While I think this piece is right in some sense, seeing it written out clearly it feels like there is something uncooperative and possibly destructive about it. To take the portfolio management case:
Why do the other fund managers prefer 100% stocks? Is this a thoughtful decision you are unthinkingly countering?
Each fund manager gets better outcomes if they keep their allocation secret from others.
I think Iâm most worried about (2): it would be bad if OP made their grants secret or individuals lied about their funding allocation in EA surveys.
Tweaking the fund manager scenario to be a bit more stark:
There are 100 fund managers
50 of them prefer fully stocks, 50 prefer an even split between stocks and bonds
If they each decide individually youâd get an overall allocation of 75% stocks and 25% bonds.
If instead they all are fully following the lessons of this post, the ones that prefer bonds go 100% bonds, and the overall allocation is 50% stocks and 50% bonds.
It feels to me that the 75-25 outcome is essentially the right one, if the two groups are equally likely to be correct. On the other hand, the adversarial 50-50 outcome is one group getting everything they want.
Note that I donât think this is an issue with other groups covering the gaps left by the recent OP shift away from some areas. Itâs not that OP thought that those areas should receive less funding, but that GV wanted to pick their battles. In that case, external groups that do accept the case for funding responding by supporting work in these areas seems fine and good. Which Moskovitz confirms: âIâm explicitly pro-funding by othersâ And: âIâd much prefer to just see someone who actually feels strongly about that take the wheel.â
(This also reminds me about the perpetual debate about whether you should vote things on the Forum up/âdown directionally vs based on how close the vote total currently is to where you think it should be.)
I think these unsavory implications you enumerate are just a consequence of applying game theory to donations, rather than following specifically from my postâs arguments.
For example, if Bob is all-in on avoiding funging and doesnât care about norms like collaboration and transparency, his incentives are exactly as you describe: Give zero information about his value system, and make donations secretly after other funders have shown their hands.
I think youâre completely right that those are awful norms, and we shouldnât go all-in on applying game theory to donations. This goes both for avoiding funging and for my postâs argument about optimizing âEAâs portfolioâ.
However, just as we can learn important lessons from the concept of funging while discouraging the bad, I still think this post is valuable and includes some nontrivial practical recommendations.
Maybe we need to flip this around. Instead of tracking how much funding was allocated to a certain cause area, we should be tracking the expected marginal opportunity in each and comparing those. I.e., what was the expected result of a marginal $1M donated to each cause area on average in, say, a given year?
This does not incentivize for making the allocations secret since the decisions are made based on the current state of the âmarketâ irrespective of any previous allocations.
Going back to the 100 fund managers example, I think Iâd much prefer them to individually recognize that people preferring the alternative allocation are just as competent in the decision as they are (in an ideal case), and as a result, apply the uncertainty to their own preference (making it 75-25 instead of 100-0/â50-50) rather than relying on an external mechanism.
While I think this piece is right in some sense, seeing it written out clearly it feels like there is something uncooperative and possibly destructive about it. To take the portfolio management case:
Why do the other fund managers prefer 100% stocks? Is this a thoughtful decision you are unthinkingly countering?
Each fund manager gets better outcomes if they keep their allocation secret from others.
I think Iâm most worried about (2): it would be bad if OP made their grants secret or individuals lied about their funding allocation in EA surveys.
Tweaking the fund manager scenario to be a bit more stark:
There are 100 fund managers
50 of them prefer fully stocks, 50 prefer an even split between stocks and bonds
If they each decide individually youâd get an overall allocation of 75% stocks and 25% bonds.
If instead they all are fully following the lessons of this post, the ones that prefer bonds go 100% bonds, and the overall allocation is 50% stocks and 50% bonds.
It feels to me that the 75-25 outcome is essentially the right one, if the two groups are equally likely to be correct. On the other hand, the adversarial 50-50 outcome is one group getting everything they want.
Note that I donât think this is an issue with other groups covering the gaps left by the recent OP shift away from some areas. Itâs not that OP thought that those areas should receive less funding, but that GV wanted to pick their battles. In that case, external groups that do accept the case for funding responding by supporting work in these areas seems fine and good. Which Moskovitz confirms: âIâm explicitly pro-funding by othersâ And: âIâd much prefer to just see someone who actually feels strongly about that take the wheel.â
(This also reminds me about the perpetual debate about whether you should vote things on the Forum up/âdown directionally vs based on how close the vote total currently is to where you think it should be.)
I think these unsavory implications you enumerate are just a consequence of applying game theory to donations, rather than following specifically from my postâs arguments.
For example, if Bob is all-in on avoiding funging and doesnât care about norms like collaboration and transparency, his incentives are exactly as you describe: Give zero information about his value system, and make donations secretly after other funders have shown their hands.
I think youâre completely right that those are awful norms, and we shouldnât go all-in on applying game theory to donations. This goes both for avoiding funging and for my postâs argument about optimizing âEAâs portfolioâ.
However, just as we can learn important lessons from the concept of funging while discouraging the bad, I still think this post is valuable and includes some nontrivial practical recommendations.
Maybe we need to flip this around. Instead of tracking how much funding was allocated to a certain cause area, we should be tracking the expected marginal opportunity in each and comparing those. I.e., what was the expected result of a marginal $1M donated to each cause area on average in, say, a given year?
This does not incentivize for making the allocations secret since the decisions are made based on the current state of the âmarketâ irrespective of any previous allocations.
Going back to the 100 fund managers example, I think Iâd much prefer them to individually recognize that people preferring the alternative allocation are just as competent in the decision as they are (in an ideal case), and as a result, apply the uncertainty to their own preference (making it 75-25 instead of 100-0/â50-50) rather than relying on an external mechanism.