Friday, February 22, 2008

Compulsory Voting with Careful Consideration of Choices

Guest Post by Amir Michail

It's hard to understand why people vote. If we put aside social pressure (e.g., what your friends and family think) and sense of duty, a rational person should be embarrassed at having wasted any time at all on voting—in the same way that a rational person would be embarrassed at having bought a lottery ticket.

To address this problem, one might consider a compulsory voting system where it takes the same effort to vote as to cast a "no vote". It's compulsory in the sense that if you do nothing, then you will be subject to a significant fine.

But even such a system is not enough as people could simply vote randomly. In fact, a rational person should vote randomly since admitting to having voted otherwise would be embarrassing given the insignificant probability of making a difference.

So how do we require people not only to vote (possibly casting a "no vote") but also think carefully about their choice?

Two ideas:

  1. Compulsory Voting in Blocks

    To vote for X you would need to write the names of at least k other people whom you know are voting for X (this applies also for a "no vote"). This would encourage more people to have discussions with their friends/family and think more carefully about their choice. However, one could argue that rational people would select a candidate randomly in blocks (since again for small k, the probability of making a difference is insignificant).

  2. Compulsory Voting with Consequences

    Voters would be personally responsible for their choice. In particular, an objective measure would be used several years later to see whether their choice was a good one and they would be penalized/awarded financially correspondingly. This is sort of like a prediction market for politics with real money.

While (2) might seem more promising than (1), at least with respect to rational people, there's the issue of the choice of objective measure.

Do you have better ideas to require people not only to vote but to also take their vote seriously?

26 comments:

  1. "In fact, a rational person should vote randomly since admitting to having voted otherwise would be embarrassing given the insignificant probability of making a difference."

    I would like to be rational. Can you prove this claim first?

    ReplyDelete
  2. "Do you have better ideas to require people not only to vote but to also take their vote seriously?"

    Here is one - don't leave it to idle academics discussing these things on blogs.

    Compulsory voting and fines for "bad" votes? Naming names? Blecchh. Sounds like good themes for Kafka-esque fiction (or a Seinfeld episode - "democracy Nazi").

    If you want to incentivize voting, there are easy ways to do it - a tiny voucher with a random number printed on it, with which you can claim a $10 tax deduction the next time you file your taxes. If you don't wish to have any record in any government archive that you voted, simply don't include that info. when you file your returns. (In the U.S., you can add a small contribution to election campaigning along with tax returns - perhaps add another component to get money back if you had voted.)

    ReplyDelete
  3. Force people to take a test on the position of candidates on various issues, and only count votes for people who do reasonably well. (This makes somewhat more sense for bills than officials, but I think it still works.) Then the number of votes counted overall is much smaller and so the vote of any one person who knows anything about the issues counts more.

    ReplyDelete
  4. I don't think (1) would ever fly in the US, as the privacy of one's vote is considered extremely important. (Unless for instance they are voting for the same person that everyone else they've ever known is voting for. But even families in the US are often divided among Democrat/Republican, with radical supporters on both sides painting the opposite side as the devil.)

    I'm not so high on (2) either. For example, in such a scenario it would no longer make sense to vote for a person with no chance to win, like Ron Paul, even though that person may agree most with Ron Paul's ideas. You are changing the game to: "who is more likely to win and do an OK job" instead of the original game which was "who do you want as president?".

    For another problem with (2), it would probably be difficult to find an objective measure of the president that a majority of people would agree on, especially if there were money involved. Let's suppose I messed up and voted for Bush. If I could get a few thousand dollars in the case that Bush does a better job than Kerry, why shouldn't I make an argument for Bush, no matter what? Here the trouble is in reasoning about counterfactuals: how do we collectively and objectively compare Bush's performance against what we believe Kerry would have done had he been elected, when half of the country didn't vote for Kerry and we don't have concrete evidence of what a Kerry-led US looks like?

    Unfortunately I don't have a good alternative. Somehow you want people to come out to the polls and be truthful in who they want. This problem looks to be of a completely different nature than bidding on items with prices...

    ReplyDelete
  5. This comment has been removed by the author.

    ReplyDelete
  6. Douglas Hofstadter mentioned something in one of his articles. The idea of being meta-rational or something like that. His article dealt with the prisoner's dillema. But the analogy still holds.

    The idea (I think) is that if everyone is ultimately rational, then everyone will make the same best choice (vote). So make a choice (vote) the way you would want everyone to vote.

    I also think that the embarrassment you talk about is small compared to the embarrassment of being labeled a geek by others who see you splitting hairs on issues so finely.

    ReplyDelete
  7. Then, accordiang to Amir, living in a totalitary state must be the most rational choice.

    ReplyDelete
  8. If rationality is so important in this conversation, perhaps we could start by justifying (rationally!) the idea that more people should vote (especially those who don't currently vote).

    ReplyDelete
  9. The entire basis of this post makes no sense to me.

