[Torie]

You got it. Nothing wrong with working with baselines, and looking at swings, but this data was so tortured, it makes the practicioners of torture by the rack during the auto de fe era, seem like compasssionate conservatives by comparison with this little study, and those who claim that Gore stole thousands of votes in Florida in 2000.

[general_re]

After skimming the math, I'm preliminarily ready to grant the general and vague proposition that there was a problem in some Florida election, but nowhere is there a magic bullet buried in the numbers to tell us which election had a problem. The authors cannot eliminate the proposition that 2004 was correct and 1996/2000 were not. Indeed, they can't, AFAIK, conclusively eliminate the hypothesis that all three counts were flawed, and that in fact none of them are correct.