What do the prediction markets know that pollsters don't?

Sorry to bring up something as profane and childish as electoral politics here (normally I try to hold myself above the usual CNN Airport Network fare), but I've noticed that despite the fact that the polls are showing Obama and McCain in a statistical dead heat, the prediction markets are still betting on Obama by a fairly wide margin. National polls show Obama's lead over McCain narrowing to between 2 and 5 percentage points, but contracts that pay $1.00 if Obama wins are going for about $0.60 on Intrade. It's possible that the polls show the individual state contests playing out so that Obama is way over McCain in the electoral college vote, but I doubt it – seems more likely that the bettors think they know better than the pollsters. And though the polls were taken before McCain's statement about not knowing how many houses he owns, the prediction markets have been at 60% for Obama for a while now. And before you blame immature and small markets for the discrepancy, note that betting volumes are much higher than they were in 2004.

And for those interested in what the markets have to say about Obama's vice presidential pick, Joe Biden's contract is trading at 41.4, Evan Bayh's at 24.2, Tim Kaine's at 12.0, Wesley Clark's at 8.7, and Senator Clinton's at a lowly 12.0. On the Republican side, Pawlenty and Romney are tied at about 25 each.