The thing is, if Romney won, Nate still would of been correct. He was not making predictions, he was making confidences. He was saying that in 10% of election trial runs, Romney would win. The fact that 90% was in Obama and not 100% was because he was modeling the probability that the polls were in error.<quoted text>
yeah I read the one questioning his brilliance and saying all he did was look at the polls.
He never claimed to do anyomore than that and that was the point.
He looked at them, he weighted them, and he made mathematical predictions.
Predictions which have proved highly accurate. In at lest two elections that I know of.
And that's good enough for me.
Sophisticated but simple.
The only way to proof Nate - would be to have those trial runs and see what the percentages of wins/looses were. Just saying Obama won as he "predicted" is meaningless. And Nate himself explained that a few times.
Its the same as saying I am tossing a coin and overall 50% of the times it will be heads. This does not preclude three tails happening in a row, nor would that invalidate.
The real lesson from all of this was not Nate, but that the polls actually correlated with polling behavior. There were serious questions about that ("likely voters", lying voters etc etc) mainly from the right.