As we have seen, Riker (1982, 122, 128) acknowledges that debate and discussion lead to similarities of judgment, that “a wide variety of rather mild agreement about the issue dimension guarantees a Condorcet winner,” that uncontrived cyclical outcomes are “quite rare,” so that “intransitivities only occasionally render decision by majoritarian methods meaningless.” His remaining objection then is that outcomes may be manipulated by strategic voting, agenda control, and contrived introduction of new alternatives and dimensions. “Manipulated outcomes are meaningless because they are manipulated, and unmanipulated outcomes are meaningless because they cannot be distinguished from manipulated ones” (Riker 1982, 237). To introduce his concern, consider the following exercise.
Suppose that a majority of 60 percent favors A over B, and a minority of 40 percent favors B over A. The minority faction could contrive a cycle if it has some idea of the distribution of preferences over original alternatives A, B, and new issue C. Issue C splits the majority. See Table 7.1. With C on the scene, A beats B by 60 percent, B beats C by 70 percent, and C beats A by 70 percent. Then, the minority can propose C against A, which C wins, then C against B, which B wins, thus, the minority view favoring B over A prevails (but only if the majority is somehow incapacitated from continuing the cycle by proposing A against B which of course A would win, ad infinitum).