Clinton and Trump are both deeply disliked — but one of them will be president. With a smarter (and more democratic) election format, we could elect more satisfying politicians. An initiative on the ballot in Maine this November would bring such a format to that state. I will describe the problem with our current format, as I see it, and argue that ranked-choice voting is a superior and viable alternative. Finally, I will invite you to comment.

We Don’t Always Get The President We Really Want

In U.S. presidential elections we vote for our top choice only — whether the field includes two candidates, three, or everyone with an iPhone 6.

The results can be illogical and anti-democratic. Think about the 2000 election, where Bush ultimately beat Gore won by 537 votes in Florida. Spoiler Ralph Nader received 97,000 votes in Florida, and polls showed Nader voters preferred Gore to Bush by two to one. Hanging chads aside, is there any doubt that more Florida voters preferred Gore to Bush?

When you order at a restaurant and the kitchen has run out of your order, the waiter doesn’t come back and say, “We’re out of that, you don’t get a second choice, come back in four years and make a better order.” But that is the experience of a third-party voter.

Given the existing system of voting for a single candidate, supporters of non-major party candidates face two bad choices:

1. Tactical Voting: You predict which two candidates are likeliest to win and vote for whichever one you can stomach. This has the distorting effect of signaling more support for the major party platforms than really exists, and it starves marginal parties of votes that would signal viability in future elections.
2. Vote your true preference and risk a bad outcome: In the absence of tactical voting, a candidate who is supported by a dedicated plurality, but strongly disliked by the majority, can win the White House.

A Better System (Too Hard in the 1700s, Easy Now)

Ranked-choice voting, as the name suggests, would allow voters to rank-order the candidates. This way, voters make their top preference known without losing their say if the final choice comes down to two other candidates. Suppose the 2016 U.S. presidential election is between three candidates: Hillary Clinton (Democrat), Donald Trump (Republican), and an independent candidate known as Mo the Wise and Ethical. Nearby is an example ballot of a voter whose top choice is Mo the Wise and Ethical, but if forced to choose between Clinton and Trump, would choose Clinton.

There are several ways to determine the winner of a ranked-choice election. An elegantly simple method that satisfies more desirable criteria than any other method is called the Schulze or Condorcet method: you simply evaluate hypothetical head-to-head races between each pair of candidates. There will usually be exactly one candidate who beats all others head-to-head, and this candidate must be the winner of the election. In the rare event that there is no such winner, some tie-breaking method would be needed. (For example, the ballot above would count as a vote for Mo in any head-to-head race involving Mo, but as a vote for Clinton in a Clinton-Trump runoff.)

Who would win between Clinton, Trump, and Mo? Suppose the election results come out as in the table below. (To keep things simple, I’m just talking about the popular vote and ignoring the electoral college, although this same process is what would occur within each state.)

I will assume that the Clinton voters (that is, the 40% of voters for whom Clinton was their first choice) would prefer anyone to Donald Trump, so all Clinton voters select Mo the Wise and Ethical as their second choice. Conversely, every Trump voter also makes Mo their second choice. Mo, the first choice on just 25% of the ballots, is thus second on the remaining 75%. I will further assume that the Mo voters are evenly divided between Hillary and Trump as to their second preference. Hillary, therefore, is the second choice on 12.5% of the ballots and third on the remaining 47.5% of ballots. Trump, likewise, is second on 12.5% of the ballots, and last on 52.5% (all Hillary voters plus half of the Mo voters).

Election Results:

 

 

 

We can generate three hypothetical head-to-head contests from these results:

• Clinton defeats Trump by 5 points, 52.5% to 47.5% (Clinton outranks Trump on 52.5% of the ballots — the 40% who picked her first and the half of the Mo voters who ranked her second).
• Mo defeats Trump by 30 points, 65% to 35% (All Clinton and Mo voters prefer Mo to Trump).
• Mo defeats Clinton by 20 points, 60% to 40% (All Trump and Mo voters prefer Mo to Clinton).

Mo should thus be president. Voters prefer Mo to Clinton and to Trump.

FairVote and Brookings Institution have made additional arguments for a form of ranked-choice voting called instant-runoff voting. I will only point out one more satisfying aspect of ranked-choice voting: there can be no more direct way to “throw the bums out” of office than to rank the offending party dead last on the ballot.

 Where Ranked-Choice Voting is Already Used

Instant-runoff voting (IRV) is used in many national elections (prominently in Australia, India, and Ireland) and in cities within the United States (Oakland, San Francisco, Minneapolis, and St. Paul). It’s widely used in the UK to decide party leaders, and it’s good enough to decide the Academy Award for Best Picture.

Drawbacks of Ranked-Choice Voting

No voting system satisfies every criteria that might be deemed desirable. This has been painstakingly proven in academic papers (See Arrow’s Impossibility Theorem or the Gibbard-Satterthwaite theorem).

As mentioned previously, the head-to-head method I laid out (a.k.a. the Condorcet or Schulze method) satisfies more criteria than any other method.

Other disadvantages of ranked-choice voting are:

• It requires some degree of sophistication on the part of voters, and
• It requires more computing power than our existing format, and will not always change the election results.

Is Ranked-Choice Voting Constitutional?

The courts say yes, although to be fair, if it weren’t constitutional, I would simply argue to pass an amendment. A very quick aside: to say that unconstitutionality in and of itself is a reason never to do something that we otherwise believe is right, is to say that we must keep doing what we previously thought we must do, merely because we previously thought we must do it.

But anyway, ranked-choice voting has been endorsed by the courts. In Stephenson v. the Ann Arbor Board of City Canvassers, a Michigan Court held that “majority preferential voting” (a form of ranked-choice voting) complies with the Michigan and United States constitutions:

“Under the [Preferential Voting System], no one person or voter has more than one effective vote for one office … The form of majority preferential voting employed in the City of Ann Arbor’s election of its Mayor does not violate the one-man, one-vote mandate nor does it deprive anyone of equal protection rights under the Michigan or United States Constitutions.”

In a concurring opinion in Williams v. Rhodes, U.S. Supreme Court Justice Harlan also suggested instant runoff voting could help decide elections.

What Do You Think?

Is ranked-choice voting a good idea? Could it happen? What have I failed to consider? Please add your voice below and invite others to the conversation who would be interested.

UPDATE TO ORIGINAL POST  (Updated July 9):

One reader emailed to point out that I am implicitly assuming that the campaign strategies and voter turnout would be the same in a ranked-choice election as in our current format. He argued that politicians would change strategies if the voting format changed to ranked-choice. As one example, under a ranked-choice format Hillary Clinton and the Democrats might encourage Bernie Sanders to run as an independent. If we assume (a) Bernie has a passionate support base that might not bother voting if Bernie isn’t the ballot, and (b) if Bernie is on the ballot as an independent, his supporters will show up to vote, and while they’re at it, they’ll choose Clinton as their second choice, then it makes sense for the Democrats to encourage Bernie to run as an independent.

I am very persuaded by this argument, and agree that campaign strategies would change if the election format changed.

I think the best policy response to all this would be to make voting so easy that everyone with any interest at all in the outcome of the election will vote. This would make get-out-the-vote efforts pointless, and we could find better uses for the time and resources that we once poured into such shenanigans.