I just took a car to the Willow Springs skid pan. I couldn't notice any difference in speed between 0.0 and 2.0. 4.0 was a tiny bit slower, 6.0 was way slower. 1.0 maybe felt faster, it certainly wasn't 1mph faster to show up on the speedo - so maybe a placebo?
So I went to Tsukuba and took telemetry at the last turn. Once with 0.0, once with 1.0, and once with 6.0. As a control I compared the lateral g on the 0.0 and 6.0 setups. I know that 6.0 is slower so I expected to see a difference in the lateral g graph. I couldn't see a difference
The chart is just too small. I wish we could export the data into Motec or something like that.
Looks like seat-of-your-pants it is.
EDIT: Also wanted to say I wonder how much the "correct" amount of camber varies car by car and setup by setup?