I suggest that modern political races in this country are generally modeled on the prisoner's dilemma. Races that are well-funded, with expert political strategists on both sides always go to the Nash equilibrium. Lesser races can yield any of the other results. For convenience, I'll state the dilemma approximately in its classic form, and then indicate how you'd restate the variables in my idea.
Two suspects are accused of a crime and held separately. Each can choose to remain silent, or testify against the other suspect (defect) for a reduced sentence. If one defects and the other remains silent, the defector goes free and the other gets ten years in prison. If they both remain silent, they'll both get convicted on a lesser charge and each serve six months. If they both defect, they each get five years in prison.
The problem assumes that there are no other costs or benefits from the prisoners' decisions, and that neither has an interest in the welfare of the other. I put the important variables in bold. How would I apply this game to politics? Obviously, in American politics, most contests for office are contested by two major candidates: they are the suspects. How do I define remaining silent, defecting, and the different prison sentences? About like this. In general, voters want politicians to promise to balance the budget, lower taxes, and provide more services. It is utterly impossible to do all three of these things simultaneously, and the candidates know this, so they have a choice: They can be honest and state their ideas about which goals they really want to achieve (remain silent), or defect: lie and state that they will do all three. This is the nature of the dilemma. If both candidates are truthful, this is tantamount to both getting six months; neither candidate compromises his ideals, and the election is fair, and based on which politician's ideas are more in line with voter priorities. If one candidate defects, i.e. lies and promises the impossible, and the other does not, the liar/defector wins by a landslide: this is ten years versus going free. If both candidates lie, it is easy to see how this is like both getting five years: they have both compromised their ideals, and the election will be close. The latter is a Nash equilibrium: no matter what your opponent does, you get more utility from defecting, so both players defect.
The only minor inconsistency I find here, in the values of my variables relative to each other, is that the going free substitution requires compromising one's ideals, which is worse (lower utility) than the six months substition of having a fair election dominated by the issues. I think it still works, however; we assume that ensuring election has much greater utility than that lost by failing to preserve ideals (for a politician), i.e. the ends justifies the means, although you'd prefer more ethical means if you didn't have an opponent. (Which makes me wonder: in races where one party fails to field a candidate, and the only candidate actually does some campaigning, does he/she bother to "defect"? There might be reasonable motives for doing so: a vulnerable, truthful platform might inspire the other party to unexpectedly enter the race, or the candidate may have no ideals, and simply be interested in popularity, perhaps as political capital for future elections, regardless of whether his/her promises make any sense. If my idea is correct, then one-candidate elections may give us valuable insight into the motivations of individual politicians: power or ideals?)
Some may object that the prisoner's dilemma game has only one turn and both players choose their move simultaneously, whereas elections play out over several months. I'd justify ignoring that aspect with the idea that damage done to one's campaign by any accusation of "flip-flopping" is sufficient deterrent that races actually are played out as essentially simultaneous one-turn games.