Why Ranked-Choice Voting cannot produce a more moderate politics
It’s in the math.
Advocates for the ranked-choice voting method claim many advantages for their voting system, not least among them producing a nicer, more moderate politics.
Something closer to the opposite occurs. In ranked-choice, you don’t vote for a single candidate in a race, but rank the candidates according to your preferences.
By doing so, advocates declare the following will happen:
With RCV, candidates also compete for second-choice votes from their opponents’ supporters, which lessens the incentive to run negative campaigns. In RCV contests, candidates do best when they reach out positively to as many voters as possible, including those supporting their opponents.
Why this doesn’t turn out to be true, we’ll illustrate with a bit of math and some common-sense assumptions.
Assume a race with 100 voters and two candidates. Let’s assume that voters’ ideological preferences are evenly distributed across the political spectrum, with exactly 50 conservatives and 50 liberals.
To win, a candidate needs 51 votes. By my definition above, the 51-vote winner would have the broadest ideological appeal, holding the votes on their side of the spectrum, while winning the votes of one of more voters on the other side.
Now let’s assume a four-candidate race with 100 voters. Let’s assume that the four candidates are, like the voters, evenly placed across the political spectrum from left to right.
Here is my theoretical outcome:
In this example, in a traditional first-past-the-post (plurality) race, candidate “C” wins over candidate “B” with a total of 40-39.
Under ranked-choice, the outcome is not immediately clear. Each voter gets two choices, 1st and 2nd. To win, a candidate would need to earn 2nd choice votes from among their opponents’ supporters.
Advocates would have you believe that this race for 2nd choices would occur in the political middle. In my example, the richest source of 2nd place votes are with the supporters of the two leading moderate candidates, B and C.
Oddly enough, these 2nd choice votes will never be counted under ranked-choice scenarios. Only the 2nd choices of the losing, fringe candidates count under RCV.
Using the above example, no candidate wins in the first round, as all are below 50 votes. The candidate with the fewest votes, “D,” is eliminated. The second choices of “D” voters are distributed to the three remaining candidates. Likely most (if not all) of the 2nd choices of the right-wing D voters went to the center-right candidate C.
Even if all 10 D voters offered a second choice, and all those 2nds went to C, we’re not done yet, as no one has crossed the 50-vote mark.
Next, left-wing “A” is eliminated. A’s voters’ second choices are distributed to the two remaining candidates, most likely to center-left “B.” And there we stop.
The second choices of the voters supporting the two finalists, B and C, are never considered.
To win under ranked choice, leading candidates must appeal to the voters of candidates who will not make the finals. By definition, it is the fringe candidates who will be eliminated, and their votes redistributed.
In my example, the two leading centrist candidates would not craft messages to appeal to each other’s voters. Each centrist candidate would appeal to their respective fringe.
Under this scenario, politics becomes more polarized, not less.
The scenario that RCV advocates think they are guarding against is the following. Assume 99 voters and three candidates, evenly distributed across the political spectrum:
Under a plurality system, the right-wing candidate “G” wins, 34-33-32. Under an RCV system, no candidate has a majority.
The 3rd-place “E” is dropped. Their votes are redistributed to F and G. Since F is the centrist candidate, F gets more second-choice votes from among E voters. F, the centrist candidate, prevails over right-wing G.
I have searched America for the above scenario. It just doesn’t happen.
More common is the first instance, where the centrist candidates have to appeal to the fringes, rather than the middle.
A good example is the most recent mayoral race for Minneapolis (2021) in Minnesota. In a 19-candidate field, the more centrist incumbent Mayor Jacob Frey had to appeal to supporters of more left-wing candidates to win. After multiple rounds of counting, Frey never achieved a majority of votes. (He topped out at 49.1 percent).
At no point in the process were the 2nd choice preferences of the two finalists (Frey and Kate Knuth) ever considered. Only the 2nd choices of the fringe candidates in the race were considered, and those candidates skewed to the left.
What RCV does is to try to simulate the two-candidate outcome in a multi-candidate field. But it has to create multi-candidate races in order to do so. It creates the very problem it is purporting to solve.