I guess I must be cranky today, because I already wrote two long complaining posts, and, yup, here's #3. I'll try to keep it short this time.
Nate Silver turns to GOP '12 today, and talks up the chances of Palin, Romney, Huckabee and Gingrich because they're currently well known, and places hypothetical bets against Thune and Pawlenty. I think he's assuming far too much. Thune and Pawlenty are unknown now -- but should their campaigns go well over the next twelve months, they'll be household names among Republican caucusers in Iowa by February 2012. Early good polling based on name recognition for weak candidates really is meaningless -- see Rudy Giuliani '08, among many others. Newt belongs in no one's top tier. And, generally, of Silver's group of four only one of them, Romney, appears to be certain to choose to run in 2012 (all four are running now, but may decide against eventually going forward).
I suppose my real point is this. Silver refers to polls as "the objective indicators." Polls are objective indicators, but they are not the only ones. Others are campaign activity so far (have they build a staff?); endorsements -- or, at this stage, near-endorsements; and fundraising and fundraising capacity.
Now, I'm not sure whether Thune or Pawlenty has a path to being in the eventual top group, as I said yesterday. I agree with Silver that it's a bit tough to see why they would break out of the pack, although I don't agree that bland is necessarily bad -- mobilizing a faction is one way to win, but usually the party decides on someone broadly acceptable to all factions, and so bland stands a chance. But I don't think that polling badly now is a real problem for either of them. If they win the support of party leaders, then they'll poll well when it counts.
At any rate: Silver also includes the Intrade odds, which I didn't mention yesterday. I'd say that Barbour (2%), DeMint (1%), and Perry (1%) are all seriously undervalued, and if I were placing wagers that's where my money would be.