I’m fairly new here, so forgive me if I’m off-topic and barging in an already-much-discussed topic, but personally I would much prefer “including some stale/broken packages” to “reducing the size of nixpkgs”. A stale derivation is a starting point for contributing a fix, and I’d rather look into that than having to hunt for other sources of derivations.
There is of course gray area here: while you can’t expect the core package maintainer to fix all downstream derivations, there’s some responsibility there to think about the downstream impact and not just randomly break things. I didn’t follow this particular change, but my impression is that that is exactly what happened. Of course it is always ‘better’ to ‘go even further’, but you have to stop somewhere.
I’m not sure more ‘formal’ quality control would have helped here, as the change appears to have been made pretty deliberately, understanding that it would break a certain class of downstream packages. Perhaps the impact was underestimated, but I don’t have the impression that it was made ‘thoughtlessly and ignorantly’.
Perhaps you could look into running Abort Qt builds that do not use a supported deriver by ttuegel · Pull Request #70691 · NixOS/nixpkgs · GitHub and taking inventory of what’s still broken?
On a more positive note, I’ve been tracking unstable since October, and only once encountered a package that did not work for which a fix wasn’t already underway. There even already was a github issue about it. From my experience NixOS is doing pretty well here.