GNOME 2.28 was supposed to preview GNOME 3.0. But it hasn't quite turned out that way, and whether what is visible will leave users eagerly anticipating or uneasy and rebellious is still anybody's guess.
The GNOME developers did their part, bringing the GNOME Shell, the basis of the GNOME 3.0 desktop, into presentable shape. Unfortunately, neither the GNOME 2.28 release notes nor many of the distributions which packaged GNOME Shell give any detailed instructions about how to swap out the existing desktop and replace it with GNOME Shell. Trying several distributions, I found packages that installed without dependencies, and at least one that conflicted with a standard package.
In the end, I found that the most successful way to install the preview was to follow the installation and compiling instructions released last spring, and forget about the distro packages altogether.
If my experience is even remotely typical, then the point of the preview has been lost. I suspect that most users will not install it, and that GNOME will not get the feedback it was vaguely counting on from this release.
If anything, distributions seem to be downplaying the preview. It is as though, having urged KDE 4.0 too early on their users, the distributions have over-compensated and made an unspoken resolution to keep the new GNOME away from users until it is completely ready.
But perhaps the inaccessibility of the preview is just as well. What I saw when I finally managed a successful install could easily go either way in users' estimation.
Change and Mobile Devices
GNOME Shell's ambiguous potential lies in the fact that it is an attempt to redraw the computer desktop. Since users neither seem greatly dis-satisfied with the current state of the desktop nor in any agreement about how it could improved, this departure is risky. Some users will undoubtedly reject it simply because it is different, no matter how innovative or useful it is, much as they did with KDE.
The GNOME Shell consists of two main parts. The first is the panel, currently glue to the top of the screen and lacking any applets except for basic ones like a clock and system tray.
Press the Activities button on the left of the panel, and the second part is activated -- the Overlay mode. The Overlay mode consists of a dynamic menu on the left, and workspaces on the right, each with its own icons and windows. The menu, workspaces, and windows for applications can all change size, up to a full-screen mode for a single workspace or application.
Actually, my overwhelming impression of the GNOME Shell is of panes continually changing sizes. I soon realized where I had seen a similar interface: on music players and phones. Apparently, computing has reached the stage where mobile devices are the norm, and they, rather than workstations or laptops, are driving interface design.
This design shift seems natural and inevitable. Adopting the interfaces that people know seems only logical. It might even have the benefit of giving new users an interface in which they can feel immediately comfortable. Yet, I could just as easily see users feeling that what is appropriate for a mobile device is less so for a full-size computer.
On mobile devices, a flurry of resizing and replacing of views seems logical. The screen is usually less than three inches wide, and needs to be constantly cleared to free up as much of it as possible. Understanding this limitation and (the convergence of email, videos, and applications on to mobile devices notwithstanding) most of our uses of mobile devices being fairly simple, we endure such interfaces without much complaint.
But could users find such activity a distraction on the desktop? If the desktop is still the place where our activities are more important and take longer, then perhaps mobile devices are as poor a model for the desktop as the desktop is for mobile devices.
Up Close and Ambiguous
My uncertainty about how GNOME 3.0 will be received continues as I zero in on the general work flow.
For instance, the unspoken consensus is that the concept of the menu needs improvement, no matter which desktop you find it on. That is why so many different alternatives are being auditioned.
The GNOME Shell offers its own alternative, complete with auto-completion and changes to meet the current context. In general, its menu reminds me of Krunner and GNOME Do, both of which pack a surprising amount of functionality for the size of their windows.
Yet, at the same time, having the menu stay semi-permanently on the screen, occupying valuable space seems a step backward for me -- even if it does disappear when a workspace or window is enlarged.
Also, while I am no fan of the classic menu that cascades across the desktop, it at least has the advantage of giving new users something concrete to explore. New users could use auto-completion to explore, of course, but that is not nearly as convenient as having everything visible at a glance, and grouped into categories. To me, the GNOME Shell menu seems more a tool for intermediate users than beginners.
The same is true of the overlay view and the workspaces. For advanced users who are used to multiple workspaces and having several windows open at a time, these features bring a much needed organization and accessibility. In my experience, however, multiple workspaces tend to be ignored by new users, no doubt because they are not standard features in Windows. Perhaps GNOME 3.0 will introduce new users more quickly to the possibilities of multiple workspaces, but, perhaps, too, it will only confound new users with a level of complexity that they neither want nor are prepared for.
But probably the element that will most determine how GNOME 3.0 is received is how much customization the final version will have. What we see now gives few concessions to personal configuration, and possibly the lack of choices seems all the greater because of the resemblance to the limited functionality of mobile devices.
Considering that what we have now is an unfinished application, that lack seems understandable. At this point, developers have not gone far beyond functionality, and no doubt have barely thought about customization. Yet, when I remember the reception of KDE 4.0 twenty months ago, I am convinced that the degree of customization is central to how GNOME 3.0 will be received.
With KDE 4.0, people might have forgiven the new features and work flows if only they had the same degree of customization as before. But they didn't, and the KDE project has spent much of its time since digging its way out of the avalanche of hostile criticism. If the KDE experience teaches anything, the lesson is that users of the free desktop value customization above every other feature.
From this perspective, the discussion among developers of the GNOME Shell as a radical simplification raises the possibility of a backlash similar to KDE's. Like KDE, GNOME could endure a period where the project is perceived like the English Puritans of the 17th Century, taking away choices like the Puritans took away Maypole Dancing and Christmas for no better reason than they are caught up with their own personal agenda.
Missing the Users the First Times Around
The present incarnation of the GNOME Shell is far from the last. At least two point releases are scheduled before GNOME 2.30, and the project has announced that it will delay GNOME 3.0 for another release cycle if necessary. With these plans, the project should have several opportunities for receiving feedback and listening to it.
The trouble is, with KDE's example always in mind, the question is whether GNOME will do. The best time for user feedback would have been last spring, when the project was conceived, yet there was almost none. Now, with the version of GNOME Shell that came out with 2.28, another opportunity seems to have lost. To date, GNOME is not even trying very hard to sell its vision of the desktop.
Add the ambiguity of what is visible, and the concern seems justified. While the new desktop design succeeds in some respects, from other perspectives, the developers might be seen as working with themselves in mind, rather than unsophisticated users. That, of course, is a time-honored approach in free software, but, these days, it is one that a project as large and as important as GNOME can no longer afford. Nobody wants to stifle innovation, yet, at the same time, GNOME needs to introduce major changes with the cooperation of its users.
Let's hope it enlists that cooperation in the next few months. So far, I'm not sure that it has tried.