Playing the numbers game 2008: number of Linux installations world wide

Tux
The number of Linux users and installations is impossible to determine. But there are several different statistical information available which can be used to at least get a rough idea of the number of Linux installations world wide.

Merging different statistical data into one number is a tricky exercise and the result is questionable at best. Keep that in mind when you read the following information. The idea is not to get exact numbers but to get a rough idea of the dimension, nothing else.

Source #1: Fedora

One of the best statistical sources regarding Linux usage are the Fedora statistics. There the number of downloaded images as well as the number of unique IPs getting software updates is counted.
The data are difficult to interpret: no one knows if a downloaded image was only used to test the new system or to bun it onto a CD and distribute it to thousands of magazine readers or thousands of company computers. The second number is problematic because one new IP can mean a big NAT network or just a dial-in user who re-connected. So flaws everywhere, but it is a interesting coincidence that the IP numbers and the downloads are rather close.

There are also the smolt data. It tracks the users who opted in to a tracking system. Currently the smolt web server seems to be lacking behind. But there are current data available for older Fedora releases: these informations say that every month Fedora still gets more than 10k new Fedora 7 users – although there are already Fedora 8 and Fedora 9 available.

So the question is how to read all the data. One way was recently suggested by Paul Frields, Fedora’s project leader: he sums up the data to be around 11.5 million. Together with currently 2.5 million Red Hat subscriptions this would result in 13 million users. Focussing on Fedora alone and leaving the Fedora 6 users Fedora would still have a user base of 9.5 million users.

Source #2: relative statistics

There are hardly any other trustfully data from other Linux distributions available. Therefore, the Fedora number does say a lot about Fedora, but not that much about Linux in general. However, there are other statistics which measure the relative acceptance of Linux distributions.

One such source is the 2007 Linux Desktop Survey done by DesktopLinux.com. There the relative importance of Fedora/Red Hat is measured with 9%. Unfortunately there is no more recent survey available. I wonder why no one has picked up such a survey in 2008. Maybe I should start one on my own? :)

The result

Given that Fedora/Red Hat has roughly 10% and also roughly 10 million users together (which in fact seems like a at least slightly realistic data base, given the facts), the total number of Linux users world wide would sum up to 100 million Linux users. Nice.

That would leave Mac OS far behind, which is however not that surprising: Mac OS is hardly used in Offices or the government outside the US, and it is far easier to give Linux a try and keep it as a dual boot option besides a Windows installation. Also, the EU governments are pushing Linux quite a lot, and many companies and governments indeed switch to Linux right now or already switched over in the client space for some of the day-to-day workstations.

Still, last year I played the “numbers game” already (unfortunately with the same relativity source, btw.) and the result said something about 20 to 30 million users. I doubt that the number of Linux users spiked that much in the last year, but think that we can safely say the number of Linux users world wide is somewhere in the middle two digit million area, somewhere around 50 million installations worldwide.

Keep in mind that this counts mainly workstations – not traffic lights, shop information terminals or any other specialized hardware. Including all these devices would result in much, much larger numbers.

The problem: discrepancy

Most numbers available guessing the number of Linux users world wide say that there are not that many Linux installations out there, not at all. Most often it is said that, in percentage, Apple has a low one-digit number, while Linux has a dot before its first non-zero number.
The statistical backup for such numbers is most often created by browser strings aggregated from Web pages. This procedure has the flaw that these strings are often faked to make it easier to access specific pages. Also, the monitored pages are only a subset of the entire web, and surprisingly often focus on the US only which is not representative for the world at any rate.

Still, I do often wonder why such numbers and my estimates are different in the order of magnitudes. I welcome any comment on that issue!

About these ads

35 thoughts on “Playing the numbers game 2008: number of Linux installations world wide

  1. There’s also NetApplications survey:

    http://marketshare.hitslink.com/os-market-share.aspx?qprid=9

    It gives Linux about 0.83%, while MacOS has 8.87%. Now, we might dispute the actual numbers, but what we can more or less agree on, is the trend. And the trend for Linux (and MacOS) is up. Linux had 0.64% in January, and now it has 0.83%.

