Chris Bowers at Open Left does the math, and finds that Obama has already won. His reasoning:

1. In order to win the election, all Barack Obama needs are
   the Kerry states, plus Colorado, Iowa and New Mexico. That adds up to
   273 electoral votes.
2. Obama leads by at least 9.5% in every Kerry state and Iowa, according to both Pollster.com and Real Clear Politics. Also, my own numbers concur with those calculations.
3. This means that in order to win the election, all Obama
   
   has to do is hold onto states where he leads by 9.5% or more, and win
   both Colorado and New Mexico. These are both states where more than
   half of all voters will cast their ballots before Election Day (source).
   In other words, the elections in Colorado and New Mexico are already
   almost over, not just beginning.
4. In Colorado, about 60% of the vote is already in.
   According to the crosstabs of the three most recent polls in the state,
   Obama leads early voters by 15% (Rasmussen), 18% (Marist) and 17% (PPP).
   Even in the best case scenario for McCain .. he
   still needs to win the remaining voters by 18.4% in order to eek out
   the state.
5. A new poll from PPP in New Mexico
   indicates that 56% of the vote is in, and Obama leads 64%-36% among
   those voters. If that is accurate, McCain would have to win the
   remaining voters by 35.7%.

So, unless one of the following occurs:

• Obama blows a double-digit lead in either Iowa or one Kerry state
  
• McCain wins the minority of remaining voters in either Colorado or New Mexico by at least 20%

Then the election is over and Obama has won no matter what happens anywhere else.

It should be noted that we knew pretty much as far back as March that Obama had won the primary, based on the same kind of mathematic analysis. Bowers’ analysis is also corroborated by Nate Silver’s latest projections, which give Obama a 96.3% probability of winning, with 338-378 electoral votes:

It’s over, folks. Time to start planning for the transition… oh wait, Obama has that covered.

UPDATE: My purpose here is to emphasize that there’s a certain point at which the cold equations of mathematics take over. This election is governed by specific rules, and polls are scientific and statistically significant samples of popular opinion. The entire system, unlike the stock market or the weather, is fundamentally deterministic. The importance of early voting here cannot be overstated, either, providing a much-needed buffer against all the attempts at voter suppression that are sure to occur, not to mention minimizing the inevitable voting logistical snafus. Thanks to Obama’s strategic emphasis on early voting, these kinds of derailing forces are largely neutralized. As a result, the election is much more predictable according to the data. The math wins.