It’s often misunderstood what polls do. Some are under the false impression that the purpose of a poll is to predict the election results. That’s not quite true, though close to the election, the difference doesn’t matter.

However this far out, the difference means everything. Polls measure the current feelings of the electorate. The further out you get from the election, the more detached the polls get from the actual results. A look back at 2011 will show how much things change.

Real Clear Politics tracked the polls last cycle. The Iowa polling on September 1, 2011 was at a pivot. The Michelle Bachmann wave was ending, and the Rick Perry surge was on. September 1 was the midpoint between the two, leaving nearly a three way tie in the polling average. Of course, neither Bachmann nor Perry would win the state. Mitt Romney did.

Contrast with now. We’ve now just seen the end of the Scott Walker dominance in the state, and now two candidates are rising: Donald Trump and Ben Carson. Only time will tell if Walker, Trump, and Carson can avoid the fate of Bachmann and Perry.

Nationally, the trend in 2011 was even more pronounced. Rick Perry’s surge was on and showed no signs of slowing. He’d just become the first candidate to knock off Mitt Romney as the national leader. Newt Gingrich, Herman Cain, and Rick Santorum would each lead at later points in time, but they were minor also rans in the polling this early.

The 2008 polling was even further off than the final results, for both parties. The Washington Post for example had Hillary Clinton leading 41-27-14 over Barack Obama and John Edwards (Joe Biden was below the Margin of Error). On the GOP side, Rudy Giuliani continues to hold his lead at 28, ahead of Thompson at 19. McCain was third at 18.

We can see the eventual winners do tend to be registering in the mix early, but name recognition plays a larger role this far out. RedState would go on to write off McCain entirely before he came back to win.

The early polls are interesting, and give us a view of the shifting electorate, but to look at them as an attempt to predict the final result is to misunderstand them. Badly.