Peter Singer's "Famine, Affluence, and Morality" [Philosophy & Public Affairs, Vol. 1, No. 3 (Spring, 1972), pp. 229-243] has been subjected to a number of common criticisms, for instance that it is too demanding (which has to be formulated very carefully in order not to be question-begging, although Singer does jump too quickly from reasonable to obligatory) or that there are problems with Singer's attempt to dismiss distance as relevant (which is right, although 'distance' here has to be at least partly concerned with difficulty and uncertainty of action rather than just geographical distance). But there is a problem with his argument that I have not seen anybody point out, one which I think is fairly significant: there is a significant gap between the principle to which Singer appeals and the supposed result of the application.
Singer's principle is (in the weaker form he gives), "if it is in our power to prevent something very bad from happening, without thereby sacrificing anything else morally significant, we ought, morally, to do it." Now, as I noted before, this moves too quickly to obligation -- that we can prevent something very bad from happening without sacrificing anything else morally significant is obviously in and of itself a reason to do it, but it is much less obvious that it is in and of itself a reason to treat it as morally obligatory -- and, indeed, this will depend on what theory of obligation you espouse. Most people would regard a principle like Singer's as true for certain kinds of domain, namely, domains for which you have direct responsibility, but would be more skeptical for other domains (domains for which you have only indirect responsibility and domains for which you are not generally regarded as responsible at all); if that is right, we need to add some principle(s) of responsibility to get it to work, but Singer wants and needs his principle both to be absolutely general and to yield obligation immediately. This is a tall order, but let's set this aside for the moment. A question that should be asked here is about the prevention.
Preventing evils requires making it so that they do not happen. Singer seems to hold that a lot of things count as prevention. Some of his examples are cases in which anyone could agree that you are preventing an evil -- for instance, saving a child from drowning in a pond. But for other examples (and, indeed, the cases to which Singer wants actually to apply his principle), the matter is not so clear, and sometimes it is not clear that there is any sort of power-to-prevent at all. Take the most obvious case: people starving to death in East Bengal is very bad. The action for which Singer argues is donating to the Bengal Relief Fund. But donating to the Bengal Fund is not, in fact, preventing people from starving in East Bengal. To be sure, some of the actions supported by money from the Bengal Relief Fund do (one hopes), but if I support the Bengal Relief Fund, I am not doing any of those things at all; I'm just giving the Bengal Relief Fund a bit of money so that (I am trusting) they will give it to people who (I am trusting) will do those things. Singer's principle is irrelevant to the case: the reason I'd be giving money to the Bengal Relief Fund (one would usually think) is that it is not, in fact, in my power to prevent the very bad thing at all; so I am transferring some of my power (buying power, in this case) to other people in the hope that, when it pools with power provided by other people, someone else will then have the power to prevent it. When you donate to prevent an evil, you are not, in general, preventing the evil; while there are exceptions, when you donate you are usually doing so precisely because you yourself don't have the power to prevent the evil; and the most common point of donation (besides, perhaps, showy appearance, which can be set aside here as irrelevant) is that we, lacking individually the power to prevent the evil, are trying collectively to create the power to prevent the evil.
Singer's argument proceeds as if the power to prevent bad things were an easy thing to find in the world and easy to use when you had it. But neither of these is true. Even for things within our immediate and ordinary sphere of action, for instance, many bad things are hard to prevent. A good example is keeping young children out of trouble, since a lot of things that for older children and adults would not be a problem can turn very bad for very young children; parents and caretakers have to put a lot of time and investment to preventing bad things from happening. While it is very difficult for us to accept it, a lot of bad things that happen to young children were not preventable given what was actually known and feasible at the time. And the more people we end up having to watch over, and the more evils we end up having to prevent, the less and less we ourselves could actually prevent. To be sure, there might always hypothetically have been someone who could have known the right things, or someone who could have been in the right place at the right time to prevent it, but hypothetically is sometimes merely hypothetically. If we are talking about ourselves, we are remarkably limited in the range of evils we have the actual power to prevent.
And even when we have the power, we might only just barely have the power, or might have it only maybe. Even if we might technically in fact have the power to prevent an evil, it might be close enough to the boundaries of what we know that we can do that it might be more reasonable to say that we have the power to do something that would be worth trying, which might prevent it.
Most of the time when we deal with bad things, we do not have the power to prevent them. Indeed, an immense amount of our time, money, and effort goes into just trying to create the power to prevent bad things, because we don't have it in the first place. Donating to a charity is not usually preventing bad things; it is trying to work with other people to create a power to prevent bad things. Volunteering time and effort is often the same. Most bad things are not like the child in the pond; the power to prevent them is not already in place to use. It has to be made, and that itself is not always easy.
Because of this, Singer's principle has a far narrower scope than he assumes (and, for that matter, than most of those reading him have tended to assume).