this is a false dichotomy: it implies that GNOME 3 isn't already stable and useful (which is patently false, given that there are people using it and being productive with it already), or that GNOME 2 was both stable and useful (hint: it was neither).
also, you're assuming that volunteer work is fungible, and that the GNOME project (i.e. the collection of maintainers, translators, documenters, and designers) can re-assign resources as they see fit.
people wanted to keep something like GNOME 2.32 (which didn't look or even behave at all like what was released as GNOME 2.0) around, so they decided to start Mate; it's perfectly fine (albeit it could have been done better, like many things) and covered by the licensing terms used by GNOME. actually, it's *encouraged* by the licensing terms used by GNOME. it was equally probably that something like that hadn't happened, and there was nothing the GNOME project could have done in that regard. at no point you can direct volunteer work to something "essential" to the project itself, let alone to what "bronson" on LWN considers essential.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds