Opinion: It’s time to rethink ranked choice voting

How should we elect our mayor?

The current Boulder City Council is likely to push for charter changes to shift our city elections to even years, so it makes sense to also rethink the process of electing the mayor that was voted on in 2020. That change necessitated altering councilmembers’ terms to address the shift from nine to eight councilmembers. But it also required indicating how the mayor would be elected. The charter language now specifies, somewhat generically, that “The election shall be conducted by ranked choice (instant runoff) voting.”

I suspect that most people voting for the change to direct election of mayor had not thought about what this language really means or what the alternatives are. Besides, it was clear at the time that the proponents of the direct election of mayor had not thought this through; their ballot issue ended up being completely rewritten by the city.

Therefore, in my opinion, it makes sense to look at the alternatives as to how this voting will be done. (Also, whether the mayoral candidates should be required to have served on the Council beforehand, so they have some experience, needs a relook.)

I should point out that none of the systems I discuss below require that you actually like any of the candidates, only that you have preferences between them.

To me, the obviously best way for us to pick a mayor is for voters to compare candidates in pairs head-to-head. This is called Condorcet voting and has been around for hundreds of years. For example, suppose we have three people running for mayor — A, B and C.

If a majority of voters prefer A over B, and a majority prefer A over C, then it makes total sense to select A as the mayor. One Condorcet system uses a ballot with rows with the candidates’ names down the side and columns (first, second, etc.) for ranking. The rankings indicate who you pick head-to-head. Equal rankings and skipped rankings are allowed. Alternative designs also exist. (There’s lots of good information at equal.vote/condorcet.)

Next best is approval voting. The only question on the ballot is whether you would “approve” of this person as mayor or not. The candidate with the most “approval” votes wins. This is simple for both the ballot design and the software. The ballot is like our current one: one oval to fill in per candidate. You only fill in the oval if you “approve” of that person for mayor. Obviously, if there are three candidates, and you “approve” of all three, or disapprove of all three, your vote becomes irrelevant. So, to have some effect, you need to “approve” of some, but not all, of the candidates.

The version of ranked choice voting most practiced today requires that you as the voter rank order all the candidates. It uses the same sort of ballot as Condorcet might use — a square matrix of ovals, with one row for each candidate and one column for rank 1, 2, 3, etc. You fill in the oval in the first column for the candidate you rank first, the oval in the second column for the candidate you rank second, etc. But indicating a tie or skipping doesn’t work here.

Where ranked choice fails is in the scoring. The standard procedure is to drop the candidate who got the least first choice votes. Then, for all the voters who ranked that dropped candidate first, their votes are reallocated to the rest of their candidates by moving that voter’s second choice to first, third to second, etc. Then scoring is repeated until only one candidate is left.

Unfortunately, this procedure can easily end up with an outcome that is obviously not what most voters want. For example, suppose there are three candidates, and, because of their positions on the latest hot-button issue, 35% of voters rank them A, B, C; 35% rank C, B, A; and 30% rank B first. And suppose, for all voters, middle-of-the-road candidate B is generally acceptable. But under ranked choice, B is dropped because B got the least first-place votes. So, A or C wins, even though B beats both A and C in a head-to-head. Clearly, ranked choice voting is not the best way to go, unless your goal is to produce the most polarizing outcome.

Rather than having to go back and fix mistakes, as happened with the initiative process rules and the direct mayoral election measure, why not set up a working group of experienced Boulder citizens to do the detailed work first?

Popular Posts

Opinion: Opportunity for the new Boulder City Council

Opinion: Is this the end of Boulder as we know it?

Policy Documents: Impact Fees and Adequate Public Facilities