If I may be permitted to point out something that seems to have been missed here, there is more to a turntable than speed stability, although speed stability is a very important requirement.
Another factor which comes into play is the effect of vibrations or other potential movements in the main bearing assembly(and thus the platter).
If stylus drag can affect the speed stability of a heavy platter or even a motor's torque, then it can also generate lateral movements of similar proportions, due to the angular nature of it's movements as it traces the groove. Externally generated vibrations can affect this also. Since any lateral movement allowed by the bearing/platter can materially affect the amplitude of the groove modulation being traced at the time of that movement, it can cause information loss or modification, based on that unwanted movement of the platter relative to the stylus. But wait, you say, the arm could be moved more easily because it has less mass, and I say that if the arm deflects, then you get similar information distortion as well.
Now, how does this affect our discussion?
Quite simply because direct drive motors generally(and I use that word advisedly) use their drive shaft as the main bearing, which typically does not have the precision tolerances of a belt-drive system's main bearing in a good turntable. In addition, making the platter part of a motor which has inherent vibrations as a natural characteristic is not conducive to ideal conditions for the stylus/groove relationship.
So, even if a direct drive does exhibit a slightly better measured speed stability, there are other factors involved besides that.
Also, wow and flutter are very different speed variations, and to lump them together as "wow and flutter" is deceiving at best. Wow is much less easily perceived by the ear than flutter, and 2 tables with the same "wow and flutter" specification can sound quite different, especially if one has mostly flutter, and the other has mostly wow component in that spec.
To sum up, all methods of analog drive systems have their pros and cons, and none is perfect, and implementation may be more important than basic concept.
To focus in on speed stability alone, with no discussion of other meaningful design aspects is incomplete and useless. I've heard very good sounding turntables of varying drive system designs, and also some terrible ones too.
Tunnel vision is not the most productive way to analyze componentry.
