I said earlier that I did not like the direction in which KDE 4.0 was going in. But the underlying cause can be perhaps found in Havoc Pennington’s post.
GNOME 2.0 and KDE 4 are bad models for change. They rewrote and broke the code, but from a user-goals perspective, they are the same thing as before. We shouldn’t feel bad; Windows Vista made the same mistake. Nobody cares about Vista, because XP allows users to accomplish all the same goals. Even if Vista didn’t have a bunch of regressions, nobody would really care about it.
I do not agree with everything in there, but Havoc gets the reason right.
I might have a bigger problem though, with the computer-is-a-desktop paradigm. I think our computers have to grow beyond it. We do so many things with it that a desktop is an insufficient and inefficient abstraction.
XFCE has already made me adopt new ways. I now minimize the applications I am running on the desktop. The application launch icons are pretty useless, as I mostly use shortcuts to launch those. In contrast to the current trend in desktop environments, XFCE helps me think about it.
Another aspect I am going to experiment with is whether it would be better to make the desktop task-oriented instead of application-oriented. It is a nascent thought, so some of you might find it completely frivolous. I have observed that people use computers to do tasks and modelling the desktop based on this might be more cognitive to use.
I also think that we can do more in the communication space between the desktop and the applications.
I have tried various desktop environments, but the current trend seems to be tilted more towards the eye-candy. One interesting project is Mezzo, but I think we really need to think beyond the desktop for new developments.