The main thing we learned last night from the massive success of the poll aggregators that I wrote about before the election is that the polls do accurately reflect the variables that have traditionally been thought of as beyond polling. The Republicans launched a massive legislative voter suppression effort that probably affected the results. The Obama campaign put together what was probably the greatest get out the vote effort in history. What we learned is that their impact was factored into the polls. Even the naive model used by electoral-vote.com did pretty well (their Rasmussen-free map did as well as Nate Silver). Forecasting the election by aggregating state polls is a winning strategy, at least for the time being.
Update: Here’s a list of the individual polling firms that most accurately predicted last night’s results. Good polling is critical, and this year’s polling was very good (as proven by electoral-vote.com), but the main takeaway is that there’s almost no point in looking at individual poll results when you can aggregate all of them.