I've done a hasty analysis of using county voting numbers and undercount totals. Frankly, it doesn't look too shabby for GWB (IF the recounting is done consistently between counties -- I know, BIG IF). Anyway, I've calculated the margin of victory in each county (votes for Bush vs. votes for Gore) and applied that to the undercounts. Yes, this assumes every undercount is a valid vote for either one of them, but the absolute total is not as important as the relative total. The "vote ratio" I calculate is Bush=22746, Gore=20112. In other words, Bush gets about 10% more recounted ...