In view of the difficulties associated with act utilitarianism, the economist [R. F.] Harrod proposed a revised form of utilitarianism, now often called rule utilitarianism. (The terms 'act utilitarianism' and 'rule utilitarianism' themselves were introduced by [Richard B.] Brandt.) Harrod suggested that the utilitarian criterion should be applied in the first instance, not to each individual act, but rather to the moral rule governing this act (and to the general social practice that would result from common conformity to the relevant rule).
Thus, an individual act should be considered to be morally right if it conforms to the correct moral rule applying to this type of situation—regardless of whether it is the act that will or will not yield the highest possible social utility on this particular occasion. On the other hand, the correct moral rule should be defined as that particular rule of behavior that would yield the highest possible social utility in the long run if it were followed by everybody in this type of situation.
For example, in most cases it will be our moral duty to keep our promise even if breaking it would yield a higher social utility than keeping it would on that particular occasion—because in the long run society will be better off if people can confidently expect that promises will be kept in all but some rather special cases. (But even the question of what exceptions should be made to the general rule of promise keeping must be decided by asking what particular set of exceptions would maximize social utility in the long run—on the assumption that everybody would know what these permitted exceptions were.)
Likewise, except for certain emergency situations, it is our moral duty not to steal, even if stealing would increase the social utility produced by us on that particular occasion—because in the long run society will be better off if people can confidently expect that private property will be respected except in certain emergency situations. Again, it is our moral duty not to take the lives of innocent people, even in order to achieve some very noble objectives—because in the long run society will be better off if innocent people need not fear being murdered for the purpose of achieving some very noble objectives. (It is bad enough that many societies cannot ensure reasonable safety from murder by criminals; but at least they should ensure safety from murder by people guided by the highest motives, and acting in the very name of morality.)
The emergence of rule utilitarianism has given a new turn to the debate about the value of utilitarianism as a moral theory. If rule utilitarianism can really avoid the undesirable implications of act utilitarianism, then it may be possible after all to retain the intellectual and the practical advantages that utilitarianism seems to provide, without being forced to give up some of our deepest moral convictions and embrace a thoroughly distasteful super-Macchiavellistic morality. But, of course, rule utilitarianism cannot possibly avoid these undesirable implications unless it has a genuinely different logical content from act utilitarianism.
(John C. Harsanyi, "Rule Utilitarianism and Decision Theory," Erkenntnis 11 [May 1977]: 25-53, at 32-3 [italics in original; parenthetical references omitted])
