political forecasting is bogus

I respect Nate Silver's work at fivethirtyeight, but how can he write without noting the wrecklessness of it that:

There has been abundance of polling in California and, the protestations of Meg Whitman’s campaign notwithstanding, it has been quite consistent, showing margins ranging from an exact tie to a small, 5 percentage point advantage for Jerry Brown, the Democrat. Mr. Brown is now a 60 percent favorite to win, according to the model; two weeks ago, Ms. Whitman had been almost a two-to-one favorite.

If you think about it, that's a really large swing in a short-time. In addition, there haven't been any "game changers" to note. Not a large scandal, major endorsement, a gaffe, etc. There's been just the regular campaign. Which should have been accounted for in the first place.

No way in a calm environment should a close to 2-1 favorate (67%) go to be a 5-3 dog. This is not credible. It is just people watching what is going on, and making up complicated formulas to really demonstrate they don't know anymore or better than me. If the polls shift again, then Whitman will be back to being the favorite, which will vindicate the original posting?