Evaluating voting systems and the criteria we judge them by
“The fact is that FPTP, the voting method we use in most of the English-speaking world, is absolutely horrible, and there is reason to believe that reforming it would substantially (though not of course completely) alleviate much political dysfunction and suffering.”
So says Jameson Quinn in “A Voting Theory Primer for Rationalists”
Considering the state of the world right now and our position in it, where America holds a colossal amount of power, it becomes clear that this may well be the central issue of our times.
Most who have dabbled in the voting science field have heard of Kenneth Arrow, who posed 3 basic "fairness criteria" and showed that no ranked method can meet all of them. Some criteria - like Favorite Betrayal and Later No Harm Criterion- which deal with strategic incentives for voters, are in effect mutually exclusive. Following up on this idea, two other famous voting scientists, Gibbard and Satterthwaite, independently came up with another theorem, which showed that “no voting system (ranked or otherwise) could possibly avoid creating strategic incentives for some voters in some situations.”
The work of these three groundbreaking scientists unfortunately didn’t answer our fundamental question. “What is the best voting system?” but their work did suggest another key point which seems to have largely been ignored by many who came after them.
A binary pass/fail approach to criteria isn’t going to get us the answers we need, and thus, in order to find the best voting system, we need to zoom out, balance competing factors, and take a more holistic approach to assessing voting systems.
Below is chart of commonly used pass/fail voting system criteria. It is impossible to pass them all.
We do not live in a black and white world
Debating who's criteria is the most important is a dead end road. Of course, each group has a strong incentive to promote the criteria their reform passes and bury the rest. Rather, it’s critical that we look through a more broad spectrum lens. What exactly are we trying to accomplish here? The Equal Vote Coalition has distilled the matter down to five pillars of a just voting system: Equality, Accuracy, Honesty, Expressiveness, and Simplicity. These five pillars allow us to group the election criteria under a few umbrellas and thus look at the considerations with more nuance. With the exception of Equality, these pillars are not mathematical proofs, they are values that are essential to consider when we fight for a more fair democracy.
Equality is a core concept in our government and our values. A voting system should be equal and the idea of the equally weighted vote and one person one vote are directly mandated by our government, despite the fact that we currently accomplish neither. A voting system should not play favorites and my vote should always be equal to yours.
STAR Voting offers perfectly equally weighted votes, while RCV favors certain types of voters and puts others- like those that prefer strong underdogs- at a disadvantage.
Accuracy is another pillar so fundamental to a good election that it is often overlooked. Of course we want to elect the leader or leaders that best represent the will of the people. In single-winner elections it’s a strong bet that a candidate that was preferred over all others should win, but in extremely close races sometimes we need more information. Since not all voting systems have a highly expressive ballot this is an area where we must turn to simulations and analysis beyond election data to get an accurate measure. In these kinds of close races we may need to know not only who was preferred to who, but also how much support each candidate had, according to each voter.
Voter Satisfaction Efficiency and its predecessor Bayesian Regret look at how often a given voting system will elect the candidate that can make “as many voters as possible as satisfied as possible with the election results.”
This kind of modeling shows clearly that our current system is the worst and that STAR Voting tops the charts in election accuracy. IRV (commonly referred to as "ranked choice voting" in the U.S.) falls about ½ way between the two and didn’t even make the top 10.
Honesty is the third critical pillar. We want to be able to simply vote our conscience. For some voters this could mean simply showing who their favorite is, but for those with a more nuanced opinion, this requires an expressive ballot where we can show how much we like multiple candidates and who we prefer to who. Voting honestly and expressively should be the primary incentive.
As Gibbard and Satterthwaite showed above, it’s impossible to create a system where strategic voting is never beneficial, but we can make blatant dishonest voting not worth the trouble. We need a voting system that is highly accurate and representative when we are honest, but we also need a system that doesn't incentivize strategic voting in the first place and that is still highly accurate even if some voters try and game the system.
STAR Voting does a good job discouraging strategic voting. VSE simulations show that dishonest voting is basically just as likely to backfire than to help your candidates and as such it’s not worth the risk. In comparison strategic voting in Instant Runoff Voting* is almost 3 times as likely to help voters than to hurt them. Specifically, voters need to be strategic in close three way races. In Burlington, the Republican voters, who knew the Republican wouldn’t win, should have dishonestly ranked their second choice in first place in order to prevent the Spoiler Effect, which ended up throwing the election to a less preferred candidate.
Even if voters were strategic in STAR Voting, the election accuracy is still as good or better than IRV in a best case scenario.
*NOTE: Instant Runoff Voting (IRV) is the single winner version of Ranked Choice Voting (RCV) that is used around the world. Many people now refer to it as RCV, but RCV is a family of voting methods which includes the Condorcet methods, which dramatically out perform IRV by many measures. For clarity we use the more specific term, IRV.
