Entertaining reading, watching things get batted back and forth.
I suspect you are both correct in elements of what you are saying.
Sometimes and answer is "All of the above."
It is pretty easy to understand the forces in a gearbox when everything is static. In the our FD the preload stabilizes the tapered roller bearing through the deep groove crownwheel ball bearing. Therefore the preload is also stabilizing the crownwheel bearing. Being a class "C" (sloppy bearing), without some axial preload the crownwheel bearing would be unstable. Additionally, this preload in stabilizing the crowngear assembly, stabilizes the relationship between the crowngear and pinion gear. The gear tooth contact pattern and gear lash are established by a shim in the input pinion shaft area and a shim under the tapered roller bearing, respectively.
When static, all the axial force on the crowngear assembly is from the preload.
When drive line acceleration (not constant velocity, but acceleration, or deceleration) is applied, additional forces will be applied to the crowngear assembly from the input pinion gear.
An engineer's analysis would reveal multiple vectors of force, but one of those vectors is certainly going to be axial to the crowngear assembly and in the direction of increasing compressive force on the crownwheel bearing, i.e. the same axis being applied by the preload but in the opposite direction. How much addtitional axial force would be applied on the crownwheel bearing during acceleration? Answer: how hard are you going to hit the throttle? It seems to me that a very aggressive acceleration would apply a significant axial force on the crownwheel bearing; a force that could easily exceed the static preload force. I suspect that this is what Dennis is referring to.
This axial force applied to the crowngear assembly on acceleration/deceleration will tend to unseat (unload) the tapered roller bearing. The preload needs to resist this unloading to the point that the tapered roller bearing remains stable. The crownwheel bearing sits between the axial force of the preload and the opposite axial force of acceleration.
Add to the static force of preload, and the dynamic force of acceleration the additional forces of pounding down the potholed roads of Vermont, two up, with the bike loaded to maximum with camping gear, and the dynamic forces on the whole FD assembly are indeed complex.
When we get past all this arcane information, the shimming needs to be correct so that there aren't damaging forces applied to components (the crownwheel bearing being the one that fails if the crownwheel assembly preload is excessive) and there is a lubrication space between parts that move with respect to each other.
I'll say again, I believe the engineers at BMW did a decent job of designing the K1200LT FD, and that manufacturing (machining) of components and assembly of those components have lead to multiple types of failures. The most common of those failures, the "classic" crownwheel bearing failure is the result of assembly error, i.e. excess preload shim thickness.
That bearing is in a pretty harsh environment: acceleration and deceleration forces, road surface and pothole pounding, etc. Add to that excess preload, the bearing surface gets spalled, the contacting surfaces are deprived of lubricant, and the bearing cries "Uncle!".