Voting is not rational (usually.)

post by araneae · 2010-11-02T19:34:28.552Z · LW · GW · Legacy · 36 comments

Today is the midterm elections in the United States, and I am not voting.

For the vast majority of elections, voting is irrational, because the individual's vote is proportionately very small. This means it cannot have an effect on the outcome.

There are, however, conditions which can lead to voting becoming rational, and these are:

  1. The number of voters approaches zero.
  2. The ratio of votes for candidates (in a majority wins, 2 person race) approaches .5
  3. The difficulty of voting becomes vanishingly small.
  4. Incentives are created to make the costs of not voting greater than the cost of voting (for instance, not voting is illegal in Australia, and incurs a fine.)
For me, as for nearly everyone, voting this year is irrational.  1 and 2 are nowhere close to true, and 3 is especially bad for me this year.  I forgot to change my address on my voter's registration until yesterday and my polling location is both a) usually overpopulated and filled with long lines and b) farther than I care to go.  
Only 3 and 4 are something that we can certainly do something about.  The web-based absentee voter system that was tested this fall is a step in that directions, but its  subsequent hacking  was unsurprising.  Is opening our system up to fraud a reasonable trade-off to get more people to vote?  Should there be an option to use paper absentee-like ballots even if you are not absent? Should the U.S. go the Australia route?

 

36 comments

Comments sorted by top scores.

comment by Larks · 2010-11-03T16:36:59.853Z · LW(p) · GW(p)

Supposing you were right that it is irrational at the moment to vote, you haven't shown why we should want to make it rational to do so.

Replies from: gjm
comment by gjm · 2010-11-04T10:45:39.931Z · LW(p) · GW(p)

The more irrational it is to vote, the stronger the selection bias in the voting population in favour of irrational voters. You might prefer (as I do) that politicians be elected by as rational a population as possible, all else being equal.

(On the other hand, the same bias that reduces voting by more rational people probably also reduces voting by more selfish people, since such benefits as there are from voting typically accrue to the population at large. I'm not sure what level of selfishness I want among the voting population.)

Replies from: Larks
comment by Larks · 2010-11-04T10:48:38.323Z · LW(p) · GW(p)

If it means my bayesian altruists are doing something more effective than queuing at the polling station, I'm happy with that.

comment by Jack · 2010-11-02T22:47:39.959Z · LW(p) · GW(p)

I find voting fun. It's like going to a sporting event and cheering loudly for my favorite team. And I enjoy the alief of power it provides. Plus they have stickers. The utility I get from the experience is a fairly good return on ten minutes.

Replies from: Yvain
comment by Scott Alexander (Yvain) · 2010-11-03T00:05:08.425Z · LW(p) · GW(p)

I find voting fun. It's like going to a sporting event and cheering loudly for my favorite team. And I enjoy the alief of power it provides. Plus they have stickers.

Speaking of stickers, I drew this the other day and have been waiting for an excuse to post it here: Sticker

Replies from: kodos96, Relsqui
comment by kodos96 · 2010-11-03T07:02:38.806Z · LW(p) · GW(p)

I drew this the other day and have been waiting for an excuse to post it

Something this awesome doesn't need any excuse.

comment by Relsqui · 2010-11-04T07:26:52.265Z · LW(p) · GW(p)

Ahahaha. That's awesome.

Replies from: CronoDAS
comment by CronoDAS · 2010-11-04T08:19:47.416Z · LW(p) · GW(p)

Indeed it is.

Replies from: Relsqui
comment by Relsqui · 2010-11-05T03:59:49.560Z · LW(p) · GW(p)

Comments like this are why I wish my inbox had inline context.

comment by sixes_and_sevens · 2010-11-03T11:00:04.727Z · LW(p) · GW(p)

Voting in the belief that your vote will influence the outcome is irrational. If your vote helps carry out some other useful societal process from which you benefit, that's a different matter.

comment by SeventhNadir · 2010-11-03T09:30:51.571Z · LW(p) · GW(p)

for instance, not voting is illegal in Australia, and incurs a fine.

