Forgot your password?
typodupeerror
Linux Business Operating Systems Software

The Problem With Estimating Linux Desktop Market Share 409

Posted by timothy
from the something-about-lying-and-statistics dept.
jammag writes "It's long been one of those exceptionally hard-to-quantify numbers: exactly what percentage of the desktop PC market is held by Linux? Doubters suggest it hovers around a negligible one percent, while partisans suggest it's in excess of 10 percent. Bruce Byfield explores the various sources of estimates, dismissers' and fan boys' alike, and guesstimates it might realistically be 5-6%. Still, he admits, 'the objectivity of numbers is often just a myth.'"
This discussion has been archived. No new comments can be posted.

The Problem With Estimating Linux Desktop Market Share

Comments Filter:
  • Guesstimates? (Score:4, Insightful)

    by FredFredrickson (1177871) * on Tuesday May 05, 2009 @09:41AM (#27829781) Homepage Journal
    Estimates are already a form of guessing. The word 'guesstimate' make me want to puke blood.
    • Re:Guesstimates? (Score:5, Insightful)

      by Akido37 (1473009) on Tuesday May 05, 2009 @09:44AM (#27829827)

      Estimates are already a form of guessing. The word 'guesstimate' make me want to puke blood.

      When I was in school, I was taught that an estimate was the same as rounding (As opposed to an "educated guess").

      Now, every time I hear the word estimate, I assume that the number started from some actual data, rather than from someone's rectum.

      • Re:Guesstimates? (Score:5, Insightful)

        by Chrisq (894406) on Tuesday May 05, 2009 @10:08AM (#27830201)
        Exactly. If I see someone I can estimate their height and weight. If all I know is your name, all I can do is guess based on sex, nationality, averages, etc.
        • Re: (Score:3, Insightful)

          by asdir (1195869)
          Actually, you can also estimate based on sex, nationality and other averages. Imagine making a regression of many people, randomly sampled, with different heigths, sex, nationality and so on. The resulting coefficients would be called "estimates". Based on those, you might try to make a prediction based on the variables you know of a person apart from the name, like sex, nationality, and so on. Since this prediction is based on estimates, it could also be called an estimate.
          What I am trying to say here is
        • Re:Guesstimates? (Score:5, Insightful)

          by commodore64_love (1445365) on Tuesday May 05, 2009 @11:52AM (#27831941) Journal

          I bet the number is only 1/10th as large as Apple's MacOS share.

          Last I heard that's somewhere around 10%, so figure 1% for long term Linux users. The reason I suspect it's so low is because many, many people have TRIED Linux but few have stayed with the habit. Just like marijuana. (ducks a spitball). I had Linux on one of my laptops, but I wiped it with the original XP Restore CD. Counting me as a "Linux user" simply because I tried it last month would be a mistake, but I suspect it's a common one made by many estimators.

          • Re: (Score:3, Insightful)

            by Carewolf (581105)

            I bet the number is only 1/10th as large as Apple's MacOS share.

            I bet you're wrong.

            And by all estimates, my bet is safer.

      • by hedwards (940851)

        That has never been the definition of estimate. If you had actual data there's absolutely no reason why one would want to estimate. Rounding has it's own term, rounding or sometimes precision.

        Estimates are often times just Fermi problems with a tad bit more information coming in.

      • by Poltras (680608) on Tuesday May 05, 2009 @10:23AM (#27830437) Homepage

        Now, every time I hear the word estimate, I assume that the number started from some actual data, rather than from someone's rectum.

        Except for estimates of colonoscopy, I guess.

      • by BetterSense (1398915) on Tuesday May 05, 2009 @10:50AM (#27830879)
        I prefer the term "swag". Scientific Wild Ass Guess.
    • Re:Guesstimates? (Score:5, Insightful)

      by sopssa (1498795) <sopssa@email.com> on Tuesday May 05, 2009 @09:44AM (#27829835) Journal

      The main problem with linux desktop usage is that all the games are made for Windows (some of them also work on Macs). I for one cant change to use linux as desktop, even if I want to and use it as server, because I like to play the games aswell (no, the freeware games on linux dont count for obvious reasons).

      Problem is that game developers neither want to develop games for linux because it doesnt have enough users, and hence it goes round and round.

      So the question is, how could we get the gaming market to linux aswell?

      • Re:Guesstimates? (Score:5, Informative)

        by HermMunster (972336) on Tuesday May 05, 2009 @10:11AM (#27830245)

        There are many Linux games. The Unreal tournament series for one, the quake series, Enemy Territory, etc. There are some solid full featured free games but I would have to say that frozen bubble isn't a game for obvious reasons as it is just an incomplete toy demo of some 3d graphics.

        One has to ask why there are no games? Would you as a developer not want to target potentially 30-50 million world-wide users?

        There are a couple of reasons for this.

