October 20th marked the seventh anniversary of the Ubuntu distribution. Anniversaries are times for reflection, so I've been thinking of how Ubuntu has succeeded and how it has failed in the last seven years.
To hear those involved with Ubuntu, the distro's history consists of nothing but triumph. Community manager Jono Bacon marked the anniversary with a blog entry full of nothing except praise and enthusiasm.
Founder and dictator Mark Shuttleworth did not refer specifically to the occasion, but he did blog that Ubuntu "is the #1 OS for cloud computing," and that the next release "will be the preferred desktop for many of the world’s biggest Linux desktop deployments."
Elsewhere, Ubuntu is presented as "third most popular operating system in the world," presumably after OS X and Windows.
But although such claims might be true, we have no context in which to judge them. Linux generally requires no registration or activation, and, in these days of virtualization, one machine may have three or four guest operating systems. Thanks to virtualization, someone may have Ubuntu installed for testing or information purposes, but not be a regular user.
As a result, the claims about Ubuntu's user base are largely meaningless, because we have no idea what assumptions they are based on. Estimates in the last few years have ranged from 10-14 million, a variation so wide that it indicates how little certainty actually exists.
Moreover, if you accept the figure of 1.8 million downloads from unique IP addresses for Fedora 15, then assume that Distrowatch's downloads for the last six months accurately represent the relative popularity of distributions (a large assumption admittedly, but perhaps the most unbiased available), then Ubuntu's actual user base is more like 2.7 million.
This figure would still make Ubuntu the most popular distribution, but only 200-300 thousand ahead of Linux Mint, whose surge in the last year to second place on Distrowatch might be read as a response to Ubuntu's switch to the Unity desktop. If you accept these figures, then Ubuntu’s success is far more modest than is usually claimed.
If nothing else, this example shows that any such figures are too unreliable to be used as a criteria for judging Ubuntu's success. Just as clearly, too, Ubuntu's success cannot be confused with the success of Canonical, its commercial arm, although the two are often intertwined. Instead, Ubuntu needs to be assessed by what it has tried to do, and the extent to which it has carried out its plans.
The Good, the Bad, and the Indifferent
When Warty Warthog was released just over seven years ago, distributions were just starting to think of usability and desktop design. Perhaps Ubuntu's earliest accomplishment was to take that vague tendency and transform it into a priority.
Through a series of small innovations, Ubuntu quickly became a desktop leader. Its improvements included a simplified menu, automatic mounting of external devices, the use of sudo when administrative privileges were needed, and increased default support for a variety of locales and keyboards.
Ubuntu was not the first with any of these innovations (nor even the second), but it was perhaps the first to assemble all of them on a single desktop.
In addition, Ubuntu quickly established a large and active community. Part of the growth of community may have been the confusion of the distribution with the Ubuntu philosophy -- a fusion you still see today among local Ubuntu user groups. Still another part may have been founder Mark Shuttleworth himself, a space tourist and a multi-millionaire eager to spend his own money improving Linux, a charismatic outsider promising to fulfill the community's dreams.
This initial enthusiasm has been challenged in recent years, but Ubuntu literally wrote the book (and shot the videos) on community building -- Jono Bacon's The Art of Community, whose second edition is currently being prepared. Meanwhile, the first edition remains a standard guide to managing the free software community.
Just about the only initial mis-step in Ubuntu's early structure was deciding to base release on Debian unstable repository. Admittedly, Debian's stable release is notorious for being out of date. But by choosing the unstable release, Ubuntu committed itself to doing the testing carried out by over a thousand Debian developers with only a fraction of their numbers.
This decision explains why Ubuntu has had periodic quality control problems, and, in its early years, was often accused of being slow to fix bugs. Probably the Debian testing repository would have a more reliable basis for Ubuntu.
As things are, Ubuntu has done its best to continue as it began, but not always as well as in its first releases. True, the introduction in the last couple of years of cloud computing, a music store, and a third party repository that includes proprietary software are just as innovative for their time as anything in Warty Warthog.
However, you might question whether these innovations improve the user experience, or have more to do with Canonical's efforts to become profitable. Certainly, no other leading distribution has apparently felt the need for such features yet -- but the verdict may be still out.
