Which is why most of it is worthless. The news itself has quoted the exit polls as usally being within .5% accurate. Here's a good article on why the difference in the exit polls are significant.
http://www.prisonplanet.com/articles/november2004/03110 ...
>> While not a guide for proving results, it can be a mechanism for ensuring voting accuracy and flagging potential fraud.
>> Perhaps more importantly, while exit polling is unreliable, the odds of ... gaining an advantage from every exit poll in swing states is an extremely improbable coincidence.
Though I think more believable numbers are the 1:50,000 — but what that's saying is that in 50,000 presidential elections we might see this once — three full factors away from reality.
That's enough to raise the question of validity of the election, and that's what we want.