    First of all, let's assume two candidates, everyone's preferences are known, and that there is no tie. Then although it is not Nash equilibrium for *everyone* to vote, it is not Nash for *no one* to vote, either. Furthermore, since voting can only help your candidate (and the "cost" of voting is pretty small), voting weakly dominates not voting.

    If preferences are unknown, the case for voting being rational becomes even more compelling.

    The above doesn't even take into account the likely fact of "coalitions" where blocks of people with the same political beliefs all encourage each other to vote.

    And why do you assign a negative utility to the "embarrassing" admission of having voted (non-randomly, or at all)? What world do you live in where this is considered embarrassing at all?

    ReplyDelete
  10. 1. I think the notion of "rational person" you are talking about does not exist in real world. It is well-known that people care about things other than maximizing their profit, otherwise things like donating to charities, or obeying laws when no one is watching is not justifiable.

    2. Read Robin Hanson's article "vote for values, bet for beliefs" (upcoming in J of Political Philosophy).

    ReplyDelete
  11. Hell, make a completed ballot redeemable for entry into an election raffle. Every election, select some random voted and give them some not insignificant amount of money.

    And probably throw in dinner with the victor in the election or something similarly photogenic.

    ReplyDelete
  12. ...And a rational person should never watch TV, given the insignificant probability of making a difference: the sitcoms are scripted and the football players can't hear your cheers. Isn't politics at least as entertaining as baseball?

    I'm also missing how sense of duty and social pressure aren't factors that a rational person takes into account. Is there some list of things that rational people need to disregard in making their decisions? I have a suspicion that a rational person would never post to this blog or study complexity theory. :)

    I apologize for treating this post like a joke, but the whole rationale for your argument, "In fact, a rational person should vote randomly since admitting to having voted otherwise would be embarrassing given the insignificant probability of making a difference," is wrong. As long as you are voting anyway (and your votes are anonymous), there is no rational reason not to vote for the candidates you like.

    The costs of voting are rather low -- google up the voter registration form for your state, submit it electronically, then mail in the absentee ballot when it arrives (41 cents). (For caucuses the costs are obviously much higher.) So voting is not one of the more irrational things that people do. That said, I do like your "Compulsory Voting with Consequences" idea, since, applied today, it would eliminate a whole party from consideration as well as one of the two candidates from the other party.

    ReplyDelete
  13. I apologize for treating this post like a joke, but the whole rationale for your argument, "In fact, a rational person should vote randomly since admitting to having voted otherwise would be embarrassing given the insignificant probability of making a difference," is wrong. As long as you are voting anyway (and your votes are anonymous), there is no rational reason not to vote for the candidates you like.

    Even if you never tell anyone how you voted, one can argue that it is still embarrassing not to vote randomly because it demonstrates a lack of understanding of probability.

    Even if this is wrong, there is little incentive for a rational person to give the matter much thought. This is a problem in the presence of compulsory voting.

    ReplyDelete
  14. On the contrary, people who do not care enough to vote should not have a say in the democratic process.

    Things are fine as they are. And the post's take on "embarrassment" is nonsensical.

    ReplyDelete
  15. "Even if you never tell anyone how you voted, one can argue that it is still embarrassing not to vote randomly because it demonstrates a lack of understanding of probability."

    How does probability even come into this? I think you are confused. If you are indifferent to two outcomes but must choose one, then it doesn't matter what strategy you choose. In this case, choosing randomly is not better than choosing according to any other strategy.

    ReplyDelete
  16. How does probability even come into this? I think you are confused. If you are indifferent to two outcomes but must choose one, then it doesn't matter what strategy you choose. In this case, choosing randomly is not better than choosing according to any other strategy.

    You may care quite a bit about the outcome but your understanding of probability would tell you that you are wasting your time voting. As a protest and to avoid embarrassment, you might vote randomly.

    But even if you don't agree that rational people would behave in this way, what's wrong with requiring everyone to vote and to also think carefully about their choice?

    ReplyDelete
  17. Is this post supposed to be a joke?? Assuming it's not, then I would like to make two points

    1) Comparing voting to buying a lottery ticket seems to me to only make sense if the way the winner of an election was determined was by selecting a random vote and whoever voted for is the winner

    2) Probably one way to get people to take their vote seriously is to explain to them how fallacious the view here is. It's a cop out. By similar reasoning a person shouldn't do anything about global warming (because nobody else will), etc.

    Am I missing something??

    ReplyDelete
  18. I'm fully aware of the fact that voter turnout is so low; my beef is that this post seems to concede that people who don't vote are being reasonable

    ReplyDelete
  19. I'm fully aware of the fact that voter turnout is so low; my beef is that this post seems to concede that people who don't vote are being reasonable.

    Take a look at those links. Whether they are being reasonable is rather complicated.

