FWIW, the SYSSTATE ARCHLVL value actually happens now to correspond to an architected value (6 is EC12). Think of the value as having started at 0 at base z/Architecture or something like that (whereas the compiler's ARCH values long pre-dated that). Being in step with "something" was a reason that ARCHLVL skipped from 4 to 6 in z/OS 2.3.
Given the number of places in code (IBM and otherwise) that "understands" (perhaps relies upon, even loosely) the correlation between the number and the machine level, any change in this area (aside from via a new option such as Jonathan implemented for HLASM) is unlikely to happen, for compatibility reasons if nothing else. But if there were a change, and if it were OK to skip numbers, then it might move everything up to the machine generation number. That would be a good value for things to converge upon. I explored that briefly a couple of years ago but there wasn't enough interest to gain traction. Peter Relson z/OS Core Technology Design