        1) Commercial developers don't understand the license--GPL and others.

        2) Microsoft created a series of "lock in" technologies. Sort of like what we went through with the OOXML/DOC thing. For nearly a decade the government and large entities public and private required that you submit your electronic files in .doc (or some other office format). This meant that say, when the court system wanted you to submit pleadings you had to submit them in .doc and that meant that you the attorney and everyone in your office had to use a proprietary tool.

        See the lock in? Well, Directx is the same way. Developers create based on Directx even though there's a near feature complete comparative technology in OpenGL. If developers developed for OpenGL then they'd have a basis for cross-platform gaming development. Some do, such as the guys that do the Unreal Tournament series. They know the value of it. Some day we may see that users are using Linux for their day in and day out tasks and switching to windows for gaming. You'll dual boot into windows like you would start up your console just so you can play the game, then you'll go back to Linux to do everything else.

        This puts us in a position of the chicken or the egg. Wait for a market to grow to justify mutliple APIs for gaming development from the standpoint of the gaming industry leaders or develop and hope you can build a gaming following.

        Yes, many of my friends have said that they play games and that's the number one reason. They won't commit to Linux unless they can game on it and it looks as good as it does under Directx.

        I personally loose site of the quality of the graphics and tend to focus on game play after the initial WOW when I first begin a game. It doesn't mean I loose track completely but my focus is on playing and not so much on the beauty of the surroundings.

        I have played some with wine and gaming and though it can work often times it has 2 failings. The first is that the games just don't look the same as they do under windows and aren't good performers. The second is that they can be problematic to get up and running. This isn't to say that all are this way. A popular game called Guild Wars is totally windows, but runs flawlessly under Wine.

        I've taken and connected one of my Linux computers to a 47" TV going from DVI to HDMI. The resolution is 1920x1080 and looks utterly awesome as a desktop. I installed wine and then Guild Wars. After a few settings adjustments it looks just as good under Linux as under Windows and it is an incredibly beautiful on that 47" TV.

        This is a tough battle to win. Only through gaining market share with Linux can we get gaming going. That's tough when dealing with a criminally convicted predatory monopolist such as Microsoft.

        • Re:Guesstimates? (Score:5, Informative)

          by Ginger Unicorn (952287) on Tuesday May 05, 2009 @10:40AM (#27830703)
          you dont need to understand free licences - there's nothing to stop you releasing proprietary software that runs on linux.
          • by Proteus (1926) on Tuesday May 05, 2009 @11:23AM (#27831401) Homepage Journal

            you dont[sic] need to understand free licences[sic] - there's nothing to stop you releasing proprietary software that runs on linux.

            That's true, but unfortunately beside the point. Many product managers and the like have such confusion over the terms of the GPL that they believe any software they write to run on a GPL'd platform (like Linux) must also have a free license.

            Or, at the very least, they believe that they'll be sued into releasing the source code.

            It doesn't really matter that their perception is a fiction: unless people who already have these managers' attention can make a convincing case ("convincing" in the PHB sense, not the reasonable-person sense), the perception won't change. And there won't be as much commercial software for Linux.

            This results in the wonderfully circuitous circumstance that consumers don't adopt Linux because the games/etc. they want aren't available for it; and those games don't get ported to Linux because there's no market share.

        • Re:Guesstimates? (Score:5, Insightful)

          by Toby_Tyke (797359) on Tuesday May 05, 2009 @10:51AM (#27830887) Journal
          1) Commercial developers don't understand the license--GPL and others.

          2) Microsoft created a series of "lock in" technologies.


          Whilst I'm sure both of those play a part, they are by no means the main reason. After all, if MS lock-in was such a huge obstacle to porting games across platforms, the 360 would have more system exclusives. There is a far more simple reason why there are so few commercial Linux games. Market share.

          Not market share in the conventional sense though. Let me explain.

          Generic Blockbuster Games inc are planning to release their new game, Mediocre First Person Shooter VII: The Shootening, this summer,and are considering investing in porting it to Linux. Is this worthwhile? Only if the investment will bring in more revenue, by selling more copies. Now on the face of it, sure it would, because Linux has, according to TFA, 2.5 percent of the desktop market. If GBG port MFPS VII, they can all buy it, right? Wrong.

          For a start, only hardcore gamers with expensive rigs can play the latest games, so only a sub-set of the 2.5 percent are potential customers. Now, ask yourself a question. How many hardcore gamers with expensive gaming rigs do you know who only play games with native Linux versions?

          You yourself are playing a game with no Linux version. How would NCsoft have stood to make any more money from you by providing a windows version of Guild Wars?

