Like Nate Silver's stuff, but don't want to cough up the price now that he's at the NYT? Go discount, with Tom Holbrook's -- hey, wait! They're both free! So dial up both sets of projections. Quick comparison: Holbrook, a political scientist, is only using district-level polling, and he bases his projections on the past relationship between past polling and past election results. Silver (whose PECOTA baseball forecasting system is letting me down this year but helped me win my fantasy league twice in recent years) throws a lot more stuff into the pot, including the national generic ballot question and expert projections of individual districts. Which approach is better? Both. What I like about Silver's projections is that he presents them as probabilities, not only as specific point estimates (it's much more useful to know that Silver projects a 57% chance of Angle winning than to know he projects her to win 49/48, and the best thing is he gives both). Silver has projections for every Senate, House, and Gubernatorial race; Holbrook so far only has point estimates, and only for Senate elections, but he says he's working on the House, too. And don't forget about Pollster.com, which has the best simple poll-of-polls out there, at least in my opinion.
Via yet another tweet from Brendan Nyhan. You do follow him, right? (And while you're at it -- you do follow me, too, right?). And, yes, face it, holdouts -- if you're a political junkie and want to really follow elections these days, twitter is, if not absolutely necessary, certainly highly useful.