This is a book built round an argument. Several variants of the argument are offered, and I shall consider but one of them. It is directed against the following act utilitarian principle:

AU: An act is right if and only if it would have best consequences (i.e. consequences at least as good as those of any alternative act open to the agent)

The argument may be freely rendered as follows. Suppose that we have an agent, Smith, in a society, S, such that the following conditions are satisfied:

C1: Smith accepts AU and attempts always to act in accordance with it

C2: Smith is rational: he understands the implications of accepting AU and applies AU correctly

C3: No other person in S accepts AU

C4: All others in S know that C1 and C2 obtain

C5: All others in S are rational: they understand the implications of Smith's accepting AU

C: All others in S are tolerant of Smith's accepting AU: they do not criticize him for acting in accordance with it

C7: Smith knows that C3 – C6 obtain

C1 – C7 collectively define the position of an act utilitarian in a non-act utilitarian society, with certain assumptions as to rationality and tolerance. Then the thesis is that under C1 – C7 Smith will sometimes fail to do the act with best consequences. His very acceptance and correct application of AU (under the other conditions) conspire to prevent the production of certain good consequences which would otherwise be possible.