Re: a definition of insanity
Sometimes the lesson isn't "try again." Sometimes it's "this is not a good idea."
It's long been known that a Jack of all trades is master of none. A UI that tries to "unify" dissimilar devices that have dissimilar UIs for a reason is not going to be as good on either of the platforms than one that was built specifically for the platform in question.
The change in GNOME seems to be hand-in-glove with Unity. Both are focused on the same "master of all" chimera that has obsessed Microsoft of late. As such, the various GNOME applets that are distributed with many Linux distros, things like Calculator, Disks, GEdit, etc., have all been infected with the hamburger menu. That abomination may be a necessary evil on mobile devices, with their tiny, clumsy, imprecise touchscreens, but there's no reason I should have to deal with them on my PC when a better, more intuitive choice (the pulldown menu bar) has been the industry standard for 30+ years.
That's the problem with these UIs that supposedly adapt to desktops or mobiles as needed-- they don't. If they did, all of the UI compromises meant to make touchscreens work (the hamburger menu is just one of them) would be blissfully absent on PC screens, but they're not. You still get the stupid stuff like tiny, disappearing scrollbars and other UI elements, sparsely-populated displays with those elements that do exist taking up way more space than they should, and all kinds of other things.
The only way I can see this ever working is to have every bit of the OS or distro, and every program that is supposed to share the convertible UI, have two fully-developed UIs for everything. In Windows terms, every program's UI, every menu, every dialog would have a native Win32 version and a UWP version, each optimized for the platform for which it was designed.
It wouldn't really be "one for all" as much as "two in one," and that's not all that creative or slick... just redundant. It's not as wasteful as it may sound; the controls and widgets used to build each dialog are already there in any case, with only a small bit of code needed to describe the UI layout. Still, that's not the way MS went, or GNOME 3 itself, or Unity. Duplicating code is inelegant; they want something more "yahh, way kool d00d" than that. What we're left with is desktop operating systems that have hamburger menus, oversized controls, menus that require more drilling and that end up with options being removed to make them less confusing, broad areas of empty space...
It's notable that both versions of Windows that were supposed to be "one UI to rule them all" have been unpopular, with 8 being rejected nearly completely (its market share is smaller than that of XP now) and 10 being so "great" that people won't even take it for free. If MS had not used malware techniques to push 10, it would almost certainly be as complete a failure as 8.
Ubuntu used to be the top "beginner" Linux distro, but now it seems to be Mint, at least judging by Distrowatch (as imprecise a measurement as that is, it's the best we've got). That change coincides quite well with the adoption of Unity and the new touch-friendly direction of GNOME 3. Mint itself is a reaction to the growing mobile-ification of Ubuntu and GNOME 3; its main mission in life is to preserve the traditional UI and eschew all of that mobile stuff on non-touch devices.
Apple seems to be the only one that has it right. Tim Cook specifically rejected the idea, stating that there would be compromises that lead to the experience being less than it should be on both platforms. I don't agree with Apple often, but this is one of the rare times when I do. I think it's a fool's errand to try to make one UI for two very different platforms. It makes no more sense than trying to graft automobile controls onto a motorcycle or motorcycle controls into a car. They have different "UIs" because they have different needs.