With all the data, Tesla has probably figured out a "fling curve" that gets better target results with smoother delivery. Versus a "smash hit" that lurches all the mounting bolts loose in the first sub-second. The smash hit always
feels faster (and maybe it is, initially for the first few feet) but the fling
gets the results. And the result Tesla is after is not the first 20 feet ... it's the
end of the dash.
Watch this:
For this experiment, the same vehicle is used (ball, size, weight, and acceleration = gravity).
The race is the horizontal position along the table.
The same car tuned differently can perform like any of these curves.
But let's assume the smash hit (track on the bottom)... is how the original P85D was tuned.
The fling curve (middle track) ... is how the P100D is tuned.
Look at that impressive "movement" (feeling) of the P85D (smash hit) ball drop in the first moment of the race during it's "free fall" phase, acceleration = gravity. But at no time is the P85D actually ahead (horizontal distance) of the P100D race.
The slower-to-start P100D (on its fling curve) handily wins in reaching the finish line goal in less time, with a smoother and more constant power delivery. Fling versus punch that do the same amount of work.
Unlike the experiment, P100D and P85D are not equally powered cars. Imagine P100D tuned to smash hit, which is more power than P85D so it would feel even more crazy at the beginning. But the P100D tuned to smash wouldn't win against itself tuned to a fling curve, like it is.
Tuning a car for a fling lets Tesla use a more consistent power delivery for a better end result for any given car. Unless they screw it up in software, like maybe they do sometimes.