On Nobel Prizes, Primaries, and Decision Theory
posted by Nate Oman
The Washington Post has a nice little column today on the basic silliness of the presidential primary system. Sebastian Mallaby writes:
Just like badly designed auctions, the primaries encourage “strategic” behavior that conceals true preferences. Some Democratic voters who preferred Bill Richardson may have chosen not to reveal that, figuring that a vote for him would be wasted. Some independent voters may have preferred Obama yet voted instead in the Republican contest for John McCain, believing that Obama would win the Democratic contest without their assistance. If voters don’t reveal their true preferences, it’s hard to reconcile them successfully.What elections ought to do is discover which candidate would beat each of the other candidates in head-to-head matchups. Eric Maskin, one of last year’s Nobel laureates for mechanism design, will suggest how a better system could do that in a lecture Thursday at Georgetown University. Maskin’s argument is that voters should list candidates in order of preference, so we wouldn’t have to guess whether Clinton would have beaten Obama in a two-person contest. If a majority of voters for Edwards, Richardson and the other also-rans put Clinton higher on their lists than Obama, she would win the contest under Maskin’s system. But if Obama ranked higher than Clinton on a majority of voters’ lists, then he would win. After all, most people would have preferred him.
Of course, my understanding is that the hideously complicated rules governing the Democratic caucuses in Iowa work on something like this system. (Although I am happy to be corrected by those who actually know something about Iowa.) According to the column:
Instead of this common-sensical system, we have a farce: On the basis of a three-point margin over Obama that tells us little about which of the two candidates voters actually preferred, Clinton has transformed her prospects.
The odd thing, of course, is that if one looks at the delegate count rather than at who “won” this or that state, then the results are far less clear cut as “losers” still get a lot of delegates. To be sure, folks still vote strategically but not as strategically as they might in a winner take all election. On the other hand, the implicit assumption of Mallaby’s argument is surely correct. The real purpose of the primaries is no longer the picking of delegates. Rather they simply serve as the focal point of a national contest carried out in the media rather than the political conventions. The real question, of course, is whether Maskin’s system would increase or decrease the number of people wearing silly political memorabilia.
[Image credit: Wikicommons]
January 14, 2008 at 10:19 am
Posted in: Uncategorized
Print This Post
Responses (4)
Drexel Student - January 14, 2008 at 1:04 pm
Maskin’s system (more accurately it is probably a variant of Borda’s system) still allows strategic voting. Voters who see that there are two clear front runners simply move the frontrunner they prefer to the top of their list, rather than throwing their single vote behind him. Under both Maskin’s system and our current system, true preferences are hidden behind strategic voting.
Greg - January 14, 2008 at 4:50 pm
I don’t think you can game Maskin’s system that way, since it’s not a Borda count. It’s a pairwise election based on preferences. Basically, you choose the Condorcet winner. So, if my main candidate is Richardson, but I’d prefer to see Obama win over Hillary, I’d rank them Richardson – Obama – Hillary. In the pairwise race between Obama and Hillary, Obama would get my vote. In the race between Richardson and either other candidate, Richardson would get my vote. The winner is the one who would win in every race. Moving Obama to the top doesn’t make it more likely that he’ll win against Hillary, just more likely to win against Richardson.
Maskin’s system can be gamed, but only if there is a Condorcet tie, where candidate A beats B beats C beats A.
Syd - January 14, 2008 at 9:21 pm
It’s not obvious to me why producing the Condorcet winner is the method we should use. It’s just one of many methods. No method is perfect.The important thing is to have a system where people can accept the winner. You could use a point total system like the Baseball MVP award if people accept it and know the rules well in advance.
Drexel Student - January 15, 2008 at 2:11 pm
Is this the same as the “garden-path” method?
Leave a Reply