screens means arranging the elements first and assigning them a size afterwards, not drawing the box first.

Bob Nystrom ·

Splitting functions into synchronous and asynchronous kinds forces that distinction into every piece of code a programmer afterwards writes.

Bob Nystrom ·

Async/await makes asynchronous code easier to write without healing the split it created, and the divide reappears as soon as you write a higher-order function.

Bob Nystrom ·

Asynchrony should be a property of how a program is run, not a colour baked into the type of every function in the standard library.

c. 2000–2014

Steve Krug ·

A user should be able to understand what a page is and how to use it without spending any effort thinking about it

Steve Krug ·

People don't read web pages; they scan them

Steve Krug ·

It matters that users understand a site, because muddling through tends to be inefficient and error-prone

Steve Krug ·

When a slight inconsistency makes something significantly clearer, choose clarity

Steve Krug ·

The number of clicks matters less than whether each click is a mindless, unambiguous choice

Steve Krug ·

Web pages should cut half their words, then half of what is left

Steve Krug ·

No one reads instructions until repeated attempts at muddling through have failed

Steve Krug ·

Design-team debates about what users like rarely change anyone's mind

Steve Krug ·

There is no average web user

Steve Krug ·

Even the worst usability test with the wrong user shows important things to improve

Steve Krug ·

Testing one user early in a project is better than testing 50 near the end

Steve Krug ·

Every web development team should spend one morning a month doing usability testing

c. 2014

Steve Krug ·

Affordances are the last thing an interface should hide

December 2014

Diederik P. Kingma ·

An optimizer that is invariant to rescaling the gradients and cheap in memory is what makes very large models practical to train.

Jimmy Ba ·

An optimizer that is invariant to rescaling the gradients and cheap in memory is what makes very large models practical to train.

Diederik P. Kingma ·

An optimizer's hyper-parameters should mean something a practitioner can reason about, and should rarely need tuning.

Jimmy Ba ·

An optimizer's hyper-parameters should mean something a practitioner can reason about, and should rarely need tuning.

Elie Hassenfeld ·

An excellent organisation is significantly more likely to succeed than a weaker one, so organisational strength belongs in the decision about where to give.

November 2014

Raph Koster ·

Social media designers routinely ignore what online-community design already established decades ago.

Dan Luu ·

The most-cited study showing that language design affects software quality rests on misclassified repositories.

Guillermo Rauch ·

Minimizing network round trips is essential to a good user experience, because bandwidth keeps improving and latency will not.

Guillermo Rauch ·

The distinction between websites and web apps will disappear: a blog and a CRM are reigned by the same UX principles.

Guillermo Rauch ·

JavaScript should act immediately on user input to mask network latency, which should remove most spinners and loading messages.

September 2014

Leonardo DiCaprio ·

The climate crisis has grown beyond individual consumer choices and now requires decisive, large-scale action by industries and governments.

Don Norman ·

For designers, signifiers matter more than affordances: affordances control what is possible, signifiers are how a product communicates what to do.

Wayne Hale ·

The goal of spaceflight is the whole solar system, not any single destination the experts are arguing over.

Wayne Hale ·

Extending consciousness beyond Earth, out to the stars, is a fundamentally good goal.

David Sirlin ·

Designing a game with unknown elements, hidden information, or randomness is a way to create a mixed solution game.

August 2014

David Sirlin ·

Subtractive design is the process of strengthening core elements by removing imperfections and extraneous parts.

David Sirlin ·

Technical limitations in older consoles likely led to the separation of save points and checkpoints.

July 2014

Rob Walling ·

A high churn rate indicates that a product has not found a market that needs it or that the product is something no market wants.

May 2014

Casey Muratori ·

Duplication is cheaper than the wrong abstraction, so code should not be made reusable before there are real examples of what the reuse has to do.

John Allspaw ·

The difference between science and engineering lies in the discretionary space engineers have in design and operation.

January 2014

Emily Short ·

The structure of an interactive story can be a meaningful part of the communication between the author and the reader.

c. 2013

Bill Gates ·

With tight budgets, governments rightly demand effective programs, which requires better tools to measure what works

c. 1988–2013

Don Norman ·

Discoverability and understanding are two of the most important characteristics of good design

Don Norman ·

It is the duty of machines and their designers to understand people, not people's duty to understand the dictates of machines

Don Norman ·

Human-centered design means starting from a good understanding of people and the needs a design is meant to meet

Don Norman ·

Good design communicates from machine to person what actions are possible, what is happening and what is about to happen, above all when things go wrong

c. 2013

James G. March ·

In organization studies, the pursuit of relevance makes scholars abandon the pursuit of knowledge too fast

Don Norman ·

Affordances determine what actions are possible and signifiers communicate where to act, and design needs both

