Democracy, Academic Philosophy

How Could There Even Be an Ethics of Voting?


(Following up on my previous post.) Thanks to everyone who pre-ordered the book! I'm looking forward to having some vigorous conversations and hearty debates with you.

In this post, I just want to introduce the point that it's puzzling, at first glance, whether there could even be an "ethics of voting".

Political decisions tend to have the following features:

  1. They are high stakes. They often are literally life and death decisions. They can greatly affect people’s life prospects and deprive them of property, liberty, and life. Bad decisions can make people lives go much worse.
  2. They are imposed by force upon (generally) innocent people who (generally) do not consent to be governed. (I won't get into it much here, but it's worth pointing out that philosophers generally regard consent theories of government legitimacy as unsuccessful. The people who advocate consent theories are more or less all anarchists!)

Insofar as 1 and 2 are true, it’s clear why political decisions are morally important.

These points also hold true specifically for electoral decisions. How we vote makes a big difference. Obviously, the good and bad governments do are not entirely attributable to how we vote.  Our voting behavior is just one of many factors affecting political outcomes. Still, on the whole, how we vote makes a difference.  Political parties have policy bents—dispositions to implement certain kinds of policies rather than others.  When voters vote for members of a party with a particular policy bent, this greatly increases the probability that those kinds of policies will be implemented. 

Despite this, the ethics of voting is puzzling. How we vote has major consequences. How any one of us votes does not. As a collective group, the German citizens who voted for the Nazis or Nationalists in November of 1932 did a lot of harm, but had any one of these voters abstained or voted differently, it would have made no difference. I’ll post more about this over the next few days. But here I’ll just note that (with a few exceptions), most people working on voting agree that the probability that your vote will change the outcome of any election—or have any other significant consequence—is vanishingly small. As Stephen Landsburg says, you're more likely to win Powerball multiple times in a row than change the outcome of an election. Even in extremely high stakes elections, the expected utility or disutility of your vote is vanishingly small.

In any major election, what will happen will happen regardless of what you do. So, why think you have any obligation to vote one way or another? If in the next US presidential election, we had to choose between a resurrected Socrates or a resurrected Stalin, voting for Stalin would have about the same impact as voting against him.

It’s thus tempting to make the following argument:

  1. The probability a vote will make difference is vanishingly small.
  2. Thus, the expected disutility of a bad vote is vanishingly small.
  3. Thus, individual votes are morally trivial. Under normal circumstances (where individual votes are trivial), you have no ethical obligations to vote one way or another. 

In The Ethics of Voting, I argue that people have no duty to vote. However, I argue that if they do decide to vote, they have rather stringent duties to vote in particular ways. Not just any vote is acceptable.

 If individual votes made a big difference on political outcomes, it would be easier to argue that people have a duty to vote and also a duty to vote well. However, if individual votes make no significant difference on political outcomes, it becomes relatively easier to argue that citizens have no duty to vote, but harder to argue they have duties to vote “well” (according to some standard of good voting).

So, the ethics of voting, on its face, puzzling. Notice that this is just an instance of a broader issue. We human beings are social animals. We often coordinate our actions together and engage in collective activities. In some of these collective activities, what we do as a group has significant consequences, but what you or I do does not.




Published on:
Author: Jason Brennan
  • Eli

    Under no circumstances would I vote for a totalitarian like Socrates.

  • plutosdad

    I think one place you really see this voting without thinking is in judge voting. So many people just put “keep” down without thinking. They even admit this and don’t care or think there’s anything wrong with blindly voting to keep a judge.

    The last few elections I actually did research and voted against certain judges that were pretty uniformly condemned by most associations. So since it was bi-partisan so to speak, across all special interest associations and general ones, that I thought “well these judges would be likely to go” since it seemed most people who cared wanted them gone. But it did nothing. I am considering just ignoring the judge voting, what is the point when most other people in the city are just playing games?

