Instant Runoff Voting – Fact and Fiction
(In response to “Island Talkback: For your consideration … ranked choice voting,” October 22)
A recent response by Preston Jordan
, a member of the Albany Charter Review Committee, repeats some common misconceptions which are often asserted as fact by proponents of Instant Runoff Voting (including its truncated form, called Ranked Choice Voting).
According to Jordan, “RCV elects candidates that elicit both passionate and broad support, both of which are necessary qualities in a leader.” That’s a common pro-IRV talking point which, despite being vague and hard to quantify in the first place, is actually not true. IRV can elect one candidate, even though another candidate is preferred to that winner by a majority and gets more first-place votes. Here’s a simple demonstration of that:
Per Jordan, “Of course, supporters of C could have strategically ranked B first, but they would have to know this was more likely to help B, their also ran, than C, their heart’s desire. This requires a level of knowledge that few if any would have before, rather than after, the election.” This is wrong for all intents and purposes. All a tactical IRV voter needs to do is identify the two most likely winners, and give a first-place rank his favorite between just those two. That strategy has a winning expected value, and causes IRV to degrade approximately into Plurality Voting. (And in practice, most experienced IRV voters do this intuitively, without even understanding this.)
The suggestion that approval voting would “tend to plurality” contradicts all theoretical and empirical data available on the subject. In fact, the tendency to vote for only a single candidate is empirically more common in IRV elections than in approval voting elections. (In a nutshell, we see this proven every time a plurality voting user votes for someone who is not his sincere favorite, e.g. when many Nader supporters strategically voted for Gore.) Data showing that is discussed in the following link, by a Princeton math Ph.D. who specializes in election theory and was featured in the book Gaming the Vote:
Preston bases that view, by the way, on multi-winner elections exhibiting plurality-like behavior. But I strongly suspect he’s misinterpreting the data. I’ve heard this same basic argument from IRV proponents for years, and it’s never withstood scrutiny.
Next Preston claims that RCV is preferable to plurality. But RCV has virtually always replaced top-two runoff (aka “TTR”, a plurality election followed by a runoff if a threshold of first-place votes is not meant), including in San Leandro and Oakland. RCV may be preferable to ordinary plurality, but compared to TTR it is questionable. One noteworthy difference is that in most of the 27 or so countries that use TTR, there are at least three viable parties, whereas IRV has correlated with two-party duopoly everywhere it has seen broad use. Here’s a more comprehensive analysis by the aforementioned Princeton math expert: http://scorevoting.net/HonestRunoff.html