Plurality Voting is Unconstitutional

Epistemic status: certain of nothing. I’m not a legal scholar, but my training in philosophy has prepared me for legalese: I believe there might be a case here? Something worth exploring? Feedback from those with legal expertise would be greatly appreciated. UPDATE: I’m ~90% certain the section on rational basis review is wrong – feel free to skip. Also, I shouldn’t have put the section on ballot design vs voting method at the end, because I think it’s one of the more interesting ideas here.

As Aaron Hamlin of the Center for Election Science has argued, improving our voting methods is a prime target for EA. Therefore, if anyone with the requisite status, money and connections thinks this case actually has a chance of succeeding, by all means, go for it. Maybe purposefully spoil your ballot in the midterms (voting for more than one candidate) and then file suit against the state?


The U.S. uses the least expressive voting method conceivable: doesn’t that violate voters’ freedom of speech? That’s what I will argue. Just because our voting method has never been tested in court doesn’t mean it shouldn’t be.

Background

This method is known as “plurality voting” in the literature (a.k.a. “first-past-the-post”, or “choose-one”) and we use it for nearly every state and federal election, supplemented by a run-off here and there. It’s a similar story at the municipal level. The two exceptions: Maine and Alaska which each use a form of Ranked Choice Voting (RCV).

Under plurality voting, voters may express approval for only one candidate. It grants no opportunity to rank candidates, score them, or even just approve of more than one. Any of these alternatives (RCV, range voting, and approval voting, respectively) would allow voters to express more about their preferences. And not in a small way: as a matter of raw math, the number of different ways a voter could express themselves under any of these alternative methods increases at polynomial or exponential rates as the number of candidates increases.

A table that compares the number ballot permutations possible under various methods, and as the number of candidates increases.
Credit: Aaron Hamlin, Center for Election Science. More details on the formulas here.

Meanwhile, that number only increases linearly under plurality voting. In fact, barring forced voting for at least one candidate, our “choose one or none” style ballot is as restrictive as it gets.

Political scientists have also identified mechanisms by which plurality voting favors a two-party system. In that sense, plurality voting is anti-competitive: by favoring duopolies, it stifles competition and the diversity of options in the political marketplace. If freedom of speech is our antitrust for the marketplace of ideas, and the marketplace of ideas shares space with the political marketplace, then our ballot is exactly where that freedom should be most vigorously defended.

Furthermore, plurality voting is the voting method most prone to vote splitting, in which like-minded voters unintentionally dilute their vote or otherwise harm their political interests by splitting their vote across similar candidates. In effect, plurality voting imposes artificial coordination problems on like-minded voters, making it much harder for them to “associate for the advancement of political beliefs and the right of qualified voters, regardless of their political persuasion, to cast their votes effectively.”[1] Thus, plurality voting might also infringe on First Amendment associational rights.

Precedent

First, a constitutional refresher. Article 1, §4 stipulates:

The Times, Places and Manner of holding Elections for Senators and Representatives, shall be prescribed in each State by the Legislature thereof…

That certainly gives state legislatures quite a lot of leeway. They are not above scrutiny however. The Court has struck down voting laws where they judged the state was infringing on citizens’ constitutional rights. The most famous cases of this revolve around the “one person, one vote” principle which was interpreted in Wesberry v. Sanders as “the weight and worth of the citizens’ votes, as nearly as is practicable, must be the same.” For example, in Baker v. Carr, the court ruled that certain redistricting practices violated the Equal Protection Clause of the Fourteenth Amendment, because they gave the votes of some citizens disproportionate weight — making their vote stronger or weaker than others. (As an aside, no, alternative voting methods do not violate this “one person, one vote” principle).