    Reliability of NetApplications is boosted by the fact that at least as far as MacOS is concerned, their numbers seem to mirror figures presented in other sources as well. It’s easier to calculate MacOS’es market-share than it is to calculate Linux’es share, and the usual figure given to MacOS is about 7-10%, with the usual estimate being 8-9%.

  2. Janne, I know these numbers – and that’s what I’m partially referring to with my question why my calculations are so far away form these numbers.

  3. One possible explanation would be that the Fedora systems are not so much used for regular office or fun tasks – i.e. browing the net (major factor for NetApplications statistics) – but that those systems are being utilized for web servers / scientific applications / [whatever].

    Not sure if the numbers of such systems are large enough to make a noticable difference, but it might at least make for a minor factor.

  4. There was a discussion on the kde-promo list recently where Aaron summed up *KDE* installations. His result was that “we hit the 100 million mark pretty easily”. Judging from that number, one even has to double your 100 million Linux installations.

  5. Another way to estimate desktop linux installations worldwide is using P2P networks, like edonkey and bittorrent, and make conclusions based on the popularity of mldonkey (linux), amule (linux& OSX) and emule(Windows). Of course, one should find a way to share files that are representative of the file “marketshare” in P2P first. The numbers I have are in agreement with the lower estimates in general.

  6. The question as stated above is if we’re talking about desktops or computers in general.

    On the desktop (and laptop) market Linux is probably around that 1%. In computers in general (web servers, etc…) it might be at 10%. It’s a huge difference.

  7. You can also guesstimate the number of Linux users user the familiarity test:
    How many people do you know that use Linux?
    If you’ll ask a random person, how many Linux users does he know?
    If Linux market share is >10%, every person should know of at least several Linux users.

    In my 3rd world country, the number is 0 for Linux, 0 for Mac, so I can guesstimate the percentage of Linux users to be close to nothing.

    What’s your results?

    1. Agreed in my 3rd world country only know one or two Linux users, event at my University. More people with Macs especially in the graphics arts.

  8. Kremer: Not necessarily, as countries do not have the same ratio of Linux users. It’s reasonable that you may see fewer techy people in a 3rd world country. In my very geeky university area, I know more Linux users than Windows. Not a good guesstimate.

    New F7 users: could that be ISPs changing IP addresses on you? A lot of them do that.

  9. @Eckhart: Aaron was referring to users using KDE, not necessarily installations (as one centralized installation could be shared in a whole network). Commercial software like MacOSX and Windows are far easier to track since every (valid) user needs to be covered by a (paid) license.

    And this is what makes me think NetApplications is only covering commercial Linux distributors. I’m not even sure if they include those nettop Linux installations with such a small percentage.

  10. @anon: Your argument just tries to prove I’m right: When people share a central KDE installation, they still need a client computer to connect to that installation. Clients connecting to KDE installations likely run Linux themselves, so the ration Linux installations to KDE installations only gets higher.

  11. All those statistic are very USA-centered. For example in my country numbers for Mac OS X and Linux should be reversed (Apple set very high prices for my country and earnings here are smaller than in USA) so for example I don’t know anbody using Mac. And I’m programmer, and studying at 5th year of Computer Science – so I gues for China/India/Russia and other developing countries these numbers would be different also.

    English – language web sites only measure english-reading users. And those aren’t even the majority of internet.

    So I’m more optimistic thant you :)

  12. Jakob, Luis, yes, I am focusing on Desktop computers here – because everything else would be almost insane: traffic lights, etc.

    Daniel, especially the missing background data of the study make it hard for me to deal with it or even discuss it properly – usually every study needs at least the context information about the sample data as well as the sample number. :/

    Eckhart, I’ve read Aaron’s post too, but he was not really relying on hard data, so I kept that apart.