  • John

    Jason, I like the basic conclusion that we want people to think about their vote and that it should have “meaning” to the person voting. At the end of the day, this does achieve the good society — and I suspect you’d say “Well, of course, I’m not claiming it does.”

    With that in mind, how would you characterize ethical voting in terms of producing the good society: necessary, necessary and sufficient (though ruled about above) or neither necessary or sufficient?

    I still think there’s an issue with how one interprets the no votes — a person may be abstaining because they haven’t done their homework or because they have and don’t see any of the candidates as consistent with the common good.

  • Jason Brennan

    Hi John,

    Good question. I’m not sure if this will exactly answer your question, but here’s a shot at it: In my view, a good society will minimize the role of politics and minimize the status we attribute to holding and exercising political power. A good society will treat presidents not as demi-gods who deserve reverence but as little more than chief public goods administrators. They will treat the right to vote not as a sign of inclusion in the national club, but roughly akin to a medical license. And so on.

    My view is often characterized as highly elitist, and it certainly is in some respects. But at the same time, in chapter 2 of the Ethics of Voting, I end up defending an unusually egalitarian and populist conception of good citizenship and civic virtue.

    As for the question of abstention–Yes, I agree that there are times when abstaining rather than voting for bad candidates is the thing to do.

  • Dan Kervick

    I don’t know what premises I could use to convince someone that they had a “moral obligation” to vote, but I can say something about the more logically tractable problem of expected utility.

    Start with some highly idealized and unrealistic circumstances. Suppose I am part of a voting population of 10,001 people, and suppose I have no information whatsoever about the voting preferences of any other members of the population, not even past voting behavior. Suppose, however, that I do have compelling evidence that all of the other 10,000 people will vote in the present election, where they will choose between two candidates A and B. As a result, each of these 10,001 voting outcomes is equally likely for me:

    A – 0, B – 10,000
    A – 1, B – 9,999
    A – 2, B – 9,998
    A – 10,000, B – 0

    Let V(A) represent the overall value I assign to A’s victory and V(B) the value I assign to a victory for B. Suppose V(A) is higher than V(B) since I prefer A to B, and that V* is V(A)-V(B), the positive number which measures the superiority, for me, of A’s victory. And suppose, to simplify matters, a tie in the election will not produce a runoff, but the rules give the victory in that case to the winner of a subsequent toss of a fair coin. So the value I assign to a tied outcome is just the average V^ = (V(A)+V(B))/2.

    In this case, the only outcome among the other 10,000 voters in which my vote for A or abstention from voting would make a difference is the case of the 5,000 to 5,000 tie, in which case my vote for A would break the tie and add significant value. The only outcome in which my abstaining from voting makes a difference is also the tie situation, whose value is V^ = (V(A)+V(B))/2 – as opposed to either V(A) and V(B) which are evenly distributed over the other outcomes.

    So as a first approximation, the expected utility of voting for A is (5,000/10,001)V(A) + (5,000/10,001)V(B) + (1/10,001)V(A) ; and the expected utility of abstaining from voting is (5,000/10,001)V(A) + (5,000/10,001)V(B) + (1/10,001)V^. The difference between the expected utility of voting for A and the expected utility of abstaining is (1/10,001)(V(A) – V^), or V*/20,002.

    However, we need to include the utility and disutility of the voting act itself. So let C be the personal cost to me of the voting act itself – which includes such things as the cost of the gas I need to drive to the polling place (or food I need to consume in order to power my walk there), as well as the sum value of the pleasures and pains I will experience during the time it takes me to vote. Let’s assume that there is no probabilistic element here since these costs are not in doubt for me. Let’s assume that D is the value for me of the utility I would generate by employing my voting time is some other activity, in the most useful way. So then a better representation of the difference between the expected utility of my voting for A and the expected utility of my refraining from voting (and doing something else worthwhile instead) is (C – D) + V*/20,002.

