After establishing that Republican operative Karl Rove is a terrible political prognosticator, Dana Milbank (Washington Post, 11/2/12) does the false-balance thing and attacks polling blogger Nate Silver:
Rove is an easy target because his motive–conveying a false sense of momentum for Republicans–is so transparent. But he has plenty of company among prognosticators who confidently predict that which they cannot possibly know.
There's Nate Silver, a statistician-blogger at the New York Times, who predicts with scientific precision that President Obama will win 303 electoral votes and beat Romney by 2 percentage points in the popular vote. He gives Obama an 81 percent likelihood of winning.
I give Silver a 50 percent likelihood of being correct.
The truth is anybody who claims to know what is going to happen on Election Day is making it up and counting on being lucky.
I suspect that for a lot of media figures–think Joe Scarborough–the same attitude that makes them think that their gut feeling about the election is more valuable than analyzing polling results is the same attitude that makes them not bother to go and check out how Silver's predictions have played out in the past. He's been doing this for two election cycles now, and the fact is he's got a pretty good track record. (By the way, Silver certainly does not "claim…to know what is going to happen on Election Day"; when he says there's an 81 percent chance that Obama will win, he really means that he thinks there's about a 1-in-5 chance that Romney will be elected president–though right now that's down to a 1-in-7 chance.)
In 2008, Silver correctly picked which candidate would win all but one state: Indiana, which he thought would go narrowly for John McCain (538, 11/4/08).
In 2010, he predicted that the Republicans would pick up 54 or 55 seats, depending on whether you wanted his average or median prediction (538, 11/1/10). The actual number was 63. (Looking at his projection for each race individually, he came up with a 59-seat gain for the GOP, but he thought the odds were that the model would be off in some of these races.)
But Milbank did go and actually check to see how often Rove's predictions turned out–a courtesy that he didn't extend to Silver. I'm just guessing here, but maybe he didn't want to spoil the symmetry.