understanding durbin-watson test in R

Question

I believe I have a misconception about how the dwtest function in the R lmtest package.

> dwtest(lm(r2 r1)) > r1 - runif(1000, 0.0, 1.0) > r2 - runif(1000, 0.0, 1.0)

Test of Durbin-Watson

lm(r2 r1) data
p-value = 0.3789, DW = 1.9806
a different perspective: real autocorrelation is higher than 0.

R2 - seq(0, 1000, by=1) > dwtest(lm(r2 r1)) > > r1 - seq(0, 1000, by=1)

Test of Durbin-Watson

lm(r2 r1) data
p-value = 0.8352; DW = 2.2123
When I'm certain I've understood everything, I first compare two sets of random integers to one another (which do not correlate - correct)

Then, I link the incrementing numbers from 1 to 1000 with themselves (which does not correlate - uhm... what)

Can someone please show out the obvious mistake I make?