Demystifying STAR Voting
STAR Voting is a new alternative voting method designed to fix our flawed election system. STAR was invented in Oregon in 2014, to blend the benefits of prior proposals such as Approval and Ranked Choice/Instant Runoff Voting.
The Background
Our democracy is seriously broken. Partisan polarization drives an ever-deeper wedge through our society, and political outcomes repeatedly favor a tapestry of special interests rather than the people as a whole.
At the root of the dysfunction is our voting method: in each election we are limited to supporting only one candidate on each ballot for political office. This limit of one choice creates a fundamental inequality in the vote that leads to the spoiler effect, amplifies the influence of money in politics, and creates a cycle of ever-increasing polarization in the political process.
Our elected leaders are taking up Oregon’s mantle of election reform leadership once again, considering multiple voting system and electoral reform measures in the legislature to reform our democratic process.
What is a Voting Method?
A voting method comprises the means of expression of the voters (ballots) and the method for tallying those ballots into an election outcome.
In the voting method we use today — known as plurality, first-past-the-post, or simply “choose-only-one” voting — all candidates for an office are listed on the ballot, and the voter selects a single candidate to support. The candidate receiving the most single-choice votes wins.
Pretty simple, right? So what’s the problem?
Here’s the bug, and it’s been with us for so long that it’s tough to see at first. The problem is that the limit of one choice creates a fundamental inequality between the voters.
If there are only two candidates, a plurality vote is equal — you vote for A, I vote for B, and the result reflects the majority opinion.
But in every plurality election with more than two candidates — every single one — the voters who support more than one candidate have a lesser weight vote than those who support just one candidate, because in the aggregate, they split their votes between the the candidates they support, while those who support fewer candidates concentrate their votes.
This inequality has really bad knock-on effects: voters enthusiastic about underdogs are encouraged by their peers to “not waste” their votes on their favorites, and instead vote for the “lesser evil” in order to prevent a worst-case scenario. Contenders are measured by money as an indicator of viability, and thus this vote-splitting spoiler effect creates an unassailable advantage for special interests.
It bears mentioning that our Constitution requires an equal vote amongst the voters, affirmed repeatedly by the Supreme Court:
“Once the geographical unit for which a representative is to be chosen is designated, all who participate in the election are to have an equal vote…” — United States Supreme Court, Gray v. Sanders
“And the right of suffrage can be denied by a debasement or dilution of the weight of a citizen’s vote just as effectively as by wholly prohibiting the free exercise of the franchise.” — United States Supreme Court, Reynolds v. Sims
The Court’s mandate is clear: the weight of our votes in each election must be equal.
How can we tell definitively that a voting method provides an equal weight vote to the voters?
The test for equality of weight is balance. To determine whether two objects are of equal weight, they must balance when placed on opposite sides of a balance scale.
This very basic principle applies to voting methods. A voting method meets the Equality Criterion if it definitively provides votes of equal weight to all the voters — if, and only if, for each possible vote expression that one voter may cast in an election, there exists another expression of the vote that another voter can cast that is in balance, such that the outcome of the election is the same whether both or neither votes are counted.
Enter Ranked Choice
The nasty knock-ons of our dominant voting paradigm have not gone unnoticed by those committed to a republic of the people, by the people, and for the people. Going back literally millennia, alternate voting methods have been proposed. Not the least recently, 18th-century French mathematician and philosopher Marie Jean Antoine Nicolas Caritat, the Marquis de Condorcet, proposed that the winner of an election should be the candidate that is preferred head-to-head against each of the other candidates, derived from the voters’ preferences between the candidates for office. (Spoiler alert, this mandate became the Condorcet Criterion.). And the Marquis de Condorcet was not alone: his contemporary Jean-Charles de Borda created a rank-based system, first published in 1781, as an alternative to the plurality method.
While there are now many dozens of voting methods based on rank-ordered ballots, the “Instant Runoff” method has gained the most traction in public elections since its first use in the late 1800s, and is now popularized by supporters as “Ranked Choice Voting.”
