There is a good article here! I have almost exclusively used Gnome ever since I started using Linux (which is a pretty long time). The transition to Gnome 2 was a wonderful one and Gnome 2 was, at the time, very innovative. However, I’m not sure Gnome is at all innovative anymore. Useful for the individual, yes. Stable, yes. Pretty even. Gnome is a solid platform, but honestly, to the average user, little has changed since Gnome 2.0. That could be either a good thing, or a bad one.
It is a difficult question as whether great innovations should be radically implemented or gradually. You write Gnome misses leadership at the moment (and thus vision). This makes either next to impossible: without leadership, no-one will sanction radical changes, and without leadership, no-one can ensure big changes to be implemented over a prolonged period of time.
Over the past few years, there have been some radical ideas about the future of Gnome (one of which was the position of Mono in the Gnome environment, iirc). Some of these were good, some of these were not (I’m still grateful for the small position Mono has now), but imho most of them were either aimed at cosmetics and / or relatively small functionality enhancements (like the recent combining of multiple copy progress bars in one window). If I understand the buzz correctly, GTK3 is about interfaces too. Interfaces are nice, but Gnome is missing key functionality. And functionality is what wins the day for you.
My personal opinion is that Gnome should aim more at corporate desktop features. I work at implementing Linux desktops a lot and I can easily name a bunch of features Gnome would profit from immensely. Some of these are readily available in KDE, by the way, which is the main reason I often choose to implement KDE, even though I personally prefer Gnome.
Gnome would benefit from:
- the possibility the really lock a desktop environment. Sabayon has great potential, my experiences with it are far from ideal. Let’s say I have yet to see it do what it is supposed to do. Apart from that, it offers too few options. KDE’s kiosktool is a lot more functional (assigning more than one profile to a user, to stack them, so to say; really locking away the commandline; locking a lot more settings, like wallpaper, theme etc.)
- Firefox and OpenOffice.org should integrate into Gnome more, possibly with a gconf backend for settings. Admittedly, I don’t know how much effort this would cost, and whether this is actually desirable, but integration would be good (for user profiles and Sabayon, again). I think OOo and FF integration is a neccessity, because Epiphany and Abiword / Gnumeric are just not good enough for a corporate desktop. The way KOffice is moving along, it might provide a viable alternative to OOo, but I don’t it will integrate into Gnome very well ;-)
- Evolution should have mapi support and thus real support for Exchange 200. Novell is working on this, I believe. This would really benefit the Linux desktop. - there is no easy way I know of to scale the used iconset in total (use smaller icons everywhere); on several occasions, I wanted a desktop with smaller icons and smaller fonts. A lot of iconsets are SVG, so this is expected functionality, ifaic.
- it’s all very cool GVFS understands webdav, but in my experience SMB / CIFS performance is poor, though that is an important feature for corporate desktops.
That’s just a few Gnome specific ones. I’m tempted to write a list with features the Linux desktop misses in general, but this probably isn’t the place.
I know for a fact a lot of governments and corporations are looking into replacing their Windows desktops with Linux or at least providing an alternative to the established Windows environment. I also know for a fact, that most of them are looking into KDE, no matter how hard different Linux vendors support & push Gnome.
To summarize, the way to go would be: aim at corporate desktop, provide good possibilities to communicate with existing infrastructure (mainly Microsoft stuff, like fileservers and Exchange) and provide solid locking / deployment tools, like a working and enhanced version of Sabayon.