I haven’t thought about this a lot, but my impression is this doesn’t work for smeared probability distributions and a medium level of risk aversion?
Let’s say Alice thinks there is still a 20% chance AGI doesn’t happen super fast / doesn’t have transformative impact and she needs to pay back a loan in 10 years. Then if she doesn’t want to take the risk of not being able to pay in these timelines, she cannot really donate all of the loan now.
On the other hand, a 20% chance that Alice has to donate a lot of money to global health doesn’t look like such a big risk, at least if she doesn’t have to donate everything right away.
But maybe the difference here is my implicit assumption that owing donations to Bob isn’t a big of a risk as owing money to a bank, because the former might cut Alice more slack and give her more time to pay.
Spiarrow
Karma: 53
[Question] What is the best way to do donation swapping?
[Question] Trade Between Altruists With Different AI Timelines?
[Question] Where are all the deepfakes?
Thank you, this is quite helpful!
Does anybody know if the same applies to OpenPhil?
If it can be spent over any period of time, this removes my biggest concern (that we wouldn’t find $100m worth of good opportunities for animals).
I am voting based on my prior that animal work is more impactful as it is more neglected and there are so many more animals.
Maybe this trade would be quite risky for Bob unless it is enforced so strictly that the risk profile for Alice is very similar to just borrowing money from a bank.