The Instant Runoff/Ranked Choice counting method works by successively eliminating the candidate with the least first choice rankings and then counting the ballots for whom a choice was eliminated for the next non-eliminated candidate.
Historical fun fact #1: according to Wikipedia, the Marquis de Condorcet considered and rejected the instant runoff method of ranked voting because “En effet, lorsqu’il y a plus de trois concurrents, le véritable vœu de la pluralité peut être pour un candidat qui n’ait eu aucune des voix dans le premier scrutin,” which Google translates as “Indeed, when there are more than three competitors, the real wish of plurality may be for a candidate who did not have any of the votes in the first ballot.” But more on that later.
Historical fun fact #2: Kenneth Arrow got a Nobel Prize back in the day for proving that no rank-ordered voting method could be mathematically “fair” — no counting algorithm for ranked ballots can meet the four criteria of unrestricted domain, non-dictatorship, Pareto efficiency, and independence of irrelevant alternatives. Arrow’s Impossibility Theorem was a major downer for most of the 20th century, as it implied that truly representative democracy was impossible.
Enter Scoring
Scoring systems depart from rank-ordered methods by allowing the voter to assign an independent weight of support to each candidate on the ballot. So instead of marking first choice, second choice, third choice, and so on, on a score ballot, voters give an independent score to each choice within the range allowed by the ballot.
The simplest scoring system is Approval Voting, where the range of choices is limited to 0 or 1 — approve or not. An Approval Voting ballot therefore looks just like the ballots we use currently, only with the single choice limit removed:
The Lay of the Land, Oregon Circa 2014
Personal story sidebar: I entered the voting method reform arena in late 2013, with a proposal crafted from more than 20 years of cogitation on our electoral ills. While I was originally a big fan of Instant Runoff, a deep look at that system’s well-documented failure modes prompted me to try something new.
At that time, and still today, the reform camps promoting ranked choice/instant runoff and the approval/score methods were engaged in heated online debate, each measuring the other’s favorite voting system against their own favorite voting system criteria.
While the “Unified Primary” measure did not qualify for the 2014 ballot, it inspired a true dialog amongst voting method reform advocates. In October 2014, luminaries in the election reform world congregated at the University of Oregon for the Equal Vote Conference, to debate the merits of a range of proposals to cure our broken franchise.
It was at the conclusion of this event that Rob Richie, Founder and Executive Director of FairVote, proposed a compromise. What if a ballot were to include both approval and preference? Could the Unified Primary be combined with Ranked Choice so as to yield a high-accuracy outcome in a single vote?
But what would that ballot look like? An approval check box and a bunch of ranking bubbles? It was only in discussion with Clay Shentrup, co-founder of the Center for Election Science later that day that the solution became clear. What if there were a simple, well-understood ballot format that included both scoring and ranking?
Enter STAR Voting
The breakthrough realization was the five star ballot. Used ubiquitously from Amazon to iTunes, stars blend the ideas of score and rank.
How STAR Voting Works
STAR Voting uses a zero to five ‘star’ ballot and elects the majority favorite between the two candidates given the most stars by voters.
Voters therefore assign stars to each candidate based on 1: how much the voter supports that candidate being one of the two finalists, as well as 2: the voter’s preference between any two potential finalists.
The count is simple. STAR is an acronym for Score Then Automatic Runoff, which describes in just four words how a STAR voting election works.
- STAR: The voters star each candidate from zero to five stars. Candidates left blank count as zero stars.
- Then: The stars for each candidate are counted up as an overall score.
- Automatic: The two highest-scoring candidates are the finalists in the automatic runoff.
- Runoff: Each ballot counts for the finalist given more stars on that ballot, or is counted as a vote of no preference if both finalists receive the same number of stars.
Said again: the star ballot allows a voter to express both how much support is given each candidate to be one of the two finalists, as well as a preference (or not) between each potential finalist.
What STAR Voting Does
STAR Voting gives each voter an equal, nuanced expression about the outcome of the election, and according to multiple independent simulation research studies, STAR produces the most accurate election outcomes amongst leading voting method reform proposals. Further, STAR’s two-step counting system allows voters to honestly express their true preferences without being given reason to strategically vote against their true favorite candidates nor fail to show any support for secondary favorites(s). At the same time, STAR minimizes any advantage a voter may gain by voting dishonestly or strategically.
