Ranked Choice Voting for Everyone?
It's not perfect but it beats our usual approach to picking winners and losers.
Its use in the Democratic primary for New York City mayor was the first large-scale use of ranked-choice voting in the United States. (It was also used in last year’s contest between Susan Collins and Sarah Gideon for the former’s Senate seat in Maine but that’s a much smaller electorate.) Not surprisingly, its use has been rather controversial.
As noted earlier this week, presumptive winner Eric Adams and his surrogates have been railing against it for weeks, arguing that the alliance between Kathryn Garcia and Andrew Yang was an attempt to suppress the Black vote. While that’s a silly argument, it’s true that the reallocation of votes from eliminated candidates moved Garcia from a distant third to very close second.
The legendary (but undeniably conservative) Harvard political scientist Harvey Mansfield argues that “Ranked-Choice Voting Is Bad for Everyone.”
Ranked-choice voting makes the common good inferior to each person’s private first choice. The common good of the country typically gets ranked second choice or below for each citizen.
Ranked-choice suffuses the spirit of systems where multiple parties vie to build coalitions after votes are cast. In the U.S., parties aspire to gain majorities through the voting rather than by secret, chancy negotiation afterward. Each looks for ways to bring together difficult companions: Republicans must reconcile the interests of evangelical Christians and libertarians; Democrats must balance the desires of progressives and moderates. Ranked-choice voting splits these coalitions and requires the pieces be brought together after the election, and not by voters. In presidential elections, the Electoral College produces a majority intended to recognize the importance of the states, one that sometimes differs from a popular majority. But in any case it is a coalition majority.
Voters in a coalition are reminded that most of them didn’t get what they wanted but avoided what they most disliked. Though this is often true, it should not be the goal. The goal should be a first choice willed as a compromise rather than a first choice abandoned for a compromise. This seemingly slender distinction makes a big difference in common trust and the way Americans think politically.
It is often thought that the sole purpose of an election is to make government accountable to the people and representative of their will. A second fault of ranked-choice voting is that it aims to perfect this idea by offering some success to as many shades of opinion as possible. But another, greater purpose for elections was intended by the Constitution’s framers: to find competent governors.
A good result from an election relies on its accuracy in representing the people’s will. The voters should get not only what they want, like consumers, but also what is good for them as citizens. Competency among those elected is never assured, but it will stand a better chance when voters are selecting the candidate who is best for the job as well as the one most representative of their will. Ranked-choice voting may bring competency along with accuracy, but accuracy is all that is asked for. Again, elections are held not to buy products but to put people in office. Competency ought to include the ability to unite the country in a majority, not just to make pleasing speeches to the voters known as “the base.”
A third fault: Ranked-choice voting rewards extremism in the electorate. Voters who make extreme choices should be punished via exclusion from the majority. Ranked-choice voting rescues them from the penalty they deserve for throwing away their ballot on an extreme first choice. One suspects that progressives like ranked-choice voting because it would allow them to vote twice: once for Bernie Sanders and once for Joe Biden.
The filmmaker and critic Noah Millman likewise argues that “The capriciousness of ranked choice voting is revealed in NYC.”
[T]he mere fact that a clear Adams plurality on Election Day is going to end as a squeaker should raise eyebrows. That’s particularly the case since it’s only a squeaker at all because of the order in which candidates dropped out. Until Yang’s votes were allocated, Maya Wiley was clearly in second place. His votes put Garcia ahead of Wiley — but only by half a percent. If Wiley had held on to second place, and Garcia’s votes had been reallocated instead, far more of them would have gone to Adams than was the case with Wiley’s. In the unlikely event, then, that Wiley wins more absentee ballots than Garcia, and comes in second in the penultimate tally, the perverse result will be to throw the election decisively to Adams — even though Wiley’s voters overwhelmingly prefer Garcia to Adams.
Ranked-choice voting is supposed to result in a consensus winner that reflects a clear majority preference. The post-election twists and turns of this race should prompt a serious rethinking of that assumption.
Meanwhile, the Washington Post Editorial Board contends “Ranked-choice voting worked in New York. More places should use it.”
Ranked-choice voting encouraged candidates to seek compromise and identify areas of common ground. This enabled voters to identify the candidates who represented the section of the political spectrum or who prioritized the issues that appealed to them.
When voters did so, the ranked-choice system came through on its most important promise: benefiting candidates who are broadly acceptable by weeding out niche candidates who would have fractured the field in a traditional vote. Mr. Adams led after the first and final tallies. But in the final round, his opposition turned out not to be left-wing champion Maya Wiley, who came in second in the initial count, but Kathryn Garcia, a less ideologically hard-edge candidate. Had Ms. Wiley prevailed in the initial tally, which was a plausible result in a fractured field of candidates, ranked-choice voting almost certainly would have saved Mr. Adams or propelled Ms. Garcia into the lead over the less broadly popular Ms. Wiley.