    People usually think of the value C as negative, and as a measure of the disutility of voting. But it need not be. For me personally, in my relatively small town, going to the polling place and voting consumes very little time. There are no lines to speak of. It is on the way home from work and so takes me only a few feet off my ordinary path. I get to see a number of old friends and other townfolk at the community center. I feel a pleasurable connection with the civic buzz of democracy, and the traditions of my community and nation. Maybe I even get a free cookie or a brownie and a cup of coffee.

    In the real world, of course, I do have some information about the likely voting behavior of the other people, which for a population of 10,001 would tend to make the denominator in the initial expected utility calculation significantly lower, and probability that my vote will have a decisive impact significantly higher. In an election involving a larger population, the denominator is much higher – but so of course is the value of V*, since the stakes are much greater. Also, in a presidential election the expected utility calculation is more complex, since it depends both on the probability that by own vote will be decisive in my state and the probability that my state’s votes will be decisive in the Electoral College.

    A good deal depends on the costs of voting, C, but also on the expected utilities of my alternatives – the things that I could be doing instead, measured by D. Again, just for me personally, there are generally few things I could do with the 5 to 10 minutes I expend on the voting act which could produce as much utility as the cheap civic buzz, camaraderie and brownie.

    This is just a loose framework for thinking about these issues. But given the stakes involved, the realistic probability distributions over the range of possible outcomes, the utility of the voting act itself and the paucity of opportunities for significantly higher utility in the same short period of time, it seems plausible to me that there would be many cases in which voting has higher expected utility than abstention from voting.

  • Jeff R.

    I’m not sure that the expected disutility of a poor vote outcome is at all small, because complicity in a bad outcome has a large disutility of its own that is not all that dependent on the probabilty of an individual vote making a difference. It seems if you do give a high value to avoiding complicity, there’s going to be a general duty to vote for the best/least bad choice most of the time, and in some cases a duty to vote strategically for the strongest competitor of a sufficiently bad candidate/option.

  • “However, I argue that if they do decide to vote, they have rather stringent duties to vote in particular ways. Not just any vote is acceptable.”

    I would re-word that. I think I understand where you’re going with it, but the current wording can easily be construed as regarding the actual vote being made, i.e., Voting for Alice is okay, but it’s unacceptable to vote for Bob. (Which is a common partisan argument.) It seems to me that what you’re aiming for is closer to: “However, I argue that if they do decide to vote, they have rather stringent duties to vote in accordance with certain ethical principles. A purely random or intentional throwaway vote is not acceptable.”

    Am I on the right track with that?

  • conchis

    For an argument that 2 does not follow from 1 in your final argument, see this paper by Edlin, Gelman and Kaplan. Basically, their point is that the disutility of a bad vote scales with population, offsetting the reduced chance of your vote being pivotal.

    From the paper:

    We demonstrate that voting is rational even in large elections if individuals
    have ‘social’ preferences and are concerned about social welfare. In a large election, the probability that a vote is decisive is small, but the social benefits at stake in the election are large, and so the expected utility benefit of voting to an individual with social preferences can be significant. What is perhaps surprising is that the expected value of the social benefit does not approach zero or even diminish as the number of voters grows large.

  • Jason Brennan

    Thanks, Conchis. I’ve seen that paper (in fact, I linked to it as “with a few exceptions”) above. Edlin, Gelman, and Kaplan might be right, but I’ll just note here that their method of calculating the probability of being decisive is much different from the mainstream method, and gives massively larger probabilities.

  • conchis

    [EGK’s] method of calculating the probability of being decisive is much different from the mainstream method, and gives massively larger probabilities

    Could you expand on this a bit (or point me in the direction of a more mainstream view)? EGK say broadly that the probability should be proportional to 1/n, with an estimate of 10/n for close elections, which didn’t strike me as unreasonable.

  • jtkennedy

    Jason, “they have rather stringent duties to vote in particular way”

    So how bad is the offense if one votes recklessly? Are victims owed anything?

  • Pingback: dna wealth blueprint warrior forum()

  • Pingback: angara fahise()

  • Pingback:

  • Pingback: angara fahise()

  • Pingback: angara fahise()

  • Pingback: bursa orospu()