At least on the face of it, utilitarianism is a very attractive ethical theory intellectually, and also from a practical point of view. It proposes to explain our multifarious and seemingly logically rather disconnected moral value judgments in terms of one fairly simple and readily understandable moral postulate, that of maximizing social utility. In principle at least, it also offers a conceptually clear practical criterion for resolving our real-life moral perplexities: it tells us what to do when two or more, prima facie equally compelling, moral rules seem to impose mutually conflicting moral obligations upon us. It even tells us how to choose between two alternative comprehensive moral codes, say, between the traditional mores of our society or of our narrower social group, and any alternative moral code we may wish to consider for possible adoption.
Yet, as many philosophers have argued, utilitarianism, at least under one of its possible interpretations, would achieve these apparent intellectual and practical advantages at intolerably high moral costs. Traditionally, utilitarianism has been usually interpreted as act utilitarianism, which holds that the utilitarian criterion should always be applied directly to individual acts. Thus, always that particular action will be morally right which here and now seems to yield the highest social utility. (Actually, it is somewhat controversial whether John Stuart Mill and some other classics of utilitarianism were in fact act utilitarians, or rule utilitarians.) But act utilitarianism seems to have many practical implications inconsistent with some of our most firmly held moral convictions, such as the belief that considerations of social expediency cannot simply override the demands of elementary justice, and cannot justify infringements upon fundamental rights of other people.
For example, according to our common-sense morality, we should keep our promises, except if this would impose disproportionally great hardships on us or on some other people. But act utilitarianism seems to imply that we should always break our promises whenever this will have better, even if only slightly better, direct social consequences than keeping our promises would.
Likewise, according to common-sense morality, in other than emergency situations, we should refrain from taking other people's property without their consent (or without a lawful transfer of ownership in some other way). But act utilitarianism seems to imply that, whenever we feel we have a somewhat stronger need for a given object than the rightful owner has, then we are morally permitted to take possession of it without any further ceremony.
Indeed, act utilitarianism seems to have much more sinister implications. According to commonsense morality, we should not take the life of another person, except in self-defense, or in a just war, or in carrying out a lawful death sentence. But act utilitarianism seems to imply that it is in many cases morally permissible—and, in fact, it is our moral duty—to kill one innocent man if this is the only way of preventing the murder of two or more innocent individuals. More generally, act utilitarianism has to deny the principle, very basic to our common-sense morality, that our moral duty to refrain from highly unjust actions (such as a killing of innocent people) normally takes absolute precedence over our moral duty to prevent other people from committing such unjust actions.
To be sure, act utilitarians have argued that act utilitarianism does not really have these undesirable implications, if proper attention is paid to the socially detrimental effects that such actions as promise breaking, taking other people's property without their consent and, in particular, a killing of an innocent person, will have on people's mutual trust and on their propensity to obey customary moral rules. But this argument is rather unconvincing. One act of promise breaking, or stealing, or even of murder, will seldom have a significant effect on people's mutual trust in general (though it may very well have a significant effect on people's trust in the person who has actually engaged in promise breaking, in stealing, or in murder). Nor is it likely to have any sizeable effect on the general propensity to obey customary moral rules. Indeed, if these acts are committed in complete secrecy, then all negative effects of this kind will be totally absent; but this does not make secret acts of this kind morally permissible.
(John C. Harsanyi, "Rule Utilitarianism and Decision Theory," Erkenntnis 11 [May 1977]: 25-53, at 30-2 [italics in original; parenthetical reference omitted])