    Kremer, out of my experiences the familiarity test fails most likely. I’ve been a help desk for years and most people do not know what they use, let aside what others use. For some of them Firefox == Internet Explorer, Winamp == iTunes, and so on. So the majority would not be able to tell the difference.
    Additionally are not aware that Windows is not part of every computer. For them Linux runs “somehow” on top of Windows. As a result they cannot (!) answer the given question if someone uses Windows or Linux, because to them it sounds like “is someone using a Bios or Linux”.
    Anyway, my results are definitely not 0/0/100 – but then again, I *am* a Linux guy and some of my friends also use Linux. And there is the Government, which is using Linux in very large installations.
    But it would be worth a try making a real survey, calling > 1000 people in a specific country to measure what OS they are using.

    roger, well, regarding Mac OSX I am actually pretty confident that there are much more Linux installations out there than Mac OS X installations, even on the desktop.

  13. The number Linux users and installs is not “impossible to estimate.” It’s maybe impossible to determine exactly, but anyone can estimate the number using various sources. The author of this blog post goes ahead and does just that.

  14. Actually, I’d say that 0/0/100 (guess who!) is probably pretty reasonable. I don’t know any Linux users outside my “sphere of influence”.

  15. @liquidat,

    “well, regarding Mac OSX I am actually pretty confident that there are much more Linux installations out there than Mac OS X installations, even on the desktop.”

    So am I, but I also think that a big part of that is because Linux runs on more diverse hardware than Mac OSX. You can basically run it on anything (more or less well though), while Mac OSX can only legally be run on Apple hardware. In a way, MacOS X is kinda like an embedded os…rather exaggerated but still. I can only speak of what I’ve seen, but I work at a university. On the server-side, the only things used are CentOS and Windows 2003 (and only for AD). On the desktop (and thin clients), there are two main desktops: WindowsXP and Gnome (unfortunately for me…). There are a couple of MacOS X workstations but relatively few.

    The students can choose whatever they want to on their own laptops of course, and it’s a rather mixed bag of what you see. I haven’t kept a close count or anything, but I have the impression that Linux is slightly ahead of MacOS X as far as the numbers go.

    And as far as “familiarity test” goes, I think that to some degree depends on the visibility of the OS. Apple is a known brand even for people not in the comp.sci camp, which is generally not the case for Linux/KDE/Gnome/whatever. Their apple stores, as well as the iPod and the iPhone to just talk about relatively recent products/concepts have made sure of that.

    Not to mention the aspect of third-party-apps…I know people who has bought Macs just for Adobe’s creative suite, even though it’s available for Windows as well. I have yet to find a Linux desktop user (home user or otherwise) who decided to use Linux for a specific program. The Linux users I know (myself included) uses it for the whole package, which is harder to market than something that is perceived to be tailored for a specific market-segment (say graphics professionals).

    Now, though: why are your numbers so different to the other statistics? Not sure, but I am highly suspicious of the extrapolation used. According to the survey, roughly 10 % uses Fedora/Red Hat. Of the people answering the survey that is. Not to downplay Fedora or anything (since the same logic applies to the other distros in the survey), but that’s only counting those who know exactly what they’re using and wants to (for whatever reason) emphasize that.

    People who just uses a Linux desktop in a government or corporate setting does not necessarily know or care what version or brand of Linux they’re using, if they’re even aware that they’re using Linux (or *BSD for that matter) in the first place.

    I’ve had people only accustomed to Windows asking me what I’ve installed to customize Windows to look like my desktop (a pretty damn standard KDE 4.2, if you can say “standard” for something that hasn’t been officially released yet). What I’m trying to say is that the only people who would vote in such a survey are people who are interested in computer technology for its own sake, not average users that uses computers as a means to an end. And as such, the percentages need to be taken with a HUGE grain of salt. Well, okay. An Imperial Stardestroyer sized grain of salt.

    I just don’t think the results from that survey are necessarily representative of the Linux-using population as a whole and as such, the mathematics don’t add up.

    Oh, and statistics gained from looking at what your browser says it is is flawed from another point of view (apart from faking the user-string) as well. Sometimes (depending on how accurate the statistics program is) a browser may be deemed to be Apple-based just because of the string “webkit”. And in my experience, people often browse the Internet from work or school where they don’t have a choice in either OS or browser. When at home, even if they run Linux on their own machine, their web-hits would be far fewer and their use of the Internet would in most cases be, as far as page-hits goes, considered a Windows-equipped machine.

  16. Just wondering, would not be simpler to look at updates downloaded by IP addresses for each of the distros from respective repositories. That could give us some numbers on the numer of desktop users.