James G. March ·

Interdisciplinary work requires full qualification in each discipline, so the people who go into it are underqualified

c. 1988–2013

Don Norman ·

A good conceptual model lets people predict the effects of their actions

Don Norman ·

However brilliant a product is, if people cannot use it it will be judged poorly, and making it understandable and usable is up to the designer

Don Norman ·

The most effective way to help people remember something is to make remembering it unnecessary

Don Norman ·

Designers understand people's physical limitations well and greatly misunderstand their mental limitations

Don Norman ·

It should not be possible for one simple error to cause widespread damage

c. 2013

Don Norman ·

Providing a good conceptual model is the most important principle for taming complexity

December 2013

Jared Spool ·

Design is the rendering of intent: the designer imagines an outcome and puts forth activities to make it real.

Jared Spool ·

Everyone who renders their intention is a designer, so the experienced designer's role shifts from owning the outcome to teaching the rest of the team to render theirs.

Arthur C. Brooks ·

Pursuing faith, family, community and work is the surest path to the share of happiness a person can control.

October 2013

John Allspaw ·

People's actions are driven by local rationality based on their specific context and goals.

Kathy Sierra ·

A presenter should aim to act as a user interface that enables an experience without drawing attention to themselves.

Kathy Sierra ·

The most important aspect of a presentation is the post-UX experience, or what happens to the audience as a result of the presentation.

August 2013

Rick Steves ·

Europeans often judge American food based on their own cuisine.

July 2013

Kathy Sierra ·

The act of making choices is a cognitive drain.

Des Traynor ·

Building a great product means being great at saying no -- not maybe, not later, no.

Des Traynor ·

Whether a feature is quick to build is a roadmap question, never a reason on its own to build it.

Des Traynor ·

A real product decision says a proposal is a great idea and still turns it down for something else.

June 2013

Chris Blattman ·

Aid is not development: transfers can relieve poverty, but the work that actually lifts a country out of it is not work donors do.

Chris Blattman ·

What keeps poor countries poor is political and economic instability, not a shortage of well-designed programmes.

Brad Frost ·

Aspects of design like color, typography, and grids are and will always be subjective.

Brad Frost ·

Interfaces can be broken down into fundamental building blocks to be worked up from.

March 2013

Don Norman · no longer holds

Design thinking is a myth: designers have no special creative thought process, and the term is a public relations label for good old-fashioned creative thinking.

Don Norman ·

Design thinking is practised by all great thinkers, but design is the one field that tries to teach it as a systematic method rather than leave it to the exceptional.

Don Norman ·

Breakthroughs come from questioning the obvious, and the outsider's stupid question is where that power lies.

January 2013

Wayne Hale ·

NASA denied Shuttle crews parachutes and pressure suits because it had convinced itself the vehicle was as safe as an airliner.

Daniel Dennett ·

Civilization may be becoming hyper-fragile through dependence on technologies that are not as reliable as they should be

Daniel Dennett ·

Most of the regularities in culture are not treasures

December 2012

Anil Dash ·

Platforms wrongly believe that giving users flexibility and control makes a product too complex to grow, and that belief is the root of most of their mistakes.

November 2012

Paul Graham ·

Build the product for yourself first: if it truly serves you, plenty of other people like you will want it too.

October 2012

Rasmus Andersson ·

Cooperative multitasking suits coherent, specialized systems better than general-purpose operating systems.

Benny Lewis · by

Speaking badly is not a failure on the way to speaking well; it is a stage that has to be passed through on purpose.

David Deutsch ·

What distinguishes human minds is creativity: the ability to produce new explanations.

David Deutsch ·

Bayesianism is a version of the mistaken induction approach to AGI.

David Deutsch ·

Expecting AGI without first understanding how it works is like expecting skyscrapers to fly if built tall enough.

Robert Greene ·

Great achievers' powers were the fruit of intense labour; genetics seemed to play only a minor role

September 2012

Karri Saarinen ·

Founders, designers and developers need to develop their taste the way a sushi chef develops a palate, and appreciating culture is how.

August 2012

Dieter Rams · by

Innovative design develops in tandem with innovative technology and can never be an end in itself.

Dieter Rams · by

The aesthetic quality of a product is integral to its usefulness, but only well-executed objects can be beautiful.

Dieter Rams · by

Good design makes a product understandable: at best it is self-explanatory.

Dieter Rams · by

Good design is unobtrusive: products that fulfil a purpose are tools, neither decorative objects nor works of art, and should leave room for the user's self-expression.

Dieter Rams · by

Good design is honest: it does not make a product seem more innovative, powerful or valuable than it is, and does not manipulate with promises that cannot be kept.

