Alex wrote,

IRV is like a post-season tournament,
where teams and players qualify by doing well during the regular season
and procede through a series of eliminations.
Actually, this is still not very IRV-ish, since each team competes in a series of pairwise tests in the elimination series. IRV would be more analogous to the current college football season, where we use the results of different (largely disjoint) athletic conferences to pick two teams to play in a national title game. The winner is declared the champion, but in some years people claim that the best team was left out of the title game. There's no way to be sure, since a team doesn't play all the other teams.

Condorcet is more like
boxing, where a person only keeps the title by being undefeated (cycles
are deadlocks where NO person is undefeated).
As you pointed out, it's not a perfect analogy, since every boxer doesn't face every other boxer in any predictable time span. Maybe a better example is the eight-team college baseball World Series, where teams play a double elimination bracket. Not every team plays every other team, but if you are knocked out you had to lose to a team that lost to the champion by extension. The cyclic ambiguity is possible, if you beat the eventual champion (giving them their first and only loss), only to be beaten by two other teams before reaching the championship game. This sort of outcome is uncommon but it does happen.

Or perhaps the race really was a good analogy for Condorcet? After all, a race does compare every candidate to every other candidate in a pairwise fashion, simultaneously. But there's no analog to the cyclic ambiguity.

Approval Voting is like
gymnastics, where each candidate is judged independently.
Or maybe a more binary scoring single event like archery. Well actually, I think Archery scores on a 1-10 scale depending on how close to the target you hit, but in the Olympics I don't recall hearing anything but 9's and 10's.

-Adam


----
For more information about this list (subscribe, unsubscribe, FAQ, etc), please see http://www.eskimo.com/~robla/em



Reply via email to