Green party presidential candidate Jill Stein recently gave an interview where she was asked, among other things (you should read the whole thing!) to talk about instant runoff voting. Her response parroted many of the incorrect claims we hear about IRV, but it's a more disappointing to hear it from Stein, since she is trying so hard, has so much against her and her party, and has so much to gain from picking a more-effective voting reform. Doctor Stein, this one's for you:
[If] you really wanted to vote for me, but you weren't sure I was going to win, it lets you go ahead and do that[.] Because it lets you vote for me as number one, if that's what you want to do, and it lets you vote your fallback, maybe your lesser evil choice, as number two. And what it does is set things up so if I don't win, your vote gets reassigned to your number two choice.
It's a win-win. It ensures there will be no splitting of the vote, that your vote might have unintended consequences, all these things that people are told to be afraid of[.] Instant runoff voting [...] ensures that the candidate that gets elected is the one that the most people can actually support.
The first paragraph contains a fair description of how IRV works, but every claim from the second one about the implications of IRV is false. An example similar to one I've used frequently:
- 45%: Romney > Obama > Stein
- 10%: Obama > Romney > Stein
- 15%: Obama > Stein > Romney
- 30%: Stein > Obama > Romney
This example (I hope obviously) doesn't reflect a likely outcome, I'd just like to use names, instead of bland identifiers (like A, B, C) to highlight what's going on here. First, let's take note that, in the example, 55% prefer Obama over Romney, and 70% prefer Obama over Stein; so, the candidate that "most people can actually support" is Obama. Second, note that the winner, by IRV's rules, is Romney. However, if Stein would have withdrawn from the race, then Obama would have won instead. That means there was a "splitting of the vote" (between Obama and Stein) which had "unintended consequences" (electing Romney).
I understand how Dr. Stein is able to come to these erroneous conclusions. If a third-party is small (as our third parties are now) and remains small (as our third parties do now) then under IRV it will always be eliminated from the election, well before it is able to act as a spoiler. That makes it easy—although incorrect—to conclude that they would never be spoilers. But, Dr. Stein, wouldn't it be nice if your party were to grow? And wouldn't it be nice if growing meant it could eventually win? That won't happen under IRV though, because if it grows, it will eventually become a spoiler, and the voters hate spoilers. Even the accusation of being a spoiler (I don't want to start a Nader/Gore argument) is enough to ruin a party's votes in subsequent elections for years.
A better voting reform target for the Green party (and indeed, all third-parties) is approval voting. With approval voting, there really are no split votes, no spoilers, and parties can grow, and grow until they win.