GNOME is a great desktop environment. Corporate users feel right at home. In many ways it is a bit like Mac OS X. The main goal of its development seems to be a simple but yet powerful interface. Gone are the toolbars with numerous buttons, the endless options and configuration dialogs which most times can confuse the users rather than help them. Consistency, simplicity and uniformity best characterize the Gnome desktop. Some Gnome technologies are even used outside the Gnome community. Nokia for example uses gnome-vfs (a virtual unification of different filesystems implemented in userspace) in their latest gadget 770.
Unfortunately, Gnome historically has targeted the corporate world. This is also reflected in its applications. Evolution, gnumeric, nautilus and friends are all great applications but bring nothing new to the table. Furthermore a great deal of Gnome technology is developed for administrators (who manage many Gnome Computers with lots of users) reminding us again that corporate users are first in line. Examples are sabayon and the notorious gconf. Applications like beagle or F-spot cannot really disguise the fact that GNOME was and will (at least in the near future) always be offered by Sun/Novell/Redhat to their respective corporate users.
GNOME suffers from the flaws of the desktop metaphor. This is true even for experienced Unix users. A lot of them use GNOME because of the simplicity and great technology behind it, and at the same time are amazed when they discover that their desktop quickly accumulates icons from "temporary" tasks. Maybe Gnome needs a "clean up desktop icons" wizard too.
The "live" desktop in Gnome comes in the form of gdesklets. The technology is powered by python and XML. The idea seems interesting but in the end most gdesklets are the usual suspects. That is, mem/sys/cpu/net load graphs, xmms frontends and weather forecasts. Great looks no doubt, but the functionality was already there before.
So where does Gnome stand today? In an interesting move by Novell several videos showing an XGL-powered GNOME have surfaced on the net. These videos look really amazing. It's GNOME like never seen before. Transparent windows? Liquid style windows? Live video thumbnails? Cube desktop switching a la 3ddesktop? It is all there. Several Unix zealots saw the videos and instantly declared that Vista is no-where near this and that the Unix desktop will never be the ugly cousin of modern desktop environments. Unfortunately they don't see the big picture. Even though these screen effects are cool, they add nothing to usability. They only thing that really deserves so many headlines is the expose/clean-up-windows effect which will certainly come handy when the desktop is cluttered with many windows. Another interesting feature is also the graying-out effect for frozen applications. This gives a visual clue to the user that something is wrong with an application. But really everything else is pure eye candy. Novell has really surprised us not with the videos themselves but rather with the direction they are going with them.
Meanwhile there are numerous other technologies around, that crave for developers and will help users when they will become mature. We mean the howl/zeroconf/avahi implementations or even the dbus/hal combo. The future Gnome desktop powered by them will certainly be better than the current one and not just look better. Sadly these technologies are not in the headlines. This is because they are considered infrastructure (=boring) work and 3D stuff is always more eye-catching.
Soon after the release of XGL, redhat/fedora proposed their solution for a fancier desktop along with the obligatory video samples. Nvidia published their own whitepaper (PDF) too giving their ideas on the situation. Most people started comparing the technologies or asking why XGL development took place behind closed doors, when the real question is what enhancements do they offer to the desktop experience.
To sum up, it is sad to see that interested parties are just focusing on the visual effects of the technologies. The single most important purpose of the graphic system is to improve the human-computer interaction and not to look cool. It is easy to create great-looking interfaces that cannot be used by humans, but rather difficult to create usable interfaces that also look good too.