Similarly, Ubuntu has had an uneasy relationship with proprietary drivers for video and wireless cards – they’ve included them, yet until recently not made them easy to find. Unlike Fedora, which has taken a firm position against including proprietary drivers, Ubuntu has seemed unsure whether to include them and improve the user experience, or to take a principled stand against them. At various times, it has been seen as doing both.
But undoubtedly the most ambiguous of Ubuntu's moves has been to take the concern with usability to an extreme. No one can argue with the small improvements of Ubuntu's early years. Nor could anyone, in theory, object to Shuttleworth's call in 2008 to make the free desktop rival Apple's for efficiency and visual appeal.
Yet, at the same time, no one could have foreseen that this call for greater usability would have resulted in Ubuntu spending months choosing a new default color palette or taking a year and a half designing the Unity desktop. Given Ubuntu's limited resources compared to GNOME or KDE, you have to wonder whether they were worthwhile -- particularly in retrospect, since the user response has been so mixed.
Even more importantly, these priorities have isolated Ubuntu from the larger free and open source community. While in 2006, Shuttleworth was calling on major projects to coordinate their release cycles, Ubuntu has moved to developing its key pieces of software mostly in-house.
That this arrangement creates labor shortages is suggested by the recent call for developers to work on Ubuntu projects -- a call that, considering Ubuntu's insistence on copyright assignment, seems unlikely to be greeted with much enthusiasm.
Acquaintances within Ubuntu deny my intepretation, but it looks very much to me as though Ubuntu in general and Shuttleworth in particular attempted to assume the leadership of free software development and became impatient when they didn't immediately get the results they wanted.
When the attempt proved slow and difficult, they withdrew. This position reduced the problem of cooperation with other projects, but at the expense of becoming more self-reliant.
In effect, Ubuntu has become the China of free software. Too large to ignore, it paradoxically remains both isolated and influential. The result is probably Ubuntu's second greatest failure to date.
The greatest failure? That would have to be the failure to develop a smooth relationship between Ubuntu and Canonical.
Unlike openSUSE and Fedora, Ubuntu has never truly become independent of its commercial sponsor. Like other distributions linked to businesses, Ubuntu has benefited from both labor and resources donated from its corporate sponsor. But the difference is that Ubuntu has never been self-governing in the sense that openSUSE and Fedora are.
When Mark Shuttleworth decides on a new direction for Ubuntu, the community has few choices except to go along with it. Individuals can question the new direction on mailing lists or by leaving comments on Shuttleworth blog, but no one has any authority to override or modify it.
This relationship seems to have caused few problems in Ubuntu's early days. However, in the last couple of years, it has fed a steady undercurrent of discontent. While the Ubuntu community is in many ways a model of open source organization (ironically, in no small part to Canonical employees), Canonical appears to have become increasingly driven by the need to show a profit in its decision.
As a result, in the last couple of years, Canonical has intervened more and more in Ubuntu, issuing arbitrary degrees and creating a resentment that is readily apparent in any browsing through the project's mailing lists. Increasingly, it seems obvious that what is good for Canonical as a business may not be ideal for Ubuntu as one project within a larger community full of traditions.
Weighing Success and Failure
You can easily assess Canonical. As a commercial company, it is judged on its ability to make a profit. Canonical hasn't done so, so you can judge it a failure so far without much fear of contradiction. Or, if that sounds harsh, you might prefer to say that Canonical is still being developed.
By contrast, no generally agreed-upon criteria exists for distributions. All the estimates of user bases are suspect, and other successes and failures can't be weighed one against the other. How, for example, do you evaluate the comparable importance of Ubuntu's early successful community-building and its current isolationism? The best you can say is that both influenced Ubuntu's development, and will continue to do so.
From the start, Ubuntu has been an amibitious project. That means that its successes and failures alike are large and often dramatic. But as Ubuntu moves into its eighth year, it defies definitive judgment. Although it almost assuredly has not succeeded as spectacularly as sometimes claimed, it has still come from nowhere to become a dominant distribution. And it’s likely to remain one for some years to come. Similarly, if it has failed in some of its efforts, so far, its failures have not been crippling.
In the end, Ubuntu proves to have been both more than anyone could have hoped seven years ago, and less than it could have been if all had gone well. That isn't the tidiest summary imaginable, but it does suggest that the next few years of Ubuntu should be even more interesting than the last seven.