Dieter Rams · by

Good design is long-lasting: it avoids being fashionable and therefore never appears antiquated.

Dieter Rams · by

Good design is thorough down to the last detail: nothing must be arbitrary, and care in the design process shows respect towards the user.

Dieter Rams · by

Good design is as little design as possible: less, but better, because it concentrates on the essentials.

April 2012

Tynan Sylvester ·

The audience responds well to well-designed systems; it is the industry’s paymasters who undervalue them in favour of authored stories.

c. 2011

Daniel Kahneman ·

Confidence in a belief reflects the coherence of the story the mind built, not the quality of the evidence

Daniel Kahneman ·

Subjective confidence can feel the same for intuitions from expertise and for intuitions from substitution

Rich Hickey ·

Simplicity is a choice, and a system that is not simple is its builders' fault.

December 2011

Jason Fried ·

Design that looks good at the end of the day often does not look good the next morning.

November 2011

Bret Victor ·

Hands feel things and hands manipulate things, and most visions of future interfaces ignore both.

Bret Victor ·

Touchscreens are pictures under glass, a transitional technology; calling them the future of interaction is like calling black-and-white the future of photography.

Bret Victor ·

The future of interaction is a choice, not something that happens to us: people choose which visions to pursue and which research to fund.

Bret Victor ·

With an entire body at our command, the future of interaction should not be a single finger.

October 2011

Patrick McKenzie ·

An employee’s startup equity grant is most likely worth nothing, because people overestimate both the odds of success and their own share of it.

Kieran Healy ·

Charismatic organizations face the difficulty of institutionalizing a founder's personal vision after they depart.

Benny Lewis · by

Everything other than talking to a native speaker is auxiliary, and none of it will get you speaking on its own.

August 2011

Holden Karnofsky ·

A world where altruists followed expected-value estimates would be one where nearly all of them helped distant strangers instead of the people around them, and that world would be worse.

Holden Karnofsky ·

The moment one cause looks far better than every other on thin information is exactly the moment it most needs sceptical investigation.

Holden Karnofsky ·

The more an argument asks you to do, the more evidence you should require before doing it.

Holden Karnofsky ·

The deeper anyone digs into a charity's cost-effectiveness estimate, the more unwarranted optimism they find, so the right prior on giving is a sceptical one.

July 2011

Derek Sivers ·

Build the product for yourself first: if it truly serves you, plenty of other people like you will want it too.

May 2011

Atul Gawande ·

Medicine trains and pays doctors to be cowboys when patients need pit crews

April 2011

Ben Horowitz ·

Peacetime and wartime require radically different management styles

c. 2010

Bill Gates ·

Innovation moves faster when different parties build on each other's work and avoid repeating each other's dead ends

August 2010

Cal Newport ·

Doing less and doing it better is worth pursuing as an end in itself, because a life with both freedom and focus is a life well lived.

June 2010

Don Norman ·

Design thinking is a myth: designers have no special creative thought process, and the term is a public relations label for good old-fashioned creative thinking.

Jarrett Walker ·

Rich people will have self-driving cars and the city will not be built around them.

May 2010

Steve Jobs · by

The mobile era is about low power devices, touch interfaces and open web standards.

April 2010

Allan Odgaard ·

Static typing does not require manifest typing — type inference lets a language avoid that without becoming dynamically typed.

February 2010

Karri Saarinen ·

A product vision cannot be transferred through specs and bullet points; the product visionary has to engage in doing.

December 2009

Don Norman ·

Design research is great for improving existing product categories but essentially useless for new, innovative breakthroughs.

October 2009

Cal Henderson · by

Web applications should be able to leverage emoji when receiving or displaying data from mobile devices.

Ben Fry · by

Technology purchasers for software systems are often several levels removed from the end users, leading to little market pressure for usability.

April 2009

Peter Thiel ·

The great task for libertarians is to escape politics in all its forms, rather than to win within it.

February 2009

Cal Newport ·

Work that matters requires tolerating discomfort and boredom rather than escaping into distraction.

November 2008

David MacKay ·

Airplanes are already almost as energy-efficient as they could possibly be, so no redesign will radically improve their efficiency.

David MacKay ·

Wave power is unlikely to play a significant role in solving Britain's sustainable energy problem, though it may suit small remote-island communities.

David MacKay ·

Using tree planting to reverse climate change is not workable because it would require country-sized land areas.

June 2008

Oprah Winfrey ·

Feelings are a guidance system for life: right decisions come from the gut, wrong ones from not listening to it

Max Bazerman ·

Better organizations recruit more effectively, because most people want to work for a high-quality organization

January 2008

Jared Diamond ·

Living standards are not tightly tied to consumption, so consuming less need not mean real sacrifice.