I'm being pedantic but it's more accurate to say that "Not marking your attendance off on the electoral roll incurs a fine". There is no penalty for then taking your ballot and submitting it blank.

comment by Manfred · 2010-11-03T04:56:10.431Z · LW(p) · GW(p)

I feel like you could put in a lot more work making this post more rational. It could use a cost/benefit analysis, or, even better, the form of the volunteering problem from the article on mixed strategy nash equilibria. Be more explicit about more things.

the individual's vote is proportionately very small. This means it cannot have an effect on the outcome.

Read that again. Do you see the same problem I see?

Perhaps reading this recent thread that's almost identical would be interesting to you.

comment by Cyan · 2010-11-03T01:40:20.653Z · LW(p) · GW(p)

I think that if you want to tackle this subject here and now, a good way to go about it would be to argue against Andrew Gelman's opposite view.

comment by David_Gerard · 2010-11-04T15:45:01.457Z · LW(p) · GW(p)

The analyses I've seen that purport to prove that voting is all but irrelevant and therefore irrational to bother with generally assume that voting is the only action that can be taken by the individual to affect the result.

This does not appear to describe real-life behaviour, where people state their voting intentions and try to get as many other people to vote that way as possible - the participants understand that having as many people as possible join in increases the likelihood of the desired effect. This can sway seats and, scaled up, whole elections.

(This is similar to the problem with the analysis of how to donate to charity: it assumes people donate and don't tell anyone. This also does not describe what happens in the real world.)

Is there an analysis purporting to prove bothering to vote irrational that explicitly includes this effect in a plausibly quantifiable fashion?

Replies from: CarlShulman
comment by CarlShulman · 2010-11-04T20:31:25.845Z · LW(p) · GW(p)

Standard analyses have several orders of magnitude of safety margin. Very few people will sway thousands of other voters through casual conversation about voting without much additional effort (which raises costs along with benefits).

comment by jmmcd · 2010-11-04T00:33:02.608Z · LW(p) · GW(p)

One doesn't just vote in order to get candidate X in (or to signal or "cheer for one's favourite team"). One also votes for X in order to, roughly speaking, live in a world where people vote for X. Search LW for "decision theory" material.

comment by komponisto · 2010-11-03T00:04:03.560Z · LW(p) · GW(p)

Should the U.S. go the Australia route?

Hell no. (No offense to the great nation of Australia, which in this particular regard is just like a number of other countries.)

The right to abstain is not often celebrated, but that doesn't make it less important. Being able to save one's vote for when it really matters, and thus signal by voting that one has a higher degree of interest in the election outcome, is no less valuable than the ability to express a preference for one candidate or party over another -- and it's something that Australians (and Brazilians, etc) obviously don't have.

Replies from: NihilCredo, Carinthium
comment by NihilCredo · 2010-11-03T00:28:47.474Z · LW(p) · GW(p)

That's not a problem, you can still vote blank or "[expletive]" for that.

Replies from: komponisto
comment by komponisto · 2010-11-03T05:19:00.188Z · LW(p) · GW(p)

In that case, the requirement to show up at the polling station seems like a pointless infringement on personal liberty.

Replies from: NihilCredo, SeventhNadir
comment by NihilCredo · 2010-11-03T14:19:57.789Z · LW(p) · GW(p)

Not pointless at all: it turns most of the cost of voting into a sunk cost, thus making it much more likely to be rational for any one citizen to cast a vote.

Replies from: komponisto
comment by komponisto · 2010-11-03T17:11:51.089Z · LW(p) · GW(p)

Well, then my original point stands, since the policy clearly diminishes the strength of the signal sent by voting as opposed to not voting.

And let me be clear: I do also object on general libertarian grounds to the government making me vote, i.e. I don't in fact believe voting is anywhere near important enough to justify such a policy. So I'm very happy not to be a citizen of one of those countries.

Replies from: NihilCredo
comment by NihilCredo · 2010-11-03T17:32:26.612Z · LW(p) · GW(p)

the policy clearly diminishes the strength of the signal sent by voting as opposed to not voting.