  17. Why doesn’t Fedora (and others) just count the downloads of one update package (e.g. the kernel-xxx). Every system is downloading this only once when updating? And natted IPs do not matter.

  18. ArtInvent, you’re of course right, I fixed that, thx.

    Michael, I run into them frequently, but then again, I look for them. But besides that, 0/0/100 misses the installed base in the government and the companies.

    Jonas, of course the star destroyer argument is right. I yet have to calculate the inaccurecy, but that would be very large. Therefore I take the numbers very carefully.
    Besides that, there is another huge difference: I give absolute numbers, and the most other surveys give relative numbers…

    Sriram, that is actually exactly what Fedora does.

    Arnd, given that Fedora is distributed by a network of mirrors this is impossible.

  19. The number one thing we can do is to encourage as many distributors to look at adopting the MirrorManager service or developing a similar service so we can start getting unique ip client hits for as much linux distributions as possible. Those sorts or numbers will be directly comparable and can be used as a baseline estimate for overall linux deployment numbers…with caveats of course with regard to NAT’d undercounts and dynamic ip overcounts.

    My understanding is that Mozilla relies on exactly the same sort of ip information to track the number of firefox installs. Remember that firefox goes out and touches a central mozilla server looking for plugin updates when it starts. That ip information is immensely valuable for trending and linux distributions need to get all on the same page and start trending that information as well by default.

    You can not rely on the web surveys like the Desktoplinux survey to give you anything at all that can be extrapolated. The methodology as to how the survey is created.. like any popularity polling… does not result in a representative random sampling.

    -jef

  20. I don’t believe in the argument of users not counted because they change the browser user agent string, these days Firefox work almost everywhere and only real geeks used to play with UA.

    What I think it may be an explanation: people who install Linux but do not use it actively, dual boot with Windows, spend a lot of the time (browsing the net and such) with Windows but get once in a while in Linux and perform some updates.

  21. For those Fedora users stats did you include OLPC deployments? Given that OLPC is based on Fedora and the numbers deployed are available I think you can add thousands or even possibly millions to that number that wouldn’t hit the Fedora download servers.

  22. Thought this might interest you. Debian has had a project called “popcon” going for a while now, its an opt-in thing they have where users who want to participate install the popularity-contest package, and each week a little cron script emails the devs what packages are installed, so they can see which packages are getting most used and which are getting ignored. Because its opt-in, the statistics are hardly definitive, but still quite interesting:

    http://popcon.debian.org

    I believe Ubuntu have a similar project going, but don’t hold me to that, as I don’t use Ubuntu myself.

    Cheers,

    Chris

  23. With the new “one laptop per child” initiative this should greatly increase the use of linux worldwide. Each laptop comes with linux, and with that many users, its going to be logical for software developers to produce adware and freeware software (as these laptops are going to children who probably cant afford 50 dollars for an application) for linux, this in turn will increase the usefullness of the OS and therefore increase the users furthermore.

    I used to run Ubuntu on my windows machine, but now i got a macbook pro and the mac os X does most of the things that I cared about that ubuntu offered (stability, better user interface). Sadly I see no purpose of installing it on my mac os x computer, but i will keep it on my old computer for experimental purposes and to keep up with the stuff. I’d like to teach myself how to program as linux is a good OS for creating your own applications and modding the system yourself.
    So those numbers of linux users may not be that large as some probably have left the platform since it is a lot more difficult to use for the novice computer user.

    I had ubuntu on my sisters P3 computer since windows 2k is inferior to anything ubuntu has. However she now has a brand new vista laptop and doesn’t see a need for it anymore, she was able to use linux but when it came to installing a printer or anything that involved drivers she needed my help. I think the biggest hold back for linux is still user friendliessness and especially when it comes to drivers. Last time i checkced they still havent resolved the broadcom driver issues, and they lack a lot of printer drivers. I was able to get my broadcom driver working but I had to do a lot of searching for the driver, since its propreitary and they wont put it on the driver list of the installer.

  24. Here’s good news from the US. My wife’s mommy blog group is about 10% GNU/Linux. While this is only 50 to 100 users, it seems to match what you estimate.

