Electoral Reform Peril
Andrew Yang is Overpromising the benefits of Ranked Choice Voting
The limits of ballot reform efforts to solve deeper political realities
Ranked Choice Voting (RCV) is the hottest electoral reform trend being considered by cities and states right now. If your state doesn’t have a chapter of activists dedicated to RCV already, it probably soon will. Wanting to “reform” our elections is understandable — many people are fed up with how politics feels dysfunctional in general and a sizable number of voters find their politician choices to be incredibly uninspiring or downright awful.
In addition to widespread discontent in the electorate, the advocacy organizations in favor of RCV are smart mobilizers and good fundraisers. In 2021 voters in New York City as well as some parts of Minnesota, Utah, and a smattering of other cities tried out RCV voting. Many voters reported that the liked how it felt to rank more than one candidate.
One of the most vocal advocates for RCV is former presidential and NYC mayoral candidate, Andrew Yang. Yang has argued that the importance of this reform is so fundamental to the future of democracy, that he even went so far as to say that he’d welcome in self-declared white supremacists in order to get more RCV changes implemented across the nation. Yang paused when asked and later tried to backtrack his position, but his gaffe demonstrates the unreasonable weight he and others have placed on RCV as a potential savior of democracy. There are risks in overpromising the ability of any single reform to revolutionize our politics. And there are very real downsides to not considering the potential negative implications of RCV.
Shortly after failing to secure the Democratic Party nomination for Mayor in New York City — which was conducted using RCV — Andrew Yang re-emerged with a new book, Forward: Notes on the Future of Democracy. In the promotion of his book, Yang took to Twitter arguing that RCV would be one of the keys to “unlocking us from polarization”, necessary to “give alternatives to the duopoly a real chance to compete — and a more resilient, representative democracy”, and simple solution to the spoiler effect. In the past, Yang has made more straightforward and less hyperbolic claims about the benefits of RCV, such as how having an “instant runoff” is way for states to save money on costly and low turn-out two stage run-off elections, but the tenor of his more recent pro-RCV statements and the newfound urgency in his appeals should give voters pause.
Having seen the adoption of RCV in NYC as well as working as a practitioner during the 2021 primaries, I wanted to see what more Yang had to say about RCV, so I used one of my favorite job perks — inter library loan — to check out his book and further get into his supportive arguments.
In the introduction, Yang states that the book will present a small menu of important structural solutions to what ails US democracy and urges readers to push for all of these reforms. He previews what is to come:
“One of them, electoral reform that establishes ranked-choice voting and open primaries across the country, I consider paramount. These electoral reforms are the skeleton key that will realign out leaders’ political incentives and make so much else possible.
With ranked-choice voting, candidates would need to achieve majority support from all the voters in their district and not just the primary voters. The duopoly would diminish in both power and intensity. Voters could express their preferences without fear of “wasting” their vote. Negative campaigning would be discouraged. Real third parties would emerge…Then — and only then — will sustained change be possible…Ranked-choice voting will be the crucial change that unlocks us from stasis and polarization.”
Describing RCV and open primaries as “the skeleton key” that will unlock us from stasis and polarization is a wild overpromise, both theoretically and empirically. This overpromising could lead reform supportive voters to be disappointed and potentially even more alienated from the seemingly unfixable political processes in the US.
As to the limits of open-primaries, there are real political difficulties in getting closed systems to open up — as the biggest political stakeholders stand to lose by relaxing party and candidate control. But even if open primaries were more widely used, historical research and contemporary experts such as Lee Drutman, routinely find that open primaries make little difference in who wins or who runs or even who votes in primaries. If anything, open primaries probably reduce third party organization.
But the real organizing drive and the most hopeful promises are often pinned on RCV. Yang presents his arguments on pages 218–220 of Forward where he says,
“the general election for each congressional seat, and virtually all of our elections moving forward, should be conducted via ranked-choice voting.”
