The Drowning Child

In 1972, while the Bengal famine was killing people at a rate of thousands a week, Peter Singer published a nine-page paper containing an argument with three premises. It has been reprinted more than almost any other article in modern moral philosophy, not because it is subtle but because it is very hard to get out of.

Premise one: if it is in our power to prevent something very bad from happening, without sacrificing anything of comparable moral importance, we ought to do it. To make this vivid he offers the case everybody remembers. You are walking past a shallow pond and a small child is drowning in it. You could wade in and pull her out. You would ruin your shoes and be late. Nobody thinks this is a difficult question, and nobody thinks pulling her out would be generous; it would be obligatory, and walking past would be monstrous.

Premise two: physical distance makes no moral difference in itself. A child dying five thousand miles away has the same interest in not dying as a child five metres away. Proximity affects what you notice and what you can reliably do; it does not change the weight of what is at stake.

Premise three: the fact that others could also help does not reduce your obligation. If there were a hundred bystanders at the pond and all of them walked past, that would not make your walking past acceptable.

The conclusion follows: people with disposable income are obliged (not encouraged, obliged) to give a great deal of it to prevent deaths they could prevent. And since the principle does not have a natural stopping point, the strict version says you should keep giving until further giving would cost you something of comparable moral importance to what it prevents. That is not a tithe. That is most of it.

Almost nobody lives this way, including most people who cannot find the flaw in the argument, and the interesting question is what that fact shows.

The objection with the most force is the demandingness objection: a morality that leaves no room for your own projects, relationships and ordinary enjoyment cannot be right, because morality is for creatures who have lives, and a standard nobody can meet stops functioning as a standard at all. Singer's response has been twofold. He notes that the objection is about what we can bear rather than about whether the argument is sound, which is a real distinction even if it is not much comfort. And he has offered a weaker practical version (give what you can sustain, and make it a fixed share) on the grounds that a demand people actually meet saves more lives than one they abandon.

The objection that does most to complicate the argument is the disanalogy between the pond and the wire transfer. The pond is certain, immediate, and yours alone: you can see the child, you know your action works, and nobody else is going to do it. International aid is uncertain in its effects, can be delivered badly, can distort local markets, and involves millions of other potential givers and institutions. None of this dissolves the obligation, but it does complicate the inference from 'save the child' to 'this particular donation'.

That objection is also, as it happens, what produced effective altruism. The response was not to abandon Singer's argument but to accept it and treat the uncertainty as an empirical problem: if interventions differ by orders of magnitude in how much harm they prevent per pound (and they do) then choosing a cause by personal resonance rather than by evidence has a cost measured in lives. Whether the movement that followed has been good on balance is a separate and increasingly contested question. The argument underneath it has not been refuted.

The most uncomfortable part of Singer is not the conclusion but a smaller claim on the way to it: that the category of charity (admirable if you do it, perfectly fine if you do not) is not a moral category at all. It is a residue of distance and invisibility. Walking past the collection tin is a choice with consequences, and the fact that it does not feel like one is precisely what the argument is trying to correct.