Yea, and Lieberman was going to lose by 13% until the day before the election and then it was 6% but it ended being 4% then the twist is he's expected to win by 20% as an Independent. If someone can figure out the method to a pollster's madness then I would love to know as long as it doesn't require a poll to figure it out.
Pollsters are less fickle than the consumers of polls; we love them when they look good for us, dismiss them as meaningless when they don't.
The poll was pretty accurate wasn't it, and polling in primaries is a tricky business, particularly when it comes to turnout models?