Should a system of voting aim to let one individual voter express their preferences?

Why should we prefer democracy as a system of government: because it indulges an individual voter, or because it produces the best results for the most number of people?

Again, the benefits of democracy have been well-documented. Amartya Sen (whose work we will review during 2022) won a Nobel Prize for showing that people in democracies do not suffer from famine. People in democracies are better protected against disease, are more likely to be vaccinated against common childhood illnesses, and tend to have better educations provided by their governments. In the long run, democracies tend to experience more economic growth than non-democracies, though at any given time there might be a non-democracy that is growing very fast. One study has shown that men are taller in democracies.

When I write about democracy, I assume the goal is to create a system that builds a healthy, happy society. If autocracy reliably delivered better results than democracy, then I would be morally obligated to support autocracy, but since democracy reliably does the job, I am morally obligated to support democracy.

What is odd is when a given system of voting is criticized for not indulging the whims of a voter. I’ve previously written in favor of approval voting. Some people are opposed to approval voting because it doesn’t always perfectly express the preferences of voters. This argument seems entirely irrelevant to me.

There are fascinating arguments for and against approval voting. For instance:

Approval voting can in fact violate “later-no-harm”. Approving an additional, less-preferred candidate creates direct competition with a more-preferred candidate. Each candidate you approve is pushed forward by the exact same amount. This potential later-no-harm violation is tough to completely avoid because of how approval voting works. You can make expressions on multiple candidates and that information is calculated and considered simultaneously. Keep in mind that this criterion violation is only a potential and not a guaranteed outcome. Inadvertently hurting a candidate on your ballot can sound frustrating, but the upside is that approval voting still always lets you support your honest favorite candidate(s). That support for your favorite will never hurt you.

This is technically correct but I’m unable to take it seriously. Why does democracy exist? Why do I respect your opinion? Why don’t I just get a gun and kill you, and then assert my opinion as the only opinion that matters? The moral argument for democracy is that it delivers the greatest results for the greatest number of people, therefore I only care what system of voting produces the best results overall. I do not care if “approval voting still always lets you support your honest favorite candidate.” I don’t care who your honest favorite candidate is. I do care which candidates best express the concerns that matter to the largest aggregations of voters, and I’ll support any system of voting that helps bring such candidates forward.

Why does the current system fail?

As we read before, Achen & Bartels have suggested that voters vote first on the basis of identity, and only secondarily on issues of policy.

Democracy for Realists, 2016

Why Elections Do Not Produce Responsive Government

By Christopher H. Achen & Larry M. Bartels

Page 310-311

Well-informed citizens are likely to have more elaborate and internally consistent worldviews than inattentive people do, but that just reflects the fact that their rationalizations are better rehearsed. For example, as we saw in the case of the budget deficit, the political beliefs of more attentive, knowledgeable citizens are often more subject to partisan bias than those of their less attentive neighbors. For most people most of the time, social identities and partisan loyalties color political perceptions as well as political opinions.

The role of political “sophistication” in analyses of this sort underlines the fact that the task of being a good citizen by the standards of conventional democratic theory is too hard for everyone.

Some people read our previous essay and responded by arguing that voters “should” vote on the basis of policy, not identity, but Demodexio aims to design a stronger system of democracy that resists autocracy while helping society respond to its challenges, and doing this with real-world voters, not ideal voters. So we must assume that issues of identity will take precedence over issues of policy, at least for a substantial number of voters.

Achen and Bartels emphasize that identity is always the central concern of voters, always, in every election:

Page 336-340

The impact of social, religious, racial, and national identities was more obvious in 2016 than in most election years. So, too, was the voters’ relative lack of interest in the candidates’ policy proposals. Because our book summarized a good deal of prior research on both these points, and added more, some reviewers found us a helpful guide to the presidential contest. But our broader argument, that every election is more about social, ethnic, racial, occupational, religious, and partisan identities than about policy judgments, may have been overlooked in the rush to “explain” what happened in 2016. Witness the post-election calls in the Democratic Party for de-emphasizing “identity politics” - as if policy-based appeals could be persuasive in a social vacuum.