In other words, the system provided much more information about what voters wanted. It turns out that lefty progressives are a substantial but minority group in New York. Ranked-choice voting makes it harder for candidates with a fervent but narrow base of support to eke out a victory.
Dave Schuler makes it a policy not to comment on other municipalities’ political choices but wishes his longtime home of Chicago were aboard the RCV train:
What would have happened in the Chicago mayoral election if ranked-choice voting had been in place? A plurality of black voters supported the most conservative candidate running. The present incumbent found her base of support among white ethnic voters on the Northwest Side of Chicago. You cannot convince me that the Northwest Side is a progressive stronghold. I’m confident that Lori Lightfoot’s pledge not to raise property taxes was dispositive. In the run-off one of the two contending candidates (Toni Preckwinkle, the “establishment” candidate) failed to carry any wards including her home ward. I suspect that what would have happened would have been that we would have selected a much less “woke” candidate than Lori Lightfoot has proven herself to be. Given her track record as mayor that would have been a good thing.
Of these positions, I find Mansfield’s arguments least persuasive. Not only, as Schuler points out, did they produce the opposite of Mansfield’s predictions in the NYC race but the notion that voters whose preferences are “extreme” should be punished makes no sense unless we’re talking about those who advocate violence or the overthrow of the system.
But, while I ultimately disagree with him, Millman makes a good point: the order of reallocation in the former of RCV in play here, instant-runoff voting, can sometimes produce perverse results.
In most American elections, we tend to think of there being two major candidates with perhaps one or two “third-party” spoilers. So, you might have something like this outcome:
In this case, roughly modeled on the 2000 Presidential election, the “right” outcome would have occurred. The Libertarian candidate is eliminated first and the second-place preferences of their voters are reallocated. I assigned the 2% equally to the Democrat and Republican, which is probably about right. Regardless, unless all of the Libertarians preferred the Democrat, it wouldn’t have been enough to create a majority winner in the second round. Thus, we move on to reallocating the Green votes to the second preference of those voters. It’s probably close to 100% Democratic but, accounting for crazy “Republicans who openly hate our policies are better than Democrats who pretend to like some of our policies” Greens, I allocated them 2-to-1 for Democrats and Republicans. Regardless, that’s still enough to push the Democrat over the threshold and, rather clearly, that’s closer to the preferences of this electorate than the only other possible outcome here, a Republican win.
But it doesn’t have to work out that way. Consider this contest:
This is a slightly simplified version of the 2009 mayoral election in Burlington, Vermont. The incumbent, Bob Kiss, won re-election despite being neither the plurality winner nor the preferred backup candidate of most voters. Looking at the rankings of those who voted, the Democrat, Andy Montroll, was easily preferred over Kiss:
- Montroll was preferred vs Kiss by a voter Majority, 4067-3477, margin 590.
- Montroll was preferred vs Wright by a voter Majority, 4597-3668, margin 929.
- Montroll was preferred vs Smith by a voter Majority, 4573-2998, margin 1575.
- Montroll was preferred vs Simpson by a voter Majority, 6267-591, margin 5676.
- Montroll was preferred vs All Write-Ins Combined by a voter Majority, 6658-104, margin 6554.
- There were no other candidates.
- Note that even Montroll’s weakest margin (590 over Kiss) was 7.8% as a percentage, which is a pretty clear win.
But, as IRV proponents rightly argue, this was still a better outcome than our traditional, first-past-the-post system. There, the Republican, Kurt Wright, would have won despite a minority of voters preferring a right-of-center candidate. And, in a traditional runoff, Democrat Montroll would have been eliminated and, presuming the same voters showed up and kept the same preferences, Kiss would have defeated Wright.
Yet there are theoretical voting models that would have picked Montroll, who is clearly the “best” choice if the goal is aggregating voter preferences. From the Wiki on the race:
- Wright would have won under plurality.
- Kiss won under IRV, and would have won under a two-round system vs Wright (under Burlington’s 40% threshold or the traditional 50%).
- Montroll would have won if the ballots were counted using Borda count, Bucklin, Coombs, Keener-Eigenvector, Sinkhorn, or any Condorcet method (Schulze, Ranked pairs, Copeland, etc.) Montroll would also likely have won using ratings systems like Score, Approval, or IRNR
I am, however, amused by this observation in the Wiki:
Kurt Wright acted as a spoiler candidate (a loser whose presence in the race changed who the winner is), splitting the vote against Bob Kiss; Wright received more first-choice votes (including promoted votes to first-choice) than Montroll due to Kiss splitting the vote against Wright
It’s odd, indeed, to consider the plurality vote-getter a “spoiler.”
Diving into the details of that race, one will note that “wasted” votes are still a thing. That is: quite a few folks don’t bother to list a second choice, much less a third and fourth. (This was apparently also true of the NYC mayoral race but it’s a bit early for analysis.)
There’s no perfect system and anything more complicated than IRV is probably going to cause more consternation than it’s worth. It is surely a better alternative than our current system, whether it’s a plurality winner-take-all approach or a physical runoff, which is expensive and relies on motivating an electorate a second time.