I see the recommended diaries about how it is just impossible for the actual results to be squared with the exit polls. But the statistical margin of error on the state exit polls easily explains the change in Ohio and Florida. In addition to the usual polling sampling error, there is the problem that no one knows until the final voting is complete what is the total turnout at the sampled precincts, so the precinct weights may be badly estimated. The late exit polls in OH and FL were 51/49 Kerry, but this was hardly a guarantee of victory.
Everyone who carefully read a description of the exit polling process (e.g. on mysterypollster.com) should have understood that 51/49 was nothing but an indication of "too close to call."
Remember that the exit poll numbers are always re-weighted after the final votes are in to reflect the actual turnout at the sampled precincts. There is no evidence of fraud in that change.
Another claim is that the results are not consistent with Zogby's predictions. But Zogby had Florida tied and Bush actually 2% ahead in Ohio. Zogby was counting on a Kerry "trend" and a big youth turnout to overcome Bush's lead. He was wrong. We all were.
Now, there may be evidence of election fraud. I hate the very idea of paperless electronic voting. I want to understand how the heck our great turnout got overwhelmed and whether there is evidence of fraud in the details.
But it is all too easy to reconcile the actual results with the "too close to call" phone polls and exit polls.
See Kevin Drum at Washingtonmonthly.com for more.
Let's focus on the actual problems, not the pre-election polls.