korrents
Adam Tornhill

Adam Tornhill

@adam-tornhill · 18 positions · On ppll ↗

Swedish programmer with degrees in engineering and psychology, author of Your Code as a Crime Scene and founder of CodeScene.

Adam Tornhill did not write this page.

We collected these quotes from things they published elsewhere, and every quote links to where it was said. They have no account here and have not endorsed this site. Quotes are word for word; the short line under each one is our own restatement, not their wording. Their own site. Is this you? Claim it or ask us to remove it. Or tell us what is wrong here.

  1. code generation

    Just like a speed reader consumes text in patterns rather than word by word, a strong manual coding background has trained us to spot algorithmic weaknesses and flaws.

    ↗Code for Humans and Machinesadamtornhill.substack.com 1st of 3 in this piece

  2. Instead, the way forward is most likely to capture experience as enforceable constraints.

    ↗Code for Humans and Machinesadamtornhill.substack.com 2nd of 3 in this piece

  3. My personal goal is to get better at working with higher-level abstractions, and accept that writing code is likely to be a dying skill.

    ↗Code for Humans and Machinesadamtornhill.substack.com 3rd of 3 in this piece

  4. 6 weeks earlier
  5. The incremental TDD steps were great for human cognition, but didn't survive the transition to agentic coding.

    ↗Code for Humans and Machinesadamtornhill.substack.com 1st of 3 in this piece

  6. What matters is that the code supports machine reasoning and change; the machine is the primary audience.

    ↗Code for Humans and Machinesadamtornhill.substack.com 2nd of 3 in this piece

  7. AI agents

    Agents are similar, and continuing to work in ways optimized for manual programming is unlikely to give us benefits.

    ↗Code for Humans and Machinesadamtornhill.substack.com 3rd of 3 in this piece

  8. 2 months earlier
  9. The reason is that AI itself changes the quality threshold required to keep systems stable.

    ↗Code for Humans and Machinesadamtornhill.substack.com 1st of 3 in this piece

  10. And still: future systems are likely to evolve at a pace that is significantly higher than today where we still have plenty of constraints and manual checkpoints in our delivery pipelines.

    ↗Code for Humans and Machinesadamtornhill.substack.com 2nd of 3 in this piece

  11. software quality

    One possible response is to aim for superhuman code quality. Personally, I don't think that's a realistic path.

    ↗Code for Humans and Machinesadamtornhill.substack.com 3rd of 3 in this piece

  12. 7 days earlier
  13. As such, naming in software goes way beyond any aesthetics. Good naming minimizes reconstruction work. For both humans and machines.

    ↗Code for Humans and Machinesadamtornhill.substack.com 1st of 3 in this piece

  14. design

    Names are cognitive compression mechanisms. Design elements with strong names stretch the amount of information you can hold in your head at once.

    ↗Code for Humans and Machinesadamtornhill.substack.com 2nd of 3 in this piece

  15. A vague and imprecise name - like Utils - will attract code of those same qualities.

    ↗Code for Humans and Machinesadamtornhill.substack.com 3rd of 3 in this piece

  16. 3 weeks earlier
  17. design

    The less we humans code ourselves, the more we need to optimize for design, architecture, and a socio-technical fit.

    ↗Code for Humans and Machinesadamtornhill.substack.com 1st of 3 in this piece

  18. So, let me go out on a limb and claim that recurring merge conflicts are a sign of socio-technical problems.

    ↗Code for Humans and Machinesadamtornhill.substack.com 2nd of 3 in this piece

  19. AI alignment

    Recurring merge conflicts are the canary telling you that there is a misalignment between a) the work you do, and b) the type of work your architecture supports.

    ↗Code for Humans and Machinesadamtornhill.substack.com 3rd of 3 in this piece

  20. 4 weeks earlier
  21. energy

    Agentic coding collapses the timeline of software decisions. What we gain in speed, we pay for in decision density and mental energy.

    ↗Code for Humans and Machinesadamtornhill.substack.com 1st of 3 in this piece

  22. Agentic coding is mentally expensive. I can usually sustain the pace for a couple of hours. Then I need a break. The pace is simply too intense.

    ↗Code for Humans and Machinesadamtornhill.substack.com 2nd of 3 in this piece

  23. Adding insult to Big Tech injury, the study also demonstrates that even expert developers overestimate the AI impact on developer productivity.

    ↗Code for Humans and Machinesadamtornhill.substack.com 3rd of 3 in this piece