On Exit Polls

The exit polls in this year's elections were unreliable. Supposedly the surveys across the country tilted towards Kerry early in the day and throughout much of the evening in some states.
Why this happened is something I have not been able to establish. It is an odd kind of mistake to make; one that causes bias in only one direction. I would have expected that some states would have had polls tilting towards Bush and some towards Kerry. Why would all exit polls tilt in the same direction, and one that was not the correct one based on the election results? I really want to understand this.

In any case, it is interesting to note that "Late in the evening, the exit polls were adjusted to reflect the official vote count."
Is this the common practice? If so, that would obviously account for the common experience that exit polls have been pretty reliable in the past. But I very much doubt that their reliability was caused by a trick like this.

This link shows some very interesting patterns in the comparisons between exit poll and final election results by state.

Voting is an important part of democracy and it's really very necessary to understand exactly what happened. If we don't understand this, our trust in the very concept of democracy will be reduced.