On the absurdity of national polls

I really can't understand why so many people are so interested in the various flavors of national Kerry vs. Bush polls. Two quick examples.

Kevin Drum finds Bush's lead of 1.7% in the latest Rasmussen poll "very small" and says the campaign is "definitely a dogfight."

Ezra Klein has written three times in the past week on various national polls.

Did we really learn nothing from the 2000 election? Al Gore won the popular vote. George Bush won the electoral vote. George Bush is president.

The national popular vote simply doesn't matter. The winner of the popular vote is not guaranteed to win the election. The electoral college is all that matters. To look at Drum's point more closely, let's suppose the final popular vote turns out to be 50.1% to 49.9%. If those percentages were identical in all states, then it'd be a clean electoral sweep, hardly the dogfight Drum sees.

Now, if you want to argue that the electoral system needs a serious overhaul, go right ahead. But you just can't ignore that it's what we have right now.

I understand that national polls can be seen as indicating general trends that could have consequences on a state-by-state basis. But with 50 states, there's just too much noise to draw any firm conclusions about the election from a national poll.

I'd love to see more discussion of state-by-state polls. Maybe it's just that not many polls are currently available. But in the absence of the useful information they provide, we shouldn't look too hard at national polls just to keep up the horse race discussion.

UPDATE: Fellow Swat alum and Earthworm Collin Peng-Sue e-mailed me with a link to Electoral-Vote.com, which has the information that actually matters. It's probably not perfect, but it's far more informative than national polls.