The Decline of Linux Diversity

The evidence is mounting: Linux is less diverse than usually assumed – perhaps far less.

Last year, I wrote how the number of Linux distributions listed on Distrowatch seemed to be declining. Specifically, the number had dropped from 323 in 2011 to 285 in December 2014.

Eleven months later, the decline seems to be continuing at about the same rate, with the number of active distributions down to 276, and the decline is starting to seem an actual trend.

Critics might argue that the apparent trend might not be a trend at all. It could be a reflection of Distrowatch's criteria for listing a distribution, or how quickly Distrowatch posts new distributions. However, given that the site regularly posts announcements of new releases for both new and established distros, there seems no reason for either to be a factor. Admittedly, Distro Hunt, a newer, similar site, includes listings that Distrowatch does not. But since projects can add their own descriptions to Distro Hunt, it’s possible that some of its entries have never had a release or disappeared without taking down their descriptions. Moreover, unlike Distrowatch, Distro Hunt provides no easy way of counting the total. The best available (if tentative) evidence, then, is that the trend exists.

But is the decline reason for alarm? That is harder to decide until you start looking at other evidence.

Diversity vs. Sufficiency

On the one hand, diversity can be considered a virtue handed down from Unix. Name any part of Linux, from core files to office productivity tools, and at least one alternative generally exists. Similarly, even the most lightweight desktop environment is full of customization options. It is probably not an accident, either, that KDE, which consistently ranks higher than any other desktop environment on user polls, has the most comprehensive set of customization options. Much has been said, too, of the healthiness of diversity as opposed to the monoculture of Windows or OS X.

Still largely free of commercial concerns, Linux can have alternatives anywhere that anyone with coding skills can implement one. Over the years, distributions have been released for Muslims, Christians, and Satanists, and so many other unlikely groups that knowing which was actually released and which is an April Fool's joke can be difficult (apparently, though, the Brittany Spears distro was a joke). People laugh at some of them, but no one questions people's right to make them. They are part of the natural consequence of using free licenses.

From this perspective, the decline in distributions is reason for alarm. Not that the concern is immediate; at the current rate, it would take twenty years of decline at the same rate before the number of distributions was below ten, and of course there is no reason yet to believe the decline will continue at all, let alone at the same rate. Yet the concern is over the principle. A decline -- any decline -- means a lessening of diversity, and never mind that few will face any immediate consequences.

On the other hand, perhaps a point comes when diversity becomes excess. Programmers new to Linux often complain about the difficulties of designing for more than one distribution. This complaint is partly due to inexperience, since, if nothing else, programmers can often concentrate on one or two major distributions, with the almost certain guarantee that free software programmers will take up the slack and adapt their work to other distributions. Still, the fewer distributions that exist, perhaps the more that programmers will be encouraged to write for Linux, so the decline is something is cheer about.

At any rate, let's be honest: Distrowatch may list 276 distributions, but how many are relevant? After the first thirty or forty, the rest tend to slide away into anonymity so far as the average user is concerned. If we only had ten active distributions, surely there would still be enough diversity to satisfy everyone?

The Lesson in the Top Ten

The only trouble with this line of argument is that, when you look at Distrowatch's top ten distributions by page view, not only is the lack of diversity obvious, but it has existed for some time already. If you look at Distrowatch's top ten distros by page view, little has changed in the past two years. Since 2013, seven distributions have been consistently in the top ten: Mint Linux, Debian, Ubuntu, Mageia, Fedora, openSUSE, and Manjaro. They have changed rankings, and since 2014, they have been joined by Arch, but otherwise, little has changed. Puppy, PCLinuxOS, zorin, elementary, and LXLE have all made their way into the top ten, but always in the bottom third, and none of them consistently.

Among the top three, Mint has always been in first position, with Ubuntu and Debian battling for second and third. All three of these distributions are based on Debian, any one of them has page views equal to at least two of the seventh through ten position.

At the same time, if you look at Distrowatch page views from 2010 and 2005, Ubuntu, Debian, and Fedora are constant, and you might equate Mandriva as the ancestor of today's Mageia, and SUSE as openSUSE's ancestor. But the other five in the top ten are more diverse, with distributions such as MEPIS, Knoppix, Damn Small Linux, Gentoo, Slackware, and Sabayon.

In other words, not only is there less diversity today in the top ten than you might think, but that has always been the case -- and, over the years, what little exists has been declining.

You might argue that downloads would be a better measurement than page views. Yet while that is true, page views still indicate interest, and are enough of a metric to justify the conclusion that the total number of distributions is no longer an isolated piece of evidence. Examining the Top Ten page views is still enough to conclude that Linux's diversity is declining, and is already less than is popularly believed.

Or, to put things the other way, if diversity is an important value for Linux users, the concern is very real. The next question is what can be done to stop the decline.

Tags: Linux, KDE

0 Comments (click to add your comment)
Comment and Contribute


(Maximum characters: 1200). You have characters left.