August 17, 2010

Desktop Linux Showdown


Last week I gave a presentation at LinuxCon in sunny Boston entitled Desktop Linux Distribution Showdown. The premise was to compare the three most popular desktop distributions to find out which is most user-friendly. It wasn't easy, and the results might (or might not) surprise you.


The concept of "user-friendliness" is highly subjective. In fact, given the nature of free distributions, just figuring out which ones qualify for the top three is as much a matter of opinion as forensic guesswork. After making an educated guess and then checking on, downloading numbers quoted by individual distros, and figuring some Tennessee windage, I determined that the top three distros in 2010 are, in popularity order: Ubuntu, Fedora, and openSUSE.

Once I determined the contenders, the showdown was on. I picked seven activities to test, and rated each distribution on each of the seven activities, using a non-weighted scale that measured three qualities: time to completion, ease of use, and aesthetics. Thus, each distro could score up to 30 points for each activity, or 210 points total.

Don't make the mistake that this was anything resembling a scientific survey. I am not a scientist or a professional reviewer, nor do I make regular contributions to desktop Linux. However, I am definitely an informed, non-partisan user who uses desktop Linux daily, so perhaps I am a good choice to judge this particular showdown after all.

Showdown Results

Like the South American anteater and the similar but biologically unrelated African pangolin, Ubuntu, Fedora, and openSUSE installations have evolved to closely resemble each other in many ways, from wizard-style installation to startlingly similar boot times. After booting, they differ in subtle ways, though overall the scores for all distros were very similar.

Specifically, here are my notes for each test:

This test examined the installation process. All three distros were very similar. I loved openSUSE's one-screen approach, and slightly disliked Ubuntu's marketingy feel.
Time to Live
This test measured startup time, power-on to usable desktop. Scores were very nearly identical, and fast! The average was less than 64 seconds.
Installing Applications
I installed TwitUX, a graphical microblogging client, from each distro's repository using the included graphical tools. All were simple if I knew what I was looking for. openSUSE annoyed me by insisting that it wanted to update my entire operating system at the same time.
System Configuration Tools
I installed a network printer. Driver search was laborious and somewhat confusing for all distros, and Fedora and openSUSE both asked for the root password multiple times. Fedora also forced me to use my IP address rather than the host's name.
Online Help
I tested system help (that came on the disk image), official help found on the distro's website, and community-authored help via search engine. Ubuntu's system help was great, but its official docs were quite outdated; community-authored help was easy to find and very useful. Fedora's system help was nearly nonexistent, perhaps because I installed the CD image rather than the DVD, but its official docs were fantastic, with community-authored help somewhere in the middle. For openSUSE, system help was again lousy, but the official docs were great and community-authored help was easy to find.
Setting Up New Hardware
I attempted to set up an older Epson scanner on all distros, and all three failed nearly equally, with no driver and confusing error messages. The closest I came to success was with openSUSE, which provided a hint that proprietary drivers might be "Out There" somewhere. (I eventually found the driver, but the point of the exercise was to compare effort, not necessarily to succeed.)
Setting Up Network Services
In stark comparison to the previous test, all distros performed admirably. Wired and wireless networking were immediately available. Samba support varied, however - Ubuntu had the nicest experience from the user's end, and openSUSE was best from an admin perspective. Fedora trailed the pack, as Samba is not installed by default on Fedora.


The following table summarizes the results, with details available on the presentation slides. In short, Ubuntu comes out slightly ahead and Fedora slightly behind, but the overall variance is only eight points. In my mind (and in the talk), I often referred to the exercise as a showcase rather than a showdown, and it should be said that all variants of desktop Linux performed remarkably well, far better than my experiences with proprietary counterparts.

  Ubuntu Fedora openSUSE
Installation 24 24 23
Time to Live 27 26 27
Installing Apps 26 26 20
System Config 22 20 21
Online Help 22 21 21
New Hardware 9 9 15
Network Services 26 17 28
TOTAL SCORE 156 144 155

So, then, what happened to the elusive Year of the Linux Desktop? First, while Linux performs very well on the desktop, there was nothing compellingly better for most users than either of the other major desktop distros.

Developers, yes, especially in terms of tools. Normal users, no. Second, enterprise application lock-in wins — when half of a company is on one operating system, it is impractical for an IT group to put the other half on a different operating system without a compelling reason (see first answer). Third, and perhaps most important to casual users, application interoperability is a big win, especially in terms of office applications. While apps don't define an operating system, to most users the operating system is only a pathway to apps. Computers, after all, are for performing tasks.

One final thought: perhaps desktop Linux adoption doesn't matter. Mobile computing is becoming the standard for people worldwide in conducting communication, interacting with applications, and seamlessly integrating computing into their lives. The desktop isn't going away anytime soon, but its relevance is being diluted — which is all the more reason for desktop distros to address the user experience, as they are all doing now.