Benjamin Hoffman recently wrote a post arguing that "drowning children are rare":
Stories such as Peter Singer's "drowning child" hypothetical frequently imply that there is a major funding gap for health interventions in poor countries, such that there is a moral imperative for people in rich-countries to give a large portion of their income to charity. There are simply not enough excess deaths for these claims to be plausible.
...
As far as I can see, this pretty much destroys the generic utilitarian imperative to live like a monk and give all your excess money to the global poor or something even more urgent. Insofar as there's a way to fix these problems as a low-info donor, there's already enough money. Claims to the contrary are either obvious nonsense, or marketing copy by the same people who brought you the obvious nonsense. Spend money on taking care of yourself and your friends and the people around you and your community and trying specific concrete things that might have specific concrete benefits.
Imagine that the best intervention out there was direct cash transfers to globally poor people. The amount of money that could be productively used here is very large: it would cost at least $1T to give $1k to each of the 1B poorest people in the world. This is very far from foundations already having more than enough money. That there are extremely poor people who can do far more with my money than I can is enough for me to give.
While I also think there are other ways to spend money altruistically that have more benefit per dollar than cash transfers, this only strengthens the argument for helping.
How does Ben reach the opposite conclusion? Reading his post several times it looks to me like two things:
- He's looking at "saving lives via preventing communicable, maternal, neonatal, and nutritional diseases" as the only goal. While it's a category of intervention that people in the effective altruism movement have talked about a lot, it's definitely not the only way to help people. If you were to completely eliminate deaths in this category it would be amazing and hugely beneficial, but there would still be people dying from other diseases, suffering in many non-fatal ways, and generally having poverty limit their options and potential. And that's without considering more speculative options like trying to keep us from killing ourselves off or generally trying to make the long-term future go as well as possible.
- He's setting a threshold of $5k for how much we'd be willing to pay to avert a death, which is much too low. I do agree there is some threshold at which you'd be very reasonable to stop trying to help others and just do what makes you happy. Where this threshold is depends on many things, especially how well-off you are, but I would expect it to be more in the $100k range than the $5k range for rich-country effective altruists. By comparison, the US Government uses ~$9M.
I do think the "drowning children" framing isn't great, primarily because it puts you in a frame of mind where you expect that things will be much cheaper than they actually are (familiar), but also because it depends on being in a situation where only you can help and where you must act immediately. There's enough actual harm in the world that we don't need thought experiments to show why we should help. So while there aren't that many "drowning children", there is definitely a lot of work to do.
(Crossposted from jefftk.com)
I think the post is more fundamentally flawed; there is a substantial funding gap under Benjamin's assumptions, even if we were to ignore GiveDirectly and other cause areas, and even if we were unwilling to save a life for any more than $5,000.
The Open Philanthropy Project started out with $8.3 billion in 2011, and presumably has less now. The Gates Foundation has an endowment of $50.7 billion as of 2017. They wouldn't be able to sustain $50 billion of annual donations for very long. As such, I think the first and second paragraphs are essentially invalid.
It sounds dubious that we could wipe out communicable diseases in a few years and have that be permanent without any further investment. The 2017 Global Burden of Disease lists some communicable diseases as follows: HIV/AIDS, syphilis, chlamydia, gonococcal infection, tuberculosis, other respiratory infections, diarrheal disease, typhoid, salmonella, malaria, schistosomiasis, dengue, rabies, other neglected tropical diseases, ebola, zika, meningitis, measles, hepatitis, tetanus, and so on.
My understanding is that rather few of these have been permanently eliminated, even in high income countries. Distributing condoms and PrEP for a few years isn't going to permanently eliminate HIV. Bed nets and seasonal chemoprevention aren't going to eliminate malaria. Measles needs ongoing vaccinations. Etc.
There are of course more permanent solutions that we can use, but these are probably much more expensive and it's unclear whether the two foundations would be able to fully fund them. In the late 1940s, the US substantially reduced malaria by draining swamps and spraying mosquito spray.¹ There's gene drives of course, but we probably need more research at this point before we can safely try to eliminate mosquitoes with that. Ending worms, diarrheal disease, or typhoid would probably require incredible improvements to the water supply. Still, HIV and respiratory infections would probably not be possible to eliminate without substantial improvements in medicine.
Also, the Gates Foundation is not particularly EA, and we should not expect it to put all its money into global health. (Nor would we assume Open Phil to do so, because it also cares about other cause areas.) In any case, even if they could fill the gap, that's not a relevant counterfactual unless they would fill the gap.
All of the above is using Benjamin's charitable, optimistic assumption that we can save a life for $5,000 up to $50 billion per year. If we consider just the room for more funding of all the top GiveWell charities better than GiveDirectly, is that low enough that Open Phil and the Gates Foundation can completely fill it? Possibly, in which case I will defer to the argument Jeff Kaufman's post.
While I agree with a lot of the critiques in this comment, I do think it isn't really engaging with the core point of Ben's post, which I do think is actually an interesting one.
The question that Ben is trying to answer is "how large is the funding gap for interventions that can save lives for around $5000?". And for that, the question is not "how much money would it take to eliminate all communicable diseases?", but instead is the question "how much money do we have to spend until the price of saving a life via preventing communicable diseases becomes significantly higher than $5k?". The answer to the second question is upper-bounded by the first question, which is why Ben is trying to answer that one, but that only serves to estimate the $5k/life funding gap.
And I think he does have a reasonable point there, in that I think the funding gap on interventions at that level of cost-effectiveness does seem to me to be much lower than the available funding in the space, making the impact of a counterfactual donation likely a lot lower than that (though the game theory here is complicated and counterfactuals are a bit hard to evaluate, making this a non-obvious point).
I think, though I have very high uncertainty bounds around all of this, is that the true number is closer to something in the space of $20k-$30k in terms of donations that would have a counterfactual impact of saving a life. I don't think this really invalidates a lot of the core EA principles as Ben seems to think it implies, but it does make me unhappy with some of the marketing around EA health interventions.
The 8.3 billion should have grown since 2011. Openphil's grants have not even totalled 800 million yet and that is the amount that the fund should have grown *per year* in the interim.
This piece does a good job at making this point: https://www.givingwhatwecan.org/post/2015/06/cost-fighting-malaria-malnutrition-neglected-tropical-diseases-and-hivaids-and/