          30 million Linux users are irrelevant. The potential market for Linux video games is vanishingly small, if you discount the people who would buy the windows version in the absence of a linux port.
          • Re:Guesstimates? (Score:4, Insightful)

            by mR.bRiGhTsId3 (1196765) on Tuesday May 05, 2009 @11:15AM (#27831249)
            I believe there is another phenomenon at work that you didn't mention. In my experience, Linux users will shy away from the latest, most powerful hardware for the simple fact that it is less likely to work properly or be fully supported. Afterall, what point is there in gettign the absolute latest NVidia card with 512MB ram and however many bajillion stream processors they have these days when it isn't going to work particularly well. For years the message I always heard was "Go Intel, it'll mostly work." Intel can't run demanding games though.
            • Re: (Score:3, Informative)

              by PitaBred (632671)
              New message is "Go ATI, it's working pretty well". Anything X1xxx and below is fully accelerated with free drivers, the HD2xxx and above are on pace for having free acceleration within a year. If you value open-source that is. Nvidia cards till perform very well with the closed source drivers, and are the de facto standard for OpenGL on Linux.
              • Re: (Score:3, Insightful)

                You just demonstrated my point though. How old is the X1xxx series? I can't remember, but still, the latest ATI cards still aren't up to par yet. So, its worth shying away from them.
          • Re:Guesstimates? (Score:5, Insightful)

            by A.K.A_Magnet (860822) on Tuesday May 05, 2009 @11:57AM (#27832007) Homepage
            Unfortunately this will be buried in the mass of posts, but I'll go anyway.

            If a game developer releases his game for Mac and Linux as well, maybe that's 10% market share but it's not the same market share! It's really disappointing to see this fact so often overlooked. Because not all users are the same. Mac and Linux users are DYING to play games. Meaning that a lower market share could still have higher game adoption.

            For instance, take MMOs. There are no decent MMOs running Linux natively. As a former Ultima Online player now reconverted into a Free Software enthusiast, I would love a good MMO to play. But I will not use Windows.

            The market for MMOs is *saturated*. If you publish yet-another-MMO, whether it's for a niche market of players (hardcore MMOs such as DarkFall of Mortal Online which I would love to play) or another WoW theme-park-MMO clone, you still have to convince Windows players to buy your game and pay the monthly fees. The Linux desktop usage may be lower, there is NO competition for MMOs. Meaning all suckers for MMOs like me would play if the game is any good.

            In fact this applies to most games because the market is saturated. It is better summed up by the Lugaru game developers [wolfire.com].

            IMO, game developers are only missing opportunities. Once they release for Linux and Mac, desktop usage will raise and more people will favor their games over their Windows-only competitors.

            Last argument is that "Linux users only want free". This is wrong. Windows users are a lot more about cracking and pirating. If Linux users see something of value, they will no doubt buy it to encourage companies to continue. Most Free Software enthusiasts have no problem with proprietary games because it can be considered a piece of art rather than a piece of software. At least art resources (graphics, music) need not be free, even RMS says so :). And we *are* desperate for good native games.

            So (to game developers), stop depending on DirectX! Use abstraction layers between DirectX and OpenGL. UnrealEngine 3 works with both, and considering how advanced it is, I don't want to hear shit about how OpenGL is not as good. It will only get worse if game developers let it die. And release NATIVE support for Ubuntu and Fedora. The community will make sure it works on other distributions.
            • Re: (Score:3, Interesting)

              by HermMunster (972336)

              Your argument does nothing more than bring up the question of the chicken or the egg. Which comes first?

              Development for Windows, when DOS was preeminent, had these same influences, and the costs were as high (in a relative sense). If target audience size were the case and the size of the audience wasn't large enough and it was an important factor, we'd have no Windows. There were other, a lot of other, influences back then and a perception that to fail to develop for Windows now meant failure in the futu

      • Re: (Score:3, Informative)

        by Mad Merlin (837387)

        The main problem with linux desktop usage is that all the games are made for Windows...

        Not all games, Game! [wittyrpg.com] for example.

      • Re: (Score:2, Informative)

        by skiman1979 (725635)

        The main problem with linux desktop usage is that all the games are made for Windows

        What about this list of the Top 25 Linux Games for 2008 [whdb.com]? There's a nice variety of games on that list from different genres.

        I haven't played them all, but I have a few installed on my Gentoo system at home.

        I'm sure there are other decent/good Linux games out there as well. You can also bring some Windows games to Linux via WINE. There are some popular games on their Top 10 Platinum List [winehq.org] and the Top 10 Gold List (scroll do

      • Re: (Score:3, Interesting)

        by FictionPimp (712802)

        I opted to stop playing computer games.

        I used the money to interact more with the outside world. Took up hobbies that improved my health and introduced me to new things and culture.

        Then I broke down and got a 360, wii, and ps3. My gaming itch is not scratched on a 50 inch screen from my lazy boy.

        I'll still buy mac games when the mood strikes. But quitting pc gaming allowed me to get rid of windows in an instant.

