Right now, there are three front-runners in the August 2022 special general election using Ranked Choice Voting (RCV) in Alaska: Mary Peltola, a Yup’ik Eskimo running as a Democrat; Nick Begich, a moderate Republican hailing from a long line of Democrat Alaskan politicians who served as a Co-Chair for the late incumbent’s previous campaign; and Sarah Palin, a polarizing Republican with a Trump endorsement who stepped down as Governor to pursue an opportunity in reality TV.
With 90% of precincts reporting, Begich trails Palin in first-choice votes by 3.2 percentage points with Peltola in first by 7.5 points. Begich’s gap to Palin is almost insurmountable at this point, which means he’ll almost certainly be eliminated after the ballots are centralized to the Director’s Office in Juneau for full counting beginning August 31st. Current polling suggests most of the votes will transfer to Palin, handing her the win.
So why is this a failure as the title suggests? Most likely, Begich beats both Peltola and Palin head-to-head. That is, if you polled Alaskan voters on two questions: “Between Begich and Palin, who would you prefer?” and “Between Begich and Peltola, who would you prefer?”, Begich would be preferred by the electorate overall in both polls, yet he will almost certainly be eliminated first among the front-runners.
How can I predict all of this? I’m not. The science is, as it has done so correctly in most of the tiny handful of RCV elections in modern US history that have had front-runners from three distinct competitive factions.
A classic example is the 2009 Burlington, Vermont mayoral election. There were three front-runners: Kurt Wright, a Republican who typically would be considered to have no chance of winning in the progressive New England city; Andy Montroll, a Democrat most conservative Burlingtonians would typically vote for out of fear a more progressive candidate might win; and Bob Kiss, the incumbent and member of the Vermont Progressive Party.
Curiously, Wright, the Republican, received the greatest number of first-choice votes, followed by Kiss, and then Montroll. After the inconsequential write-ins and fourth candidate were eliminated, Montroll was next on the chopping block. After Montroll was cut, a sizable portion of those votes predictably transferred to Kiss, who ultimately beat out Wright for the win.
However, further analysis of the full ballot data revealed something peculiar: head-to-head, Montroll was ranked higher than either Wright or Kiss on more ballots than the other way around. Sound familiar? Montroll was preferred by the electorate overall and should have won, yet he was eliminated first among the front-runners.
The craziest part of this is that if a few hundred conservative voters had either dishonestly ranked Kiss above Wright or simply just stayed home and not voted at all, then Montroll would have won instead of Kiss. It’s likely that progressive Alaskan voters are in the same position, whereby either dishonestly ranking Palin above Peltola or just staying home, Begich would end up winning instead of Palin.
So this is rare, yes? Proponents of RCV will state Burlington 2009 was an anomaly, but that’s only true if a crucial detail is ignored: the Burlington election had three distinct front-runners, just as Alaska does now. As I stated before, there have only been a tiny handful of RCV elections with that dynamic in modern US history and a majority of them exhibited the fundamental mechanic that caused the Burlington 2009 failure, many leading directly to repeals of RCV. This mechanic, called non-monotonicity, means that ranking a candidate higher can hurt them and vice-versa. Even our current, terrible Choose-One Voting is impervious to this problem. The practical issue of non-monotonicity appears rarely under RCV only because RCV does not foster competitive elections, just like our current Choose-one Voting. Neither voting method is stable when there are more than two competitive factions, which, it turns out, is a trivial bar to overcome. Just as the science tells us the flaws with RCV, it tells us about other options that lack those flaws, and there are many. Right now, I’ll focus on just one that I think will be palatable to RCV supporters: STAR Voting.
STAR Voting was invented to address the problems of RCV. It’s simple and empowers each voter to expressively vote their conscience while electing winners who accurately reflect the will of the people by ensuring every vote is counted. Under STAR Voting, voters score candidates from 0 up to 5 stars. The two candidates with the highest scores are finalists. Each voter’s one full vote then goes to the finalist they scored higher. The finalist with the most votes wins.
Add up the scores, then add up the votes. Simple.
To get a high total score, candidates need to get broad support from across the electorate. To get lots of votes, candidates also need to build a strong base of supporters. This hybrid approach ensures STAR Voting elects candidates who achieve both and levels the playing field for all candidates, fostering competitive elections, unlike RCV. Because all of the data voters express on their ballots is guaranteed to be counted and is counted simultaneously, STAR Voting ensures candidates receive support from all of the voters who express it, unlike RCV. Because each voter gets one full vote toward the finalist they prefer, STAR Voting ensures votes cannot catastrophically backfire on honest voters, unlike RCV. Because the tally is performed with simple addition, STAR Voting results can be determined the night of the election through the same secure decentralized process we currently use rather than waiting for physical ballots to be trucked to a single point of failure to be tabulated, unlike RCV.
You can learn more about the science supporting STAR Voting at starvoting.org.
Of course, we won’t know for sure that this set of predictions about this Alaskan special general election are true unless and until the full ballot data is released, but if they are true, that’s another point in favor of the science that tells us that RCV does not attain the goals it sets out to achieve while voting methods like STAR do.
