Reply #1: Interesting.

[Clark2008]

National Election Pool pollsters have acknowledged that their polls deviate from official totals by 1.9% nationwide (a 3.8% shift from Kerry to Bush) and intimated that this deviation was caused by disproportionate numbers of Bush voters refusing to participate in the polls. Analysis of the available data and theory, however, strongly suggests that at least part of the discrepancy is due to miscount. Moreover, a review of 2004 election processes suggests little reason for confidence that the count reflects either the intent of the electorate or the way that the votes were cast.
I expect this claim will meet with incredulity and anger. The idea of mass-scale electoral fraud in a US Presidential election may be difficult to fathom - or to stomach.