Nate Silver gives Romney a ~20% chance of winning because there's a chance that the polls are systematically wrong. Each polling firm comes up with their own li

Nate Silver gives Romney a ~20% chance of winning because there's a chance that the polls are systematically wrong. Each polling firm comes up with their own likely voter model and participation model. If they're are correct in the aggregate, there's essentially a 0% chance of Romney winning, but they might not be correct (and there were times in the past where they were systematically wrong). Basically, don't confuse uncertainty about whether the polls are accurate with uncertainty about the election if they are accurate because if they're accurate then Obama will win.