    There’s a good explanation for the jump you see too. Many system administrators, such as the person who runs the BRLUG, use GNU/Linux on their serves but have had a Windows desktop for compatibility with devices like Palm Pilots. This latent desktop demand may finally be shifting thanks to Vista’s failure.

    One thing is sure, the future belongs to *nix. Anyone that really loves computers and wants to get real work done, knows and uses GNU/Linux. All the interesting development is going on in the *nix world. Eventually, the world will follow them. Some non free Unix may survive but Windows will vanish.

  25. Chris, thx for the hint, but I already knew it. Actually, one of the linked post describes the tool.

    To the others mentioning the OLPC: there are hardly any numbers about the amount of sold OLPCs . and there is an increasing number of cheap laptops running WinXP…

  26. The whole counting can be only a theorethical, because we have different situations what we should calculate.

    1) New computers with X installed on them. Every year there is Y amount of computers sold. Calculation is easy by these.

    2) Every year there is X amount of new OS products sold. Calculation is easy by these numbers, how many byed OEM version and how many a retail version. But no computer.

    3) Old computers what has X installed on them, every year there is Y amount of users who upgrade the X tonew version what they has buyed from store (part 2) is needed). We have problems because we can not know are they selling old one, making dualboot or what

    4) Old computers what has X installed on them, every year there is Y amount of users who replace X with Z, what they have downloaded or buyed. Problem is that you can not get counts how many machines the one CD/DVD is used as installation media.

    5) New computers what are buyed as parts from multiple different computer stores, user builds custom computer and use old installation media from X, buyed new upgrade version or Z from web or from friend etc.

    6) Old computers what gets upgade or replacement from disk what is got from friend, web or other older broken machine.

    7) We have lots of old computers and new computers and it is hard to calculate what can be running on these all computers combined.

    So we can get information only from some computers what is just a quess of the installed amount.

    My personal quess by amount of Linux users what I see, I would say that Linux has 10-20% market share. Including all the old computers running old version of one the distributions of LInux OS or new computers running distribution of Linux OS or multiple of those.

    I dont care what is the official marketshare of Linux OS when counted all it’s distributions together. I just care that I can get devices what I want, running in the Linux OS just by bluggin it in, knowing that driver exist in OS so I can upgrade the OS or move to other distribution if needed. As long I dont need to code my own driver to get devices work, I am happy :)

  27. Fri13, regarding the drivers: with the newest kernel at least Wlan and Webcams should be no issue any more. The eonly things really bugging me still are graphics drivers and printers.
    But you are right – as long as no one forces me (by market share or by force) to use anything specific I’m happy :)

  28. There’s also the linux counter: http://counter.li.org/
    It seems to be down at the moment but it’s been running for quite some time now. It’s also an opt-in so perhaps the numbers are not really representative. They have an API and scripts to post information semi-automatically so I suppose it could be integrated into a distribution to register a machine on first start if allowed by user.

  29. A much more interesting number to have, would be new computers bought neither with Windows nor with MacOS.

    This could be corelated with the overall sales statistics for Windows and MacOS.

    The difference between new computers without OS beeing bought and the sales numbers could give you a relative figure of the growth factor of the market share for Linux/Unix, since there are not many alternatives to Linux except Win/Mac.

    One would also have to estimate the number of new computers beeing assembled by the buyer itself. This could be done by looking at the numbers of sold CPUs, RAM, mainbords… This should not be too hard, because the harware manufacturer are mostly happy to provide the statistics.

    This would also factor in servers that run on desktop hardware. Additionally there are servers running Windows. Servers should be more easily identifiable than desktop computers…

    tragor

  30. ~1.3 billion windows (all versions) stated by microsoft
    ~2.5 billion windows incl. all russian, chinese, piratez, else
    50 million Linux: thats 2 percent of 2 billion. No discrepancy “in the order of magnitudes”.

    Fu**, can’t find source for windowsfigures anymore. Sorry for that…

    Ciao, Salvatore

  31. Anonymous – would be interesting to have the source for such numbers. But if they are right they are really out of sync with the numbers from Apple…

Comments are closed.