Commentary & Analysis

Was Ranked Choice Voting Really a Success in NYC?

As soon as the last votes were in, advocates for ranked-choice voting (RCV) were ready to celebrate the first New York City RCV election as a success. But was it?

If merely deciding a winner is a success for a voting method, then drawing names from a hat should do just as well. So we must think about what it means to be successful. The challenge with RCV is that it can be hard to know.

The most obvious question is, “Did RCV get the winner right?” The answer is, well, maybe. 

Eric Adams, despite that he openly opposed RCV leading up to the election, won the Democratic mayoral primary. Adams received just over 50% of the votes between the last two candidates and among the remaining ballots with active rankings. 

Of course, that percentage dips well below 50% when you consider the over 100,000 votes (about 15% of all voters) that were no longer being used by the last round. RCV tries in vain to find the majority winner in elections.

Unfortunately, that majority winner does not always exist and no voting method can change that.

Another way to look at whether the winner was right is to see if the winner matches the person who could beat every other candidate head-to-head (called the Condorcet winner). Without the raw ballot data, we have no idea who that person is.

That said, Citizen Data conducted a poll using ranked-choice voting that suggests that Adams would beat all the other candidates head-to-head. The challenge with that is that it also limited respondents to five candidates—just like the New York City ballot.

Unfortunately, when voters are forced to limit their rankings either through a truncated ballot like New York City’s version or fatigue from ranking, this can hide who the real head-to-head winner is.

Even if we had the raw ballot data, it could give us the wrong answer—particularly in a close election. This has independently been shown experimentally (I’m one of the authors).

It’s also worth being cautious about listening to RCV advocacy groups on whether the election was successful. For example, a 2009 RCV election in Burlington, Vermont, failed in every way imaginable.

Firstly, it incorrectly eliminated the clear head-to-head winner among the candidates. Worse, some voters who ranked their favorite candidate as first got a worse outcome overall. In yet another twist, if that election’s winner had actually gotten more votes from a particular subset of voters, it would have caused them to lose instead.

This just goes to show that RCV can behave bizarrely. RCV advocates, despite all this, declared that election a “great success.” So if you find yourself skeptical of claims from RCV advocates, you’re probably justified.

This isn’t a call to get rid of RCV in New York City or to do a traditional choose-one election with a runoff. Indeed, traditional runoffs can behave just as oddly as RCV.

A 1991 gubernatorial election in Louisiana denied the candidate who could beat everyone head-to-head. Instead, that candidate, Buddy Roemer, was eliminated and the race came down between an openly corrupt politician, Edwin Edwards, and a klansman, David Duke. The openly corrupt politician won and would later go to prison for felonies committed while in office. Because the race was between three candidates, we know that RCV would have given the exact same result.

At the very least, this history should be a cautionary tale for other cities and states who are considering changing up their elections. It’s clear that complicated is not the same as better.

Fortunately for other cities and states, there is an alternative that is both easier and better than RCV. And that method was recently used in two very different places—Fargo, ND. and St. Louis, MO. Those cities now allow voters to select—not rank—as many candidates as they want. Imagine the equivalent of giving each candidate an up or down vote and you get the idea. This is called approval voting, and it is spreading quickly.

Approval voting is nice because it’s so simple.

The ballot is the same. You’re still just adding votes. And the winner is still the one with the most votes. You’re just letting voters select all the candidates they want. Now, that’s not to say the New York City Board of Elections couldn’t mess up an approval voting election as well. But it would be an even more embarrassing bungle if they did.

Approval voting is also more predictable.

You’re not forced to wait half a dozen rounds to see if the winner changes in some startling way. As for electing strong winners, approval voting does an excellent job at selecting the appealing candidate who can beat everyone head-to-head and consensus candidates in general. But it doesn’t randomly punish your vote in ways that RCV does.

Both elections in Fargo and St. Louis were actually uneventful.

Voters chose the candidates they wanted without splitting their vote, and all the candidates got the clear support they deserved. The focus was on the issues and candidates themselves. That’s how seamless approval voting was. It did the job without needing complexity to draw attention to itself.

It’s a simple solution waiting to be picked up by cities and states across the country.