Voter registration is the key to the November election.
The anonymous statistician "Poblano" behind fivethirtyeight.com runs a statistical model that aggregates a large number of public polls, giving different weights to pollsters based on their historical accuracy. The model is very fascinating and worth a look. Does it work? Well, Poblano beat other pollsters in making a dead-on accurate prediction of the Indiana-North Carolina primary results.
Polbano also uses his statistics to tackle interesting questions that other pundits only speculate about such as, "does the drawn-out Democratic primary help or hurt Obama?" The most interesting articles on his website are collected here.
So here is the interesting observation of the day: Poblano points out that Obama could follow a completely novel route to electoral victory. Rather than just fighting it out with the GOP over traditional battleground states (Ohio, Pennsylvania, Michigan), Obama is kicking off a 50-state voter registration drive that could put a large number of other states into play. Poblano observes that a relatively small increase in voter participation in certain groups would significantly tilt the election in Obama's favor in many regions of the country from North Carolina to Wisconsin to New Mexico.
The Poblano voter turnout analysis was done together with Progress Illinois, which has an excellent article about it here.