Great post, and sorry to hear about your dark night Tyler. I think one thing that has given me pause in EA has been the explore-exploit tradeoff. Crucial Considerations, and the idea of working smarter rather than harder, means that it’s unlikely to be optimal to go all out on exploiting any given opportunity. Related is the idea of giving now vs giving later, which should probably relate to time as well as money—i.e. it could be better to grow your skills before you spend them down in a sprint/burn-out, and perhaps better to not sprint/burn-out at all given new Crucial Considerations may always be forthcoming.
I think these are true, and mitigate some of the problem of “going all in”. As I see it, though, they don’t mitigate the potential (psychological) harms from treating most things as a means to impact.
I guess a failure mode of this is just going all out when exploring as well as exploiting! And optimising your pauses for reflection to maximise both. Perhaps ultimately there is no substitute for a well-cultivated garden of ends.
As mentioned in a footnote of the OP, an exception to this might be if your AGI timelines are short. Even then though (as is the case for me), there is still uncertainty over what is the right action. I guess there is always more reading to do to figure things out better, but that always requires time to digest and think. It’s still better to work smarter rather than harder in this case. And Tyler makes a point here about more well-adjusted and happy people potentially being better at coordination.
Great post, and sorry to hear about your dark night Tyler. I think one thing that has given me pause in EA has been the explore-exploit tradeoff. Crucial Considerations, and the idea of working smarter rather than harder, means that it’s unlikely to be optimal to go all out on exploiting any given opportunity. Related is the idea of giving now vs giving later, which should probably relate to time as well as money—i.e. it could be better to grow your skills before you spend them down in a sprint/burn-out, and perhaps better to not sprint/burn-out at all given new Crucial Considerations may always be forthcoming.
I think these are true, and mitigate some of the problem of “going all in”. As I see it, though, they don’t mitigate the potential (psychological) harms from treating most things as a means to impact.
I guess a failure mode of this is just going all out when exploring as well as exploiting! And optimising your pauses for reflection to maximise both. Perhaps ultimately there is no substitute for a well-cultivated garden of ends.
As mentioned in a footnote of the OP, an exception to this might be if your AGI timelines are short. Even then though (as is the case for me), there is still uncertainty over what is the right action. I guess there is always more reading to do to figure things out better, but that always requires time to digest and think. It’s still better to work smarter rather than harder in this case. And Tyler makes a point here about more well-adjusted and happy people potentially being better at coordination.