Exactly. Just think about how much work it was to move from X to Wayland, how Nvidia didn't want to play ball, how different the architectures are, in a world where ossification is very much a thing. AND STILL there was enough momentum to get all the major players on board. I have never done any X programming but it must have been bad when people would so readily jump ship and put all that work in. Same for systemd.
Wayland was simpler because it was just a compositor. It meant people interested in graphics acceleration could focus on just that, iterating faster without the friction of the rest of the stack. But now that the Wayland ecosystem has to reinvent everything else, it's clearly at least as much work, if not more so. The initial switch moved the tightly knit rendering developers to a better local optimum, at the expense of a worse global optimum.
It forces some kind of modularity. Maybe that is a good thing. We don't have all features of X11 in Wayland but we also have features in Wayland that are missing in X11. I have been on Wayland for a while now and I can't say I am really suffering. There are some tings I'd like improved further but the protocol is evolving all the time. I think we'll get there. Also the decision is already made by the big distros anyways. They seem to be mostly happy.
> It forces some kind of modularity.
Wayland is the complete opposite of modular; that's why the compositor has to handle everything. The claimed problem with X11 was it was too modular, easy for different clients to control various aspects of the environment, but perhaps too easy, lacking a security policy framework to limit client access. (Or at least a framework that was fully baked. Wayland put an end to such initiatives.)
The lack of modularity in Wayland has been widening the rift between the desktop environments. It's as-if KDE and GNOME each forked X11 and went their own ways.