I can’t actually well-model the sorts of things the house will accomplish once fully formed
My best guess is that having a highly specific plan that includes steering/replanning capacity and then totally abandoning it when the wheels hit the road because it turns out to be the wrong thing is way better than having a generic plan.
I had some thoughts, like running a talk series at prestigious universities, publishing a book or a movie, creating an org to teach rationality to middle- or high-schoolers and then doing it, building a robot car, trying to develop Veritaserum
I’d love to see how you’d design a house specifically for any one of these goals. Robot car is the one that I think would give you the most feedback from your internal models during the planning stage, followed by publishing a book or movie. “Create an org” is a bit recursive, and a talk series is probably either too easy or too vague. Not sure what you mean by develop Veritaserum but it seems to strongly overlap with some of Leverage’s most plausibly successful research.
I claim with moderate confidence that simply walking through how the house as currently planned might go about building a robot car would substantially improve not just your plans for particular object-level capacity, but general capacity. “How will this organization change its mind?” might be a lot harder to cash out usefully than “How will this organization change its mind about valve design for the fuel injector?”.
re: your best guess, that makes sense. It’s possible I should just choose one of those plans above (many of which actually have lots of fairly detailed planning behind them already) and run with it for now.
Eli Tyre strongly agrees with your last paragraph, and is (correctly, and appreciated-ly) pushing for the first large-scale project to be determined sooner rather than later.
My best guess is that having a highly specific plan that includes steering/replanning capacity and then totally abandoning it when the wheels hit the road because it turns out to be the wrong thing is way better than having a generic plan.
I’d love to see how you’d design a house specifically for any one of these goals. Robot car is the one that I think would give you the most feedback from your internal models during the planning stage, followed by publishing a book or movie. “Create an org” is a bit recursive, and a talk series is probably either too easy or too vague. Not sure what you mean by develop Veritaserum but it seems to strongly overlap with some of Leverage’s most plausibly successful research.
I claim with moderate confidence that simply walking through how the house as currently planned might go about building a robot car would substantially improve not just your plans for particular object-level capacity, but general capacity. “How will this organization change its mind?” might be a lot harder to cash out usefully than “How will this organization change its mind about valve design for the fuel injector?”.
re: your best guess, that makes sense. It’s possible I should just choose one of those plans above (many of which actually have lots of fairly detailed planning behind them already) and run with it for now.
Eli Tyre strongly agrees with your last paragraph, and is (correctly, and appreciated-ly) pushing for the first large-scale project to be determined sooner rather than later.