One-way and two-way doors

Their wordsIf you make the wrong decision, if it's a two-way door decision, you pick a door, you walk out and you spend a little time there. It turns out to be the wrong decision, you can come back in and pick another door. Some decisions are so consequential and so important and so hard to reverse that they really are one-way door decisions. You go in that door, you're not coming back. And those decisions have to be made very deliberately, very carefully.

Their wordsAlways try your stupid ideas if you can do it cheaply and reversibly. Jumping off a bridge is not a reversible decision. Not talking about that. I'm talking about stuff like this where you're just like, "Here's a stupid idea." 99 times out of a 100 it'll fail. But that one time you won't have any competition cuz nobody else is stupid enough to try this idea.

Their wordsYou cannot just reverse that, right? It's just not working. So, I do think that we need to build out the whole guardrails for the reversibility of actions because that actually where things get really really scary.

Their wordsThis is especially severe for open weight models, where deployment is essentially irreversible and mitigations can’t be improved later.

Their wordsSmall decision? Reversible? Easy to learn from? Both sides equally informed? Then leave people alone — that is exactly the territory where freedom to make our own mistakes is how we learn, and where paternalism has no business.

Their wordsAnd we get to this point, these are the hardest decisions people ever wrestle with. And the paradox is if the advantages and disadvantages of the two options are almost perfectly balanced, it doesn't matter which one we pick. And yet we will spend days and weeks going around in a funk, which one am I going to choose?