Well, I saw how bad pollsters did in the Alberta election earlier this year, and have been weary of them ever since. However, that election had a high number of undecideds and represented a huge change in the electoral landscape there. This election does not have a lot of undecideds. The only worry I have is the effect of all of this voter ID BS on the election.
I was thinking the same, but I should note unlike Canada, the US election cycle is much longer so votes tend to stablize more towards the end. Lets remember even if the final week numbers were still moving around in Alberta and Canada federally. In the case of 2006 federal election which was 8 weeks vs. the normal six weeks the polls were a lot closer. Also Americans are far more polarized than Canadians so it is a lot harder to get someone to switch parties. Still it is possible they could be wrong. Anybody know more about 1948 when newspapers read Dewey defeats Truman when in fact it wasn't even close. Mind you US politics up until the last decade was far less polarized and the swing vote was much larger than today