it does appear that some Motorola and some Collins transmitters were sensitive to which generation of 6146. I've never had a tx that was, though, and certainly none of the ones I designed were - except that substituting for the B version needed the drive cutting back to stay within ratings on RTTY.
I agree with that. The problem comes in when someone writes a big long white paper with impossible technical explanations.
It all comes back to inventing a story in one's mind why something happens, and then presenting it as fact. The idea neutralization "drifts" is just silly to anyone who understands what neutralization is and how neutralization and tubes work. Making it worse, someone not fully grasping neutralization should probably not author a white paper claiming RCA tube engineers made a mistake.
I fixtured a bunch of 6146 evolutions and they all were very similar over the normal operating frequency range.
My best guess about why certain select things are 6146 tube-brand or tube generation sensitive is the devices were already at the edge of problems, most likely way up at high VHF. For example, Collins wrapped feedback around multiple tuned circuits across two stages in an attempt to add negative feedback. That's all well and good at the operating frequency if everything is resonant, but feedback phase will never be stable over a wide range or if a tuned circuit in that path is off-resonance. I'd hardly consider a system like that being sensitive to tube types an indictment of RCA engineering.
If we examine the 6146 tube generations, we find the anode to anode cap path changed. A few base leads changed, also. This has no measurable effect on normal operating range performance, and neutralization is the same, but if a system has marginal VHF stability or depends on tube internal transmission line effect at upper VHF, changes could cause problems.
I think if I had a rig that lost stability with a tube generation or brand change, I'd diagnose the problem and correct the engineering in the gear. While a 150 MHz rig might always be sensitive to lead diameter, routing, and length, any HF transmitter unforgiving of that has an inherent VHF stability flaw.