Final Report on Bush-Kerry Exit Poll
This sounds about right to me, but that makes it no less annoying. Via Washington Monthly. The exit polling firm has discovered that the error rate was more pronounced the younger the interviewer was, the farther away from the polling place the interviewer was required to stand, and the less often the interviewer made requests.

Nothing to do with the type of voting maching or the company/method doing the counting. The only question I have left is why was the error always favoring one candidate? I suppose Kerry voters were proud of their vote and wanted to tell everyone who asked. Bush voters were not and wanted to hide and run away. Not sure if the whole report addresses that, and not sure if I have the heart to read it anyway.