We benefitted from this attempt. Even the reasonable pundits say the first exit polls should be ignored, that they are extremely unreliable. IMO, it motivated our base.
1. They're, um, POLLS. Polls have sampling errors or they wouldn't be polls.
2. They are subject to any number of biases: geographic (which locations did they happen to exit-poll?), personal (they polled way more women than men for some reason), time (an exit poll in the a.m. is by definition polling only the people who vote in the morning, not those who vote after work), etc.
3. They will necessarily be based on some *model* of the overall electorate. If they poll some blacks and their *model* says blacks are 60% of all voters then they'd always show the (D) way on top. But that model would be wrong. It looks like the model was wrong yesterday too - not as wrong as mine, but still wrong.
So my solution: disregard "exit polls" in the first place. They don't mean squat.