Whereas my own theory would make the social welfare function the arithmetic mean of individual utilities, Rawls . . . has proposed a social welfare function based on the maximin principle, and always measuring the welfare level of society by the utility level of the worst-off individual. This means mathematically that Rawls's social welfare function would always assign infinitely more weight to the interests of the poorest, or otherwise least fortunate, members of society than it assigns to the richer, or otherwise more fortunate, members. In contrast, my own social welfare function would always assign the same weight to equally urgent needs of different individuals, regardless of their social or economic positions. But of course since, typically, poor people have many more unfilled urgent wants than rich people do, in practice in most cases my own social welfare function will lead to similar policy decisions to Rawls's, because it will give much higher priority to poor people's needs. Only in those, rather special, cases where some rich people may have even more urgent needs than some poor people do, will my social welfare function lead to opposite policy decisions. However, precisely in these cases I feel that my social welfare function would lead to the morally right decisisons [sic] while Rawls's would lead to morally wrong ones.
For example, if a philanthropist has to decide whether to give $100 in cash, or the same value in food, or in clothing, etc., to a poor man or to a millionaire, it is clear that he should choose the poor man since the former will have a much greater need for extra cash, or for extra food, or for extra clothing, etc. But if he has to choose between giving a life-saving drug in short supply to a poor man or to a millionaire, then the only relevant consideration must be who needs it more, i.e., who could derive the greater medical benefit from it. Surely, it would be highly immoral discrimination against the millionaire to refuse him a life-saving drug, even though he has the best claim to it from a medical point of view, merely on the ground that he happens to be a millionaire. Yet, Rawls's social welfare function would force us precisely to engage in such immoral discriminatory practices in some, rather special, but perhaps not-all-too-rare, cases.
(John C. Harsanyi, "Nonlinear Social Welfare Functions," Theory and Decision 6 [August 1975]: 311-32, at 318-9 [endnote omitted; ellipsis added; italics in original])
