I’d distinguish between two ways in which a report can ‘be’ cause-neutral:
1. Whether its domain of focus/​cause area was chosen purely through cause prioritisation
2. Whether its contents are of value from a cause-neutral perspective
Now I agree that this report is not cause-neutral on (1): it was written at least partially because many of FP’s community members are interested in women’s empowerment.*
However, note that cause prioritisation is just a heuristic to restrict our domain of search: what you want to compare in the end are the (donation) opportunities themselves, not which cause/​domain they happen to be in by some categorisation.
Maybe you don’t think women’s empowerment should be the first domain to check when you are looking for the highest-impact charities overall, but you should at least agree that it is valuable from a cause-neutral perspective to know what the best charities within this particular domain of search are. You might then be surprised that they are actually better than you thought, or you might find that your intuition of other areas having better opportunities is confirmed.
As the methodology of this report allows you to compare the charities to those in other areas (we don’t use outcome measures that are restricted to women’s empowerment/​the analysis is done in a cause-neutral frame), I think it to be cause-neutral on (2). And I hence think it’s very much worth discussing (from a cause-neutral perspective of course!) its contents on the EA forum, e.g. how do the recommended charities compare to other near-term welfare opportunities, such as those recommended by GiveWell?
Lastly, I don’t think this research provides a post-hoc justification for women’s empowerment: in my view it could have as much provided a justification to not donate in that area (if the best charities turn out to be worse than in other areas) as a justification to donate in that area. At FP we do research into areas not to justify our member’s initial preferences, but to be recommend high-impact opportunities tailored to those preferences (if high-impact opportunities are available), as well as to be able to make a solid, justified argument to focus on other areas (if higher-impact opportunities are available in those other areas).
*This does not mean that the choice of writing this report was a non-cause-neutral choice: for FP to do the most good we obviously need to take our community’s preferences into account. Neither does it mean that one couldn’t arrive at women’s empowerment as a high-potential cause area through cause prioritisation.
Maybe you don’t think women’s empowerment should be the first domain to check when you are looking for the highest-impact charities overall, but you should at least agree that it is valuable from a cause-neutral perspective to know what the best charities within this particular domain of search are. You might then be surprised that they are actually better than you thought, or you might find that your intuition of other areas having better opportunities is confirmed.
I agree that comparisons of that type are valuable, but I don’t think that this report helps me much in doing that kind of comparison. This report did no comparative analysis of the interventions against other near-term welfare interventions, and you used denominations that make that comparison quite difficult (as SiebeRozendal pointed out in another comment).
See for example this:
This suggests that Village Enterprise’s programme can bring about nominal gains in consumption of about $0.99 for each $1.00 donated. Adjusting for purchasing power, this is equivalent to gains of $2.18 for each $1.00 donated.
I don’t know how to compare an increase in consumption with other near-term interventions, so as long as this number isn’t shockingly high or low it’s quite hard for me to judge whether this is a good intervention. So while your analysis helps me a bit in comparing Village Enterprise to other near-term welfare charities, it really doesn’t help me much and I still need to put in the vast majority of work, which consists of building models about the world in how things like increases in consumption compare against direct reductions in disease burden (and then how those compare against increasing or decreasing the speed of technological progress, and other major methods of impact). The analysis has some use, but I think it’s relatively minor for the cases I am interested in.
Lastly, I don’t think this research provides a post-hoc justification for women’s empowerment: in my view it could have as much provided a justification to not donate in that area (if the best charities turn out to be worse than in other areas) as a justification to donate in that area.
I think the current framing of the post and report does not allow for the possibility of a negative recommendation, and I expect the casual reader to walk away with a mistaken sense that this has been chosen as a promising cause area comparable to other top cause areas. De-facto, even though the numbers seem on a first glance a lot worse than other top GiveWell recommendations, the post does not give a negative recommendation. I recognize that the report was written for a different audience than the core EA community, but I think that’s what makes it lose most of its value to me.
Hi Habryka, just wanted to draw your attention to the update above, which is in part referring to some of your comments that have been incorporated in the new version of the report. Thanks for those!
I’d distinguish between two ways in which a report can ‘be’ cause-neutral:
1. Whether its domain of focus/​cause area was chosen purely through cause prioritisation
2. Whether its contents are of value from a cause-neutral perspective
Now I agree that this report is not cause-neutral on (1): it was written at least partially because many of FP’s community members are interested in women’s empowerment.*
However, note that cause prioritisation is just a heuristic to restrict our domain of search: what you want to compare in the end are the (donation) opportunities themselves, not which cause/​domain they happen to be in by some categorisation.
Maybe you don’t think women’s empowerment should be the first domain to check when you are looking for the highest-impact charities overall, but you should at least agree that it is valuable from a cause-neutral perspective to know what the best charities within this particular domain of search are. You might then be surprised that they are actually better than you thought, or you might find that your intuition of other areas having better opportunities is confirmed.
As the methodology of this report allows you to compare the charities to those in other areas (we don’t use outcome measures that are restricted to women’s empowerment/​the analysis is done in a cause-neutral frame), I think it to be cause-neutral on (2). And I hence think it’s very much worth discussing (from a cause-neutral perspective of course!) its contents on the EA forum, e.g. how do the recommended charities compare to other near-term welfare opportunities, such as those recommended by GiveWell?
Lastly, I don’t think this research provides a post-hoc justification for women’s empowerment: in my view it could have as much provided a justification to not donate in that area (if the best charities turn out to be worse than in other areas) as a justification to donate in that area. At FP we do research into areas not to justify our member’s initial preferences, but to be recommend high-impact opportunities tailored to those preferences (if high-impact opportunities are available), as well as to be able to make a solid, justified argument to focus on other areas (if higher-impact opportunities are available in those other areas).
*This does not mean that the choice of writing this report was a non-cause-neutral choice: for FP to do the most good we obviously need to take our community’s preferences into account. Neither does it mean that one couldn’t arrive at women’s empowerment as a high-potential cause area through cause prioritisation.
I agree that comparisons of that type are valuable, but I don’t think that this report helps me much in doing that kind of comparison. This report did no comparative analysis of the interventions against other near-term welfare interventions, and you used denominations that make that comparison quite difficult (as SiebeRozendal pointed out in another comment).
See for example this:
I don’t know how to compare an increase in consumption with other near-term interventions, so as long as this number isn’t shockingly high or low it’s quite hard for me to judge whether this is a good intervention. So while your analysis helps me a bit in comparing Village Enterprise to other near-term welfare charities, it really doesn’t help me much and I still need to put in the vast majority of work, which consists of building models about the world in how things like increases in consumption compare against direct reductions in disease burden (and then how those compare against increasing or decreasing the speed of technological progress, and other major methods of impact). The analysis has some use, but I think it’s relatively minor for the cases I am interested in.
I think the current framing of the post and report does not allow for the possibility of a negative recommendation, and I expect the casual reader to walk away with a mistaken sense that this has been chosen as a promising cause area comparable to other top cause areas. De-facto, even though the numbers seem on a first glance a lot worse than other top GiveWell recommendations, the post does not give a negative recommendation. I recognize that the report was written for a different audience than the core EA community, but I think that’s what makes it lose most of its value to me.
Hi Habryka, just wanted to draw your attention to the update above, which is in part referring to some of your comments that have been incorporated in the new version of the report. Thanks for those!