As discussed in this talk from Hilary Greaves, “simple cost-effectiveness analyses”:
Measure the immediate intended effects of the interventions, such as the number of deaths averted based on randomised controlled trials.
However, omit further future effects (potentially down the millennia), such as the long-term consequences on economic prosperity, population size, and politics.
Are there any “uber-analyses” of GiveWell/ACE top charities? These would correspond to “Response three: Make bolder estimates” of Hilary’s talk:
In principle, one could build a model that takes into account all distant future effects (e.g. effects on population size, and value of changes to the population size), although there would be questions where there is relatively little guidance from evidence, and where one could feel very much like guessing.
Addressing Global Poverty as a Strategy to Improve the Long-Term Future has been discussed, and improving animal welfare could also impact the long-term future via e.g. expanding the moral circle. However, I am not aware of estimates of the total impact of GiveWell and ACE top charities. I think these would be valuable:
Although we cannot take expected value estimates literally (even when they are unbiased), we may learn from the process of trying to estimate the total expected impact as accurately as possible.
“We [GiveWell] generally prefer to give where we have strong evidence that donations can do a lot of good rather than where we have weak evidence that donations can do far more good”, but there does not seem to be strong evidence about the total expected impact of GiveWell top charities.
Michael Dickens did something like that in 2016 here. The web app doesn’t seem to work anymore (you can see how it looked here) but you can still access the spreadsheet from the internet archive here.
Thanks, I will have a look!
See also Gordon Irlam’s Back of the Envelope Guide to Philanthropy. Maybe not exactly what you are asking, though it’s in the vicinity.
Thanks for the feedback! I am aware of that guide, but it seems to lack the incorporation of bayesian reasoning (it is highlighted here as an example of an explicit expect value approach).