There are too many DE projects, so the already scarce human resources are scattered among them. The final result is the slowness in bug fixing, where not in the development itself.
Then there's a lot of focus on "user experience" which I translate with "eye candies" and "cosmetic features" and not enough focus on "real user experience" which I translate with "real life use" and "meat".
Try reading the latest release notes for
KDE and GNOME (both
core and
apps).
A few examples.
NetworkManager GNOME's front end is quite usable. KDE's is not working properly, especially with system wide connections.
CD/DVD burning KDE's (K3B) can do almost anything you need, while GNOME's (brasero) is too basic.
Then you have a number of GTK+ (GNOME) pieces of software with no real competition in QT (KDE) and vice versa, And a few which don't use either and are real leaders like Mozilla Firefox 5.
And, finally, the bloatware is spreading everywhere.
It's almost impossible to run KDE without running MySQL at the same time (bacause of the Akonadi PIM)!
In the end, XFCE still needs bits from GNOME for full functionality. LXDE and friends are either too embryonal or are actually toys.
The same seems to happen with Linux distributuions.
The only thing to fear is that the whole Linux world will be exiled to servers and not spread on desktops and portables, where the DE is among the main components.
DE developers, unite!