Wasn't it established that the blank area method of setting anti-skate was inaccurate because it didn't take into consideration the dynamic grooves of actual recorded vinyl?
Actually that only comes on top of it - and also poses a certain problem, 'cause which degree of modulation would one preferably set one's anti-skating for: rather for a very high degree, so that it formost helps to avoid mistrackings during the most demanding passages - or rather for an average degree of modulation, so that it would foremost help to avoid uneven wear?
However, there already is a systematic difference between the "tip of tip on blank surface" and the "tip in groove" scenario, because in the former scenario the "normal force" for the created friction is pretty exactly equal to the tracking force (minus a tad due to vertical skating), whereas in the latter scenarion it is pretty exactly equal to two times the tracking force divided by the square root of two (= roundabout 1.4 times the tracking force). So even compared to a tip in a completely unmodulated groove the blank disc method would result in ca. 30 % undercompensation. Theoretically. In practice the results might as well be more all over the place, as the very tip of the tip isn't really designed to come into contact with the record whilst playback - or, more physically speaking, different coefficients of friction may apply to the two scenarios.
Hence I'd rather vote for the test record/modulation test rack method as the more scientifically correct approach. However, that would of course at first result in a setting for a, depending on the capabilities of the cartridge (or cart/arm combo), pretty to very high degree of modulation. So the next question would be, whether one would like to reduce that setting and by how much - to suit a more average degree of modulation or for some sensible compromise in between...
Back to friction: A part of the problem is that what one learns about friction in school physics lessons is pretty idealised/simplified. I.e., basically our "school book table of coefficients of friction" only applies to "flat, smooth surface against flat, smooth surface" and a couple of other conditions - and wherever these conditions don't apply, things become quite a bit more complex and complicated, so that rules like that contact area size or speed would have no influence on the friction also don't really apply anymore.
Greetings from Munich!
Manfred / lini