        A much more tangible side effect? I'm off the upgrade mill. I don't have to spend money on new vi

    • Re: (Score:2, Funny)

      by bodger_uk (882864)

      There it is! The most intelligent 1st post ever created.

      Although the analogy of the consequences of using a, admittedly daft, word is a little over the top.

    • by Canazza (1428553)

      Estimate to me implies some data goes into a process, and comes out the other side, now, either the Process used to process the data was incomplete (for example, it was a Taylor series) or if the data itself was only a sample of a larger population (Like a TV ratings list)

      A Guesstimate, that shudderingly horrible portmanteu, would say to me that you had no data, or no process, and really, just imagined what the numbers were.

  • by FudRucker (866063) on Tuesday May 05, 2009 @09:43AM (#27829819)
    since most all Linux distros can be downloaded anonymously for free from many servers/mirrors around the world there is no way of knowing for sure...
    • Re: (Score:2, Insightful)

      by corsec67 (627446)

      Plus that one download could be used to install an unlimited number of computers, so even counting people that complete the download might not be correct.

      • Re: (Score:3, Interesting)

        by LWATCDR (28044)

        Or you could be like me and have 8 differn't ISO that you run in VMWare just to keep up with what they are doing.
        Heck I don't even know where you would count me. I run Linux and Windows on my desktop. If your a Windows Fan I guess you count me as a Windows user if your a Linux fan I am a Linux user.

      • by Rary (566291) on Tuesday May 05, 2009 @10:08AM (#27830205)

        Plus that one download could be used to install an unlimited number of computers, so even counting people that complete the download might not be correct.

        Plus there's people like me who download multiple different releases of multiple different distributions just to try them out, or to use them on servers, but still use Windows on the desktop.

    • by Dystopian Rebel (714995) * on Tuesday May 05, 2009 @09:52AM (#27829961) Journal

      It would be foolish to count downloads for this purpose. However, Canonical could surely count update requests to repositories, for example.

      • Re: (Score:2, Informative)

        by Tribaal_ch (1192815)
        I did set up a mirror for all of our company's workstations (32), so canonical would see us as one user...
      • Re: (Score:3, Insightful)

        by greenbird (859670) *

        It would be foolish to count downloads for this purpose. However, Canonical could surely count update requests to repositories, for example.

        Why? That's how Microsoft counts their Vista sales. Units sold no matter how many were down-graded to XP. Oh...wait...you said it would be foolish.

    • Re: (Score:2, Funny)

      by ausekilis (1513635)
      Just ask MediaSentry to look up the number of *nix distros flying around torrents. Then we'd have a good 30% market share.
  • by Anonymous Coward on Tuesday May 05, 2009 @09:43AM (#27829821)

    Can't wait until Wolfram Alpha goes online. This question will finally be answered once and for all.

    • by hedwards (940851)

      That's a good point, we might also be able to figure out when the year of Linux is going to be and have something more reliable than netcraft to confirm things.

    • by melikamp (631205)

      Will we finally find out the question to 42?

  • While you may claim it prevents the self-fulfilling "tipping point" of everyone switching to it because everyone else is using it, I have no complaints with Microsoft and Apple thinking that they have nothing to worry about from Linux until it's too late. What do big dogs do when small dogs start to threaten their dominance? They try to kill them. I actually prefer the "slowly but surely until it's too late" scenario.
    • I tend to think of Mac's & *nix being around the same market share, perhaps around 7-8% but both growing by the month as Microsoft piss off more and more people. It will be a while before either get to a significant chunk of what Microsoft once had but took for granted, but slowly it's getting there.
    • Forgot to add, that it's gonna vary from country to country too, just like Firefox numbers. Some countries have more resistance to Microsoft bullies than others.
    • by onion2k (203094) on Tuesday May 05, 2009 @10:04AM (#27830159) Homepage

      It's the analysts who are (probably) underestimating Linux. You can be absolutely certain that both MSFT and AAPL are very aware of their competition. They'll both have labs full of Linux installs (plus OSX and Windows respectively) where they examine what new things are added, old things removed, what's fixed and what's left broken. These are companies with billion dollar budgets. Spending maybe a million (20 staff plus a big office) to research your competition is obvious.

    • by melikamp (631205)

      I agree. I especially like it when some government agency or other switches without much ado, or comes out saying that they have it running the back end. Because one thing I do not see in the future is people converting from an established Linux solution back to Windows: the costs would appear to be astronomical.

    • Re: (Score:2, Insightful)

      by noundi (1044080)

      While you may claim it prevents the self-fulfilling "tipping point" of everyone switching to it because everyone else is using it, I have no complaints with Microsoft and Apple thinking that they have nothing to worry about from Linux until it's too late. What do big dogs do when small dogs start to threaten their dominance? They try to kill them. I actually prefer the "slowly but surely until it's too late" scenario.

