Originally Posted by
conspiratemus1
Your circle-in-circle description doesn't explain where the force comes
You're right, I didn't explain that. The "rotation" is actually a rotating radial stress. The pedal loading pushes on the back and bottom of the cup, and the opposite pedal induces the same stress through the rest of the pedal cycle. In other words, there is an outward force on the cup wall, and that force moves around in a circle. It does not apply a significant circumferential force, but rather a radial force. The cup simply rolls around the inside of the shell. (Note, it is not required that the force be the same through the entire pedal cycle. It is required merely that it rotate.)
Originally Posted by
conspiratemus1
There may not be any rolling friction between spindle and bearings, or between bearings and cup, but there is thrust, from the "grip" between the moving parts.
If friction from the spindle transmitted to the cup through the bearings was the reason for the cup moving it would tighten a right-handed thread on the right side of the bike. It doesn't do that.