    ReplyDelete
  20. I agree with Adam: the suggestions must be a joke.
    Or Amir wants us to move to a totalitarian system (or Kafka-esque fiction). His suggestions are completely unrealistic and there is no way any of them would fly in the US.
    One thing is to think how to improve the low voter turnout and another is to force people to vote.
    I think, long time ago, in Albania (?), there was 100% (or more ;-) turnout in one of the elections!

    ReplyDelete
  21. If rationality is so important in this conversation, perhaps we could start by justifying (rationally!) the idea that more people should vote (especially those who don't currently vote).

    From http://knzn.blogspot.com/2007/12/compulsory-vs-voluntary-voting.html:

    "To begin with, economic logic will indicate that voting is, for the individual, irrational, unless it is mandatory. Therefore (here’s the political part), under a voluntary voting regime, only irrational people will vote. Do we really want government by the irrational, for the irrational, and of the irrational?"

    ReplyDelete
  22. It's hard to understand why people vote. If we put aside social pressure (e.g., what your friends and family think) and sense of duty, a rational person should be embarrassed at having wasted any time at all on voting

    No. Your logic may hold if you're talking about monetary returns from voting (say you believe Bush would extend tax cuts while Kerry would not). However, you are leaving aside the ethical dimension. Having voted for Kerry in the last election gives you the feeling of having voted against war and torture. And that feeling should be worth a lot.

    BA

    ReplyDelete
  23. quote: Do you have better ideas to require people not only to (a) vote but to also (b) take their vote seriously?

    How about this scheme:

    Ballot contains 2 sections, one contains the usual names of candidates. Second section contains the same candidates but with any identification removed and replaced by their quotes, voting records, and some other relevant information (such that the candidate's name cannot be inferred easily unless the voter knows the candidate).

    Voter is required to vote for a person in the first part and then for the set of relevant characteristics in the second part. Voter is fined if the person voted for in the first part does not match the characteristics voted for in the second part (or alternatively awarded if it does match). Casting no vote is assumed to be equivalent to mismatch for the fine/award purposes.

    Now, this should:

    (a) Encourage to vote given fine or award.

    (b) Encourage informed vote given fine / award for knowledge of candidates.

    Now there are counting options:

    1. Ignore mismatches, this should theoretically make only informed votes count.

    2. Do not ignore mismatches, count either the named candidate, characteristic candidate, or both (this would also require matched votes to be double counted). If we count only the named candidate, then we fail. If we count the characteristic one or both, then we have the same result as in #1, that is, voting set includes those that are either informed, or uninformed+random. Assuming non-zero informed votes, this is better than just uninformed+random.

    There is of course a big problem with this scheme. Who decides on the "characteristics" in the second section of the ballot? Skew of large proportions if one candidate can be easily recognized compared to the other. Complex solutions may be possible... but this comment is already too long...

    ReplyDelete
  24. There are serious modeling problems that this "economic logic" suffers from.

    Buying a lottery ticket is "embarassing", or irrational if you make two (strong) assumptions: 1) a linear utility function, and 2) risk neutral agents. While these are assumptions economists make because math becomes hard when you get rid of it, the assumptions are not valid in many situations.
    My utility for a million dollars is more than a million times my utility for a dollar, since I can retire now if I win a million dollars. A small probability of being able to retire now is worth more than giving up a few dollars today.
    If your model suggests that buying lottery tickets is irrational, a good explanation for that is that your assumptions are violated, not that people are irrational.
    (Not that I'm saying that people are always rational, only that this particular instance is not necessarily an example of irrational behaviour)

    In the voting case, there is utility associated with feeling that you have a voice in the political process, feeling that you have done your duty as a citizen, not regretting later that if you had voted, thousands wouldn't be dying in war. Your economic model fails to account for this utility, hence the fallacious conclusions.
    --kunal

    ReplyDelete
  25. I know this post may have been a joke, but I'm just amazed at how far removed from reality it is. Have you ever wondered why theoretical computer science (and science in general), has been receiving so little funding recently? I know this is starting to change, but the lack of participation in politics from scientists and engineers has profound consequences. Just consider this: we're spending $100+ billion in Iraq each year and only $6 billion a year on the NSF. Ever wonder why? Could it have anything to do with the people we elect to public office? I can't believe we're even discussing the merits of voting here. If we are to talk about reality, we should be really be discussing whether or not we should hire some lobbyists to advocate for more science and technology funding. Then we could do some real cost/benefit/risk analysis.

    In case you're not convinced somehow about the merits of voting, let me attempt to debunk your argument from a more mathematical prospective. I assume you say that voting has an "insignificant probability of making a difference" because in a given election with millions of voters, the likelihood that your single vote effects the outcome is relatively low. It's true that if you're only considering this one election and nothing after that, you probably shouldn't vote. But voting and politics is a repeated game. If you don't vote now, the politicians won't pay attention to you during the next election season. They won't care about funding your NSF project, but they will care about all the other powerful special interests (unions, corporations, etc.) that can and will get them elected.

    Henry
    PS Sorry for taking this post too seriously, perhaps.

    ReplyDelete