      Fair analogy, although, while we're doing animal analogies, I would look at MS or Apple as the "big dogs" and Linux as a shitload of bees holding the (important but not cruical) hive together. The difference being that even if a bee is lost, or even the hive itself, it's not over, whilst the dog is one.

  • Confusion (Score:4, Interesting)

    by 99BottlesOfBeerInMyF (813746) on Tuesday May 05, 2009 @09:46AM (#27829863)

    Sadly the article seems to confuse install share and market share, not just confusing the phrases, but using them concepts interchangeably. For some uses, this does not matter, while for others it matters a great deal. That and the fact that the article ends with a cop out, "We have no way of knowing which is closest to the truth" makes this pretty useless.

    • Re: (Score:3, Insightful)

      by value_added (719364)

      Sadly the article seems to confuse install share and market share, not just confusing the phrases, but using them concepts interchangeably.

      I'd go farther. The term "market" is sufficiently ambiguous and using it invites all sorts of connotations that simply aren't applicable, or are relevant only in narrowly-defined circumstances. For the vast majority of downloads and installations, there is no money changing hands so there is no "market".

  • Most of these types of figures are always done with sales, and since Windows is the only pre-installed option on a new PC and counts as a Windows sale when you buy it (even though you were buying a PC, not Windows) it's always gonna skew the figures. Even when your PC leaves the store with XP on it, it's on the books as a Vista sale. The game is rigged. The fact that Linux is not available in many outlets as a purchase it will never gain any parity.

    You could look at downloads, but not every download is inst
  • Easy solution (Score:5, Insightful)

    by MyLongNickName (822545) on Tuesday May 05, 2009 @09:49AM (#27829911) Journal

    Go out on the street. Talk to about 1000 people. Ask them what operating system they have on their home computer.

    My prediction on the results

    Huh?: 45%
    Windows: 25%
    No Computer: 20%
    Mac: 8%
    Linux: 2%

    • Re: (Score:2, Insightful)

      by downix (84795)

      Mine would be:

      Windows: 35%
      Mac: 9%
      Dell: 3%
      IBM: 4%
      What's an Operating System?: 40%
      Linux: 2%
      I don't have a computer: the rest.

    • by Big Nothing (229456) <big.nothing@bigger.com> on Tuesday May 05, 2009 @10:02AM (#27830125)

      My prediction would be:

      Huh?: 50%
      Word: 10%
      Internet: 10%
      Windows: 10%
      No computer: 10%
      Mac: 8%
      Linux: 2%

      There probably should be an option with ISPs in there, but I can't be bothered.

      • by Jestrzcap (46989) on Tuesday May 05, 2009 @10:17AM (#27830353)

        Man did this make me regress. Back when I was doing tech support and I had to ask what ISP people used I tended to get all kinds of wonderful answers.

        Me: "I just need to ask you a few questions to better understand your problem"
        Them: "Ok"
        Me: "What internet service provider are you using?"

        Them1: "Netscape"
        Them2: "Internet Explorer"
        Them3: "Windows?"
        Them4: "I don't have one"
        Everyone else: "AOL"

        Me: "What operating system are you using"

        Them1: "Dell"
        Them2: "Netscape"
        Them3: "AOL"
        Them4: "I don't have one"

    • by Swizec (978239)
      In most cases a "Huh" can be resolved with the question "What computer do you have?" and they will answer either Mac or Windows, but most probably Windows. Remember, people are buying COMPUTER most of them don't understand that there's an OS running on it, it's just a computer, much like a radio.

      Do YOU know what OS/firmware your television/radio/refridgerator/telephone/dishwasher/washingMachine/etc are running? I didn't think so.
    • You forgot "Microsoft", "Microsoft Office" and "Internet Explorer".

      Seriously. A surprisingly large number of people are pretty unclear on the OS/application distinction.
    • by Chrisq (894406)
      Yes, I think that the usage of the Huh? operating system is seriously underestimated.
    • "I don't have an operating system on my computer. I use the Internet."

      You'd be amazed...
    • by Rary (566291)

      It all depends on the neighbourhood you choose.

      In the inner city, "No Computer" could hit 80% or more. In an artsy neighbourhood, "Mac" might get up into the 20% range. In a suburb with lots of teenagers living in the basement, "Linux" might even creep up into the double digits.

      As a general rule, whatever the majority of Slashdot visitors think about desktop Linux use, the reality will be significantly smaller. We're a particularly non-representative demographic.

    • What about people with more than one computer?

      Do you want to know about only their "main" computer, or all of them?

      If it's only their "main" computer, what about people who use two machines equally? One vote for one, no vote for the other? Half-votes?

      If I can only vote once, how come, since I bought more than one computer?

      If I get to vote more than once because I have more than one computer, how many votes do I get?

