You following the Coleman-Franken recount in Minnesota? The latest headlines say Coleman is ahead by 120 votes but if you read a little further, you realize that number is kinda meaningless because there are more than 3000 challenged ballots that haven't been figured into the total. No one knows how many of those will eventually yield countable votes…and those aren't even random ballots which one might assume would break roughly along the same ratio as the already-tallied ballots. These are ballots that the Coleman or Franken observers specifically had pulled out of the piles for later examination. There are also hundreds — perhaps thousands — of disqualified absentee ballots that have never been counted and some of those may wind up being tallied.
In some cases, it appears that the two campaigns are playing a little game here, trying to manipulate current news reports. The premise here is that if you can look like you're ahead (or gaining) for a while in the news, that will give you a certain legitimacy as the winner, above and beyond what the final totals say. If Coleman loses when the vote is ultimately certified, he'll be out there arguing that the results are a lie and saying, "I was ahead for weeks! Suddenly, at the last minute, they claim I lost!" But of course, he may be "ahead" right now because of which ballots have been removed from the count and put into that "challenged" pile or which absentee ballots simply have been wrongly disqualified and therefore uncounted. Franken's side is probably challenging ballots with something similar in mind — to win the daily press coverage, claiming that the gap is narrowing, as a means of winning the election.
There are a lot of statisticians out there, trying to predict the outcome. Nate Silver, who had a superb track record in forecasting the election, is seeing a Franken win by between 48 and 136 votes…but it seems to me that there are too many unknowns (including which ballots the judges will admit) for any prediction to have much gravity right now. And even Silver says his analysis could be off by "at least" 200 votes…which kinda means he thinks Franken will win unless the other guy does.
I don't think anyone knows as much about who will win as they think. All I think we can say is that Al Franken's in a good position. He's either going to get a Senate seat or a very interesting book out of all this.