Hey Nate!

Query Nate Silver:

Suppose that a particular state flips contrary to model . . . for example NH goes Romney.

Can you then recalculate the chances that other states follow in the same direction?

It seems like if your model is underestimating chances in NH for Romney, it may well be underestimating his chances elsewhere as well.

Also, when you run the simulations, do you account for the possibility that your model may be biased in a particular direction? In other words does your model make a contrary result in WI influence the likelihood of a similarly contrary result in MI or PA?

Finally, why don't you run 100,000 run sims? Or even 1,000,000 run sims? Since no doubt you can afford the processing power.