This seems like a good way to separate the more “subjective” parts of judging interventions (cause-effectiveness, which depends on what you care about) from the “objective” parts (leverage ratios, which people should reach a consensus on given enough data).
An organization like GiveWell could compile estimated leverage ratios for different charities, while leaving “cause effectiveness” up to individual donors to decide. They (or others) could then provide multiple estimated cause-effectiveness tables that vary based on common differences in donors’ values (e.g. how much they value the future compared to the present). Then donors could pick which “value system” they liked best and decide donations based on that.
Would you say the Global Priorities Project is primarily focused on estimating L or C? One obvious thing your factorization suggests is that projects estimating C tend to complement projects estimating L (like GiveWell?).
I agree with the thrust of what you are saying here. I’m not sure the “objective/subjective” division quite fits, though. Of course questions of values do come into cause-effectiveness, but I also think there are a lot of questions about matters which are in theory objective, although in practice very difficult to be sure about (for instance “will health or education lead to more economic growth?”).
Would you say the Global Priorities Project is primarily focused on estimating L or C? One obvious thing your factorization suggests is that projects estimating C tend to complement projects estimating L (like GiveWell?).
In this domain the focus has been on improving our comparisons of C—exactly as you say, because it’s a better complement to things that are already being done. We’ve also been exploring some things (e.g. engaging with policy) which aren’t about estimating cost-effectiveness at all. Seb Farquhar will be joining our team in the new year, and we’ll spend some time after that thinking through what the best avenues to pursue going forward are.
I’m not sure the “objective/subjective” division quite fits, though. Of course questions of values do come into cause-effectiveness, but I also think there are a lot of questions about matters which are in theory objective, although in practice very difficult to be sure about (for instance “will health or education lead to more economic growth?”).
Good point—if you just list the effectiveness of a bunch of causes (C1, C2, C3, etc.) the shared downstream effects (like your example of economic growth) probably make them non-independent, unless you have very unusual values.
This seems like a good way to separate the more “subjective” parts of judging interventions (cause-effectiveness, which depends on what you care about) from the “objective” parts (leverage ratios, which people should reach a consensus on given enough data).
An organization like GiveWell could compile estimated leverage ratios for different charities, while leaving “cause effectiveness” up to individual donors to decide. They (or others) could then provide multiple estimated cause-effectiveness tables that vary based on common differences in donors’ values (e.g. how much they value the future compared to the present). Then donors could pick which “value system” they liked best and decide donations based on that.
Would you say the Global Priorities Project is primarily focused on estimating L or C? One obvious thing your factorization suggests is that projects estimating C tend to complement projects estimating L (like GiveWell?).
I agree with the thrust of what you are saying here. I’m not sure the “objective/subjective” division quite fits, though. Of course questions of values do come into cause-effectiveness, but I also think there are a lot of questions about matters which are in theory objective, although in practice very difficult to be sure about (for instance “will health or education lead to more economic growth?”).
In this domain the focus has been on improving our comparisons of C—exactly as you say, because it’s a better complement to things that are already being done. We’ve also been exploring some things (e.g. engaging with policy) which aren’t about estimating cost-effectiveness at all. Seb Farquhar will be joining our team in the new year, and we’ll spend some time after that thinking through what the best avenues to pursue going forward are.
Good point—if you just list the effectiveness of a bunch of causes (C1, C2, C3, etc.) the shared downstream effects (like your example of economic growth) probably make them non-independent, unless you have very unusual values.