StatCounter Finally Admits 3% GNU/Linux in USA

While IDC had counted GNU/Linux as 3% of desktop PCs back around 2003, StatCounter today showed that level on USA daily webstats:

It took a while and the rapid change is questionable but StatCounter reached 3% today for USA GNU/Linux

I still have no idea what caused these two “jumps”. It doesn’t look like any migration I have ever seen but StatCounter swears it’s not a glitch…

It’s huge, though. Along with this substantial rise in GNU/Linux, they show a huge rise for MacOS and a drop from 82% M$ to 72% M$. If that’s real, it shows consumers want anything but M$’s “8”. If XP’s heartbeat means anything it’s that business just will not allow it to die… I would have thought folks would migrate from XP to GNU/Linux but it seems they are migrating from “7” to GNU/Linux. Whatever works… 😉

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology and tagged , , , , . Bookmark the permalink.

28 Responses to StatCounter Finally Admits 3% GNU/Linux in USA

  1. notzed says:

    “And let me help you with basic numeracy – 2.95% is still not 3%. So my point stands.”

    Actually 2.95% is 3%. Since the 3% doesn’t specify any decimal places. Even 3.0% would be the same, but 3.00% is not.

  2. oiaohm says:

    Robert Pogson forgetting to pay bill with statscounter you don’t get the results it still send data home.

    Webdeveloper screwing up and removing the statscounter reference from somewhere important would be possible.

  3. oiaohm wrote, of referring to StatCounter, “Its a common screw up.”

    I was thinking a bit about this today. Considering how many sites are monitored by StatCounter, there is a real issue of the time it takes to collect the data. It would be a lot of overhead to do it per transaction so they likely collect the data some way on the server and then send it in to HQ periodically or by polling. This phenomenon would then be consistent with uneven distribution of share amongst the sites or regions. It could also be then that StatCounter’s customer base changed suddenly as for example a bunch of accounts expiring/beginning on the same day.

    I think the sharp changes indicate that there are some heavy targets for FLOSS now included in StatCounter’s sites. That may offset the bias that firewalls may provide.

    Hey! How about this: M$ forgot to pay its bill…

  4. oiaohm says:

    Ted Please note most MS trolls as well quote Statscounter numbers using the dynamic numbers as well.

    Its a common screw up.

  5. oiaohm says:

    Ted Galaxy S4 default browser in those does not report a User-agent string that can be mixed up with general Linux.

    Sorry this is my problem most of the volume of Android could not be.

  6. Ted says:

    “The “daily” page is dynamic.”

    It may be better for you not to use StatCounter data in articles until the reported day is over, then. Linking to data that may not support you by the day’s end could at best be slightly embarrassing.

    BTW – here’s another factoid that supports my hypothesis – Samsung has sold over 10 million units of the Galaxy S4. That’s not exactly small change if they’re being miscounted.

    http://www.engadget.com/2013/05/22/samsung-galaxy-s-4-10-million/

  7. oiaohm wrote, “Its not just the Linux number that has moved without clean explanation.”

    Yes. That complicates the interpretation. I just don’t have any good idea what happened. In about a month GNU/Linux went from 1% to ~3%. There’s no way that’s likely to happen. There aren’t that many GNU/Linux geeks around and there’s no way to get them to install a bunch of systems on a given date. There’s no single organization large enough for a roll-out that large either. It’s a mystery. I like mysteries.

    The key feature of StatCounter’s methodology is to use page-views. I suppose an organization could have set up scripts to hammer StatCounter’s sites but you’d think they would have blocked them. It wouldn’t have taken a month or more to figure that out. Even so, the mix of changes would be hard to achieve. It would require a bunch of MacOS and GNU/Linux bots or faking User-agent strings mixture. Why would a GNU/Linux fan do that? Why would a MacOS fan do that? Are there enough M$-haters who are not fans of other operating systems? It’s a puzzle.

    Again, the list of StatCounter sites is hidden. I suppose someone could probe the web looking for responses from StatCounter but that would be a lot of work. Would that be worthwhile for any normal human? I don’t think so and I am an open hater. M$ can fall under its own weight. I don’t need to give it a push.

  8. Ted wrote, “If you deride their numbers when used against you, you cannot then use that same source to prove your own points. Or at least you can’t with any semblance of integrity or honesty.”

    StatCounter is reasonably transparent with their methodology. Assuming they are not lying, changes in their data mean something. That’s all I’m saying. The change in their data, not the magnitude of their data is what’s interesting here. Unless they are fudging their data, what can possibly explain their result? A new client that somehow attracts GNU/Linux users? A major roll-out? What’s your honest opinion besides that I’m dishonest?

  9. oiaohm says:

    Ted flick over to mobile OS stats as well you see the Android graph. That shows no major movement.

    If what you are saying is correct what browser on Android is being miss detected Ted.

    There is a few scary possibilities.
    http://gadgets.ndtv.com/mobiles/news/first-firefox-os-phones-now-shipping-worldwide-357946

    Firefox OS phones going like a bat out of hell. Ok I kinda not believe this but it is a possibility for the timeline.

    Ted maybe some Linux powered TV’s running like baba turning up on-line.

    Basically its Linux something does not have to be Android.

    Ted I have seen statscounter display one number than revise it later up or down. Basically Robert Pogson made a mistake with statscounter. Rule don’t trust the last number until another number appears. Last number it a statscounter guess some of the time.

    Ted nothing really explains the overall market share reduction of Windows. Some is just strange like the number of XP machine turning up has increased.

    Something happened. Something that has rattled all of statscounters numbers.

    Its not just the Linux number that has moved without clean explanation.

  10. Ted wrote, “Go to the page you linked to, download the CSV file, then look at the “Linux” column in it. Show me where this “little over 3%” is.”

