<$BlogRSDUrl$>

Rhinocrisy

12 October, 2006

Kinky elections

I'm coming back from my 2-year hiatus! Why? It's election method time!

2006 is an election year for Texas governor. Rick Perry (the one with the hair) is the incumbent, polling 34%. There are three other major candidates on the ballot: Chris Bell (D), Carol Keeton Strayhorn (ex-Republican, decided to take her chances in the general rather than the primary), and Kinky Friedman (comedian). All 3 are polling about 20%.

It's quite possible that none of the other 3 could beat Perry in a one-on-one race. If this is the case, then our current system comes to the same result as virtually any other decent method. On the other hand, it's quite possible that any of the other 3 candidates would beat Perry in a one-on-one election. That means Perry could be elected despite 2/3 of the voters preferring anybody else to him.

In that circumstance, IRV would offer an improvement: Perry would not be elected. However, it wouldn't be perfect: it's still possible that the vast majority of the electorate would prefer one of the losing candidates to the one that won. For example, Strayhorn might end up with less first-place votes than Bell or Friedman because most Republicans vote for Perry, but be much preferred to either because most who voted for Perry prefer her to Bell or Friedman.

A far worse case for IRV, though, is if only one or two of the candidates can beat Perry. If Strayhorn got fewer first-place votes than Bell or Friedman, she'll be knocked out immediately. But it might well be that she's the only candidate among the 3 who can beat Perry head-to-head.

While this might seem like a strange what-if brought on by the special circumstances of a 4-way election, circumstances like these are latent in virtually every election we hold. After all, if John McCain had been running head-to-head against George Bush in the general election in 2000, he would have won a landslide. Under a sane voting system, he wouldn't have been forced out of the race because a majority of a minority preferred another candidate.

Comments

Is there some metric for available information about a candidate? I.e., if there are enough people in a field, it becomes difficult to know who is interesting and why, and to conduct meaningful debates. (LD-style debates are right out.) This is one appeal behind a primary process, which means McCain must always lose to Bush. 

Posted by saurabh


I don't know what the limit on candidates is before you get information overload, but I have a feeling it's more than 2. Already, the polls and elite (media, money-givers, etc.) rule candidates out as unelectable (take the Libertarian candidate in TX-Gov or virtually anywhere for that matter) and their presence on the ballot doesn't detract from the rest of the candidates. You might not have a series of 1-on-1 debates, but then again, most debates these days suck and it might be that the debates would benefit by having candidates with a wider range of ideas. Multi-party democracies in Europe sometimes have debates between heads of leading parties.

The primaries could still be there, but instead of being mostly for a nomination, they'd be mostly for legitimacy and resources, much like the Sierra Club endorsement, only more democratic.

CT-Sen and TX-Gov both have shades of that this year, where candidates (Lieberman, Strayhorn) decided they could do better in the generals than they did or would have in the primaries. 

Posted by Dan


The real elections where this would seem to matter is elections like Arizona, District 8 , where the Democrats helped elect an extremist Republican who will now lose to the Democrat. IRV could possible help in districts like this, though it could also possibly fail, if the non-extremist was eliminated first. 

Posted by Dan


This page is powered by Blogger. Isn't yours?