Off the Beat: Bruce Byfield's Blog

One of the frustrations of writing about the free desktop is the lack of information about the distributions and apps that people are using. That's why, when the latest issue of the online Linux Journal came out, I immediately turned to the results of its Reader's Choice Awards in the hopes of observing the latest trends (despite writing on a competing magazine's site)

Of course, like all such surveys and questionnaires, Linux Journal's Reader's Choice Awards are an imperfect indication of what's happening. Linux Journal's audience probably consists of moderately experienced users, so the results are unlikely to indicate what recent immigrants from Windows are using, or the preferences of someone who just bought an Ubuntu-preloaded laptop from Dell.

In addition, those who submitted their answers are self-selected, so no one has any idea of whether they are even a representative cross-section of Linux Journal Readers. 

Some people would argue that these are good reasons for ignoring the results altogether -- and, statistically, they are probably right. However, while the results shouldn't be accepted down to the decimal point, or even to within more than a few percentage points (which is why I have rounded off the figures that the Linux Journal actually gives), they should give some very general indications of trends, especially when several polls agree with one another.
If you want the full details immediately, you'll have to subscribe, but here are a few highlights that caught my eye:

Distributions and Desktop Environments
No one will be surprised that Ubuntu (with all its versions counted together) and Debian took the top two places. However, the fact that Ubuntu registered only 30% may be one indication of Linux Journal's relatively experienced readership. A more general survey might show Ubuntu more popular, and mention fewer distributions in the results.

However, what might be surprising is how thoroughly .deb-based distributions dominated the results. The three most popular distributions, Ubuntu, Debian, and Linux Mint accounted for 58% of the votes. By contrast, Fedora, openSUSE and CentOS, the three most popular .rpm-based distros, received only 16%. We appear to be in a largely Debian-descended world, although the differences between package management formats has long ago ceased to matter to most people.

In the netbook category, Ubuntu and Debian also dominated, although Ubuntu declined to 25% while Debian remained at 15%. Third was Android, with the rest of the field being dominated by distros designed for minimal systems, such as Puppy Linux, or ones for cloud-based computing, such as Chrome OS and Jolicloud. Although Ubuntu does have its own cloud storage system, cloud services would seem to have limited appeal, even among this sophisticated group of readers.

On the desktop, the diversity that has been increasing in the last year is clearly indicated. KDE remained first at 26%, but GNOME 3 registered 22%, GNOME 2 (presumably including Mate and Cinnamon) 15%, and Xfce 12%. 
Strangely, Unity is listed as only having received a few write-in votes, which seems inconsistent with Ubuntu's popularity. Are Ubuntu users really switching to another interface as soon as they finish installing?

I also have to wonder if GNOME 3 is really so close to KDE, or GNOME 2 lagging so much behind GNOME 3. I suspect that these results are the strongest indication of some unknown bias in the poll. GNOME's technology is more popular than any single GNOME interface, but, historically, GNOME's and KDE's results have always been extremely close, no matter what the poll.

Application standings
Advocates usually claim that one advantage of the free desktop is its choice of apps. However, looking over the various categories, I have to wonder whether that ability to choose is becoming an illusion. 

Whether you are talking about desktop or cloud apps, in many categories, there is one application with half or more of the votes, followed by the second most alternative with one-half or even one-third as many. The closest thing to a monopoly is in Best Office Suite, in which LibreOffice received 73%, followed by Google Docs at 12%. 

However, other results were only relatively more balanced. For instance, under Cloud File Storage, Dropbox had 54%, and was followed by Ubuntu One at 15%. Similarly, under Best Virtualization Solution, 56% of the votes went to VirtualBox, and 18% to VMWare, while 63% went to Git in the Best Revison Control category, and 19% to Subversion. Only a few categories such as Best Web Browser showed several apps in close competition.

I have reviewed my share of distributions over the last year, so these figures only confirm my impression. Just as people used to talk about a LAMP stack (Linux, Apache, MySQL, Perl/Python/PHP) on the server, so now we could talk about a standard array of desktop tools (I'll leave someone else to coin the catchy name).

But why many apps on the free desktops are becoming standard -- and why a few are still resisting the trend -- is less obvious. One possible answer is that the areas with healthy competition are those that have existed for a long time, and in which the competitors all began over five years ago. Maybe in some basic categories, Linux Journal's experienced readers are simply sticking to long-established preferences?

That answer wouldn't explain the closeness o fthe competition in the Best Web Browser category, where Chrome is recent competitor. But it does fit the results of the Best Programming Language category, where Python won 28%, followed by 19% for both C+++ and C, or of Best Scripting Language, in which Python received 36%, Bash 24%, and Perl and and PHP 14%.

Polls and voices
With 2012 drawing to a close, Linux Journal's Readers' Choice Awards are only the first of the surveys we can expect over the next couple of months. I'll be looking at them closely, both for their own sake and to check how they compare with each other.

If nothing else, these polls are useful as a reality check, to ensure that impressions made from my daily scanning of news sites and blogs have any accuracy, and aren't just the result of being influenced by the loudest voice. Although they all need to be taken with a healthy dose of skepticism, together they are better than no indicator at all -- if sometimes just barely.