      Do I get to vote for my old Sun3 that I haven't switched on in years?

    • Re: (Score:2, Funny)

      by noundi (1044080)
      Not to forget: "Operating System? No wait, I'm not a surgeon."
  • by hattig (47930) on Tuesday May 05, 2009 @09:51AM (#27829937) Journal

    It's just tipped above 1% for consumer systems that are used for internet usage. http://techreport.com/discussions.x/16860 [techreport.com]

    Munging together servers and clients is a pointless benchmark. Linux could have 30% of the server ecosystem, but that would make a 0.001% indent on client share.

    Regardless, 1.02% is a far cry from 5 or 6 percent, never mind 10%. Who would even say that a Linux machine makes up 1 in 10 machines on the web, haven't they seen all the Windows machines, all the business machines, etc?

    • Re: (Score:3, Informative)

      by gbjbaanb (229885)

      Or you could use these stats [w3schools.com], which show 4% from browsing OSes.

      You'd think this showed more desktop usage, as most people don't use a server OS (that's used for servers) to browse the web - hence the Windows 2003 server showing at 1.7%

  • what's a desktop? (Score:5, Interesting)

    by xzvf (924443) on Tuesday May 05, 2009 @09:51AM (#27829939)
    Seriously, how to you define desktop today? Linux holds a decent share of the POS/retail market. Are point of sale devices desktops? How about thin-clients? Some have a small Linux OS that RDP's to a Windows server. Is that a Linux or Windows desktop? I just finished a project where the thin clients were diskless and hosted totally on servers. Do I count the servers or the thin clients as desktops? At home I'm 80% Linux, 10% Mac and 10% Windows, but from the outside how am I counted.
    • by Anonymous Coward on Tuesday May 05, 2009 @10:47AM (#27830817)

      Seriously, how to you define desktop today?

      On a typical day, when you sit down to surf the web, what operating system is your browser application running on top of? That's your desktop OS.

      Linux holds a decent share of the POS/retail market. Are point of sale devices desktops?

      No.

      How about thin-clients? Some have a small Linux OS that RDP's to a Windows server. Is that a Linux or Windows desktop?

      It's not a desktop.

      I just finished a project where the thin clients were diskless and hosted totally on servers. Do I count the servers or the thin clients as desktops?

      Neither.

      At home I'm 80% Linux, 10% Mac and 10% Windows, but from the outside how am I counted.

      As a nerd.

  • Just as with any evaluation system it's not that different to assess Linux use as it is any other.

    You can count the number of hits a series of websites get that come from a unique address over an extended period of time. The sampling of sites has to be pretty massive. You couldn't just use a few and the types of sites would have to vary significantly to get a good cross-section. That variations would also have to understand the Linux users will visit sites, at times, primarily targeted at Linux users.

    I'm

  • The problem with being a Linux user is that you tend to find other people who use it and remember them more because it is unusual, then your own estimates of people who use Linux will run higher because you overvalue the data points you know about, and you lump all the other users out there into 1 data point, even if there are far more of them.

    I can say that in my law school class of a little less than 200 people we are above the 1% mark... because there are 2 of us that use Linux as our regular OS. If any

    • Hopefully you'll take that Linux use with you to your job as an attorney and then encourage the use of it and open standards via your employees and the court system, thus encouraging competition and ensuring that vendor "lock in" is minimized.

      It's more important that we have competition than we have Linux but both would be just cool.

      One thought that came to mind was regarding the intelligence of the people in your class. If they were presented with Linux would any switch? Would they just say that they don

  • I think a more important question is, how many people use Linux as their primary desktop? I know I have several Linux machines and only one Windows machine at this point, however, I tend to use the Windows machine much more frequently because it has far superior hardware. My Linux machines are more or less on-going projects typically.
    • I do, I do. /Raises hand.

      I can say that I have also put it on many casual users systems over the past year that are using it as their primary desktop. Everyone that does this is a win for Linux as it spreads things out a lot. People took their computer use home. They then demanded better computers at work. This meant that the average person drove the direction we are in. Getting the average person aware of Linux now that software has matured on both sides of the fence will be the catalyst to growth in m

  • If you took reports from major websites (Google, ESPN, Yahoo, MSN, etc, etc), I think that would be the best metric for filling in any gaps.

    That would give you a percentage of an OS actually used.

    Oh, numbers are objective. But raw facts do not come with their own correct interpretation.

    • by hitmark (640295)

      to bad google trends do not have a way to show browser usage...

    • by grumbel (592662)

      While it would be true that it would give you a percentage of actual usage, it still wouldn't be a valid estimate of actual Linux market share, as a lot of desktop users have Linux, but don't use it as their primary OS, instead they dual boot into it. I wouldn't be surprised if the actual usage count is at 1%, while the install base is more around the 10%.

