Why I Didn't Vote

This is not advice

My Vote Had Zero Causal Impact

To make it clear how little power my single vote has, I’m going to give you a superpower.

I’ve dipped your ballot in radioactive waste and it has mutated into a super-vote that’s ten times as strong. In addition, it can be cast in any district and state across the country so you can send it to hotly contested swing states, and it can travel back in time to vote in any past election and change all subsequent events.

You now have a vote that’s thousands of times more powerful than any other in the history of American democracy, but how much impact could you make with this superpower?

There are zero presidential elections you could tip. The closest election in American history was the 2000 race between George Bush and Al Gore. This came down to a few hundred votes in Florida, but it would still have come down to a few hundred votes no matter where you cast your super-vote.

There haven’t been any senate races within ten votes since 1976 either.

There were only two house of representatives races close enough to be tipped by ten votes out of over ten thousand since 1976.

There aren’t even very many state or local government races that are decided by ten votes or less. Zero gubernatorial races since 1865 were decided by less than ten votes and only 8 have been decided by less than 100.

Many more local government races are decided by less than 10 votes but still not that many. In this database of 57,000 local elections in 1,747 cities, counties, and school districts from 1989–2021 there are 5 mayors, 4 sheriffs, 3 prosecutors, 3 school board members, and 1 county executive whose elections were decided by less than ten votes.

There are definitely some elections you could tip with a 10x time-traveling super-vote. Unfortunately, we’re only given the normal stay-at-home single-vote. My personal ballot counts only in Virginia, which is not a seriously contested state so it has zero causal impact on the election whether I cast it or not.

No set of beliefs about how much of democracy is at stake in this election matter for my decision to not vote given the inability of that vote to affect the outcome. These beliefs might push me to donate to or publicly advocate for a candidate, as these actions can change many votes, but personally adding one vote in an uncontested state is only symbolically relevant to the outcome.

Back of the envelope calculations by Maxim Lott here suggests that if you have some certainty over the magnitude and direction of the difference in value between the two candidates and you’re in an important swing state like Pennsylvania, your vote can have significant social value on the order of thousands of dollars. If you’re not in this situation then one vote just isn’t very important. Down-ballot contests are more likely to be decided by your vote, but this is still very rare and the increased likelihood is offset by decreased importance.

Voting is Not a Normal Collective Action Problem

The usual response to this argument is that while each individual vote may have negligible impact, if everyone thought this way and abstained from voting, democracy would collapse. This is like littering or shopping cart return. I might be able to get away with saving time and leaving the cart behind unnoticed, but this is just selfish shirking from the social good; voting is no different.

The problem with this response is that in a winner-take-all system like American elections, all of the social good happens in a single bit-flip moment when the winner changes and nowhere else. When you put a shopping cart away or stop to pick up some trash, you incur a small personal cost and leave the world a little bit better for others. Contributing to these and other public goods is commendable and important.

However, if I vote for a candidate in a state they are already certain to win (or lose), the world doesn’t get a little bit better (or worse), it just doesn’t change at all. Voting isn’t a collective action problem like climate change, pollution, or shopping carts. All of these problems have smooth changes in marginal costs and benefits as more people contribute or shirk. In voting, everything comes down to the 50/50 threshold. If you’re away from that threshold, small changes simply have no effect on the world.

I Don’t Want Either Candidate to Feel Popular

One response to the above argument is that while the outcome of the election looks like the binary relationship on the right, things like voter turnout and polling percentages are more continuous like the other collective action problems on the right. Adding to voter turnout counts is an important way to contribute to the stability of democracy and adding to a candidates polling average can help them even when they’ve already won or lost.

The problem here is particular to this election: I don’t think either candidate should feel more popular. Both candidates have major policy proposals that I think are terrible so I don’t think bumping their polling percentages is a contribution to the social good at all. No matter who wins, I’d prefer for them to win in a constrained way that pushes them to moderate their campaign promises. Abstention is one way to signal this.

Understanding Voter Irrationality is Important

If what I’m arguing here is true, then why do people vote at all? Usually when there are collective action problems where each individual can get the same outcome and less costs by shirking, we see a lot more shirking. Sure, lots of people don’t vote, but most of them do!

The only explanation that can explain both the fact that each individual vote has next no causal impact and that most people vote regardless is that people aren’t voting for the causal impact at all. It’s true that each individual voter can be essentially certain that they can get the same election outcome whether they vote or not, but there are lots of other things they will miss out on if they don’t vote. This is called voter irrationality but it’s really just rational optimization over a more complete set of costs and benefits.

If someone doesn’t vote, they are sure to get the same president but they will miss out on the feeling that people get when they cheer for their team alongside other fans, a feeling so strong that people often cheer alone at home watching on TV. Or, they might fail to fulfill their sense of civic duty or they will risk embarrassment when their friends or coworkers notice their missing sticker.

Understanding that these alternative motivations are driving voting behavior leads to different predictions for how democracy works. These alternative motivations may lead people to vote against their narrow self-interest in terms of the candidate’s actual policies. E.g a Texan farmer who personally benefits from cheap migrant labor might still vote for a border lockdown candidate because the emotional returns to nationalism drive his voting behavior rather than his causal impact on the next four years of policy. In general, these “less rational” motivations for voting would lead to elections that are based more on personality, looks, entertainment, and emotion rather than the net difference between each candidate’s policies.

While there’s no direct relationship between my abstention and the popularity of Bryan Caplan’s public choice theories, publicly and shamelessly talking about not voting often draws people in to talk about the topic.

I’m Not Convinced by Rationalist Decision Theories

Some rationalists will agree that the causal impact of my single vote is zero, but that I should still vote anyway because if I don’t vote then people who are using decision algorithms similar to my own are also more likely not to vote.

This is connected to problems like Newcomb’s box: Imagine a super-smart AI gives you two boxes, one labeled $1,000,000, one labeled $1,000, and a choice to take just one or both. If the AI predicted that you would take just one box, the $1,000,000 labeled box really does have the money in it and you can take it. However, if the AI predicted that you would take both boxes, only the $1,000 labeled box has any cash.

Causal decision theory observes that no matter what the AI predicted, taking two boxes is always better than taking one. But the rationalist decision theory says that having a decision algorithm that leads you to picking just one box is the best way to get the million.

The one-boxing argument in Newcomb’s paradox has always seemed tenuous to me and even more so in the case of voting where there is no near-omnipotent AI that’s predicting my decision making process and changing the world based on that prediction.

Conclusion

This post is definitely not arguing that more people should abstain. Trying to sway the votes of lots of people immediately undermines all of my arguments for why it’s unimportant for one person to vote. Everything is predicated on only changing one vote while holding all the others constant.

But I don't think it's some big transgression or betrayal of the social good to not vote. A single vote almost never matters; elections are decided by everyone else with margins of thousands. There are much more important ways to improve the world that people treat as far less sacred.