October 2007

Peter Bevelin ·

Acquiring insight into how things work is a worthwhile effort in itself, not only for any payoff

May 2007

Charlie Munger ·

Acquiring wisdom is a moral duty and not only something done to advance in life

Charlie Munger ·

Objectivity maintenance routines are totally required for correct thinking

February 2007

Jeff Atwood ·

FizzBuzz-style screening is needed because many people applying for programming jobs cannot program.

December 2006

Martin Nowak ·

Evolution could not have built new levels of organization at all without cooperation.

November 2006

Daniel Dennett ·

There are solid grounds for believing that intercessory prayer simply does not work

October 2006

Joel Spolsky ·

Interviewers should hear nothing about a candidate beforehand and should not compare notes until each has decided independently.

August 2006

Matt Cutts ·

In the on-page space, one should think more about words and variants and less about keyword density or repeating phrases.

June 2006

John Maeda ·

More emotions are better than fewer: when emotion comes first, do not be afraid to add ornament or layers of meaning.

May 2006

Bruce Schneier ·

The right to privacy is fundamental; without it people cannot feel fully free and protected.

March 2006

Bret Victor ·

For most software, interactivity is a curse for users and a crutch for designers.

Bret Victor ·

Information software design is graphic design: a software designer who decides what the user will see is doing graphic design whether she realizes it or not.

c. 2005

Charlie Munger ·

One should never think about something else when one should be thinking about the power of incentives

Charlie Munger ·

Because of incentive-caused bias, a professional advisor’s advice deserves distrust, especially when it is especially good for the advisor

Charlie Munger ·

Reaching a conclusion quickly and then resisting any change in it naturally causes many errors of cognition

August 2005

John Ioannidis ·

The more analytical freedom a field allows in designs, definitions and outcomes, the less likely its published findings are to be true.

June 2005

Kathy Sierra ·

Continuing to add new features eventually leads to a decline in usability and maintenance.

Steve Jobs ·

Time is limited, so one should not live someone else’s life: dogma is living with other people’s thinking, and one’s own intuition already knows what one wants to become.

January 2005

Kathy Sierra ·

Effective teaching involves designing interactive situations rather than just delivering content.

Kathy Sierra ·

Authors who focus on proving their own technical prowess can inadvertently make learners feel less intelligent.

August 2004

David Spiegelhalter ·

Non-risk-adjusted mortality data cannot support patient choice and only signal that a fuller review is required.

May 2004

Kurt Vonnegut ·

The most vocal Christians demand the Ten Commandments in public buildings but never mention the Beatitudes

February 2004

Jeff Atwood ·

Computers change but people do not, so understanding people is how you make better software

c. 2002–2003

Daniel Kahneman ·

Predictions should be less extreme than the information they rest on, an idea that is deeply counterintuitive

September 2003

Edward Tufte ·

The standard PowerPoint presentation elevates format over content and turns everything into a sales pitch.

Edward Tufte ·

Visual reasoning works best when information is shown side by side, and the more intense the detail, the greater the clarity.

Edward Tufte ·

Chartjunk is a clear sign of statistical stupidity.

Edward Tufte ·

Audience boredom is usually a content failure, not a decoration failure: if the numbers are boring you have the wrong numbers.

August 2003

Martin Fowler ·

Lines of code cannot measure productivity, because better-designed code is shorter.

May 2003

Paul Graham ·

You need a good sense of design to judge good design, and there is no correlation, except possibly a negative one, between people's ability to recognize it and their confidence that they can.

Paul Graham ·

Big companies have software designed by committee to avoid disasters, so a startup wins design wars by having the same people design and implement the product.

Paul Graham ·

Premature design, deciding too early what a program should do, is as dangerous as premature optimization; specifications should be allowed to change on the fly.

Paul Graham ·

Great software requires a fanatical devotion to beauty: the parts no one is supposed to see should be beautiful too, because unseen details become visible in the aggregate.

Paul Graham ·

Empathy is probably the single most important difference between a good hacker and a great one, because great software has to be seen from the user's point of view.

January 2003

Paul Graham ·

Design is for users, but you have to design what the user needs rather than what he says he wants.

Paul Graham ·

You are most likely to get good design if the intended users include the designer, because looking down on the user corrupts the designer.

Paul Graham ·

To get good design you have to get close to your users and stay close, calibrating your ideas on actual users constantly.

Paul Graham ·

It is hard to design something good for users you look down on, because you cannot stay interested in something you do not like yourself.

Paul Graham ·

Good design requires a dictator: design by committee is a synonym for bad design, because a design has to be all of a piece.

December 2002

Susan Sontag ·

What looks like callousness toward images of suffering comes from the unstable attention television is built to arouse and satiate

February 2002

Paul Graham ·

Taste is not just personal preference: there is such a thing as good and bad design, and a maker's taste can be wrong.

