08.19.11
GNU/Linux Desktop Market Share
20 years old kernel finds itself everywhere
Summary: The usage of GNU/Linux on the desktop is not as tepid as the corporate press has some people believe
WHEN this site was a lot younger and well before it required a cache server to offload pressure, monthly “market share” statistics were occasionally posted to show that about 40% of the visitors of the site use GNU/Linux. We have a wiki page about this subject, still. Many actually used SUSE, perhaps because we covered SUSE quite routinely.
After the DDOS attacks of 2009 we needed to move to a server which was not shared. Whenever people attacked the site this impacted some other sites. It was then that we lost cPanel and AWStats. Later in the year we also added server-level cache for handling the load and for protection against attacks (common around December of that year). This was not CMS-level cache, which continued to be used. Varnish basically divided, based on some criteria, requests that could be served from a local file (static) and those that needed to be passed for processing the usual way by Apache and the underlying CMS (with or without the database, depending on cache). The outcome of this was that statistics ceased to be meaningful. The logs contained requests from Varnish rather than from the users who accessed pages. Having said that, it is reasonable to speculate that about 40% of the visitors are still GNU/Linux users, putting Android aside.
What might be worth noting here is that there is a population bias that affects how people perceive the “market share” (usage) of GNU/Linux. Users of this operating system are drawn into niche sites that have a strong privacy policy and would not give away logs for the sake of someone’s business model. Moreover, there are uneven distributions of OS use, typically based on geographical and lingual factors. Those factors are rarely or never accounted for by quick ‘facts’ vendors that just use brute force to output some ‘magic’ numbers, never bothering to study the population in question.
The bottom line is, as far as a site like Techrights is concerned, GNU/Linux as a desktop operating system is massive. It keeps growing, too. █
Content is available under CC-BY-SA
Nate said,
August 19, 2011 at 10:20 pm
I administer a server for a small business based in MA, USA. I occasionally check the server-wide logs to formulate statistics, but unfortunately they haven’t really impressed me. I just did a quick check now, and only 0.34% of 48,496 form submissions to approximately 200 websites from 2011-01-01 to 2011-08-19 have been from a system identified as a Linux (excluding Android). Most of our clients are small businesses located in the northeatern United States. The submissions I am referring to are from such as general contact forms, reservation request forms, payment forms, and shopping cart order forms.
Of course, I know that statistics will be different for different audiences, but it’s disappointing to me not to be able to see better numbers from the only source I can personally trust.
Dr. Roy Schestowitz Reply:
August 20th, 2011 at 2:25 am
Yes, this validates exactly the point that I was making. The users of GNU/Linux tend to located in other geographies (Brazil for example) and visit particular types of sites. My other site, schestowitz.com, has 10%-20% of the visitors running some form of “Linux”. Those sites are never included in statistics from the firms the corporate press loves to quote (US-based firms).
See
http://royal.pingdom.com/2008/08/21/linux-popularity-across-the-globe/
http://royal.pingdom.com/2011/05/12/the-top-20-strongholds-for-desktop-linux/
The US is not even listed in the top 20.
NotZed said,
August 20, 2011 at 12:55 am
My blog is but a piddle of a puddle but I see a fair bit of Linux. Not entirely surprisingly given my past and blog posts, but I don’t have too many regular readers from previous connections and most hits are just from web search results unrelated to gnu/linux (e.g. #1 hit every month for the last year has been: “java fft”).
So most of those are for quite technical programming stuff – OpenCL, scientific programming with Java, beagleboard related, or obtuse performance stuff even I can’t remember writing much about. That probably means the demographic is largely students and other engineers who are probably more likely to be using a decent engineering platform like gnu/linux. Last month I had about 20% gnu, 65% microsoft and 5% apple.
What I find more interesting is where the people are from. USA is by far the biggest on it’s own, and then usually a bit over 1/3 as many hits from Germany, with Australia and India a bit behind, and then UK trailing somewhat. But the USA is only about 1/3 of the total too, so it isn’t even a majority.
