Trying to make sense of the recent history of the Linux desktop, I realized recently that the main perspective was wrong. The fragmentation of user share between desktops isn't a new trend -- it's simply a reversion to coding as usual in free and open source software.
That the fragmentation is happening is beyond dispute. Five years ago, GNOME and KDE together accounted for roughly eighty percent of Linux desktops, yet today the picture is more complicated. GNOME 3, Ubuntu's Unity, and Linux Mint's MATE and Cinnamon have divided GNOME's user share, while Xfce has leaped into second place after KDE, according to the 2011 LinuxQuestions.org Members Choice Awards. Other desktops have also become more mainstream -- LXDE, for instance, is now the basis for the Ubuntu variant Lubuntu.
Conditioned by proprietary software, where a single desktop per operating system is the norm, many pundits worry about this situation. Some worry about the inefficiency of the same efforts duplicated many times over, or whether such variety might discourage hardware manufacturers from supporting the Linux-based ecosystem as a whole.
The concern is understandable, but mistaken in its assumption. For eight to ten years, GNOME and KDE dominated the Linux desktop, so much so that those who started using free software in that period must be forgiven for thinking that was the norm.
Now, the unique circumstances that produced this dominance have diminished, and desktop environments have now returned to the diversity of other free software categories.
The Rise of KDE and GNOME
KDE and GNOME dominated the Linux desktop for so long because of the circumstances in which they began, and later because of their rivalry.
Neither was the first interface for Linux or other Unix-like systems. Window managers and proprietary desktops like the Common Desktop Environment (CDE) were widely available long before KDE was founded in 1996.
However, before KDE, widgets and applications lacked a common look and feel. Nor could they share libraries and other resources. In other words, the situation was similar to the days of DOS, and far behind what was available on Windows or the Mac. In working to provide these advantages, KDE and later GNOME were carving out a new space on the Linux desktop, like nothing that had come before them.
Given the relative complexity of desktop environments and the much smaller size of the community in the late 1990s, even GNOME might not have begun development, except for one thing: KDE's use of the then-proprietary Qt toolkit. In response, the free software community began GNOME in 1997 in order to provide a completely free-licensed desktop.
This response was the beginning of the legendary rivalry between the two main desktops. Traces of the rivalry remain today, and you can still find users of one desktop who know nothing about the other one.
But compared to the flame wars around the turn of the millennium, the current situation is mild. Back then, not only the ethics of free software were matters of passionate debate, but also related issues such as whether Debian should continue to include the then non-free KDE in its repositories.
At the time, the rivalry seemed unnecessarily divisive. A time would come when GNOME and KDE developers would try to work together to ensure cross-compatibility with organizations like freedesktop.org and joint conferences.
Still, although the rivalry might have been counter-productive in many ways, it helped to assure that, in many users' minds, KDE and GNOME were -- if not the only Linux desktops -- at least the only ones worth considering. In the middle of the highly partisan flame war, other alternatives were all too often overlooked. As early as 1998, during the Dot Com Era when the computer industry first discovered Linux, the only desktops being discussed were GNOME and KDE.
In opposing each other, the two rivals helped to ensure that other choices were rarely acknowledged, let alone considered. Throughout the first decade of the millennium, the two desktops vied closely for dominance, often exchanging positions at the top of the opinion polls, with alternatives like Xfce a distant third to both.
The Decline and Fall of the Giants
With the rivalry more or less stalemated, both KDE and GNOME developers eventually settled down to incremental releases, each group adding features without altering the basic look and feel of their software. In the short-term, this situation satisfied everyone, making most users more or less satisfied and disinclined to look at alternatives.
However, in the last few years of the decade, the limits of the existing software had been reached. Although sometimes rough around the edges, GNOME and KDE had both matched their proprietary equivalents. Leading KDE developers had begun to be interested in expanding the desktop metaphor, while leading GNOME developers wanted to implement usability theory more thoroughly.
What neither project had counted on was that users would no longer follow quietly where developers led. While developers wanted something new to keep their interest strong, most users were content with what already existed. If anything, the long stasis of incremental releases only increased user's tendency to conservatism.
Consequently, when KDE 4.0 was released in 2008 (and included in distributions, despite being a developers' release), nothing less than a user revolt took place. So far as I know, no one ever measured how many users KDE lost as a result, but opinion polls suggest that it might have been as high as ten percent, mostly to GNOME.
KDE managed to maintain the rest of its users through a combination of publicly discussing the situation and by adding over the next couple of releases -- as its developers had planned all along -- the features that users missed most.
It helped, too, that, the KDE 4 series restructured the desktop without changing its basic nature. Although users might struggle with setup, once everything is configured, the functionality of the KDE 4 releases differs little from that of earlier releases.
When GNOME came to make its changes, it lacked similar advantages. First, Ubuntu became more distant from GNOME, adding its own features unilaterally, and finally settling on Unity for an interface. Then GNOME 3 produced its own user revolt, thanks to an over-reliance on abstract usability theories and radical changes that allowed fewer choices of work flow for users -- and never bothered to discuss the dissatisfaction, or even admit that it existed.
For KDE users who had fled to GNOME, the result was history repeating itself. For half or more of GNOME users (judging from surveys), the combination of unwanted changes and an unresponsive development team was more than they cared to endure.
While the old rivalry may have kept many GNOME users from investigating KDE, they did look into Xfce, Cinnamon, and Mate, searching for as close a recreation of the GNOME 2 series as they possibly could.
In less than a year, the fragmentation was well established. As things stand now, the chance of it ever reversing seems unlikely.
Alternatives or inevitability
In studying this transformation of the Linux desktop, you can easily see possible turning points. What would have happened if the KDE 4.0 release had been delayed until it had more features? If Ubuntu had been more patient about its changes getting into GNOME? If GNOME 3 had been less radical, or user complaints addressed? If some or all of these events had occurred, then maybe GNOME and KDE would have remained as dominant as ever.
However, I doubt it. More likely, other incidents would have caused a similar fragmentation sooner or later, no matter how anyone acted.
The fact is, free licenses encourage fragmentation by definition. Small items like desktop wallpapers or browser extensions number in the thousands on the Linux desktop. The more complex the general category of apps becomes, the fewer the alternatives, but there are still at least half a dozen major choices for email and web browsers, music players, and most other standard desktop items.
Even if fragmentation could be proved harmful -- which I doubt -- nobody could do much about it anyway. Even projects like the Linux Standards Base have a hard time encouraging uniformity.
The only exceptions to the diversity are either specialty tools such as FontForge, which is used for the development of typefaces, or software released under unusual circumstances, such as OpenOffice.org. When Sun Microsystems released the code in 2000, OpenOffice.org was already a complete office suite, and years ahead of free software's tentative efforts. In fact, OpenOffice.org was so far ahead of any alternatives that the tentative GNOME Office was abandoned, leaving AbiWord and Gnumeric as sideshows.
Yet even Sun's efforts to control development could not prevent off-shoots of OpenOffice.org like NeoOffice and Lotus Symphony, let alone LibreOffice, which has now replaced it so thoroughly that OpenOffice.org may never catch up.
There's no reason to think that KDE or GNOME would have been any different. Because of unusual, mutually reinforcing circumstances, the two projects dominated for nearly a decade, but the situation was an anomaly that could hardly be expected to endure and is unlikely to be repeated.
The fragmentation of the Linux desktop may turn out to advantageous. It may turn out to be harmful. However, more likely, it was simply inevitable -- and, no matter what anyone thinks, everyone simply needs to get used to it.