Anecdotal, personal experience. I personally am very sensitive to key pressure, and bottoming out actually hurts. I do have a low-force rubber-dome keyboard that I bought from another poster in this thread (hi!) and although one _could_ theoretically use only 35 grams of force to press the keys, in reality that would require limiting the stroke to exactly the 2mm travel to achieve. In practice, (35 grams * 2mm) ~ (0.035 Newtons * 0.002 meters) = 70 μJ. That is, I am pressing 35 grams for the whole 2mm. My Cherry Brown board, though, activates at 55 grams of force but _getting_ to the 2mm activation point requires less work because much less force is needed at the beginning of the stroke. Let's assume linear, some calculus and we have 55 μJ. Not much less work.
But here is the kicker: one cannot press down exactly 2mm. As humans we are inaccurate, and if our inaccuracy is +- 0.5 mm then we must aim to overshoot to 2.5 mm so that there will be no missed keystrokes. At 3 mm of travel (the high end of the inaccuracy) the Cherry Brown board requires only 60 grams of force, but the rubber dome shoots up to infinity! Even at the 2.5mm mean travel, the rubber dome keyboard requires orders of magnitude more force on the fingertips due to having bottomed out. This is why, so long as one can prevent bottoming out, the mechanical keyswitches are much more gentle on the fingertips.
In summary, human fingers aim for distance and the force on the keyboard is a function of that distance. Once that is recognised, then the need for key travel after activation is obvious. How much key travel is necessary is a matter of how accurate one can set the distance for their fingers to travel. The 2mm of after-activation on the Cherry Browns seems to be my limit, as I do occasionally bottom out. So a 55-gram switch that allows for 2mm of travel after activation requires less work and less pressure on the fingertips than a 35-gram switch with zero after-activation travel.