The elephant in the room with GTK2.

GTK3 and GNOME3 were not designed around touchscreens and predate the stratospheric rise of modern touch-based smartphone UIs. Indeed G3 was explicitly designed for keyboard-centric (if not keyboard-exclusive) use, and works far far better than G2 in that respect.

G3's primary goal was minimizing distractions; acknowledging that one-fullscreen-application-at-a-time was how most folks actually used their systems, so they made multiple dynamic desktops a first class feature to make switching between said applications easier, along with the overview for a quick way to see everything, and navigate between or launch new things.

Other stuff like putting controls/buttons into the title bars were mostly by wide aspect-ratio screens with poor vertical resolution (eg the 1280x720 on relatively large screens that was the unfortunate baseline until pretty recently). UI elements to make touchscreens easier (eg larger buttons, thick window titles, elimination of window decoration, etc) had their origins in accessibility -- larger (and fewer) targets are easier to hit with a mouse when you have poor motor skills (or a touchpad!) and so forth.

> Now, from the perspective of most open-source developers, working on something like GTK2, or a modern equivalent, isn't interesting

It's not a matter of _interesting_ so much as that porting an application from GTK2 to GTK3 requires a ton of _work_. And folks reasonably expect some sort of compensation for that work (of course, "learn new marketable skills" can be one of those rewards, but in that respect it makes more sense to learn a web-centric UI toolkit instead)

Meanwhile, dragging GTK2 into the Wayland era will require substantially rewriting most of it, and dropping what's left. In other words, a process similar to what GTK3 already went through.

(As an aside, GTK3 also predates wayland, so you can't blame Wayland for its design either. But the design of both was influenced by many of the same failings of X11 leading to moving entirely to client-side rendering/decorations, among other things. GTK2 had a lot of X11-isms baked into it, and it was/is common for GTK1/GTK2 applications to have their own direct interactions with the X server. This is one of the things that made GTK1/GTK2 applications such a portability nightmare..)

> What most of those users actually want is essentially boring security updates and tooling updates to something that's still working fine for them, not fundamentally new paradigms or a big change. They want essentially the same kind of stability a lot of big enterprises want, only they aren't in a position to pay a lot of money for it.

I'd correct that last bit to read "only they aren't willing to pay _any_ money for it."

> I think if something like that existed, people might be willing to put in the work to migrate, but as it stands, they prefer to just stick with the old thing as long as they can, because they don't actually need or want anything newer GTK provides

To analogize, their comfortable middle-class lifestyle has been heavily subsidized by other folks' hard work. Now that those other folks have largely moved on (often in a very _final_ sense) these users are finding out just how much work is no longer getting done.