I’ve been suggesting, however, that plurality voting violates First Amendment rights, notably the right to free speech and associational rights. Have voting laws been struck down for violating these rights? Indeed they have: in an oft overlooked case, Anderson v. Celbrezze, the court ruled that discriminatory regulations placed undue burdens on independent candidates and their supporters, violating “the right of individuals to associate for the advancement of political beliefs and the right of qualified voters, regardless of their political persuasion, to cast their votes effectively”. More generally, in Doe v. Reed, the majority opinion states “the expression of a political view implicates a First Amendment right.” What could be a clearer case of political expression than one’s vote?

So there is precedent that might bolster the case against plurality voting, but that doesn’t mean there aren’t obstacles. To begin with, as civil rights attorneys know only too well, voting has never been fully recognized by the court as a form of speech (or petitioning) that would receive full protection under the First Amendment. As such, laws that regulate access to voting (e.g. voter ID laws) are not subject to the strict scrutiny applied to laws that directly limit speech. Instead the Court takes a much more deferential stance towards voting laws, generally putting the states’ interests above the voters’. This is why virtually all recent voting regulations have been upheld by the Court.[2]

The present case is different to these voting rights cases in one key way however: this is not about who has access to the ballot, but to what degree voters are allowed to express themselves on the ballot. Jurisprudence involving laws that regulate access does not therefore settle the matter: the case against plurality voting is not just another voting rights case, dead on arrival.

Tests

The Anderson test

Following Anderson, voting laws must pass a test that balances voters’ interests against the state’s interests in regulating elections to ensure elections are “fair and orderly.”[3]

The fact of the matter is, alternative voting methods exist (namely, approval voting) that are just as transparent (easily audited), cost the same, and even tend to decrease the number of ballots spoiled (i.e. invalidated for clerical reasons, such as voters failing to follow instructions properly). In other words, they are just as fair and orderly, if not more so. This means that significant gains in the voters’ interests can be had without compromising the state’s interests at all.

As far as I can tell, a ban on plurality voting concretely translates into, at the very least, the inclusion of two additional words on non-compliant ballots such that they read “vote for one or more.” Again, no requirement for new voting machines, new run-offs to organize, or any other new expensive administrative procedures. At a minimum, states must spend a few more ounces of ink. That cannot outweigh an exponential increase in voters’ freedom of expression, and an effective increase in their ability to associate for the advancement of political beliefs.

Whether this line of argument is enough to pass the Anderson balancing test remains to be seen. Over time, the test has slowly come to resemble the very permissive rational basis review which places the full burden of proof on those pressing charges against the state.[4] States need only provide a rational relation, actual or hypothetical, between the law in question and how it advances a legitimate state interest.

Anderson cannot collapse into a rational basis review: there is a clear implication of fundamental First Amendment rights in voting that demands at least an intermediate level of scrutiny. However, even if it did so collapse, plurality voting still might fail to pass it.

Rational basis review

According to Derfner and Herbert, under a rational basis review “[t]he question whether there are less restrictive means of achieving the state’s purpose is irrelevant.”[5] While this may be true in some sense, it is impossible that there be no comparison in play. States must show how a legitimate state interest i is advanced by a law l. The key predicate “is advanced by” is a three place predicate, putting in relation: an interest i, a scenario in which law l is in force, and (implicitly) a set of alternative scenarios in which law l is not in force. Just as no one can be said to be taller or smarter simpliciter, an interest cannot be said to be advanced by a law simpliciter, i.e. without there being some understanding of what other state(s) of affairs it is being advanced relative to.

The other state(s) of affairs could simply be the absence of the law in question, without replacement: this would be to evaluate the law’s effect ceteris paribus. In the cases cited by Derfner and Herbert, this seems to have been the alternative in play (whether conscious or not, I do not know). But in any case, there is always a set of alternative scenarios at play. There must be an answer to question “furthered relative to what?” or else our judgment of whether or not the interest is furthered by the law is confused.

So the question becomes, what is the relevant set of alternatives? As far as I know, the Court has never outlined a procedure for determining this set. In the present case however, I argue the relevant laws cannot be judged strictly ceteris paribus. The alternative in play cannot be the total absence of any ballot design whatsoever: one design must be replaced with another. It would be a grossly unfair characterization of our case to say we want blank pages for ballots. Instead, we are asking for a minimum standard: states cannot use the most restrictive ballot design there is. Not only does anything else go, but something else must go in its place.

