Wednesday, November 19, 2008

Shooting Down IRV

I hope that, by now, I've convinced you that preferential voting is a good idea. Now, onto phase II: convincing you that preferential voting is a bad idea. Or rather, that it's most popular form, Instant Runoff Voting (IRV) isn't such a good one.

Consider the following election. There are three candidates--A, B, and C--and 9 voters. Candidate A leans strongly to one side (left, right, or whatever you prefer), while B and C lean the other way, although B is a bit more centrist. Let's suppose then that four voters would prefer A the most, and prefer the more moderate B over candidate C; three would go in the precise opposite direction, prefering C over B with A in a distant third; finally two prefer candidate B, followed by C, followed by A. Like this:

4: A > B > C
3: C > B > A
2: B > C > A

Who wins? By plurality, A does; but the supporters of B and C will point out that A only won because their candidates split the vote; clearly their camp should have held a primary! But IRV would proceed by eliminating the candidate with the fewest first-ranked votes, in this case B, and passing those votes to the voters' next-higest ranked candidate, in this case C, making C the winner. This is the same result that would have occured had the "BC" camp held a primary to decide on who their candidate should be. Which is some progress; perhaps we could eliminate the need for primaries by moving to IRV, and allow supporters of minor-party candidates to speak their honest opinion instead of being forced to choose between the lesser of two evils.

But consider this: what if A's supporters had decided to disingenuously rank them at the bottom of their ballot? Then we get this:

6: B > C > A
3: C > B > A

Now, B wins. Which, for A's supporters, who unanimously prefered B over C, is an improvement! We've traded minor-party supporters lying about their preferences for major-party supporters lying about their preferences (in both cases, in order to avoid what they see as an even worse outcome.) And while some might bitterly remark "Good, let them see how it feels!", upon calm reflection I can't say that this is a good thing. If we're going to change our voting methods, we should do so with an eye towards eliminating these sorts of insincerity-encouraging effects as completely as possible, rather then push them off to some other segment of the electorate.

Unfortunately (and this will be covered in more detail in a future post), it has been mathematically proven that no reasonable voting method is entirely immune to these incentives. While plurality performs quite poorly--allowing two major groups to gain a stranglehold on all political discourse across a nation--IRV doesn't do that much better. In a highly-polarized election, it can (and mark my words, it will!) tend to eliminate moderate choices, and if it doesn't, the decision of which moderate candidate (when there is more than one) ultimately wins, becomes very sensitive to the order they are eliminated, which can hinge on an incredibly thin sliver of voters. If you thought re-counts were bitter fiascos now, just wait to see the chaos of a close 4-or-more way IRV race. In the most pathological case, consider an "ultimate compromise" candidate, one who is everyone's second choice. Under IRV, such a candidate has no chance of victory, as he has no first-place votes. But I can't help but think that such a compromise candidate would be a very good choice as an election winner.

Next up, an alternative preferential system.

No comments:

Post a Comment