Now let's apply the same nonparametric test to the data with the machine counts flipped. The result is quite different:

        Wilcoxon rank sum test

data:  (Clinton..d - Obama..d)[machines] and (Obama..d - Clinton..d)[!machines]

W = 234, p-value = 0.7086

alternative hypothesis: true mu is not equal to 0

95 percent confidence interval:

 -0.10751315  0.07106908

sample estimates:

difference in location

            -0.0175131

This means that one cannot reject the hypothesis that the hand counts and the reversed machine counts have the same distribution.

We have met the enemy, and he is us — Pogo