I'd like the outcome in your world. I think the best thing to keep in mind is that not all polls identify likely voters, which is key if the goal is to try to predict the outcome. Gallup can be good at calculating likely voters, although in the most recent all they reported were registered voters.
And like it or not, people lie about whether they vote. Remember Bennifer in 2000? He was huffing and puffing about algore and then it came that he could not show that he had ever voted in an election (which is great for us).
All the polls are getting more and more iffy in an age of cell phones and message machines. Coors won by a margin of 22pts--in a race that was supposed to be tied. What does that tell you? There are huge problems with all the polls. The big news recently was the MO gay marriage referendum turnout--it was huge. The polls predicted a narrow victory for the right. Instead the vote was 70%+. The polls are missing something important--voter turnout. Which, by the way, is the result of superior organization. They missed the GA win for the senate in 2002 for the same reason. Ralph Reed was responsible for that.