To support his statement, he provides the following arguments:
1. Ranked choice voting leads to majority winners
2. Ranked choice voting means you don’t need to worry about spoilers
3. Ranked choice voting diminishes the incentives to campaign negatively
4. Ranked choice voting favors coalition building and reaching out to different constituents
Let’s consider these points in turn. Does RCV lead to majority winners? It’s complicated. In the pro-RCV camp advocates say that winners only emerge after getting 50%+ of votes after successive rounds of elimination. Does that mean a winner is supported by 50%+ of all voters casting a ballot? No, because as more candidates are eliminated in RCV elections so too are ballots of people who picked losing candidates. What’s more, because RCV elections tend to field more candidates at least in the first few elections after the change is adopted, ballot exhaustion rates are higher as the candidate pools get wider.
There are at least three reasons for this. First, some voters simply don’t prefer the final two candidates and thus rank others. Second, voting is already mentally taxing, asking voters to determine and order their top 3–5 candidates is a big ask, and many voters rank fewer than their fully allowed amount. Finally, because every US version of RCV has a hard limit on how many candidates a voter can rank (between 3–5) when candidate pools exceed that number, there will be greater numbers of ballots exhausted that don’t contribute to the “majority” in the final winner determination because they are thrown out of the denominator all together.
In the 2021 NYC Democratic primary, the average city council race had 6 candidates and the mayor’s race had 13 candidates! With only 5 possible rankings permitted there, it’s no surprise the the majority of races were won by candidates who did not have the support of the majority of voters. In the case of the Brooklyn Borough President, only 37% of voters casting ballots that ranked the winner, Antonio Reynoso.
Now some might say that RCV makes it more likely that the eventual winner is supported by a majority of voters when compared to standard, pick one elections. But again because of the incentives for more people running fueled by the belief that RCV removes the potential for any one candidate to play the “spoiler” role, for many elections the likelihood that any candidate gets to a majority is heavily influenced by the number of candidates running. Elections elsewhere have found the same thing, RCV does not lead to majority winners. Bottom line: RCV is not an assured way to get a winning candidate supported by a majority and this is an over promise.
Does ranked choice voting means you don’t need to worry about spoilers? Again, the answer is complicated. In the classic election spoiler story, there is 1 candidate who earns a number of votes that is greater than the difference between the winner and the runner up. And for this person to be considered a “spoiler” one must assume that their voters would have agreed to rank the runner up as their second choice in a hypothetical RCV counterfactual.
But a hypothetical RCV counterfactual with just those same 3 candidates is not the right frame of reference. RCV elections tend to field more candidates because people are told they have a better shot, and they can’t be considered spoilers. Because of this, pinning the spoiling on any one candidate becomes much harder, but the possibility of spoiling remains and might even be exacerbated but just better hidden across multiple candidates.
In a RCV election a spoiler can be a group of candidates that receives a sufficient number of votes that, if those votes had all gone to the second-place candidate, could have changed the result of the election. This result can happen accidentally or in a coordinated way. For the accidental explanation consider a race with 7 candidates, 5 liberals and 2 conservatives (this could be within a party for a primary or a general). Now, most elections are low information events, meaning voters don’t know a ton about all of the candidates. In standard elections, voters just need to know their favorite, vote for one and be done with it.
Because of this dynamic, candidates who are less likely to win and fear being a spoiler in local or state elections typically drop out or coalesce around 1 candidate. In an RCV election, where candidates are told they can’t spoil the outcome, they tend to stay in until the end — that was the experience in the NYC city council elections conducted with RCV. So,[lc1] in the 5 liberal and 2 conservative, low information election hypothetical, say there was one leading liberal and one leading conservative. The 5 liberals may draw from different parts of a constituency, perhaps from women, people of color, disability advocacy groups, environmentalists, school equity issues, etc. Voters would probably prefer any one of the liberals to the conservatives, but because it’s a lower information environment, many will not know more than just their favorite candidate. Voters preferring a conservative only have two to choose between, but if one is leading and there is just one other, the rankings are relatively more straightforward than voters having to sort through 5 candidates.
