So keith taunted me with a little e-mail about my previous post on the US election. He thinks, as he put it, that my track record of misguided election predictions will be intact.

Hm. Maybe. I wrote a long note back to him, and am going to restate the least funny parts here.

The presidential race is a complete toss-up. Any last-minute surprises (cf. Bush’s DUI snapshot of the last weekend of the 2000 vote) could easily tilt the election dramatically. Turnout (as mustered by the “get out the vote” or “GOTV” plans of both parties) will probably be the decisive factor in the really close states.

Slate has updated its Election Scorecard today, and is still calling the race for Kerry. But it’s incredibly tenuous. I think that Real Clear Politics has a view of the state-by-state race that makes the undecided nature of the race more clear. They show 116 Electoral College votes as a complete toss-up, not counting the 82 EC votes that are “leaning” one way or the other.

It’s at this point that the sheer toss-up nature of the vote is clear. The will of the voters, for all practical purposes, is evenly divided. I am reminded of my feeling about playoff sports: a season of play tells you which team is the most consistently superior, with a reasonable level of statistical meaningfulness. A playoff is more or less a crap-shoot, even in the case of 7-game series, but especially with a one-game runoff. You can always resort to the argument that one team was chosen as the best on that day, but it’s unsatisfactory to me that for some reason that is seen as the team to remember. One rarely recalls the team with the best season record, unless that team also wins the play-offs.

Back to elections, the American two-party arrangement at the federal level makes every competition a close one, and one decided every 2-6 years, depending on the office. This isn’t as bad as it might seem to countries with multiple federal parties of substance. The key is that both parties are seeking to win half the voters. This necessarily makes the parties avoid severe extremism in their presidential candidates (like it or not, neither W. nor JFK represent a political extreme, either within or without their respective parties), that being reserved for locally elected candidates like congressional representatives.

With such a close, and essentially randomly decided race, I almost wonder if it wouldn’t be better to use some sort of sortition, and make the essentially random nature of the process be made explicit. One variation of this election-by lottery that has been proposed is to randomly draw the candidate from a single ballot.