Slate's breakdown of major polls gave Kerry a 276-262 Electoral College edge on 10/19 (above), Bush a 271-267 edge on 10/24

This is (duh) all a crapshoot, not least because in many cases more people hang up on pollsters than talk to them (as I've covered before; more this week from the NYT and NYkr [Update: Slate too]), because of the ambiguity of who is a "likely" voter, because of last-minute voting decisions or changes of mind, and because the election will probably go into the courts for a few weeks again. The AP outlined a few scenarios that will make the election anything but cut-and-dried:

For example, if just New Hampshire and Nevada (or West Virginia) shifted from favoring Bush to the Democrats this time, there could be a 269-269 tie, leaving it to the House to pick the next president and the Senate to pick the new vice president come January. That would leave open the jarring possibility of a Bush-Edwards or Kerry-Cheney pairing, depending on the political leanings of the new House and Senate.

More likely is the chance that results from one or more states could be up in the air for a while because of a recount, challenges to provisional or absentee ballots or lawsuits related to other voting problems. Both parties have lawyers primed to pounce at any target of opportunity this time. And the opportunity for challenges has grown under a new federal law requiring all states to allow people to cast provisional votes if their names don't appear on registration rolls. ...

Michael White, the federal official responsible for coordinating certain aspects of the Electoral College, says he'll be keeping an especially close eye on Colorado, where voters are considering a referendum to divide the state's electoral votes proportionally among the candidates rather than using the existing winner-takes-all formula. A lawsuit is virtually guaranteed if the referendum is approved, meaning the state's nine electoral votes could be a lingering question long after Election Day.