Looney Labs Icehouse Mailing list Archive

Re: [Icehouse] IGDC Winter 2009?

  • FromDale Sheldon <dales@xxxxxxxxxxxxxxxxx>
  • DateWed, 18 Mar 2009 17:57:18 -0400 (EDT)
On Wed, 18 Mar 2009, Christopher Hickman wrote:

I forget what it's called, but I like a method I once saw for a board game award. Everyone submitted their ratings, which were converted to ranks. The total of ranks was added up, and the lowest one dropped. Then the ranks were recomputed, and the process repeated until the winner was the last one left. Somewhat complex, but it eliminated the difference in ranges of the ratings.

That sounds like Instant Runoff Voting, a system that guarantees a majority winner instead of a plurality winner, but it seems to me that it is designed for single winner contests, not good for determining overall ranking of entrants.

*GAK*ARGH*SPASM*

* Instant Runoff Voting is a derivative of the Single Transferable Vote method, which was designed for multiple-winner elections; IRV is simply STV where the number of open seats is equal to one. And you could use the order of elimination to determine the full order (e.g., if it were a three-seat STV election, candidate C would be a winner, but if it were a two-seat STV election, they would not; therefore candidate C came in third). And STV actually works better for what it was originally intended.

* IRV does not guarantee a majority winner; first is the likelihood of exhausted ballots (e.g., in an election with candidates A, B, C, D, & E. I vote "A>B," ignoring C, D, & E, but both are eliminated before a winner is declared; my ballot is exhuasted and it's possible no candidate will have a majority victory if there are enough such ballots.)

** But even ignoring this, the widely-circulated claim that IRV *guarantees* a majority winner is false. By the end, the winner has more than half the votes, but that doesn't mean that a majority of the voters prefered them over every other candidate. Such a failure just occured in the Burlington, VT, mayoral election; Andy Montroll was preferred by an 8% margin over the IRV winner, Bob Kiss.

* It sounds like a weird combination of IRV and Borda Count. Intriguing. But yes, complex.

--
Dale Sheldon
dales@xxxxxxxxxxxxxxxxx

Current Thread