Paul Graham ·

Good design is simple, because being forced to be simple forces a maker to face the real problem and deliver substance instead of ornament.

Paul Graham ·

Good design is timeless: if something is ugly it cannot be the best solution, and someone will eventually find a better one.

Paul Graham ·

Good design solves the right problem; a lot of bad design is industrious but misguided.

Paul Graham ·

Great designers make it look easy, and the ease is mostly an illusion: the conversational tone of good writing comes on the eighth rewrite.

Paul Graham ·

Good design is redesign: it is rare to get things right the first time, and experts plan to throw away early work.

Paul Graham ·

Good design can copy: the greatest masters decide it is more important to be right than original.

Paul Graham ·

Great design comes disproportionately from a few hotspots, because nothing is more powerful than a community of talented people working on related problems.

July 2000

Jakob Nielsen ·

Users spend most of their time on other sites, so they prefer yours to work the same way as the sites they already know.

Jakob Nielsen ·

A feature that works the same everywhere needs no teaching: a first-time visitor can use it by transferring what they learned on other sites.

March 1999

Edsger W. Dijkstra ·

In computing, elegance is not a luxury: it decides between success and failure.

Edsger W. Dijkstra ·

A design has to be kept simple for reliability and intellectual control, but nobody yet knows how to reach simplicity systematically.

May 1998

Steve Jobs ·

It is really hard to design products by focus groups, because people often do not know what they want until they are shown it.

January 1998

Yoshua Bengio ·

With suitable architecture, gradient-based learning can form decision surfaces that classify high-dimensional patterns such as handwritten characters with minimal preprocessing.

Yoshua Bengio ·

Convolutional neural networks are specifically designed to handle variability in two-dimensional shapes.

c. 1996

Mihaly Csikszentmihalyi ·

A website that promotes flow should be built like a gourmet meal, but most are built like a cafeteria

Mihaly Csikszentmihalyi ·

Clear, hierarchical goals turn a random walk into a chase

February 1996

Steve Jobs ·

The Web spread so fast largely because of its simplicity, and making it more complicated too quickly would be a loss.

Steve Jobs ·

The Web is an incredible democratizer: a small company can look as large and be as accessible as a big company, neutralizing the advantage of costly distribution channels.

Steve Jobs ·

What is wrong with education cannot be fixed with technology.

April 1995

Steve Jobs ·

What ruined Apple was not growth but values: executives who cared more about their own glory and wealth than about making great computers for people to use.

May 1990

Tom Peters ·

The relative development of brains is the only remaining distinguishing feature for national competitive advantage.

August 1984

Edsger W. Dijkstra ·

Simplicity takes hard work to achieve and education to appreciate, and complexity sells better.

c. 1979

Christopher Alexander ·

The timeless way of building brings order out of ourselves and cannot be attained, only allowed to happen

Christopher Alexander ·

Working in the timeless way, a town can grow without architects or planners

Christopher Alexander ·

Wanting to bring a building or a town to life is a fundamental human instinct

Christopher Alexander ·

Used well, the method of pattern languages ends by freeing us from all method

Christopher Alexander ·

There is a central quality, objective and precise but nameless, that is the root criterion of life in buildings, towns and people

Christopher Alexander ·

A lifeless place almost always has a mastermind behind it whose will leaves no room for its own nature

Christopher Alexander ·

The more living patterns a room, building or town has, the more it comes to life as a whole

Christopher Alexander ·

A place comes alive not through one isolated pattern but through a whole interdependent system of patterns

Christopher Alexander ·

The quality that makes places alive cannot be made, only generated by a process

Christopher Alexander ·

The process that makes towns alive must let their people shape rooms, houses, streets and churches for themselves

Christopher Alexander ·

Patterns decide not only a building's shape but how far it comes to life

Christopher Alexander ·

Pattern languages are the source of beauty and of ugliness in buildings

Christopher Alexander ·

To let a pattern form itself in design, you must give up control and let the pattern do the work

Christopher Alexander ·

To make a building egoless, the builder must let go of willful images and start with a void

c. 1978

Herbert A. Simon ·

The social sciences should be hardened, and natural and social scientists should work together on public policy

December 1978

Herbert A. Simon ·

Solving problems like hunger, pollution and poverty requires deeper knowledge of both nature's laws and human behaviour

c. 1977

Christopher Alexander ·

Towns and buildings can only become alive if all the people make them, sharing a common pattern language that is itself alive

Christopher Alexander ·

A design pattern describes a recurring problem and the core of its solution, usable a million times without repeating itself

Christopher Alexander ·

No design pattern is an isolated entity

Christopher Alexander ·

There is abundant evidence that high buildings make people crazy

Christopher Alexander ·

The more monolithic a building is, the more it prevents people from being personal and making human contact

Christopher Alexander ·

Everybody loves window seats, bay windows and big low-silled windows with chairs drawn up to them

c. 1970

Mortimer Adler ·

Temperance is the will-power to resist good fortune that offers too much of a good thing

Mortimer Adler ·

Moral virtue is a settled habit of preferring a good life to a good time

Mortimer Adler ·

The virtues needed to lead a good life cannot be formulated as rules

c. 1968

Melvin Conway ·

No design group can be both organised and unbiased.

