Polling on the night of the 31st (as Zogby has done in IA and Suffolk University has done in NH) is completely crazy. But is polling on the night of the 1st really any more problematic than polling on any other night between Christmas and New Year's? Are people any more likely to go out on the night of the 1st (when most of them probably have work the next day) than on the night of Saturday Dec. 29th or Sunday Dec. 30th?
Oh, and four years ago, there was a Time/CNN national poll conducted on 12/30 & 1/1 (12/31 was skipped):
http://www.pollingreport.com/wh04dem.htmSo it's not totally unprecedented.