A lot of folks out there are trying to score how well the pollsters did in calling the election. The scoring seems useless to me since it's usually a matter of matching a pollster's last prediction to the final score. If you were an inept pollster, you could just make wild forecasts with no basis in reality for months. Then, just so long as you got close the day before voting, your reputation wouldn't be tarnished, at least by the kind of measures now being applied.
What I get in hindsight is, first of all, no one poll is without its big misses, and some of them try to have all bases covered with multiple surveys that may differ. To the extent there's something to be learned there, you have to look at a consensus over some period of time, especially the rolling average. If your world rocks each time Zogby says your guy is up or Rasmussen says your guy is down, you're letting yourself be jolted by near-meaningless noise. This year, and maybe it's this way every year, daily fluctuations and outliers gave us a very inaccurate view of what was going on with the electorate but the long-term averages of all the major pollsters were probably close to reality. Which, of course, we'll all forget next time as we hang on each day's polling headlines.