Consider a standard prisoners’ dilemma between two groups, A and B, with members of these two groups, or their leaders (who can be either elected or self-appointed) considering whether or not to lobby for political advantages. Given the groups and the two choices for each, there are four possible outcomes—neither group lobbies, both groups lobby, group A doesn’t lobby and group B does, and group A lobbies and group B doesn’t. The payoffs for the four possibilities are shown in Figure 1, with the first number in each cell representing the payoff to group A, and the second number in each cell representing the payoff to group B.
Collectively the best outcome is for neither group to seek political advantages, concentrating on creating new wealth rather than fighting over existing wealth. In this case both groups make the cooperative choice and receive a payoff of 100, illustrated in cell 1 of
Figure 1. But if one group chooses to lobby for advantages (behave noncooperatively) and the other does not, then the former realizes a payoff of 110 while the latter receives the suckers’ payoff of 75, as shown in cell 2 of Figure 1. So, no matter what the other group is expected to do, the best action for each is to behave no cooperatively by lobbying. Unfortunately, when both lobby, the worst collective outcome is realized, with both groups receiving a payoff of 90, as shown in cell 4.
If the prisoners’ dilemma in Figure 1 were a complete representation of the relevant payoffs, then the dominant strategy for each group would be to lobby. But notice that we have described the prisoners’ dilemma in Figure 1 as “initial” to indicate that it does not give a complete picture of the possible payoffs beyond time period one. In situations in which there is a sequence of prisoners’ dilemmas over time, the choices made in one play of a prisoners’ dilemma can affect the payoffs from future plays, and this