Expressiveness and Simplicity
As we saw above, expressiveness is a key component to both honest voting and accurate results, but it’s important to balance this pillar with it’s cousin simplicity. A voting system must be simple and transparent enough that voters understand the system and trust it with their votes. For some, our current First-Past-The-Post system may seem to be the epitome of simplicity. Anyone can figure out how to fill out a ballot- but the implications can be extremely complex. Many voters understand that our system plays favorites and devolves us into a polarized caricature of our better selves, and many people feel that the system is rigged so badly it’s not worth voting.
A truly simple voting system will be user friendly to both the voters, and the elections officials, and will instill confidence by yielding results that make sense. STAR Voting is precinct summable, unlike IRV, and produces clear and transparent results. After an election we can know what percentage preferred the winner over the other finalist, and we have the overall scores and average scores to compare the other candidates. IRV’s lack of precinct summabilty means that the system doesn’t scale well and could be more vulnerable to fraud and talking with elected officials this is the major reason many withhold their support.
Widespread bias in the election reform advocacy world
It is of course natural for individuals and organizations to prefer the metrics at which they excel. Having an opinion on voting systems is generally unavoidable, especially once we have dug into the issue and done our homework. Our priorities differ and there are legitimate reasons to prefer various reforms over others.
Unfortunately the election reform field has a long history of advocacy work that in some cases has oversold the reforms, cherry picked criteria, and used widespread misinformation and misrepresentation to fend off competing proposals. The Equal Vote Coalition strongly opposes this "the ends justify the means" approach. To those who are committed to making an unbiased review we encourage you to include information from both reputable cardinal and ordinal voting advocacy groups in your review.
What are Cardinal and Ordinal Voting methods? Put simply, Rating and Ranking methods. Cardinal methods use rating, scoring or approval ballots that can be tabulated by adding up the scores or votes. Ordinal methods use ranking ballots and are generally tabulated in elimination rounds.
Reputable Cardinal Voting Advocacy Groups:
The Center For Election Science, Counted, The Center for Range Voting
Reputable Ordinal Voting Advocacy Groups:
Ranked Choice Voting Resource Center
*NOTE: FairVote and their founder Rob Richie have been under fire for many years for propagating numerous false and misleading claims about both IRV itself and numerous other reforms. Even as they have now begun to walk some of the false claims back, misleading arguments still abound and have been repeated by many other groups that they work in coalition with. Sources cited are in no way comprehensive of relevant examples and reflect a quick search only. This is common knowledge throughout the voting reform movement. Even highly reputable groups like Represent.Us and the LOWV regularly cite common debunked talking points about IRV. Our movement needs to do better.
Many of the largest election reform advocacy groups around the country, led by FairVote, have historically supported Ranked Choice Voting. In fact many Equal Vote Coalition members are longtime advocates as well. If you are a supporter as well we are not asking you to withdraw that support. We support IRV in Benton County, OR which was passed in 2016, in part, by Alan Zundel, who later became a Chief Petitioner for "STAR Voting for Lane County". We would support any reform which is an improvement or stepping stone, but we do believe it is critical to take a step back and withdraw support from one specific voting system criterion that many believe has severely compromised the voting reform movement.
Later-No-Harm, aka LNH, is a much loved criterion at the core of FairVote, Sightline, and other groups endorsement of Ranked Choice Voting, as both FairVote's Rob Richie and Sightline's Kristin Eberhard have explained to us. Later-No-Harm states that honestly ranking or rating your down ballot choices will not hurt your favorite, which is of course desirable. Unfortunately this criteria is at direct odds with two even more critical goals.
In order to pass this criterion, voting systems like IRV in practice sacrifice too much: Later No Harm requires more expressive voting systems to ignore some voters’ rankings. Stated another way, in every Instant Runoff election, some voters will not have their next choice counted, even after their first choice is eliminated. This is fundamentally unfair and directly compromises voter equality, imposes implicit bias into the system, and denies every voter and equally weighted vote.
Later No Harm leaves us wide open to the "Spoiler Effect" and in practice requires us to fail an even more important criteria called Favorite Betrayal, which states that it’s safe to honestly rank your favorite first.
The Spoiler Effect is the number one reason we are trying to reform the voting system in the first place. Concern around Vote Splitting and the Spoiler Effect has been echoed by the dozens of elected officials and their policy directors who I have spoken with about election reform in the last year. Voters have of course been echoing concerns around “Lesser-Evil” voting and Vote Splitting since long before candidates like Ralph Nader and Ross Perot brought the issue to the national spotlight.
“Whether they’ve realized it or not, folks who tout Later No Harm as the holy grail of voting systems criteria are actually saying that The Spoiler Effect is not a problem they think is important to fix. We respectfully disagree, and thus said goodbye to Later No Harm.”
- says Emily Dempsey, BA in Mathematics and BS in computer science from Truman State University, in her article “A Farewell to Pass/Fail: Why We Ditched Later No Harm”