I think the claim is that while this intervention wasn’t as cost effective as we thought, it’s still more cost effective than most interventions. The belief regarding the cost effectiveness (about a third as good as we thought) is itself supported by evidence, so if you think the best options are 10× as good as the average ones, you would still expect this one to be “above average”. Not sure about 90th percentile but I do think there’s a fair amount of NGO work whose effectiveness is statist indistinguishable from zero, and a minority that’s arguable actively harmful.
I think the claim is that while this intervention wasn’t as cost effective as we thought, it’s still more cost effective than most interventions. The belief regarding the cost effectiveness (about a third as good as we thought) is itself supported by evidence, so if you think the best options are 10× as good as the average ones, you would still expect this one to be “above average”. Not sure about 90th percentile but I do think there’s a fair amount of NGO work whose effectiveness is statist indistinguishable from zero, and a minority that’s arguable actively harmful.