On election day, when everyone can stay in and not fear being a spoiler, there may be more voters who prefer a liberal representative than voters who prefer the conservative, but they will also be split in many more directions when filling out a ranked ballot. And in low information environments, people are less likely to use their full rankings. As ballots are exhausted and the race comes down to the leading liberal and the leading conservative, those who chose other liberals that were eliminated can only have their preferences expressed if they knew to put the leading liberal as their lower ranked candidates, but many won’t and instead more ballots will be exhausted. Thus, a conservative candidate in an RCV election can emerge as the victor, despite the fact that more voters preferred a liberal (just not all the same liberal).
In an election system that discourages people from “spoiling” elections, candidate drop out or coalesce; this is exactly what we saw in the NYC Manhattan DA race because it was a plurality ballot. But when candidates are told they can’t spoil in an RCV system (and NYC public funding also reduces the incentive to drop out) candidates will run all the way to end, even if that accidentally means a victor from a different ideological orientation. Are all those candidates spoilers? It’s harder to figure out who should have dropped out or coalesced, but this issue certainly doesn’t go away under RCV, it becomes harder to identify.
Does ranked choice voting diminish the incentives to campaign negatively? It’s possible, but only for a few election cycles. Advocates tell voters and candidates that staying positive is the move because you’ll always want to be considered as a possible #2 in the rankings for someone else. The voices and guidance of RCV advocates are heeded in the first RCV elections because of general unfamiliarity with first time candidates and political consultants operating in relatively uncharted waters.
But the political consulting class quickly learns. Trying to secure a #2 spot only makes sense if you believe the person a voter wants as #1 will lose or at least do worse than your own candidate. Trying to be #2 for someone who is a front-runner or is expected to do better than your candidate in round 1 is foolish, because you’ll never get the impact of that voter’s expression of you as their second choice unless their top pick is eliminated. Going negative against the front-runners will be more of a tactic moving forward in NYC RCV elections and will be in other places after the first few election cycles.
Another political consulting lesson is that contesting an RCV election is mostly the same as contesting a single, choice plurality election. In about 95% of all elections conducted under an RCV system in the US, whichever candidate is winning the first round, goes on to win in the final round. The best strategy is to get the most people to vote for you as number one and leave them to make up their down ballot choices on their own. There is not good evidence that voters listen to candidates who tell them how to strategically rank others, and for winning trying to game out a come from behind victory built on the efforts of strategic voters is unworkable.
Finally, does ranked choice voting favor coalition building and reaching out to different constituents? This is the most positive potential upside of any reform that allows voters to think about more than one candidate as their choice.
There are, of course, benefits to adopting RCV. All systems have at least some specific benefits, but like all other systems RCV has downsides and there are enough reasons to believe the purported benefits are less revolutionary and less enduring than Yang supposes. By overpromising a set of voters and activists, we risk even great disillusionment when we find ourselves in the same muck of issues, having swapped for a more expensive, time consuming, and less helpful system of balloting. We’ve also done nearly precisely this cycle of frustration leading to electoral reform, and then to reforms failing to live up to promises, and then repeal and even deeper frustrations than before. Jack Santucci has a forthcoming book that describes in detail exactly how this happened in the middle of the 20th century in different parts of the US.
“The” solution to the myriad of challenges that plague US politics cannot and will not be fixed by allowing voters to rank more candidates. The promise that adding a more expressive way to determine single winner elected officials will bring about revolutions in the party system, permanent changes in the campaign landscape and a kinder, less fractious US politics is a temporary myth.
Yang has obvious reasons to be upset with the dominant political order in the US. The first part of Forward details how not knowing enough party insiders, connections between media and elected officials and longtime loyalty politics all conspire to make most outsider entries into office unsuccessful. Yang experienced this on the national and state level within the Democratic Party. Yang and his wider network do not get enough credit for their efforts in the Georgia run-off election giving the Senate to the Democrats and he doesn’t really get credit for his work on behalf of electing Joe Biden. But overpromising the potential of RCV to “fix” American democracy is not the answer and may, in fact, be counterproductive.
When the system is adopted and doesn’t deliver the promised results, the “do something” reform crowd will understandably be disillusioned. But importantly, getting municipalities to change the way the ballot and conduct elections is not something that is simple to undo or modify after the fact. We need to have robust discussions about the benefits and the limits of different reforms, versus a race to advocacy on the route to adoption. Building up false expectations is not the way to a better politics.