    • by bcrowell (177657)

      If you took reports from major websites (Google, ESPN, Yahoo, MSN, etc, etc), I think that would be the best metric for filling in any gaps. That would give you a percentage of an OS actually used.

      Counting web hits is a very common technique. Here [w3counter.com] is a recent survey showing 2% market share for Linux, and here [hitslink.com] is one showing 1%. That shows that the technique is very crude -- uncertain by at least a factor of two. There are all kinds of reasons for that uncertainty. Many user agent [wikipedia.org] strings are bogus, often

  • say, for or against gun control

    and both sides trot out numbers, facts, that support their assertions

    when the truth of course is that various quantities out of context can be twisted or misunderstood as to meaning

    simply put, when dealing in the hard sciences, numbers rule. but when you get into politics, religion, sociology: numbers mean shit

    but try telling this to a committed partisan when you debate them on various issues. they take your avoidance of numbers and their dubious meaning as some sort of implicit admission of defeat

    when in reality, the issues are one of logic, reason, and principles, not bullshit numbers and their essential uselessness in supporting what you think they support

  • One of the primary problems of estimating the number of users who use Linux or the BSDs is that, if they use them there, they likely use Windows at work. So at work they tally in the Windows column, even though it is only because they have to.

    I think in order to get a better picture, they need to estimate the number of Windows users who are using it at work and then cut them out. The comparison of what is left would give a better idea, I would think, about what people use when they have a choice in the ma

  • One problem with estimating Linux desktop market share is there's no one definition of market share. Is that worldwide share, English-speaking world share or USA share? Is it a share of operating system licence revenues, support revenues, the cost of hardware on which Linux is installed or is revenue irrelevant? If it's usage-based, do you count physical machines or virtual machines? Does it matter how much a machine is used; if so, do powered-up unattended desktops count? Or is web usage the best metri

    • Re: (Score:3, Interesting)

      Agreed. Even the words "market share" are almost meaningless for Linux. "Market share" is the share of the market...how exactly do you count sales for something that's given away for free?

      If I buy a PC with an OEM Windows license, then download and install Linux on that box, what does that mean? I've given money to Microsoft in exchange for a product, and no money to any of its competitors. Obviously, a market share point in MS's favor.

      The Net Applciations numbers track "usage share" (the percentage of peop

  • by danhuby (759002) on Tuesday May 05, 2009 @10:36AM (#27830647) Homepage

    I run a couple of sites that probably cover both extremes in terms of Linux desktop market share. The stats are as follows:

    Site 1: A local community site based in the UK; so the profile here is 'UK home user' (I find similar figures for other UK home focused sites I manage).

    Windows 92%
    Mac 6%
    Linux 1.5%

    Site 2: A site for an open source business application; the profile is therefore 'global IT worker / developer'. The picture is very different.

    Windows 60%
    Mac 30%
    Linux 9%

    The actual figure is between 1.5% and 9% then, depending on the ratio between home/office workers. As I imagine there are more home desktops than work desktops, my leaning would be towards the lower end of the scale.

    3% to 5% seems like a reasonable estimate.

    Dan

  • What counts? (Score:2, Interesting)

    by Cyner (267154)

    Part of the problem is also establishing what counts? I personally have 4 "Desktops" around the house with a Unix-like OS. Do those all count toward the total? Or should they count for two since only two people use them?
    And what about the boxes I have that I no longer use? Most of them are also non-Windows PCs.

    I can see where 1% of users might be Linux, and a much higher number (though 10% seems darn high) of boxes are Linux.

  • by wjousts (1529427) on Tuesday May 05, 2009 @10:42AM (#27830731)

    The number is somewhere between 0 and 100%

    This being the internet, I look forward to somebody disagreeing with me.

  • Ask Google (Score:3, Insightful)

    by berpi (1187131) on Tuesday May 05, 2009 @10:47AM (#27830805)
    No, really. They know.
  • by Pecisk (688001) on Tuesday May 05, 2009 @10:54AM (#27830925)

    First of all, there is clear notion that statistics can be "lying", or even better, people are drawing wrong conlutions from them. That's fine, because decrypting stats is daunting task and can require full-time team of specs to do that.

    I personally don't care about TOTAL number, because it is not all about market share. As lot of people have already pointed out, most people DON'T care about what OS they use, they care about APPS. So question is more like - do Ubuntu has nice DVD player with Tango niceness and integration with rest of desktop? No? Vola! Afaik, Gstreamer guys works on one so it could be available commercially for OEMs and people who cares about legitimacy of DVD playback on computer. Do Linux has Visio replacement? Of course it doesn't. It is so hard to do? No! (let's be honest, it's not a web browser). So why then anyone ignores it?

    Because everyone waits for some kind of grand sign to come out! :) Guess what - unless Linux Foundation don't create some kinda of OEM sales counter, Linux sales will and will stay a mystery.

