Hitec is correct, of course, because they are explaining the electronic function. Measuring pulse widths, you'll get exactly what they describe. But, as soon as you introduce the ROTATIONAL movement near the servo...it is quite different. If we had linear servos (like once upon a time in the EK Logictrol era) it would be perfect.
In essence, I guess my suggestion is that if you start with ZERO exponential with servos that have arms...it's already more sensitive near center. To get to the point where you actually "feel" expo in the air, you need to be somewhere in the 20-30%'s "softer". Then, you're actually starting with mechanical "feel" of the airplane at neutral.