Disagree. If the cost of voting is negligible (because you're going to be in the voting booth anyway) and you still vote blank or F.Y., that is a much stronger signal.

Someone who stays at home most often still has a favoured candidate, but he may be lazy, or the polls may have been so skewed that he figured it wasn't worth showing up. But someone who shows up and votes blank can only have done it because, for some reason, he had no preference in the election - and someone who writes F.Y. on the ballot can only have done it because he wants to express his spite at the whole system.

Replies from: komponisto
comment by komponisto · 2010-11-03T18:16:58.087Z · LW(p) · GW(p)

Right, well I think voting should be the costly signal, and not voting the default. I don't think abstention should be a signal of spite or contempt for the system; rather the opposite (signaling a sufficient level of contentment that one is indifferent to the outcome).

If you really want to express contempt, you can voluntarily show up and vote "F.Y".

comment by SeventhNadir · 2010-11-03T09:40:13.539Z · LW(p) · GW(p)

It's annoying sure, but what other strategies are available to them? Voting is anonymous process after all. They've succeeded in getting you all the way to the polling booth which is a start. If the voters can't even be bothered to fill out a simple form, that suggests to me deeper problems.

Not that I'm one to talk, I vote for Optimus Prime.

comment by Carinthium · 2010-11-14T02:40:15.804Z · LW(p) · GW(p)

What evidence do you actually have for Australia being a great nation? If none, then why did you say it?

comment by Oscar_Cunningham · 2010-11-02T20:07:02.421Z · LW(p) · GW(p)

This has nothing to do with refining the art of human rationality, worse it's a political issue, which are well known to be hard to discuss rationally. Just because you've used the word "rational" in the title, doesn't mean that this post belongs on this site. If you had discussed the biases or logical fallacies that lead to people irrationally voting your post would have been worthwhile.

Sorry for being so scathing, maybe non-rationality related posts should be allowed in the discussion sectioned (personally I believe that they should not), I guess I'm particularly irked because political discussions tend to go downhill so fast.

In case I'm wrong and the community does want this discussion, I shall post a link to a rather nice analysis of when to vote: http://gowers.wordpress.com/2010/05/03/is-the-british-voting-system-fair/

Replies from: ronnoch, Carinthium
comment by ronnoch · 2010-11-02T21:20:15.962Z · LW(p) · GW(p)

Is the question of whether to vote really a "political issue" in the same Blues vs Greens way that, say, abortion or gun control is?

Replies from: Oscar_Cunningham
comment by Oscar_Cunningham · 2010-11-02T23:06:24.968Z · LW(p) · GW(p)

Definitely, though perhaps to a lesser extent. I think it still fulfils all the criteria that lead to mind-killingness.

ETA: Time will tell.

comment by Carinthium · 2010-11-14T02:41:32.091Z · LW(p) · GW(p)

Shouldn't rational websites try to improve themselves with regard to rational discussion of political issues?

Replies from: Oscar_Cunningham
comment by Oscar_Cunningham · 2010-11-14T09:09:50.180Z · LW(p) · GW(p)

No, in fact, I don't think LessWrong should discuss any issues at all. The point should be that we discuss rationality and learn it and get better at it. Then we can go out and make our own decisions.

If we decide what's rational by treating each subject in turn, we will never be done.

Replies from: Carinthium
comment by Carinthium · 2010-11-14T09:22:24.501Z · LW(p) · GW(p)

The reason to discuss it is that one of the strengths of a rationalist website is (more) rational discussion of issues in which there is a high degree of irrationality (e.g- politics). To the extent people on such a site are more rational they are more likely then politicians or political debaters to get accurate answers to the problems on which there is public debate.