    Anyway, numbers does matter to check progress. But it is only one of things. We, Linux devs and active users, have still lot to do. But let's not forget that that's OS for us. We do this for us. And rest of bunch are just invited to join :)

  • I did a quantifiable survey. On my desk, I have two machines running Linux, one machine running Vista and one running XP.

    2/4 machines are running Linux.

    Therefore, Linux adoption is 50%.

    (The margin of error for this survey is +/- 50%)
  • by woboyle (1044168) on Tuesday May 05, 2009 @11:18AM (#27831287)
    I somehow suspect these numbers (1% Linux market penetration, and such) are for systems that are shipped with the OS pre-installed by the manufacturer. That would seem about right to me. However, many systems cannot be ordered without MS Windows of some sort pre-installed, yet people remove that and install Linux, or dual-boot their systems with Linux. Even my grandson, who got a Windows system last year (my old Dell D600) switched from Windows to Linux after his Windows system disc blew up, and he is LOVING it! So, my best guestimate about actual market penetration of Linux is probably about 5-6%. It seems about right to me. Right now, I only have 2 programs that I must use which are Windows-only, so I mostly run Windows in a VM on a 64-bit Linux host. I have just installed Ubuntu on my laptop and will only run Windows in a VM there as well, as soon as I finish setting it up. Even my bluetooth wireless headset and Skype work fine on the Jaunty Jackalope (Ubuntu 9.04)!
  • by rkhalloran (136467) on Tuesday May 05, 2009 @11:36AM (#27831603) Homepage

    There's two data points the surveys are likely to miss, though one is VERY small and unlikely to skew the results.

    I rolled my own desktop system by purchasing the various components (mobo/CPU/RAM/...) and assembling; said box has been through three versions of Linux and never seen an MS install disk. Is this somehow being tallied in? Doubt it.

    The "scrub the pre-installed Windows and reload" scenario is probably more prevalent, but still unlikely to be in the counts. I'm looking at a netbook, and probably one with an internal HD vs. flash storage. Most of those come preloaded with XP. If I get one, the first action is plugging in an external optical drive and reloading with some netbook-friendly distro. Do they count the preloaded XP I was sold, or the Linux I'm actually running with?

    SCOX(Q) DELENDA EST!!

  • by SocietyoftheFist (316444) * on Tuesday May 05, 2009 @11:45AM (#27831807)

    Reality distortion fields are very prevalent among believers. I use Windows, Mac OS X and Linux. In the past I have used FreeBSD, OpenBSD (they're dead you know!), OS/2 and others. In my everyday life, working with co-workers, interacting with friends, paying attention to machines in use at the bookstore or coffee house, I've never seen a Linux machine in use outside of work or my home. I do have one co-worker that says Linux is his primary OS at home with a Windows machine only for gaming. One thing I have noticed is a surge in Mac usage. Last weekend I actually had a period of several hours where I only saw Macs in use on a street mall. At every coffee shop or sandwich shop you'd find at least one person with a laptop and I only saw Apples, I was actually incredibly surprised. I think the fact that more software houses are writing for the Mac shows where people are migrating too.

  • by advocate_one (662832) on Tuesday May 05, 2009 @12:15PM (#27832335)

    there has to be some significant penetration now if supermarkets are devoting shelf space on the magazine racks to Linux magazines...

    competition for shelfspace in those racks is cutthroat... if they don't sell, then they get dropped for titles that do sell.

  • 10% WTF? (Score:5, Interesting)

    by Brandybuck (704397) on Tuesday May 05, 2009 @12:53PM (#27833039) Homepage Journal

    Partisans suggest 10%? WTF? That sounds like someone needs to get out of their parent's basement and start living in reality. Perhaps they know nine other people in the world, and so assume that 10% of everyone uses Linux. But it' simply not true. 10% of the people my company use Linux. But we're a Unix development shop! In my circle of friends, 2% use Linux, and we're all geeks and nerds.

    You simply cannot extrapolate your narrow slice of the world onto the whole.

    But on to the good news: It doesn't matter what the market share for Linux is. All that matters is that you choose to use it. I don't use Linux, I use FreeBSD. It doesn't matter to me that fewer people use it than use Windows, or Mac, or Linux. It's my choice and that's all that matters. I don't have a need to use the same software everyone else is. I don't need to drive a car the same color as my neighbor. I am free to be an individual. So choose your own operating system, your own distro, your own pick of packages. Build it all from source if you want. Use something polished like Ubuntu, or hardcore like Slackware, bleeding edge like Arch. Or think outside the box ad try FreeBSD or OpenSolaris.

    The key is to put yourself in charge, not the market share.

Those who do things in a noble spirit of self-sacrifice are to be avoided at all costs. -- N. Alexander.

Working...