The Germany and Australian figures surprise me a bit, particularly Australia with it’s much smaller population than every other source. Perhaps googlesearch.au adds some bias to the results based on location/the language I use (unfortunately, you never can tell …).
A few of the others are surprising too (although the numbers are so low it’s hard to be definitive), for example in the last month Malaysia has more more hits than Russia or France, and Thailand (just) more than China (maybe malaysia/thai are hitting the cooking/gardening posts!).
But back OT … personally I think the ‘year of the linux desktop’ was about 8 years go. That people chose not to use it is their own problem – one that often costs them dearly – I definitely wouldn’t even consider using anything else right now .
As far as I can tell it did everything back then that all the ‘desktop’ systems do now, which for the most part means automatically mounting a usb drive when you insert it, or configuring a pcmcia card (or whatever the current iteration is called) automatically. Oh except sound actually worked out of the box …
I think all the work since then has been purely cosmetic (or big steps backwards, e.g. trying to put a touch interface to a keyboard+mouse system, or packagekit, or pulse-audio). On the other hand the quality and breadth of software beyond the core so-called ‘desktop’ just continues to improve in bounds and leaps.
Dr. Roy Schestowitz Reply:
August 20th, 2011 at 2:32 am
Last night my connection was down (had been down for 14 hours), so I went through my file system looking at old screenshots of my older Linux boxes (I love to document these). I saw Ubuntu 4.10, Mandrake 9, SUSE 8, and a few more. It was then that I reminded myself they had to compete with XP. Nowadays XP is still “the one to beat” and GNU/Linux is considerably better. I installed Queeze last week (with KDE4) and I am still finding new features that astound me. Carla put it well last week.
Based on capability, GNU/Linux has had its “desktop” year a long time ago.
Michael Glasser Reply:
August 30th, 2011 at 8:22 pm
Shuttleworth speaks of Apple’s OS X as the one to beat. As you and I discussed, at least in the areas we covered, desktop Linux, as represented by PCLOS, has not yet hit that mark.
From the audiocast when you kindly let me on your show:
Snit: 24:45
—–
When you have “Save” on the left and “Save” on the right -
people click on the wrong button. And it’s user error -
that’s what it’s called – but it’s not; it’s an error of
the system. [About older PCLOS] looking at its default
programs … and there is a mish mash [quit/exit, save
dialog with "Save" on left or right]
—–
Roy: 25:32
—–
I’ve used it before and I don’t think it’s true what you
are saying now.
—–
And:
Snit: 32:28:
—–
[Of the newest PCLOS] I will post – I will make screenshots
or maybe even a quick video and we can look at it. Now I
haven’t used the newest one … but let me look at PC Linux
and I can pretty much guarantee you there will be a mix of
quit and exit, there will be a mix of hot keys, there will
be a mix of save dialogs, there will be a mix of print
dialogs, … some programs will lose the clipboard when you
quit some programs won’t…
—–
Roy: 33:15
—–
[talks about how development happens... ] Interesting that
you raise this point … when you are using something like
a file dialog you might say that will be inconsistent but
that is just not true … the way it works in KDE I would
be quite surprised if it is very inconsistent. I would be
very happy if you found some cases where it’s not
consistent – maybe even help developers.
—–
In both cases, though, you were shown to be wrong.
http://tmp.gallopinginsanity.com/PCLOS-OSX-comparison.pdf
Michael Glasser said,
August 30, 2011 at 8:20 pm
I responded to this post here: http://trw.gallopinginsanity.com/2011/08/27/gnulinux-desktop-market-share
Dr. Roy Schestowitz Reply:
August 31st, 2011 at 1:55 am
Please refrain from spamming this site. You have just posted 7 non-content self-promotion links in a matter of 5 minutes, Your trolling and stalking elsewhere have verged the pathological too.