Look at it this way.
Let's say there are 1,000 people in a poll.
To get results like this, Bush would get 490 votes, Kerry 480, and Other/Undecided 30.
The day before, it was 500 to 470 to 30.
Does anyone see the problem with this?
A move of 10 people one way based on a NEW sample means nothing.
Good work.