The "gearing" is special, in that it is essentially non-existent, at least between the motor and the main mechanical drive for the focusing mechanism. In order for a traditional (low-frequency or DC-driven) syncromotor to have enough control fineness (or even control finesse), it needs to be geared down significantly. The mechanical forces involved in gearing down are slight, but working backwards against the system (manually focusing while the motor is trying to drive the system) would mean applying enough torque to damage gears, bushings/bearings, and probably strip the mechanical coupling between the gears and shafts, etc. (Not to mention that if there's a worm drive in the system anywhere, it won't work backwards at all.)

High-frequency systems (and systems with enough poles, even if they operate at lower freqs) can be directly controlled with a lot more fineness because the phase angles involved can be larger for shorter physical intervals. Because they don't need to be geared down, there's no extra mechanical stress on any of the components when you physically oppose the motor.

I don't think it would be viable to have direct drive from the in-body focus motor. The screw drive would need to run at a very different speed for a lens meant to be driven directly at or around 1:1 gearing than it would for an older, geared-down lens. (I don't have a Pentax to play with, but my Nikon screw-drive lenses seem to be at about a 8-12:1 drive rate.) If the motor has enough torque to drive a 1:1-geared lens effectively, then it's probably got enough torque to strip the drive screw on an older lens (or if the driver and receiver are hard enough not to strip, then to destroy some even more expensive parts, like the attachment points on lens barrels or body castings).