I'm not someone who had a hand in GNOME 3 so I don't know the rationales. Nor have I followed most of the flames about it, nor been immersed in Linux user feedback lately, other than myself. So I've been trying to stay out of any debates about how GNOME 3 works in specific. I don't have a lot to say other than "it works well for me" and I know I do a lot of things differently than most people.
From general knowledge I would expect that anyone switching to GNOME 3 will hit speedbumps and initially be annoyed. I'm sure your wife would be. This is the cost of any big change. Whether the change is worth it in this case requires knowledge that I don't have.
It's easy to say CADT but it's also easy to point to countless technologies that became old and stale and were crushed by newer replacements. There isn't some easy guideline, like "never" or "always" change things. "It depends."
fwiw, I know enough about the tech to tell you that GNOME 3 is not from scratch by any stretch. GNOME 1.x -> 2.x preserved a lot less code. GNOME 3 is essentially a new UI in the window manager (but keeping a lot of existing tricky WM logic from Metacity), dropping some deprecated stuff, etc. What changed is very visible but not necessarily that huge code-wise. See also http://www.joelonsoftware.com/articles/fog0000000356.html
The issue is that the pixels changed.
I'm not trying to tell anyone that GNOME 3 will definitely succeed. There are just certain arguments about it that I think are wrong.
The bottom line is that success or failure, just as with GNOME 2 or anything else, will come down to detailed judgments about specifics. The developers will have to get it right, or not.
I wouldn't venture to say whether it's on the right path, on balance, without better visibility and immersion in the strategy and the feedback. For GNOME 2 I had the daily feed of raw information, now I just have my own usage and occasionally reading an article like this one or whatever.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds