This is the type of thing that speaks to me aesthetically but my guess is that it wouldn’t pencil, though I haven’t done the math myself (nor do I have a good sense of how to model it well). Improving business psychology is just not a very leveraged way to improve the long-term future compared to the $X00M/year devoted to x-risk, so the flow-through effects have to be massive in order to be better than marginal longtermist grants. (If I was making grants in metascience this is definitely the type of thing I’d consider).
I’m very open to being wrong though; if other people have good/well-justified Fermi estimates for a pretty large effect (better than the marginal AI interpretability grant for example) I’d be very happy to reconsider.
(sorry for pontificating when you asked for an actual envelope or napkin) upside is an externality, Ziani incidentally benefits but the signal to other young grad students that maybe career suicide is a slightly more viable risk seems like the source of impact. Agree that this subfield isn’t super important, but we should look for related opportunities in subfields we care more about.
I don’t know if designing a whistleblower prize is a good nerdsnipe / econ puzzle, in that it may be a really bad goosechase (since generating false positives through incentives imposes name-clearing costs on innocent people, and either you can design your way out of this problem or you can’t).
My guess is still that this is below the LTFF bar (which imo is quite high) but I’ve forwarded some thoughts to some metascience funders I know. I might spend some more free time trying to push this through later. Thanks for the suggestion!
Is this a retroactive grant situation?
This is the type of thing that speaks to me aesthetically but my guess is that it wouldn’t pencil, though I haven’t done the math myself (nor do I have a good sense of how to model it well). Improving business psychology is just not a very leveraged way to improve the long-term future compared to the $X00M/year devoted to x-risk, so the flow-through effects have to be massive in order to be better than marginal longtermist grants. (If I was making grants in metascience this is definitely the type of thing I’d consider).
I’m very open to being wrong though; if other people have good/well-justified Fermi estimates for a pretty large effect (better than the marginal AI interpretability grant for example) I’d be very happy to reconsider.
(sorry for pontificating when you asked for an actual envelope or napkin) upside is an externality, Ziani incidentally benefits but the signal to other young grad students that maybe career suicide is a slightly more viable risk seems like the source of impact. Agree that this subfield isn’t super important, but we should look for related opportunities in subfields we care more about.
I don’t know if designing a whistleblower prize is a good nerdsnipe / econ puzzle, in that it may be a really bad goosechase (since generating false positives through incentives imposes name-clearing costs on innocent people, and either you can design your way out of this problem or you can’t).
My guess is still that this is below the LTFF bar (which imo is quite high) but I’ve forwarded some thoughts to some metascience funders I know. I might spend some more free time trying to push this through later. Thanks for the suggestion!