Setting the limit at 80 doesn't make everyone drive 80, either.
Evidence, please.
We did this experiment and the evidence is against you.
Fatalities did not significantly increase for any of the states that increased their limits after the 55mph thing ended and there was some weak evidence that they went down.
In addition, if you look at the fatality and injury numbers for urban interstate they went down for 65MPH and up for 55mph and 75mph. That makes no sense.
Basically, the statistics are sufficiently close to a wash that the confounding factors (like texting or rerouting due to more convenience) matter more than the speed limit.
And it's the large differentials in speed which create the most dangerous conditions and accidents.
Even rigorous enforcement can't catch everyone, and the high speeds quickly become the norm; with people going 65 the dangerous exceptions.
If the break-even point isn't 1 million people, where is it? 622k? 100 billion? 3^^^3?
This is Eliezer Yudkowski's dust speck argument all over again. http://lesswrong.com/lw/kn/torture_vs_dust_specks/