Replies from: Vaniver
comment by Vaniver · 2010-11-14T09:58:35.522Z · LW(p) · GW(p)

The risk here is in several dimensions. First, I expect the rational answer is often "I don't have enough information to have an informed opinion." Second, it seems easier to develop rationality on simple, painless issues before moving to complex, painful issues. Telling people about the dangers of identity in the abstract, letting that sink in, then having them find examples in their life is one thing; introducing them to the concept with something fundamental to their identity is another entirely. I haven't seen much talk about religion here- partially because the groups are self-segregating, and partly because it's simply not a good idea to tell someone "ok, step 1 to being an atheist" instead of "ok, step 1 to weeding untruths out of your life." Third, these things tend to be eternal discussions if you have influxes of new people (which, as a newbie, I consider likely / a good thing). Fourth, coming to conclusions is something individuals should do, not groups, as much as possible. For example, imagine that the official policy position of Less Wrong was "Everyone should sign up for cryonics"- even if rational considerations lead one person to that decision, it seems unlikely those are sufficient to convince everyone.

There could be a sequence leading up to the conclusion that cryonics is great, but then we run into the issue of dogma: if someone asks why Bayes' theorem matters, we can point them to the sequence and it won't feel dogmatic. If someone asks a question about cryonics, saying "oh, we hashed that out in 2010, look here" doesn't seem like it's valuable, and similarly having cryonics as a perpetually open issue doesn't seem like it's valuable. And, for most people, cryonics simply isn't relevant enough to make it a particularly good rationality exercise.

To take the example of voting- I know how to divide. I know about gerrymandering. I still consider it worth the time to vote- why? Because I think the second order effects are worthwhile, even if there are no first order effects. I think that registering as a Republican, voting for the most liberty-friendly candidate in primaries, and voting for Libertarian party candidates in general elections shows support and increases the credibility of those people and that party. The effect is tiny- but it's there, and it will have more effects than gambling on being the one vote that changes the tide.

Would a discussion of just the first-order effects be helpful? And is it worth going to second or third order on discussions where the group might lack sufficient experience? I haven't run the numbers on how effective my support will be long-term, like I have on how much value I expect to get from the chance I change the current election. So maybe I have a pretty rationalization rather than rationality, but the expected value of doing that calculation is too low to follow through on.

comment by araneae · 2010-11-02T19:51:15.286Z · LW(p) · GW(p)

Wow, guys. I would have expected more of the less wrong crowd.

You emotionally dislike the fact that voting is irrational, so you've downvoted this post, but you haven't even attempted to reason with me.

If I'm wrong, tell me why. Don't just have an emotional reaction.

Replies from: michaelkeenan, Servant
comment by michaelkeenan · 2010-11-02T20:22:05.145Z · LW(p) · GW(p)

You emotionally dislike the fact that voting is irrational, so you've downvoted this post, but you haven't even attempted to reason with me.

If this post is getting downvoted, it's probably because it shows a lack of awareness of a zillion previous discussions on voting, e.g. this one. Many LWers like the decision theory called Timeless Decision Theory, in which - if I understand it correctly, a dubious assumption - you consider yourself to be making a decision not just for yourself, but for all similar minds. One advantage of this is that TDT agents cooperate on Prisoner's Dilemmas. Some people consider voting to be an instance where TDT agents decide to vote.

Your comment complaining about being downvoted was downvoted for complaining about being downvoted, and for making a probably-inaccurate accusation about why.

comment by Servant · 2010-11-02T19:56:54.914Z · LW(p) · GW(p)

The reason may be less that the LW community believes voting is rational...but that the LW community already understands that voting IS irrational, and the LW community is just reacting negatively to having the same idea be brought up again.

I understand you provided some solutions to try and make voting 'rational', though, however, I do disagree with the idea that #3 and #4 would actually work. Voting only becomes rational when your vote does sways an issue, and not only is this chance incredibly small, but it would likely provoke an automatic recount, meaning your vote is again proven useless.

Providing one extra vote to your side when your side is already is winning/losing is inherently irrational (at least in terms of swaying policy, not signaling views).

Replies from: araneae
comment by araneae · 2010-11-02T20:03:24.858Z · LW(p) · GW(p)

4 does work. In Australia they have a near-perfect voter turn out. It makes voting rational, in order to avoid the fine.