Melvin Conway ·

Any system design will someday be bettered by someone.

Melvin Conway ·

An organisation that is not fully flexible in how it communicates stamps an image of itself on every design it produces.

Melvin Conway ·

The larger an organisation, the less flexible it is and the more its designs copy it.

Paulo Freire ·

Having internalized the oppressor's image, the oppressed fear freedom, which would demand autonomy and responsibility instead

Melvin Conway ·

A small, well-chosen design team gives a better system; reasoning that works for peeling potatoes fails for design.

Paulo Freire ·

Freedom is acquired by conquest, not received as a gift, and must be pursued constantly and responsibly

Melvin Conway ·

Probably the biggest common cause of poorly designed systems is a design organisation in need of work.

c. 1963–1968

Karl Popper ·

Every genuine test of a theory is an attempt to refute it

c. 1968

Melvin Conway ·

Military-style organisations design systems that look like their organisation charts.

Melvin Conway ·

Organisations that design systems are constrained to produce designs that copy their communication structures.

Melvin Conway ·

A design effort should be organised according to the need for communication.

Paulo Freire ·

Apart from inquiry and praxis, individuals cannot be truly human

Melvin Conway ·

Because the first design is almost never the best, organisational flexibility matters for good design.

Melvin Conway ·

Adding people to a design effort does not simply add productivity.

Paulo Freire ·

Knowledge emerges only through invention, re-invention and restless, hopeful inquiry people pursue in the world and with each other

Paulo Freire ·

Any situation in which some people prevent others from engaging in inquiry is one of violence

Paulo Freire ·

True humanists cannot use banking education to pursue liberation, which it would negate, nor may revolutionary societies inherit it

Paulo Freire ·

What distinguishes revolutionary leaders from the dominant elite is not only their objectives but their procedures

December 1968

Garrett Hardin ·

Freedom in a commons brings ruin to all

Garrett Hardin ·

Other, more precious freedoms can be preserved only by giving up the freedom to breed, and soon

c. 1963–1964

Hannah Arendt ·

Denmark's people and government behaved uniquely among all European countries in the story of the Danish Jews

c. 1964

Hannah Arendt ·

Political responsibility exists apart from individual deeds and can be neither judged morally nor tried in a criminal court

c. 1962

Rachel Carson ·

Known alternatives to chemical spraying should be used more widely, and ingenuity and resources devoted to developing new ones

Rachel Carson ·

Modern insect control overlooks that the really effective control of insects is the one nature applies, not people

c. 1961

Frantz Fanon ·

Impetuous spontaneity bent on quickly settling scores with the colonial system is destined to fail as a doctrine

Frantz Fanon ·

Nationalism that is not explained, enriched, deepened and quickly turned into social and political consciousness leads to a dead end

Frantz Fanon ·

We should not pay tribute to Europe by building states and societies modelled on it; humanity expects more than that emulation

c. 1959

Isaac Asimov ·

Making the cross-connections that produce new ideas requires a certain daring

Isaac Asimov ·

Creativity, in the writer's feeling, requires isolation

January 1958

Frank Rosenblatt ·

Understanding higher organisms' perceptual recognition, generalization, recall, and thinking requires answers to three fundamental questions about sensing, storage, and behavioral influence of information.

c. 1955

Albert Camus ·

It is legitimate and necessary to ask whether life has meaning, and so to face suicide squarely

December 1954

Ernest Hemingway ·

What a writer puts into the work eventually becomes clear, and that plus craft decides whether the writer endures or is forgotten

c. 1944

Friedrich Hayek ·

Voluntary agreement can guide state action only in areas where people actually agree

Friedrich Hayek ·

Observing the Rule of Law is what most clearly separates a free country from arbitrary government

Friedrich Hayek ·

A policy of freedom for the individual is still the only truly progressive policy, as in the nineteenth century

c. 1942

Albert Camus ·

Suicide does not follow from revolt; it is not revolt's logical outcome

Albert Camus ·

Suicide, like the leap, is acceptance at its extreme

Albert Camus ·

If life is only absurd and freedom bounded by limited fate, what counts is the most living, not the best

c. 1930

Sigmund Freud ·

People seem to measure by false standards, seeking power, success and riches while undervaluing what is truly precious in life

