Polls

In case you're wondering, I checked and at least some of the numbers in these polls appear to be outdated or otherwise not quite right.

The Coulson poll, for instance, appears to be a House Republican benchmark survey from August - long before the House Democrats began pouring in mail and troops.

Curiously, the head-to-head numbers published for that Coulson race don't quite match up with what the Republicans had back then. Her opponent's number is the same, but Coulson's number is several points higher than the benchmark results. One top-level HGOP source claimed "the numbers were slightly modified in some instances," (on their end, not on Arch's).

When asked what the heck that meant, the Repub source said, "Let's just say, the margin of error would be 9-10%, instead of 5%."

Take it for what it's worth.