So what exactly is going on in the Minnesota recount, and can Al Franken overcome GOP Sen. Norm Coleman's statistically tiny lead from the initial vote count?

The answer is a definite maybe. Here's why.

Franken got a big boost yesterday when the first wave of recounting cut Coleman's lead from 215 votes down to 174, with over 80% of ballots left to be sorted out. Franken's gains came from ballots that weren't marked well enough for a machine to clearly pick up but were obvious to the human eye (what is also called an "under-vote").

So what's next?

Today's count will continue, and there's no telling what could happen -- today could plausibly be good for Coleman just as yesterday was for Franken.

Both campaigns are also challenging ballots that their election observers think should be counted for their side or are too ambiguous to be clearly counted for the opponent. Those are as a result of ballots that seemed to have markings for more than one candidate, and as such were rejected by the machines (known as an "over-vote).

After all the precincts are done recounting, those challenged ballots will head to a single state canvassing board for review in one place. And it seems very likely that there will be more of those ballots than the final Coleman-Franken margin from precinct-level recounts.

By the way, if you want to see what these challenged ballots look like, Minnesota Public Radio has posted photos. Some of them seem like they'll be easily called for either Coleman or Franken -- possibly more of them for Franken -- once they go to the canvassing board in December.

So how should we expect all these under-votes, over-votes and challenged ballots to break? Well, a lot of the roughly 35,000 under-votes will probably be found to be intentional -- people who cast a vote for president, then skipped the Senate race. But many other will be found to have some kind of discernible vote.

A Dartmouth study found that voters in Democratic areas were more likely to fail to properly register their votes in a way that could be picked up by the machines, analyzing not only this race but also the 2006 Minnesota elections as a baseline. The authors then predicted that the recount would resolve those votes with a small break in favor of Franken -- but they did not venture to guess just how much of a break it would be, and whether it could swing the race.

Nate Silver calculated early on in this whole sordid process that it was actually quite difficult to figure out Franken's odds -- depending on the model used and the set of assumptions about those ballots, a Franken win was anything from a shoo-in to a lost cause. But a Franken victory, assuming the resolved rejected ballots break in his favor as everyone expects, it not out of the question.