Bertrand Russell ·

Humans depend on co-operation, and nature has given them, somewhat inadequately, the instincts for the friendliness it needs

Bertrand Russell ·

A happy life must be largely a quiet life, because true joy can live only in quiet

Bertrand Russell ·

Happiness that needs intoxication of any kind is spurious; satisfying happiness comes with the fullest use of one's faculties

Sigmund Freud ·

There is no recipe for happiness that suits everyone; each person must find their own means

Bertrand Russell ·

A woman with professional skill ought to be free to keep using it despite motherhood, for her own sake and the community's

Sigmund Freud ·

The price of progress in civilization is paid in lost happiness, through a heightened sense of guilt

Bertrand Russell ·

Except in very rare cases, happiness does not simply fall into one's mouth through fortunate circumstances

Bertrand Russell ·

Admitting at least one painful truth to oneself every day is as useful as a daily kind action

Sigmund Freud ·

Humanity's power over nature has grown so great that people could now easily exterminate one another

c. 1925

Bertrand Russell ·

The good life is one inspired by love and guided by knowledge

c. 1921

Ludwig Wittgenstein ·

The world of the happy is quite another than the world of the unhappy

c. 1918–1920

William Strunk Jr. ·

Concise writing does not mean short sentences or leaving out detail; it means making every word tell.

June 1920

Max Planck ·

Research needs the pursuit of a specific aim, which guides it through early failures

Max Planck ·

Although only entropy differences are measured, it is useful to define an absolute value of entropy

Max Planck ·

The quantum of action has not yet produced a genuine quantum theory

Max Planck ·

Bohr's atom rests on hypotheses every physicist would have flatly rejected a generation ago

Max Planck ·

Science will master the quantum dilemma, and what looks unsatisfactory now will prove harmonious and simple

c. 1918

Bertrand Russell ·

The world worth seeking is one built on the impulse to create rather than the desire to keep or seize possessions

c. 1917

Will Durant ·

Perhaps the most valuable information is the kind that guides the search for information

May 1905

Nitobe Inazō ·

Bushido may vanish as a separate code of ethics, but its power and light will survive

February 1903

W. E. B. Du Bois ·

Black Americans need schooling more than ever, both practical training and higher culture

W. E. B. Du Bois ·

Making the blacksmith a scholar is almost, but not quite, as silly as making the scholar a blacksmith

c. 1902

William James ·

Healthy-minded optimism is inadequate as a philosophy, because the evil it ignores is part of reality and may be the key to life's meaning

c. 1890–1902

Peter Kropotkin ·

It would contradict all we know of nature if defenceless early humans had progressed by reckless competition rather than mutual support

c. 1902

William James ·

No account of the universe can be final if it disregards forms of consciousness other than the ordinary one

William James ·

Human experience as a whole, as I see it, points beyond narrow 'scientific' bounds; the real world is more intricately built than physical science allows

c. 1898

Elizabeth Cady Stanton ·

The early case for liberal divorce was not wrong, only made before people were ready to hear it

February 1867

John Stuart Mill ·

Universities exist to make capable and cultivated human beings, not to train people for a particular livelihood

John Stuart Mill ·

What most distinguishes one mind from another is the ability to judge evidence correctly

John Stuart Mill ·

A university ought to be a place of free inquiry, and the better it does its other duties the more surely it will be

c. 1861

John Stuart Mill ·

The main ingredients of a satisfied life appear to be tranquillity and excitement, either of which is often enough alone

c. 1859–1860

Florence Nightingale ·

Nursing should mean the proper use of fresh air, light, warmth, cleanliness, quiet and diet, not just giving medicines

October 1859

Charles Darwin ·

Animals descend from at most four or five ancestors, and analogy hints at one ancestor for all life, though analogy may mislead

Charles Darwin ·

This theory will give psychology a new foundation and throw light on the origin of humans

c. 1854

Henry David Thoreau ·

Most people lead lives of quiet desperation, and what is called resignation is confirmed desperation

Henry David Thoreau ·

Most luxuries and many comforts are not only unnecessary but positive hindrances to human elevation

Henry David Thoreau ·

The cost of a thing is the amount of life that has to be exchanged for it, now or in the long run

July 1852

Frederick Douglass ·

No nation on earth is guilty of practices more shocking and bloody than the United States at this time

c. 1851

Arthur Schopenhauer ·

Cheerfulness is the direct currency of happiness, and securing it should be the supreme aim of the pursuit of happiness

Arthur Schopenhauer ·

The value people set on the opinion of others is out of all proportion to anything it can gain them

