Political Attack Ads and the Prisoner’s Dilemma
Anyone who follows politics can tell you that political campaigns tend to be rather nasty. Candidates run ads attacking their opponents’ policies, pasts, and personal lives, all in the pursuit of winning an election. Many members of the public (and, indeed, the politicians themselves) would likely prefer for campaigns to reintroduce civility and tone down their rhetoric. Why, then, are the campaigns still so mean? And can anything be done to make them less so?
One plausible answer to the first question uses game theory, specifically the prisoner’s dilemma. When two candidates, A and B, are running against each other, each can choose if he wants to utilize attack ads. Assuming that neither A nor B has any moral qualms with running attacks against his opponent and that, without attack ads, the two candidates have equal chances of winning the election, the payoffs for both of the candidates are as follows. A’s payoff is highest when he uses attack ads but B does not; in this situation, he can inflict major damage on B’s campaign while keeping his own on track. In this case A would have quite a good shot at winning the election. The second-best case for A would occur when neither candidate uses attack ads, as his own campaign would be unaffected and he would retain a 50/50 chance of winning. Next-best for A would be when both candidates run attacks against each other. Because both campaigns would be hurt (hurt equal amounts, we assume), A would still have an even shot at winning the election. The attacks would, however, probably damage his public image and adversely impact his chances in future races. A’s worst-case scenario occurs when he opts not to use attack ads but B does; here A’s campaign would suffer and his odds of winning would decrease. B’s payoffs in each case would be analogous to A’s because of symmetry.
Now that the candidates’ choices have been described, how does the model predict they will behave? If we start at the outcome where neither candidate runs attack ads, neither A nor B would see his campaign harmed. This outcome should be fairly favorable to both candidates. Either of the candidates could, however, improve his position by choosing to run attacks against his opponent. The opponent would then be able to improve his own position by also airing attack ads, allowing him to even up the race. When both politicians are running attack ads, A or B would have nothing to gain and a lot to lose if he were to stop running them. This demonstrates that, even though both politicians would prefer attacks airing from neither side to airing from both, attacks from both candidates is the only stable possibility because it is a Nash equilibrium.
If there is a near-consensus that civil campaigns are preferable to mean ones, is there any way to modify the system so that attack-free campaigns are plausible? In a blog post, Jonathan Sallet describes how recent changes to California’s election system could do just that. Beginning with the 2012 election cycle, candidates running for Congress in California will run in an “open primary.” What this means is that all candidates from all parties will compete in a single primary, and the top two vote-getters will run in the general election. This system, as Sallet explains, could reduce campaign viciousness. The presence of independents in the primary voter pool would likely have a moderating effect on the discourse. Politicians typically cater to their party’s base in primaries, and this base would be more likely than independent voters to positively respond to attack ads. In order to avoid alienating more moderate independent voters, candidates would have to be less malicious in their criticisms of opponents. We will have to wait several more months before we can see whether this actually happens, and the recent trend of increased polarization in American politics may be tough to overcome. Perhaps, however, it truly is possible to avoid suboptimal outcomes in the prisoner’s dilemma, both in this particular situation and in others, by changing the conditions under which the game is played.