The problem with air-bearing arms is that the lateral mass is often a multiple of the vertical mass. So you can be tracking at the right weight for the cartridge, but the lateral mass can cause visible left-right motion in the cantilever. When this happens, the tracking error is significantly higher than that of a radial tracking arm!!
The solution is a cartridge with a lower compliance. This might not work out all the well when it comes to simple tracking, as the vertical component (mechanical resonance) might not be satisfied.
I used to run a heavily modified Rabco arm many years ago that got around these problems- but then had other problems, most notably mechanical resonance in the track in which the arm ran. I do think linear tracking is the way to go, but have yet to see an embodiment that solves all the engineering problems.