For the next 72 hours, tens of thousands of people are going to spend a lot of time and energy worrying about what will mostly turn out to be statistical noise. Don't be one of them.
Most national and state polls will be moving around inside their margin of error. As such, in the absence of consistent movement across the board, you shouldn't worry about it. And even then, polls are only accurate to a degree. The combination of late-breaking undecideds and the varying effectiveness of the candidate's GOTV operations means that the final national polls are usually still off by a few percentage points on average. In a race this close, that means we're going to be in the dark until Election Night.
Three things really matter:
-The ground game - Who will get out their people? A surge in turnout for Kerry could make the difference.
-How the undecideds break - The much-debated "rule" that undecided voters tend to vote against the incumbent may or may not hold (I tend to think it will). If it does, Kerry will be in a strong position; if not, he's in big trouble.
-The Electoral College math - What combinations of states get you to 270?
You'll hear all kinds of nonsense before this is over, but remember, nothing else matters except the above three things. Silly prognostication won't change that.
Update: Just to illustrate how dead-even the polls are, UNC political scientist Jim Stimson's methodology for combining national trial heat polls and stripping out statistical noise shows Bush at 50.4% of the two-party vote and Kerry at 49.6%. But using an alternative methodology that only employs polls from the last two weeks and weights them differently, he has Kerry ticking up to 50.4% of the two-party vote.