Ranked choice voting vs approval and runoff

Disclaimer: I am not completely unaware of the intricacies of voting systems, but there’s a lot that I don’t know.

Once upon a time I was a fan of ranked choice voting, but I’ve decided that for a single-winner election we would be better off switching to a primary+runoff where the primary is done with approval voting, and the runoff (between the top 2 from the primary) in the usual way.  Understand, I am in no way a fan of the first-past-the-post status quo, I merely think that we ought to seriously consider another alternative.

There are three reasons.

First, complex ballots are a problem, and people can screw them up.  A ranked choice ballot is more complex and there are easy ways to get it “wrong” so that you no longer have a valid vote, and then what happens?  There’s also many more bubbles to mark — if 5 candidates, 20 choices, of which only 4 may be marked.  Approval voting, 5 candidates, 5 bubbles, you can (sensibly) mark as many as 4 or as few as 1.  Screwing up one bubble leaves the others still valid, whereas screwing up one bubble on a ranked choice ballot may invalidate the ranking.

Second, the process of tallying up the vote is complicated and difficult to explain.  I’ve seen the explanation, and I agree with how it works, but it is complicated and plenty of people will be confused by it, perhaps not trust it.  Approval voting is easier to explain; each candidate gets the number of people who thought they were “okay”, and the primary will be between “most ok” and “next most ok” from the approval vote.

The combination of these two problems means that sore losers may challenge elections, and it will be difficult to convincingly demonstrate that they are just more losers.  (Of course, Republican sore losers will challenge any result that does not go their way, but with a simpler system, they’ll look more ridiculous). More ballots will be spoiled, and recounts will be tricky and annoying.

Third, approval and 2-way runoff have the property that counting can be split by precinct, and then the sub-results combined.  Ranked choice doesn’t work that way; reasoning about partial results is far harder.

One thing that I think inclines people towards or against ranked-choice is different models of voter motivation.  One advantage of ranked choice is that expressing a secondary preference does not harm your first choice (or so I understand).  If I like Alice then Bob, then Carl, I just express that preference, my 2nd-choice vote for Bob does not harm Alice’s chance.  This is very important to some people.  In contrast, in approval voting, if I “approve” both Alice and Bob, that vote for Bob may reduce Alice’s chances; if voters are focused on a favorite candidate winning (“Alice or bust!”), this may lead them to try to do the sort of strategic voting that we hate in the current First-Past-The-Post system (“I really like Alice, and Bob and Carl are both okay, but I think a lot of other people like Bob so I’ll only vote for Alice, and Carl just in case”).  On the other hand, if your goal is just to get good candidates into the runoff, then you’ll vote for anyone who doesn’t suck.  And if you wish to merely vote against a particular candidate, approval voting makes that easy, just vote for everyone but the bad guy, where ranked choice might require more strategic thought.

There’s another alternative voting system that is more expressive than approval voting and shares some of its good properties, and dodges Arrow’s Impossibility Theorem.  Instead of approving or ranking candidates, voters assign “scores” to candidates, for example, numbers between 0 and 9, subject only to limits on the maximum and minimum scores.  This is called “score” or “range” voting, and its proponents are very enthusiastic.  Tabulating votes can be done in parallel, and it is harder to spoil a ballot, though the ballot is more complex than an approval ballot.  The difficulty, to me, of score voting is that it is hard to reason about what a range of scores means — if I prefer Alice to Bob to Carl, is that 9-8-7 or 9-5-1?  Do I have an easy way to measure my hypothetical satisfaction?  I suspect that it many cases it may devolve to 9/0 approval voting.  One advantage of score voting is that it can be done in a single election; one disadvantage is that it might be vulnerable to accusations of “complexity” or “voter confusion”, though not as much (in my opinion) as ranked choice voting.  Wikipedia calls these “cardinal voting systems”.