Speaking of the governor's race in Minnesota, the latest poll shows Dayton and Emmer tied, with Horner trailing by a considerable margin, though still almost within range given that a full fifth of the respondents are undecided. If that is really the state of the race, Emmer is not being penalized for his failure to state clearly how he will address the state's budget problems. He will "redesign government"--what does that mean? And how is it worth a few billion dollars? Does he think our current governor, Republican Tim Pawlenty, has been insufficiently vigilant at cutting taxes and waste? He won't say.
Since Emmer may have a solid 30% base, but a ceiling not much above that, he could win a close three-way race. The problem is that the sane part of the population has two choices, while Emmer is the only option for those suffering from a reality deficiency. Which brings me to the real subject of this post, instant runoff voting. It's needed. Suppose that on election day the undecideds in the poll split their votes evenly among the three main candidates. Then Emmer and Dayton would finish in a dead heat and we might have a crackpot inflicted upon us by 38% of the voters--a coalition of the crazed and confused. In an instant runoff election, however, this dismal outcome would be avoided. The first round of counting would eliminate Horner, but the next governor would be the candidate named by a majority of his supporters as their second choice. If there is anything to my view that two of the candidates are reasonable men and one isn't, then presumably the second choice of one sane man's supporters will be the other sane man--Dayton.
Under instant runoff procedures, it is hard to see a path to the governorship for Crazy Tom Emmer. That is a good advertisement for instant runoff voting. Here are a few more.
(1) Campaigns would be cleaner. Candidates could not risk alienating an opponent's supporters, whose second-choice votes might decide the outcome.
(2) Candidates with extreme views would be at a disadvantage. See (1) above. Of course, truth may reside toward one end of the spectrum. In that case, we shall trust it to be the first choice of at least half the voters--or the first choice of 48% and the second choice of 3%.
(3) No one has to cast an insincere, "strategic" vote. In the Minnesota governor's race, for example, Horner's poll numbers would likely be higher (and the undecideds lower) if no one had to consider the possibility that a vote for him might help to elect Emmer. A vote for Horner would be a vote for Horner, and it would help keep everyone else below 50%.
(4) In general, permitting voters more of a voice--allowing them to convey more information on their ballots--is a result devoutly to be wished.
The only point in your list which is true is #4.
1.) You need to look at Australia, which uses IRV to elect it's house of representatives. The campaigns there are not noticeably any "cleaner"; but cleanliness is a somewhat subjective issue, so I won't focus on it.
2.) IRV actually has a bias towards extreme candidates. The proof of this is in computer simulations: http://rangevoting.org/IEVS/Pictures.html
From those graphics, IRV's preference for less-moderate candidates becomes clear.
3.) Strategic voting is still necessary when their is a strong third-party candidate in an IRV election. Consider:
45%: A > B > C
10%: B > A > C
15%: B > C > A
30%: C > B > A
If this election is just A vs. B, B wins, 55% to 45%. But when you add candidate C, the winner changes to A, just like in plurality. Therefore, C spoiled the election for B.
IRV LOOKS like it should fix the spoiler problem, but that is only because all our third parties are currently so pathetically small. When there are three competitive candidates in an election though, IRV fails in the same way that plurality does. Spoiler elections, lesser-of-two-evils tactics, and a two-party-dominated government.
4.) You're are correct that better expressiveness would be better.
A better answer is approval voting (or even better still, score voting), which ACTUALLY has no spoilers in three-candidate races, and ACTUALLY has no bias toward extreme candidates; and so it actually leads to better electoral outcomes. And score voting would allow more expressiveness too.
If you're curious, check out William Poundstone's book "Gaming the Vote", or my blog, The Least of All Evils, at http://leastevil.blogspot.com
Posted by: Dale Sheldon-Hess | September 08, 2010 at 03:40 PM