Quite frankly, even if
the choice of C were to do *nothing* but keep the C++ programmers out,
that in itself would be a huge reason to use C.
(Linus Torvalds)

I just happened to see this, which is about a rant by Linus Torvalds, about Gnome 3. They conclude that XFCE will probably be more popular in the future. And I think that is a correct conclusion. XFCE is not that bloated, but still versatile. However, I am afraid that XFCE will join the fate of KDE and Gnome soon or later (especially when I see how the Xubuntu-people raped it).

The Linux desktop seems to be "stuck" somehow. The systems get more and more complicated, having a lot more little tweaks popping out of edges when pressing keys, but the usability does not increase in any way. That is, you can get used to the desktop environments, but especially advanced users want to adapt the computer to their needs rather than the other way around.

I think there is a simple fact that we just have to admit: The Linux desktop is good as it is.

No matter how many tweaks you add, the basic design has not changed and just leaves a few alternatives which are commonly used. There are some niche-products like tabbed and tiling window managers, there are the usual window managers with a task bar, there are docks, there is stuff like a common main menu for all applications. There are a few tweaks like key combinations that activate some other bars to appear. And except a few newer graphic tweaks (like window-previews in the taskbar or mac's expose) that need GPU acceleration, everything has been there for years. And still, the old Windows 98 Desktop default configuration is much better for both beginners and advanced users than much of the new stuff, and even the design may not be the most beautiful, but is acceptable.

Of course, a desktop environment does not only consist of the window management. But the same goes with file managers and editors: There are several approaches for having an UI, but all of them have been there for years. There is probably always some space left for innovation, but at least currently I do not see any.

Therefore, it seems like all this blabber about new versions of desktop environments is mainly about the recombination of already existing concepts, and about design. I would appreciate if the responsible people would concentrate more on making the software stable than making it new. In that sense, I like projects like Trinity very much - but of course, for a programmer, it is more prestigious to create something entirely new than just keeping the old stuff stable.

So here is a suggestion for something really "new": As I pointed out, I think all desktop programming is mainly about recombination of existing features. There is a lot of expierience with these features, and it applies to most operating systems. So why not use this knowledge to build a "meta-framework" for defining how features can be combined, platform-independently? What I am thinking of is some simple scriptable API with which one can program and configure his own environment. For example, I could think of something that bases on HTML and CSS with some extensions that allow to include program windows, but without having to care about how exactly the graphic backend works. Where you can include program windows into your DOM-tree by a special element, like you would do with images and stuff. HTML and CSS, because these are formats that have been here for years, are well-tested, well-supported and well-maintained, but of course, any other similar format would do as well.
Such a thing is definitely not trivial to implement, but I think it should be possible, and it would be nice.

And it would be better than bloating up existing environments.