Instead tablets peaked, then bigger smartphones ate their share instead of tablets taking out the remaining use cases for traditional form factors.
And pushing that agenda tooth and nail.
KDE seems to keep it together. KDE plasma 5 seems an incremental upgrade from what came before, getting rid of some of the weirdness and converging on something sane. Progress at least. I can respect what they're doing.
Microsoft, in all it's bullshit, every few years seems to give the users what they want. Licensing is a headspinner, and privacy is questionable, but otherwise they still make an operating system that feels good to use.
I've been using Gnome now for about three years. I find the hot corner UI to be intuitive (and I find myself trying to use that and the meta-key expose function on other OSes). The favorites dock and the search are things that I use frequently, and the title bar menus make sense to me.
I was a Gnome 2 user for years as well, and yes it was polished and predictable. But it wasn't attractive. I think Gnome 3 was the next logical step for the project as far as contemporary UI patterns go, and it feels very "ergonomic" to me.
I would suggest that Pantheon (the DE for Elementary OS) is more polished compared to Gnome 2. Cinnamon (the DE for Linux Mint) is another solid choice if you wanted incremental improvements over Gnome 2.
Instead it seems like in the main, people have (once again) decided that they don't really care to switch between a PC modality and a tablet modality [ADDED: on a single device], however appealing the concept seems on paper. And, as you say, the smartphones that everyone (to a first approximation) owns grew to their approximately maximum size and those are good enough tablets for a lot of people.
Windows 8 was designed before the iPad was announced, but that 3 year ship cycle this article discusses came back to bite Microsoft. Their vision was a hybrid system but then Apple came out and solidified themselves as the standard tablet. Microsoft lost their ability to define the market and the market liked Apple's definition better.
For example the new ideas for the UI were quite interesting, but it was quite obvious they had ship it way before it was ready. They were interesting, because they solve one (IMHO) big issue: ordinary users have hard time taking full use of their screen real-estate. Windows are hard to manage and people end up switching between full screen apps, while it would be in many case more effective to use them side-by-side. A tiling window manager would help here.