Presumably then, the relevant set of alternatives is of all scenarios in which readily available alternative voting methods are implemented. Relative to this set of alternatives, we can demonstrate there are ballot designs which yield elections that are just as fair and orderly, at no additional cost or complication: this state interest is not furthered by the use of plurality voting.

Perhaps another legitimate state interest could be found which is only advanced by the use of plurality voting specifically. But this is doubtful. Thus, plurality voting might fail to pass even a rational basis review.

Restoring public trust

In closing, we should note that freedom of expression on the ballot is not just a matter of individual liberty, but also of keeping our democracy true. It’s no secret that public trust in American institutions has been in steady decline for several decades, even hitting a record low this year. Part of the blame likely falls on our dysfunctional voting method: both theoretical and empirical results indicate that plurality voting has a poor record of choosing winners that maximize voter happiness compared to other methods. For this and other reasons, plurality voting has drawn unanimous disapproval from voting theorists.

Unlike elections in private associations (from which we can withdraw membership, taking our business elsewhere), citizens are forced to adhere by the results of government run elections. And what, after all, legitimizes them? Their transparency, impartiality, clear-cut results, and immunity to tampering, certainly. But more fundamental than any of these, is their ability to accurately record the will of the people — that will which bestows legitimacy on our elected government in the first place. All other qualities are for naught if the elections fail to fulfill this basic function. But how can they succeed when the people’s ability to express themselves on the ballot is so severely and arbitrarily abridged?

Further details

Are run-off elections enough?

Plurality voting supplemented by run-offs is equivalent to a form of RCV in which voters rank their top two choices. (Indeed RCV’s other name is Instant Run-off Voting). Plurality voting supplemented by run-offs could therefore be spared, judged to pass the minimum standard we seek.

Are we trading freedoms?

Alternative voting methods need not force a voter to be more expressive on their ballot. In each there is a simple way of casting one’s vote that is identical to the “choose-one” paradigm of plurality voting. In and of themselves, alternative voting methods only extend voter freedom, allowing them to express more of their opinion on more candidates, should they wish to do so. Of course states are free to implement laws and ballot designs that do force expression — but such forcing is not inherent to any alternative voting method I know of.

Why now?

The Founders were aware that various voting methods existed, and surely understood that determining which was best was no trivial matter. With a spirit of compromise and experimentation, they left it to states to choose which method to use. Unfortunately, we have seen very little of that experimentation hoped for. (Notable exceptions: Maine and Alaska’s recent implementation of RCV, as well as Illinois’ use of cumulative voting from 1870 to 1980). Arguably, this stagnation is due at least in part to plurality voting’s aforementioned anti-competitive effects. Favorable to duopolies, plurality voting is a friend to major parties in power; voting reform through the legislative process becomes an exercise in bootstrapping; plurality voting gets locked-in. Hence this request for the Court to step in.

But these are not solid legal grounds. The real question is: was plurality voting always unconstitutional? What changed? Not being a legal scholar, I don’t know how to answer the first question, but to answer the second: what changed is the evidence. We now understand voting methods in much greater detail — their advantages and disadvantages, and their comparative expressiveness. Where before the gains for voters relative to the costs for states were unclear, they are now very clear, and clearly lean in favor of voters.

Laboratories of Democracy

The outcome sought would not force states to adopt any particular voting method: it would merely prohibit the use of the least expressive voting method that exists. A minimum standard should be established — no more. States should be free to experiment with any other ballot design: the point isn’t to smother our laboratories of democracy, but to catalyze them where they have calcified.

This raises a question: have we settled on plurality voting because it is perhaps simply the best? Has our experiment simply run its course? Given that the majority of other Western democracies have moved away from plurality voting, I find this explanation hard to believe. To my mind, the much more plausible explanation is some combination of: our ballot design becoming invisible to us, something we don’t think twice about; regulators being reluctant to tinker with the system that got them elected; the major party duopoly having no incentive to reform the ballot (they have everything to lose and little to gain from making room for third party or independent candidates).

Why stop there?

The standard sought is principled in that it is the weakest possible: it merely bans the use of the most restrictive ballot design that exists. Any requirements more specific than this quickly become onerous and unwarranted, placing costs and limits on states that cannot be justified by the more marginal gains in voter freedom which might be had by implementing one alternative ballot design instead of another. Besides, experts disagree over which alternative is best — there is only consensus on the poorness of plurality voting.

What about the Electoral College?

All alternative voting methods are compatible with the Electoral College, and this minimal standard for ballot design does not tie the hands of the states regarding how they appoint their Electors. They still could, for example, revert to having state assemblies choose their Electors, instead of relying on popular vote. All that is demanded is that whenever and however states rely on a popular vote, the ballots used must guarantee a minimum degree of freedom of expression. Any less is inexcusable.

More generally, we could potentially leave states free to decide how they use the data collected by the vote to determine winners of their elections. But at a minimum states can’t place arbitrary limits on how much data voters can provide about their preferences.

Ballot design or voting method?

This last comment raises a question some may have been wondering about. I have been using “ballot design” and “voting method” interchangeably, when in fact the are very different, and suggest two very different possible rulings. Let “ballot design” refer to the layout of physical ballots, and how much voter preference data they are able to accurately record. I’ll understand “voting method” as the combination of a ballot design and a tabulation method, an algorithm that takes voter preference data as input and yields a winning candidate as output. While there is a natural correspondence between certain tabulation methods and certain ballot designs, some recombination is possible. So for example, we might conceivably deploy a ranked choice ballot design (“Rank your top 3 choices”) while still only using a simple plurality tabulation method (count first choices → candidate with most first choice votes wins).

The ideal outcome of the case would be a ban on plurality voting as a voting method, which is to say a ban on the combination of a “choose one” ballot design and the simple plurality tabulation method. However, if not this, then at the very least “choose one” ballot designs must go: ballots must allow voters to enter more information about their preferences into the public record. Tabulation could be spared any restrictions.

The latter outcome, though better than nothing, is far from ideal for at least two reasons. First because it permits the miscarriage of democracy: the will of the people could be satisfactorily recorded, yet only an arbitrarily amputated interpretation of it would be respected. Second, because it would likely mislead voters. If the ballot design does not make crystal clear how the voters’ preferences will actually be used, voters could very well be tricked into voting against their interests. (For example, by tricking them into voting their honest favorite first — even when doing so is not rational, as is often the case with plurality voting).

That said, the Court may prefer this solution for cleaving closer to Article 1, §4 while still remedying the infringement on voters’ freedom of expression. (It would not, however, totally remedy the infringement on their associational rights: vote-splitting would still pose an artificial coordination problem). Just having that extra data entered into the public record could do wonders for the political marketplace. Third party and independent candidates would finally get an accurate measure of support, giving them a chance at the podium and spurring healthy competition. In the long run, I think public trust in our elections would be helped as well. Maintaining a simple plurality tabulation method would likely become politically perilous: media outlets would no doubt run alternative tabulation methods on the public data, making plain to all how their will was being truncated on a government whim. This could provide the impetus and leverage voters need in order to pressure their government for true reform.

  1. ^

    Anderson, 460 U.S. at 787 (quoting Williams v. Rhodes, 393 U.S. 23, 30–31 (1968)).

  2. ^

    Armand Derfner & J. Gerald Hebert, Voting Is Speech, 34 Yale L. & Pol’y Rev. 471, §III (2016).

  3. ^

    Id at 480.

  4. ^

    Id at §III.

  5. ^

    Id at 477.

Crossposted from LessWrong (6 points, 20 comments)