Although the state of Minnesota has completed the first stage of its recount process, counting by hand each of more than 2.9 million ballots cast in the state's November 4 senatorial election, we are really no closer to knowing who the winner might be.
The quasi-official count provided by the Star Tribune shows Norm Coleman 192 votes ahead. This count, however, is not very useful. This is because it treats challenged ballots as nonvotes, essentially giving the benefit of the doubt to the challenger, when in fact the vast majority of challenges are liable to be rejected.
An alternative would be to add back in the number of challenged ballots made by the opponent to each candidate's vote total. As Norm Coleman has made more challenges than Al Franken, this would reduce Franken's deficit to 97 ballots. This measure too, however, suffers from a significant flaw. Namely, not all types of ballot challenges operate equally on an opponent's total; some types of challenges result in a deduction to the opponent's count, while others do not. Without having information on the types of challenges issued by either campaign, this measure too is ambiguous.
A third alternative is presented by the Franken campaign itself. This method assumes that all challenges will be rejected, and that the initial ruling of the local elections judge will prevail. Under this accounting, the Franken campaign claims to lead by the grand total of 4 ballots. In theory, this is the best way to do things; the vast majority of challenges almost certianly will be rejected by the Canvassing Board, and so to give the benefit of the doubt to the challenger -- as the Star Tribune count does -- does not make a lot of sense. Of course, it requires us to take the Franken campaign at its word, as it relies on information gathered by the campaign that is not available to the general public.
We have also, at various times, released the results of an inferential statistical model, which is described in some detail here and here. Various iterations of this model now project a net gain by Franken of somewhere between 132 and 227 ballots once challenged ballots have been processed. Given his initial deficit of 215 ballots, this implies that he'd finish somewhere between 83 ballots behind Norm Coleman and 12 ballots ahead of him:
On average, the various versions of model show Al Franken finishing 34 ballots behind Norm Coleman. However, the margins of error on the model are high -- at least +/-200 ballots -- and so they do not equate to much of an advantage for Coleman, perhaps implying that he is a 60/40 favorite to prevail once the challenged ballot process is complete.
None of these various estimates, however, account for two significant X-factors, one of which could potentially work to Coleman's benefit and the other to Franken's. The first X-factor is the disposition of ballots in Minneapolis's 3rd Ward, 1st Precinct, where 133 ballots have gone missing. As the votes in this precinct heavily favored Franken on November 4th, they would have a deleterious effect on Franken's total if not counted, harming him by a net of 46 votes. Minneapolis, however, is still looking for these ballots. In addition, even if the ballots are not found, the state could make a decision to accept the original results from this precinct -- or Franken could sue to try and motivate that result.
The second X-factor is rejected absentee ballots, which the Franken campaign has been trying to get counted. Although most rejected absentee ballots were discarded for perfectly valid reasons, some minority of them -- estimated at between 500 and 1,000 ballots by Secretary of State Mark Ritchie -- were rejected in error. These ballots have essentially never been counted, either during the initial count nor during the recount. The state has instructed local officials to sort through their absentee ballots, and identify those ballots that may have been rejected improperly, a potential first step to having such ballots counted. If such ballots are counted, they are likely to result in a net gain of 25 to 100 votes for Franken.
As you have probably inferred, if the current standing of the recount is Coleman +34 (as estimated by our statistical models), or Franken +4 (as estimated by the Franken campaign), then a swing of 46 ballots in Coleman's favor from Ward 3, or 25-100 ballots in Franken's favor from the rejected absentees, could quite easily be decisive.
The good news for the Franken campaign is they are arguing in each case for more rather than fewer ballots to be counted, a stance which is probably superior from a public perception standpoint. It is hard to make a compelling case, for instance, that Franken should be punished because Minneapolis misplaced some of their ballots, or that he should be punished because some absentee ballots were rejected erroneously and never counted in the first place.
Winning the battle of public perception, however, may not necessarily equate with winning such balltles in court. It is quite likely that if either the Ward 3 ballots or the absentees would alter the outcome of the election once all challenged ballots have been considered, the losing side will sue to seek the opposite outcome.
It is even possible that the US Senate itself will get involved, as under the Constitution, the Senate ultimately serves as judge and jury on adjudicating electoral disputes before its chamber. If, for instance, the state certifies Coleman as the winner by 20 votes, but the missing ballots from Ward 3 have not been counted, the Senate could decide to seat Franken -- or, as happened in the 1974 Senate Race in New Hampshire, to order a re-vote.
Minnesota's recount is a long way from over. If you put a gun to my head and asked me to predict the winner, I would tell you to shoot me.