>That would be okay. I try an figure how they can call a winner with only 30% of the vote counted. At one point in Michigan they show Bush ahead 48-45% with 32% reporting and called the state for Gore. I would like to borrow that crystal ball for about 5 minutes and I promise to return it.

Heck, they call most states without any counts at all, just a few verifying exit polls for the pre-election polls in most states.

Technically, the VNS folks (the projectors) are scientifically very good at this, as witness the accurate early calls of Michigan & PA (but not FL :) They have key, early-reporting precincts that are excellent indicators in close states, when used with a combination of exit polls and verifying actual tabulations, and also normally presidential elections aren't so close that every state matters. There have been very close states that weren't called quickly in the past, too, but no one cared because there was a big margin by one candidate in all elections since VNS was created. This year the many very close states were deciding states, and much pressure was on to call them, to the misfortune of the projectors. No easy paycheck this election.


Obviously from this lesson, these attempted projections must be thrown out whenever votes are extremely close, as in FL and also now add NM, whether the states are critical to the electoral count or not. And, these projections are still subject to human reporting errors, apparently the cause of the first bad call in FL (but the second bad FL call has no technical excuse I've heard yet ("oh, it looked like Bush had enough votes to win" duh!) Nothing scientific about that bad call, they were just tired of waiting around all night and anxious to be first to call the election. Whoever finally wins, both those states are still too close to call 6 days later, and I think there will be a lasting memory of this in the media - this really trumps "Dewey beats Truman" as the all-time classic media blunder.