Comparisons, Contrasts, and Criteria
Majority outcomes: STAR always elects the majority favorite between the two most supported candidates overall. Ranked Choice Voting also claims to create majority outcomes, but this is a stretch. In competitive elections between three or more candidates, RCV’s peculiar counting system discards the secondary preferences of some voters and as a result, the “majority” choice in the final round can be between two candidates who are not well-supported overall. A real world example is the 2009 Burlington RCV mayoral election that led Burlington to repeal RCV the following year. Learn more.
Later-No-Harm: Systems that meet the Later-No-Harm criterion guarantee that a voter’s expression of support for a secondary favorite cannot, in any circumstance, cause the voter’s favorite choice to lose the election when the absence of that expression would have resulted in the true favorite’s win. Ranked Choice Voting is one system that meets this criterion.
No Favorite Betrayal: Systems that pass this criterion never create a worse outcome for voters who express maximum support for their true favorite candidates. Ranked Choice Voting does not pass the No Favorite Betrayal Criterion — putting a factionally-strong but overall weak candidate in first position on a Ranked Choice ballot can lead to the worst outcome for a voter. The aforementioned Burlington election is a textbook example of this flaw. Approval Voting is an example of a system that complies with the No Favorite Betrayal Criterion.
STAR Voting balances Later-No-Harm and No Favorite Betrayal
Like Ranked Choice, STAR allows the voter to safely express support for secondary preferences. Two candidates advance in STAR Voting, so offering token support to a second candidate carries little risk. If nominal support for a secondary candidate knocks one’s favorite from the runoff, it means one’s first choice is a truly weak contender.
Like Approval Voting, STAR gives no reason to offer anything less than full support to one’s true favorite in the race: if one’s first and second choice are vying for second seat in the runoff, whichever makes that spot is the stronger, and therefore deserving of appropriate support from the voter.
The Chicken Dilemma
In the case of a competitive three-candidate race where two of the three are from a single majority faction, it may benefit die-hard supporters of one or the other same-faction candidates to give only token support to their second choice. Not a big deal either way. We suspect a relatively equal number of voters will make this choice and so the correct election outcome is likely to occur. Regardless, this is one of several cases where STAR encourages more honest and/or reciprocal behavior in subsequent elections.
The Dreaded Turkey Raise
Turkey-raising isn’t just for pilgrims. In voting science it means a voter’s promotion of a weak third choice over a voter’s strong second choice, in the hope that the voter’s first choice will have a better chance of winning as a result.
Turkey-raising is one of several forms of strategic voting that STAR’s counting system protects against. Why? Because if you fear you second choice is stronger than your favorite, your favorite, in your mind, is playing for second seat in the runoff. In that case, the dumbest thing you could do is add support to a candidate you like even less. That support is therefore more likely to knock your first choice out of contention rather than helping your favorite win.
What If ‘Average/Low Information/Minority’ voters are confused?
A number of well-recognized and often certified Smart People have suggested that “average” and/or “low-information” and/or “disadvantaged” and/or “minority” voters just won’t be able to figure out the whole “0 stars bad, 5 stars good!” aspect of STAR — and therefore might not use the full range of star options in any given election, meaning an advantage for those who understand 0 stars bad, 5 stars good.
STAR petitioners have explained this system to many thousands of “average voters.” Comprehension and enthusiasm are high if STAR is explained with clarity.
Yeah, but has it ever been used in a real election?
Yes, STAR got its first test in a statewide binding nominating contest in the 2020 Independent Party of Oregon Primary Election. This contest was a perfect first stress test of STAR: voters were unfamiliar with STAR and the contests were lopsided. Promising data points: STAR score (approval) ratings nearly exactly matched public approval polling data for the leading candidates, and ~90% of voters utilized the full range of the STAR spectrum.
When we really think about it, the move to STAR is obvious. A simple well-understood five star ballot, a precinct-summable and auditable counting system. A democracy in dire need of new solutions.
