Today's article on why Romney is contesting PA at the last minute contains the most interesting chart I've seen this election cycle.
It's here: State by State Nate Silver chart
Basically, what it shows is that from 1988 through the 2008 election in ALL the states where there was significant polling (basically swing states), there have been only been THREE instances where the Polling Average has been wrong!
THREE. Total.
So, for the GOP spin to be correct, Romney would have to equal or exceed that number of incorrect Polling Average calls in the swing states since 1988 on Tuesday -- ALL BY HIMSELF! I guess that explains Nate's 85% prediction certainty!