Next week voters in New York City will go to the polls to select the Democratic candidate for mayor using the ranked choice voting (RCV) system. Using examples of RCV’s recent performance in Maine, Brian J.Gaines writes that while it is often touted as being more democratic, it is not a guaranteed fix for problems like low turnout, a lack of candidates, or polarization.
New Jersey and New York have primaries two weeks apart this year. They are separated as well, by electoral rules. US Representative Mikie Sherrill will be the Democratic nominee for governor in the Garden State this fall. On June 10, she topped a field of six. The June 24 Democratic primary for mayor of the Big Apple is also crowded, but the race will be held under “ranked choice voting” (RCV). Voters will be permitted to rank as many of the candidates as they like, and the vote tabulation will proceed in rounds, with last-placed candidates successively eliminated, and the votes on those ballots then transferred to whoever is next in the rankings (if there is anyone). Eventually, the winner will be whoever has a majority in the final count.
Is Ranked Choice Voting more democratic?
In advance of the New Jersey vote, some worried that Democratic primary voters would not be able to handle the “mess” of six candidates vying for their party’s gubernatorial nomination, with no clear front-runner in the polls. How could ordinary people know which candidate to back? The campaign, moreover, was “brutal” and “contentious.” RCV was allegedly a solution to these woes.
Fans of democracy might welcome a crowded field and lively debate, without worrying that voters will somehow be overwhelmed by too much choice. Why should voters fret about polls and horse-race coverage anyway? Each individual voter’s odds of making or breaking a tie are minuscule, so simply picking a favorite is sensible. RCV’s solution to the alleged problem is to give voters a more demanding task. Instead of picking one from many, voters are asked to rank all the contenders, which requires more information, and makes greater demands on voters’ scarce time and attention. If the rankings are part of a transferable-vote system, the winner will, so the argument goes, eventually, be able to claim majority support.
So what? Backers of RCV engage in a bit of sleight of hand, taking their own rhetoric too seriously. “A beat B with over 50 percent of the vote in the two-way comparison of round 5” is not the same as “A enjoys the full support of more than half of the electorate in this race against B, C, D, E, and F.” The process of ranking, eliminating, and re-counting does, automatically, create a majority vote-share at some point. But this is a manufactured, highly conditional majority, and it does not ensure, for instance, that backers of E and F, who viewed A as the second-worst candidate, better only than godawful B, will rejoice over A’s eventual win.
Some surveys show that voters in places that have already adopted RCV like the system. But how much actual experience do they really have with its operation? RCV is irrelevant if there are not more than two candidates, and the rankings are ignored in a multi-candidate race in which the top vote getter obtains a majority. In the New Jersey Republican race, former state Assembly member Jack Ciattarelli romped over four rivals, with 68 percent of the vote. Had RCV been in effect, the rankings would have been ignored.
RCV in Maine
Consider Maine, an RCV pioneer. In 2024, with 151 House and 35 Senate seats up, there were 372 major-party primaries for the state legislature. There were no more than two candidates in 370 of these, while only two of the contests were three-way. In both of those races (Democratic primaries for the 118th and 123rd House seats), the winner took a majority of the vote on the first count, so the rankings were irrelevant. In 2024, Maine voters used RCV even for the presidential race…sort of. There were only two candidates in the Democratic primary, as Joe Biden walloped the then Minnesota member of Congress, Dean Phillips. On the Republican side, Donald Trump creamed four others, making the use of RCV invisible (the state GOP had previously stated that they would not recognize the result if went past the first round of voting). Unusually, Maine awards some presidential electors according to vote totals in US House districts. In the general election, the Harris-Walz ticket exceeded 50 percent in the first congressional district and statewide, thus taking three electors, while Trump-Vance won more than 50 percent in the second congressional district and thereby secured one elector. No vote transfers were required, and rankings were, again, ignored.
“Zohran for Mayor” (CC BY 2.0) by edenpictures
The race for the US House seat in that 2nd district was a nail biter, and it is the lone contest for which the Secretary of State provided rankings. But even there, no vote transfers took place: Democrat Jared Golden edged out Republican Austin Theriault, with a smidgen under 50 percent of the vote until write-ins were discarded as “invalid,” at which point the official result became 50.35 percent – 49.65 percent. With only two real candidates in the race, if voters enjoyed filling out RCV ballots as Golden-Golden-Golden rather than Golden-blank-blank, who is to begrudge them their fun? But RCV is a distraction without many candidates, and too few candidates is what ails American electoral democracy much more than too many. So far, Maine does not prove that changing the way people vote brings aspiring political candidates out of the woodwork.
How RCV can make losers out of winners
An oddity with RCV, well-understood by specialists, but not often discussed in public debates, is that it is non-monotonic. That jargon means that getting higher rankings can convert a winner into a loser. Examples in academic articles or textbooks abound—the key is that not all the second, third, etc. preferences are used in tabulation. Which preferences are used, and which are ignored can matter a great deal. Are there many examples of perverse outcomes, in which being persuasive and popular harmed a candidate? Most jurisdictions that employ RCV do not release the data that answers that question. But odd and disturbing reversals of this kind are more likely in races with many candidates of roughly equal strength, like the Democratic contest in New Jersey and potentially the race for the Democratic mayoral nominee in New York city.
Whether disgraced former New York governor Andrew Cuomo passes the first test of his comeback attempt, to gain the country’s most visible Democratic mayoral nomination, may depend on how seriously voters take informal alliances between some of his rivals, who are offering suggestions for rankings after the top spot. Is that sort of elite-level scheming somehow better for democracy than heated argument?
Political scientists love experimentation with electoral rules, so that they can obtain more data on how various systems work, rather than in mathematical simulations. But beware those who promise that ranked choice or any other innovation tagged a “reform” is sure to solve some real problem, whether low turnout, uncontestedness, negativity, or polarization. There is no perfect voting system; they all have pros and cons, benefits and costs.
- Subscribe to LSE USAPP’s email newsletter to receive a weekly article roundup.
- Please read our comments policy before commenting.
- Note: This article gives the views of the author, and not the position of USAPP – American Politics and Policy, nor the London School of Economics.
