GUEST EDITORIAL: More Problems With Ranked-Choice-Voting

By Rick Hogaboam

Let’s say, hypothetically, that the jurisdictional matters I raised yesterday (you’d have to go back and read them) are small potatoes, because the ends justify the means — and the ends are so amazing that it’s worth it. It’s here where I would turn to my mathematical and philosophical concerns. I won’t describe in detail monotonicity, Condorcet winner, Borda count, independence of irrelevant alternatives or Kenneth Arrow’s “impossibility theorem,” but you can look all that up (and make sure to have a cup of coffee while doing so), as alternative voting methods have been debated for a long time. In as simple terms as I can try to elaborate: RCV is not an alternative worth the costs and radical change to our current election system in Idaho.

Though rare, RCV can create an outcome where the winner actually isn’t the preferred candidate over the opponents when measured by 1:1 preference for each matchup. This is a brief summary of the Condorcet Method:

(Pardon me as I do try to summarize a bit the concepts I said I wouldn’t elaborate on.)

In a recent Alaska special election in 2022, the candidate who unanimously won in head-to-head preference matchups with all candidates was actually Nick Begich. Since he received the fewest first preference votes, he was eliminated as the first loser, ironically, and his voters got transferred to remaining candidates if they marked a second choice. Herein is another irony: only some second choices count, and they count the equivalent of other first choices, but not all second choices count. Who’s on first? (Cheesy pun alert.) Second choices are all or nothing in the algorithm, and herein is cause for even more confusion: Begich would’ve won if more Palin voters instead voted for the winner, Peltola.

How in the world can the winner lose by getting more first place votes and the loser win by not receiving a single vote more relative to the top preference? Well, this is an explanation of the violation of monotonicity, the principle that votes for a candidate shouldn’t hurt the candidate relative to an alternative outcome in which they receive fewer votes.

If more Palin voters switched to Peltola, and Palin was eliminated first, then her voters would’ve transferred, and Begich would’ve won. Begich would’ve won, not by receiving a single additional vote, but the winner actually receiving more votes to the extent Palin would’ve dropped to elimination first. Again, some second choices get counted and some don’t. When you run alternative scenarios and get different results, you can now detect that the majoritarian consensus RCV advocates highlight as a feature is not exactly as advertised in some scenarios.

If secondary rankings count all-or-nothing, then a fairer system would be to count all secondary rankings with a weighted total of some sort. Scoring every candidate for every voter who ranks their candidates would be a more equitable alternative. In such a system, Begich would be the likely winner, because, well, he was kind of everyone’s second choice. Even scoring second choices with a lower score than first choices, he would likely win.

The added irony, relative to RCV proponents advertising how this will moderate our elections and create a more majoritarian outcome reflecting the consensus of the people, RCV in the 2022 Alaska election did the opposite. The two most polarizing candidates who were ranked either first or last by most voters were the final two standing. The moderate who was the consensus second choice to most and the preferred candidate over each alternative when measured by head-to-head ballot preference, didn’t receive enough first choices to make a runoff. All those second choice ballots mean nothing if you don’t get past elimination. Of those who ranked all candidates in order, here’s the last-place vote tally:

62,874 had Palin last
61,102 had Peltola last
8,297 had Begich last

Another way to describe this is that the two most liked candidates (by number of first-place rankings) are also the two least liked candidates (by number of last-place rankings). The result was a “center squeeze,” where the actual consensus candidate was squeezed out by the two most polarizing and partisan options.

34,049 Palin voters had Begich as their second choice. Only 3,652 Palin voters had Peltola as their second choice. By an almost 10:1 margin, it was clear what alternative Palin voters preferred. Another advertised feature of RCV is that it encourages “sincere voting” with assurances that the spoiler effect is mitigated and you have your backup choice counted in case of elimination. RCV actually creates a system where the spoiler effect can generate outcomes that are the least preferred by the voters who cast a “sincere” ballot. The paradox is Palin voters would’ve been 10x happier with the outcome if they stayed home and didn’t vote for Palin (or, as mentioned, actually voted for their least preferred candidate).

Monotonicity Failure:

If Peltola were able to gain the support of 5,825 Palin voters, she would’ve lost to Begich. Receiving more votes shouldn’t decrease your odds of of winning. In this case, it would’ve happened. Begich would’ve won if more Palin voters actually voted for the winner.

Participation Failure:

If 5,825 Palin voters that preferred Begich to Peltola had abstained from voting, Begich would’ve won. We’re not talking about them switching their first vote to Begich. In this case, the second place finisher receiving fewer votes means the third place finisher would’ve won first place.

Consistency Failure:

If 5828 Palin>Begich voters, 2915 Begich voters, and 2914 Peltola>Begich voters were removed from the election, Begich would’ve won — and if you counted the removed votes, Begich would’ve won.

In addition to my procedural concerns, as I shared yesterday, are these philosophical concerns. RCV doesn’t actually guarantee the result that many are advertising and may paradoxically deliver the exact opposite. If we’re going to violate all of the advantageous aspects of our election system in Idaho, then it, one, better be worth it, and, two, needs to more carefully address many logistical matters in statute. I’m not advocating for alternative voting methods, to be clear, but IF we were to ever pivot from our current system, RCV is not the alternative worth tearing down the house for. I get that the current plurality system with the first-past-the-post is susceptible to spoiler effect, etc., but the current system is easier to audit and provide localized tabulation. Spending more money, introducing longer ballots, potentially splitting ballots, all of it, for a system that actually adds more dynamics that violate long-standing criteria for preserving not just election integrity but the will of the voters, well, this would be a costly step backward, not forward. There are superior alternatives IF Idaho was to decide to change its elections system, and if so, it should be carefully constructed through the legislative process. I’m not advocating for alternative voting systems, to be clear, but simply admitting that RCV is not the best alternative IF we were to entertain options.

Rick Hogaboam is the Canyon County Clerk. Originally posted on Facebook, republished with permission.