So, as we said in the previous essay, we favor approval voting because it allows voters to send a lot of important signals. The example we gave in the earlier essay was:

[[Edited to add: Achen and Bartels suggested that most people vote primarily based on obvious markers of identity, such as race and religion, so in opposition to that sad fact, this next example explores the possibility that a system of voting can discover what people’s real policy concerns are, even when the voters are still biased to first vote on the basis of race and religion.]]

Assume a voter who is a Catholic Hispanic male working a low-wage job, who perhaps wants a candidate with these attributes:

1. Hispanic

2. male

3. Catholic

4. supports a higher minimum wage

But instead there is a candidate who is:

1. Hispanic

2. male

3. Catholic

4. a businessman who supports a lower minimum wage

So the voter votes for this candidate because this candidate has 3 of the 4 attributes that the voter is looking for. One attribute is forfeited. If the voter could vote for 50 people a month, then the voter could find different candidates who separately have the attributes that the voter wants. That is:

1. one candidate might be a Hispanic

2. another candidate might be Catholic

3. another candidate might be male

4. another candidate might support a higher minimum wage

There is no longer the need to find every attribute in one candidate.

So what would be the best argument against approval voting?

The only interesting argument against “later-no-harm” would be if the system of voting actually sabotages the voter’s ability to communicate their knowledge to the overall system. That is, as we wrote previously, we hope every voter will cast many votes, each signaling a small part of the knowledge they hold, and all of their votes, combined with everyone else’s votes, will communicate something important to the highest levels of the political system. What if violations of “later-no-harm” mean that voters are unable send their most important signals? That would be a disaster, yes?

But how could an individual voter know which of their signals is their most important signal? That is precisely the thing they don’t know. They don’t know the truth of the overall system, so they can’t know which part of their knowledge is the most crucial knowledge from the point of view of the overall system. Their not knowing is (part of) why Kenneth Arrow was so skeptical of rank voting. Their not knowing is what makes Achen and Bartels occasionally sound cynical about all forms of human government.

Some might argue that score voting would be better than approval voting. As a practical matter, I think either system of voting would be a large improvement over most current systems of voting, and the differences between these systems would not be large in the real world. But having said that, the case for approval voting remains strong. After all, in a system of Machine Learning, we eventually want to weight the signals that survive each round of filtering, so to refine the difference between signal and noise. But if the individual knew how to weight their own vote, then rank voting would work, and that is precisely what Kenneth Arrow showed to be false — his Impossibility Theorem should leave us skeptical of the idea that the voter will know how to weight their own vote. Also, as a practical matter, Netflix has already run this experiment for us. To power their recommendation engine, they used to use a 5 star system. Then they switched to a simple up or down system. In other words, Netflix tried score voting, but then they eventually switched to approval voting, because people are surprisingly bad at weighting their own preferences in movies.

So, to put this as strongly as I can:

The alternative to democracy is civil war. I can go and get a gun and try to kill everyone who disagrees with me. So we need to clarify why I would care about a voter's honest favorite candidate. What if I disagree with that preference? Wouldn't it be simpler to kill the voter and then not have to worry about their honest favorite candidate? That is, what is the utilitarian case that would want me to honor the votes of others? The answer, based on all of the evidence we have regarding democracy, is that I will end up happier, wealthier, healthier and safer if I honor the outcome of a good system of voting. Over the long term, democracies tend to deliver higher economic growth, more safety and stability, less corruption, longer life spans, lower childhood mortality, less arbitrary action, and less incidents of national crisis. But if I follow the utilitarian argument, then I'm only interested in the best outcome of the best voting system. I do not have any interest, at all, in the "honest first preference" of any voter. If approval voting smothers a voters "honest favorite candidate" I could not care in the least. What I do care about is that the voter participates in the election, sends signals to the system via their votes, and thus helps improve outcomes for everyone.

So we shouldn’t care what the voter happens to think is their most important (most widespread) signal. We shouldn’t care who their “honest favorite candidate” is. We should simply ask voters to send a lot of signals (by voting for any candidate who strikes them as reasonably good) and we should let the higher levels of the system do the work of aggregating together everyone’s signals, to reveal the most urgent concerns that society is facing.