A couple quick thoughts on what happened tonight:

1.) As I mentioned earlier, I think John Edwards played more than a trivial role here. Between double-teaming Hillary at Saturday's debate and punching her in the nose (rhetorically) after her brush with tears, I think he evoked a certain amount of sympathy for her, particularly among women. That dovetailed with voters'--again, particularly female voters'--reluctance to see her get bounced from the race so quickly, which the media told them was about to happen (and almost seemed to be rooting for).

That Hillary did so well, according to the exit polls, among voters who made their decision in the last three days (whom she narrowly lost) and on Election Day (whom she won) would seem to support this theory. You'd obviously have expected Obama to do better among voters who made up their minds recently, given his Iowa momentum.

2.) Moreover, by doing relatively poorly in New Hampshire, Edwards probably coughed up a significant number of working-class voters, which the exit polls show favored Hillary by a significant margin.

3.) Speaking of working-class voters, one big question is how Obama did so well among them in Iowa, but so poorly among them here. A number of people have suggested that racism was a factor: people want to be perceived as racially progressive in a public forum like a caucus, so it wouldn't have been a problem for Obama in Iowa. But in the privacy of your voting booth, you can let your prejudices run amok.

That may be part of it. I'd suggest a slightly different theory: The working class is hardly monolithic. It's entirely possible that the white working class in Iowa and elsewhere in the midwest is simply more progressive on race than the white working class in New England. Des Moines, after all, recently had a two-term mayor, Preston Daniels, who was black and whose geographic base of support wsd the working-class precincts in the East and Southeast part of the city. I don't know as much about the success of African American policians among working-class whites in New England (Deval Patrick would be instructive here), but obviously this demographic has a reputation for being somewhat racist. (There's a reason Barry Bonds said he could never play in Boston.) The reputation could be unfair, but it's certainly worth exploring further.

The big question going forward, I think, is whether black voters in South Carolina look at the Iowa rersult and conclude that whites will happily support an African American presidential candidate, or whether they look at New Hampshire and conclude that they won't. Or something in between.

4.) My back-of-the-envelope calculation is that about 40 percent of registered independents voted on the GOP side, versus about 60 percent on the Democratic side--which is on the low end of what most pundits and pollsters were expecting. This probably made some difference at the margins.

--Noam Scheiber