multifoliaterose:

If you have a good argument that "it's illusory to believe that it can be avoided by some simple precautions such as those advocated by GiveWell" then I'm interested in hearing it. But at the moment your implicit criticism of the efficacy of donating to GiveWell's top recommended international aid charities appears to be totally ungrounded.

Maybe I should state my claim more clearly. What I mean is that while criteria such as those used by GiveWell can eliminate certain modes of failure in aid projects, they are by no means sufficient to eliminate the possibility of numerous other non-obvious failure modes -- and given the actual historical record of various humanitarian aid programs, it seems pretty obvious to me that such failures are a rule rather than exceptions. I would say that this constitutes enough evidence to shift the burden of proof on those who argue in favor of supporting such projects.

Note that even if it's proven beyond reasonable doubt that a certain program has saved so many lives, it is by no means a guarantee that its overall long-term consequences are positive by any standard. For example, reducing mortality among people who are stuck in a Malthusian equilibrium in a way that doesn't force them out of this equilibrium will only increase the amount of suffering in the medium-to-long run, which can be alleviated only by ever increasing amounts of aid, creating a diabolical positive feedback process that results in an ever greater dependence -- possibly making it even more difficult for them to escape the Malthusian condition. Or, to take another example, aiding the subjects of a disastrously bad government will increase its stability and grip on power in a way that may easily allow it to make an even worse subsequent mess. Such scenarious, and various other equally depressing ones, have happened on innumerably many occasions in modern history, and keep on happening.

On the whole, when someone claims that some project will improve the lot of distant strangers that are outside of your regular sphere of attention and comprehension, you need an awful lot of evidence to be reasonably sure that these claims are true -- certainly way more evidence than even GiveWell is capable of providing, even if they really are the best source of information on these matters.

There may be causes that are much more cost-effective from the point of view of maximizing positive social impact than giving to improve international health. People who believe that developing world aid is not cost-effective should consider donating a sizable fraction of their income to an organization that supports a cause that they prefer, or placing a sizable fraction of their income in a donor advised fund for future charitable use.

I have no particular argument with this, as long as such attitudes are not based on the absurd Singerian idea that distant anonymous strangers have an equal claim on my altruism as people stuck in some accidental trouble right in front of me. An essential part of human nature is that people care more about family and friends than about strangers, and more about strangers that are closer by various criteria than about more distant ones -- and there is absolutely nothing wrong with this by any reasonable standard, regardless of how much Singer and his ilk smugly chastise normal people for not falling in line with their ideology.