Instant-Runoff Voting Has a Viability Problem
Long-term adoption of RCV is set up to fail due to its reliance on IRV which can fail to elect the majority winner. Condorcet-consistent methods never fail that test.
Ranked Choice Voting (RCV) is having a moment. While many don’t realize it, there are several methods (algorithms) for selecting a winner in an RCV election. The movement method is called Instant Runoff Voting (IRV). Unfortunately, IRV can fail to elect the candidate preferred by a majority of voters. In technical terms, IRV can fail to elect the Condorcet winner.
Imagine a soccer tournament held in a round-robin fashion where every team plays every other team, and imagine one team emerged undefeated. Now imagine that the trophy was awarded to a different team—a team that lost to the undefeated team. That's what happens when IRV fails. If you are a little outraged by this hypothetical, imagine how voters react when it happens to their preferred candidate in real life.
You don't have to imagine it actually. It has happened twice in the US in the last twenty years: the Burlington, VT mayoral race in 2009 and the Alaska special congressional election in 2022. Both times it led to efforts to repeal RCV—successful in Burlington and on the ballot in two weeks in Alaska!
IRV advocates1 and o2 dismiss concerns about this kind of failure by arguing that IRV “almost always” picks the Condorcet winner (if one exists). They cite the low historical frequency of IRV failure: of 185 IRV elections in the United States 2004–2022 in which no candidate received a majority of first-choice votes, the Condorcet winner won all but the Burlington and Alaska races.3
Of course, the likelihood of failure is only half the equation because the expected harm of IRV failure is proportional to its probability and the severity of the consequences. While IRV hasn’t failed very often, those failures have had existential consequences due to the repeal efforts that followed.
To use an analogy, arguing that it is OK that IRV "almost always" picks the Condorcet winner is no more reassuring than being told that the planes an airline flies "almost always" land without crashing. Those failures are never acceptable.
Repeal efforts undermine the RCV reform movement, and history indicates that repeal efforts follow IRV failures to elect Condorcet winners.
Fortunately, there are RCV methods that are guaranteed to pick the Condorcet (aka undefeated) winner when one exists. These Condorcet-consistent voting (CCV) methods insulate the RCV movement from IRV's vulnerability and its inherent risk of repeal.
It's important to recognize that when repeal efforts are launched, they are not launched to replace IRV with a CCV method, which would preserve the benefits of RCV. Instead, the repeal efforts seek to remove RCV altogether.
Today, many people equate RCV adoption with the IRV method because IRV is by far the most commonly adopted RCV method despite this vulnerability. Transitioning the RCV movement to a CCV method—and there are many good ones—would be a challenge, but it might be a good investment in the durability and viability of RCV advances.
This post is the first in a series comparing CCV to IRV. For the record, we want RCV to succeed—we prefer RCV to plurality “first past the post” (FPTP) elections.
For example, FairVote RCV FAQ @ [https://bit.ly/3AgQzY0].
For example, see Stephanopoulos, Nicholas, Finding Condorcet (March 18, 2024). Washington and Lee Law Review, available @ https://scholarlycommons.law.wlu.edu/wlulr/vol81/iss3/4/.
See Stephanopoulos. The FairVote RCV FAQ noted above claims a much bigger “nearly 500” number, but that bigger pool contains races in which there was a simple majority winner and races in which there were only one or two candidates.
You are cherry picking. As in:
1) You wrote: "history indicates that repeal efforts follow IRV failures to elect Condorcet winners," but this is a ridiculous statement, since in only 1% of RCV races has the non-Condorcet candidate won. There was nobody running around Burlington or Alaska saying, "Heavens, did you know the non-Condorcet candidate won this election? Whatever are we to do. The sky is falling, sky is falling."
2) The reason RCV was repealed in Burlington had nothing to do with the non-Condorcet candidate winning, it was all due to local politics. A Progressive Party candidate beat out the Republican candidate, and the progressive mayor took some positions while in office that alienated him from voters. The losing Republican candidate fanned the flames of RCV repeal, blaming the unpopular mayor on RCV.
3) Similarly in Alaska, the MAGA Republicans lost AK's lone US House seat to a moderate Native Alaskan female Democrat, and blamed it on RCV instead of them running an unpopular candidate in Sarah Palin. Plus, under the impacts of RCV, moderate Republicans took over the state senate and decided to make a governing coalition with moderate Democrats instead of with MAGA Republicans. This caused the MAGAs, from Trump on down, to push for a repeal of RCV and the Top 4 open primary. It had nothing to do with non-Condorcet candidates winning (and BTW, Alaska was filling a vacancy, which it did by using the RCV/Top 4 hybrid first in the special election and then also in a second election (in the same year) to permanently elect that seat. The non-Condorcet candidate only won the first time in the special election. In the second election to permanently elect that seat, the RCV winner was also the Condorcet winner. Same election year, same three major candidates in both elections (Peltola, Palin and Begich) and the second time Peltola was also the Condorcet winner. You were saying?
3) Yes, Burlington voters were so upset about the non-Condorcet candidate winning that they have now fully restored RCV/IRV elections, this time not only for mayor but also city council! You were saying?
4) you have seriously messed up the terminology and etymology around ranked choice voting (RCV) and instant runoff voting (IRV). There are not different types of RCV. There is only one type, that is IRV. I know this for a fact because I am one of the people who came up with both of those names. I ran the campaign for RCV in San Francisco when I was with FairVote (then known as the Center for Voting and Democracy), and in the late 1990s we changed the name from "majority preferential voting" to "instant runoff voting" because MPV was a political science term that was a terrible name for educating voters. San Francisco already had a two-round "delayed runoff", with the first election in November and a runoff election in December if necessary. So we just invented the name instant runoff voting to compare it to the "delayed runoff" in December.
In the IRV ballot measure in San Francisco, March 2002, the city attorney helping us to draft the charter amendment recommended that we include the term "ranked choice voting" along with IRV because they thought that, from a legal standpoint, it would be more descriptive and more defensible if it came to that. At this point, no other place in the country was using IRV/RCV (Cambridge was using STV), and the city attorney wanted to make sure that there was no confusion. So we went along with that added name.
Flash forward two years in 2004, when we were implementing it in San Francisco for the first time -- the director of elections unilaterally decided to change the name to ranked choice voting because he did not want to feel any pressure that he could provide "instant" results! We the advocates had little say in the matter. But it kind of worked, so for the next few years, we interchangeably used both of those names, depending on the place and context. Ranked choice voting was useful sometimes because it described what voters had to do – rank their ballot. Instant runoff voting was also useful for explaining to voters in those cities that were replacing a two-round "delayed runoff" with the "instant runoff."
Both of those names are still used today. In Richmond, California, this reform is on the Nov 5 ballot and they are calling it "instant runoff voting" because they are comparing it to a two-round "delayed runoff." In the four states voting on this method on November 5, Colorado, Oregon, Idaho and Nevada, they are all calling it ranked choice voting for their own reasons.
But they are the exact same system. Instant runoff voting = ranked choice voting = instant runoff voting = ranked choice voting. The history of those names makes it clear – there is no space at all between those two names, they are the exact same system. So...go get your own name for whatever Condorcet variant you want to peddle, because Ranked Choice Voting and Instant Runoff Voting are already taken!