November 1844

Margaret Fuller ·

Men and women constantly pass into one another; no man is wholly masculine and no woman purely feminine

c. 1843

Ada Lovelace ·

The analytical engine gives analysis a new, vast and powerful language, making its truths more quickly and accurately useful to mankind

c. 1838

Harriet Martineau ·

Many women must support themselves without access to varied work or education, so they are raised to see marriage as their one object

December 1790

Mary Wollstonecraft ·

The passions are necessary helpers of reason

Mary Wollstonecraft ·

The only rightful security of property is the right to enjoy and bequeath what one's own talents and industry acquired

March 1789

Olaudah Equiano ·

Christians who sell enslaved families apart betray their own rule to do unto others as they would be done by

c. 1785

Immanuel Kant ·

In the kingdom of ends, what can be replaced by an equivalent has value; what is above all value and admits no equivalent has dignity

July 1783

Benjamin Franklin ·

Money spent on wars would have added vastly more to human comfort if spent on public works

March 1776

Adam Smith ·

No society can flourish when most of its members are poor, and those who feed and house everyone deserve to be decently fed and housed

Adam Smith ·

By pursuing their own interest, people frequently serve society better than when they set out to serve it

c. 1748

David Hume ·

Custom is the great guide of human life; it alone makes experience useful by leading us to expect the future to resemble the past

David Hume ·

No testimony suffices to establish a miracle unless its falsehood would be more miraculous than the fact it seeks to establish

c. 1733

Voltaire ·

By resisting their kings, the English alone have limited royal power and built a government that can do good but is restrained from evil

c. 1713–1726

Isaac Newton ·

Never give up the evidence of experiments for dreams and fictions of our own devising

c. 1704–1718

Isaac Newton ·

The properties of light should be proved by reason and experiments, not explained by hypotheses

Isaac Newton ·

Induction from experiments demonstrates nothing, but it is the best way of arguing that nature allows

Isaac Newton ·

Perfecting natural philosophy will also enlarge the bounds of moral philosophy

c. 1694–1697

Mary Astell ·

Since the soul was made to contemplate truth, keeping women from knowledge is as cruel and unjust as keeping them from good

September 1656

Blaise Pascal ·

Violence cannot weaken truth, and truth cannot stop violence; each only strengthens or provokes the other

c. 1651

Thomas Hobbes ·

Words are tools the wise reckon with, but fools treat them as money, valuing them by the authority of famous names

Thomas Hobbes ·

Nature made people so equal in body and mind that no one can claim a benefit another cannot also claim

Thomas Hobbes ·

Without security, there is no industry, arts or society, only fear of violent death, and life is solitary, poor, nasty, brutish and short

Thomas Hobbes ·

The papacy is nothing but the ghost of the dead Roman Empire, sitting crowned on its grave

June 1637

René Descartes ·

Having a vigorous mind is not enough; what matters most is applying it rightly

René Descartes ·

If anything can ever make people wiser and more ingenious, it is probably to be found in medicine

c. 1597–1625

Francis Bacon ·

Seeking, knowing and believing truth is the highest good of human nature

c. 1513–1514

Niccolò Machiavelli ·

Advice should follow how people actually live, since acting wholly on professed virtue among so much evil leads to ruin

c. 1258–1273

Rumi ·

Whoever has doubt or disquiet in the heart is secretly a denier and a philosopher

c. 1126–1195

Ibn Rushd (Averroes) ·

The business of philosophy is to study creation in order to be guided to its Creator

Ibn Rushd (Averroes) ·

We should study what the ancients wrote, accept what is true and be warned by what is wrong

Ibn Rushd (Averroes) ·

No one can discover alone everything a science or an art requires

c. 1190

Moses Maimonides ·

The Law aims at the well-being of both soul and body, the soul's through correct opinions suited to people's capacity

c. 1106–1111

Al-Ghazali ·

Defending Islam does not require condemning the exact sciences, which religious law neither approves nor condemns

c. 450–500 CE

Thiruvalluvar ·

Using harsh words when kind ones are at hand is like choosing unripe fruit over ripe

c. 89–135 CE

Epictetus ·

Education is learning to apply general notions to particular cases and to tell what is in our power from what is not

Epictetus ·

Wish for things to happen as they do, not as you would like, and life will flow tranquilly

c. 62–65 CE

Seneca ·

Philosophy shapes the soul and guides conduct, showing what to do and what to leave undone

c. 341–270 BCE

Epicurus ·

Being content with little is a great good, and those who least need luxury enjoy it most

Epicurus ·

Of everything wisdom provides for a happy life, by far the most important is gaining friends

c. 450–250 BCE

Laozi ·

No calamity is greater than discontent with one's lot, and the sufficiency of contentment endures

The same record, other ways: every position as a chronology, or person by person.