Dan Davies at Crooked Timber points out an inconsistency in a common argument for voting for a major party:

The key point I want to make here is that when major party activists put the guilt-trip on supporters significantly to their left, they engage in what looks like very fallacious reasoning. The point is that a voter considering a protest vote against the Dems from the left has three options on election day:

First, stay at home
Second, vote for their minor party or abstain
Third, vote Democrat

And the thing is that the major party activist has to steer them between the Scylla and Charybdis of the first two choices, both of which might superficially look more attractive than voting for a candidate you don’t support. To do so, they need to make two contradictory arguments.

Obviously the problem to overcome in getting you to drag your ass (note American spelling) down to the polling station is the Paradox of Voting. Which isn’t really a paradox; it could more accurately be titled “The Actual Extremely Low Expected Value Of Voting”. This requires an appeal to your civic sense of duty; remember Martin Luther King, etc. In other words, they need you to see it as your duty to society to vote, or alternatively to see your vote as an important form of political expression.

However, once your ass is duly dragged and you’re in the voting booth, the last thing they want you to do is your civic duty (which would be to vote for the candidate you think is the best; that’s how voting systems work, strategic or tactical behaviour is a pathology of a badly designed system) or political expression (which also wouldn’t have you voting for their guy). Once you’re there, they want to argue in purely instrumental terms – you have to vote for the Democrats because if you vote for your minority party, you have no chance at all of being the marginal voter.

I considered the same quandary way back in 2004, and maybe it’s worth revisiting now. This is a problem that generalizes far beyond voting to a whole class of collective action problems or “tragedies of the commons” that have something like the following structure: There’s a system in which each individual’s rationally self-interested choice is what we’ll call the “defect” option—don’t bother to vote, overgraze the pasture, pollute, whatever. Moreover—and this is a difference from the traditional prisoner’s dilemma—holding everyone else’s choices constant, no individual’s choice to defect makes a difference to the collective outcome. My voting won’t change the outcome of the election; my constraining my livestock’s consumption won’t prevent the pasture from being overgrazed; my factory isn’t going to measurably effect the scope of climate change.

One traditional way out of this—the Kantian option—is to say that everyone has a moral duty to act on the maxim that they could will universally. In other words: Figure out what course of action would be best if everyone acted that way, and do that. This is both intuitively appealing, and for some cases seems to work pretty well, at least in theory. You can figure out the sustainable level of grazing or emissions and some procedure—whether simple or complicated—for determining everyone’s fair share of the total, and say that’s the limit everyone ought to observe. Enforcement, of course, is another matter, and determining both the correct collective limit and the right mechanism for apportioning shares may be technically quite tricky, but the general shape of the solution is relatively straightforward.

With a case like voting, though, it’s another matter, because it’s a premise of the system that people are supposed to register their presumptively diverse opinions about which candidate or policy would be best. If you actually acted on the Kantian maxim in the voting booth—selected the candidate you genuinely believe to be best—you’d probably end up with a lot of people writing in a huge variety of names. You can, of course, argue that there’s some one best candidate, and selecting that person is the uniquely moral choice. But the premise of a democratic system is precisely that we don’t have a ready collective answer to that question—voting is the process by which we try to approximate an answer by aggregating people’s different opinions. And even if you think you know the ideal candidate, given this fact of disagreement, voting for thatperson can seem like “defecting” in another sense: Given that, realistically, the winner is going to be one of two (or occasionally three) contenders, the “cooperative” outcome is going to be the one where people with roughly similar views converge on a compromise candidate rather than one’s personal first choice. We can think, by analogy, of driving: It might be that for some obscure reason having to do with human physiology, a system in which everyone drives on the left side is actually slightly safer—but that doesn’t make it the moral thing to do if the local norm is to drive on the right!

Yet here’s the paradox Davies is pointing out. Once you start modifying your rule to take into account what everyone else is likely to do—asking what rule you can will given that, in fact, many or most people will not act on the rule you’d ideally universalize—it’s not clear where you stop. In other words, once you allow that your own action is contingent on the predictably non-ideal actions of everyone else, why aren’t you back where you started, acknowledging that your own vote almost certainly won’t affect the outcome, and so your optimal response is to stay home?

One way out of the paradox, of course, is to have a voting system in which everyone ranks as many candidates as they please, allowing everyone to genuinely act in the way they would universalize without generating a perverse result. (For the most part, anyway: As Arrow proved, no voting system can entirely eliminate perverse results.) But that’s not much help for someone trying to decide what to do here and now.

The solution I gestured toward in the 2004 post was a kind of quasi-universalization principle, where you don’t ask what rule you’d want everyone to act on, but rather what rule you’d universalize within a symbolic community of like-minded folks. Alas, the same problem recurs at this stage. Which is the relevant community? If you’re a progressive, and a relatively conservative Democrat is likely to win anyway, should your imagined caucus be left-of-center folks, or people more closely aligned who might register a protest vote without changing the outcome?

In a system of perfect transparency and frictionless negotiation, people could just actually negotiate this—arguably that’s what the point of the primary system is, and sites like Nader’s Traders back in the 2000 election took the idea a step further, though ran into legal difficulties. Still, in principle at least, it looks like a coordination problem that’s intractable as a matter of abstract moral logic may become manageable—thanks to the Internet—if we leverage the more general moral obligation to abide by commitments with actual negotiation. If that’s right, then maybe Davies gets it backwards when he says that the choice between abstaining, voting Democratic, or registering a third-party protest vote, is “up to the conscience of the individual voter to make.” Maybe, instead, the way to go about it is to begin with large communities of broadly shared ideals and abide by the rule governing the symbolic sub-communities that emerge from pre-electoral deliberation. In theory, even at the deliberative stage, you can run into rapidly alternating equilibria, or suboptimal but sticky ones, but since people aren’t all making simultaneous choices about which sub-community to align with, people can probably work it out in practice.

That’s be the best I can come up with, anyway; better thoughts welcome in the comments.