The left and the Obama campaign have been pooh-poohing the notion that Mitt Romney has any chance of winning Pennsylvania, despite devoting millions of dollars in ads and visits by the candidate himself to the state this weekend. David Axelrod has gone so far as to say he will shave his moustache off if Romney takes the Keystone State.

But there is a method to Mitt's supposed madness, as Jay Cost so ably explains:

Broad context: PA outside of Philly County has been trending red for 20 years. It has so far been checked by Dem turnout in Philly County, but Philly County's population has been flat. So turnout increases in the county are from turnout machines/enthusiasm alone. At some point, that could breakdown.

So assume:

(a) Total PA turnout is up 3% over 2008. Philly County comprises 11.5% of total PA electorate (similar to 2004, less than 2008).

(b) Romney wins non-Philly county 54-46. (Slightly better than Bush '04, who won 52.5 to 47.5)

Then:

Obama MUST net 433k votes out of Philly County to win. In 2008 he netted 478k votes. In 2004 Kerry netted 410k votes. In 2000 Gore netted 350k votes.

Tweak the assumptions to lower Philly turnout, increase non-Philly turnout, increase Romney share of non-Philly. And Keystone State goes red.

The electorate has changed since 2008 in several states, including Pennsylvania, Minnesota, and Colorado. These states are all trending red and liberals may be in a for a very rude surprise on Wednesday morning.