I just did a comparison of 2000 and 2004 Presidential results by county in my state, Iowa. Bush gained in the range of 1-3.5 percentage points in most counties, with some +5 point gains, a few blip gains of less than half a point, a loss of percentage of the vote in one(1) county, Jasper. In Jasper he lost 1.56 points.
The electronic systems used in Iowa can be found
It is painstaking but not at all hard to compare the data using the two links above. Oh, yes, and turnout was up in almost every county.
The electronic systems used in Iowa, county-by county, are listed here. The optical scan systems made by ESS are the most used statewide. That, I think, means that a recount could be worthwhile. Iowa's recount provisions allow for a recount days after the county canvas, and many counties start their canvasses tomorrow.
I have no statistical background, and I am not registered at DU, and their registration page is unavailable. Can somebody post this stuff on the PhD statistician thread at DU?
I do not believe that Bush could have held or gained his share of the vote vs. 2000 in every county. Not this year.