I share this opinion. Hell, lots of uses of stateVersion are arguably incorrect. Testing is very rare, only one test currently does anything with it and I’m like 90% sure that’s a mistake. stateVersion as designed sucks.
Sadly, that does not change the design. By changing it, you’re exposing yourself to the very complexity it tries to hide, which is fine, that’s your choice.
But then you throw your hands up and say you don’t want to deal with the complexity after all? You can’t both expect the distro to deal with the complexity for you and also circumvent it to deal with it yourself.
Even more so because you seem to refuse to read the release notes, which tell you how to deal with changes that are not hidden behind stateVersion. You have to give people a chance to help you if you want help!
Anyway, I guess ultimately your opinion just boils down to the fact that stateVersion is generally a poor way to deal with managing state, and that you would like to see this changed. That’s fair enough. You have many allies, hopefully we can come up with a better solution.
This recent thread talks about it, and an interim solution (that @waffle8946 mentions) that will in fact make dealing with it yourself more achievable was proposed and seems likely to be merged. Probably in use by 27.05 or so, given release timelines.
Various people over two threads have also already come up with many scenarios and edge cases for full deprecation. The final solution will probably have multiple different replacements for different types of state, likely encoded in policy rather than nix code.