    The “daily” page is dynamic. The count changes throughout the day. I viewed it early in the day.

    Here is the tail-end of US daily … 2013-05-25:
    2013-05-18 49.26 10.81 9.3 13.44 6.19 2.28 6.08 0.53 2.1
    2013-05-19 48.44 10.64 9.26 13.77 6.54 2.34 6.36 0.5 2.15
    2013-05-20 48.16 14.03 8.04 14.04 5.29 2.71 5.46 0.51 1.78
    2013-05-21 47.97 14.16 7.92 14.35 5.3 2.68 5.37 0.49 1.76
    2013-05-22 48.27 13.6 8.08 14.28 5.46 2.43 5.47 0.49 1.91
    2013-05-23 48.92 13.4 8.23 14.13 5.22 2.47 5.36 0.49 1.78
    2013-05-24 48.96 13.11 8.4 13.85 5.32 2.63 5.41 0.5 1.83
    2013-05-25 48.05 9.67 8.78 14.54 6.78 3.14 6.33 0.41 2.31

    There it is, 3.14%

  11. Ted says:

    @bw

    “There is a Linux line, but there is no Android line at all.”

    StatCounter only show the top 7 and “other” as lines.

    Download the CSV and feed it into the spreadsheet of your choice. You’ll see “Android”.

    “I suspect that there is next to no Linux activity at all in their sample base and they are just counting Android phones as Linux.”

    That’s the explanation I’ve suggested. It seems to fit other facts the best.

  12. Ted says:

    “When I wrote 3% the number was a little over 3%. So my point stands.”

    Go to the page you linked to, download the CSV file, then look at the “Linux” column in it. Show me where this “little over 3%” is.

    You’ve linked to data that disagrees with you.

  13. Ted says:

    “StatCounter’s numbers are nonsense”

    Yet they are solid enough for you to claim a sudden huge increase in Linux market share in the US. Or to show rises in South America. Or to crow about Sunnyvale.

    But if someone quotes “1%” to you, then they’re nonsense.

    If you deride their numbers when used against you, you cannot then use that same source to prove your own points. Or at least you can’t with any semblance of integrity or honesty.

  14. oiaohm says:

    Robert Pogson you were caught by statscounter rewriting there stats voodoo. I have seen that voodoo be up to 10 percent.

  15. Ted wrote, “let me help you with basic numeracy – 2.95% is still not 3%. So my point stands.”

    When I wrote 3% the number was a little over 3%. So my point stands.

  16. Ted wrote, “So which is it? You can’t have your cake and eat it.”

    StatCounter’s numbers are nonsense but assuming they don’t change their methodology there should be no dramatic shifts as are seen here. That means something changed “out there”. That’s what interests me. What changed? The last time there was something this big it was Google switching to Goobuntu. That shifted NetApplications’ numbers for California. They have stopped giving out details for $0 so I don’t know whether they can give us better localization.

  17. bw wrote, “I suspect that there is next to no Linux activity at all in their sample base and they are just counting Android phones as Linux.”

    I suspect they know how to parse User-agent strings. GNU/Linux usually has X11 and linux. Android/Linux usually has Android and linux. Since Ubuntu GNU/Linux is so much larger than other GNU/Linuxes there should be no confusion.

  18. oiaohm says:

    bw mobile OS is a stat option. That shows android. That also shows android running very stable from the start of year.

    The huge collapse in Windows 7 numbers happen around the same time.

    Fairly much there is no explanation for what is going on. As I repeatedly say I don’t trust web stat collection method full stop.

    This is why strange crap just happens to them all the time.

  19. bw says:

    What strikes me as odd is that there is a bunch of lines for various Windows releases and a Mac line and an iOS line. There is a Linux line, but there is no Android line at all. But there should be some level of Android use on a par with iOS, one would think. There is an “other” category, too, but it is small.

    Other stat gatherers show Android and PC Linux as separate categories and show Android on a par with iOS.

    I suspect that there is next to no Linux activity at all in their sample base and they are just counting Android phones as Linux.

  20. Ted says:

    “I think whatever the reason for the increase, Microsoft will “tap Statcounter on the shoulder” and Statcounter will issue a correction.”

    StatCounter have been showing Linux at ~1% for years (globally, it’s STILL ~1%). This was enough for some to accuse them of being bought and paid-for corrupt and of being in Microsoft’s pocket. This despite almost every other web-counter showing similar results. Anyone who quotes those results when others claim 5%, 10% or even 19% Linux share is just a troll, and told that 1% is a “myth”.

    Now that StatCounter show Linux on the rise, StatCounter are now Gospel Truth.

    So which is it? You can’t have your cake and eat it.

  21. Ted says:

    “Here, let me help you use their site:”

    Pardon me for using the link YOU provided to link to the stats. When you posted this article on May 24th, the highest number was 2.71%.

    And let me help you with basic numeracy – 2.95% is still not 3%. So my point stands.

  22. matchrocket says:

    I think whatever the reason for the increase, Microsoft will “tap Statcounter on the shoulder” and Statcounter will issue a correction. There will be some reason given for the “erroneous” increase and that will be the end of it until the market increase simply can’t be hidden any longer. It will shoot up again but this time it would be suicide to try to hide it second time.

  23. Ted, disagreeing on just about everything else, wrote, “PS. The Statcounter numbers do not mention 3% for Linux anywhere. It’s 2.71% at highest.”

    Here, let me help you use their site:

  24. Ted says:

    “I don’t understand the explanation.”

    1. Lots of new “Android” devices are released. Presumably, more than a few are being sold, considering the blanket coverage in print, TV, and cinemas. Not to mention advertising and marketing by the mobile phone carriers.
    2. StatCounter show NO significant rise in “Android” devices.
    3. StatCounter DO show a rise in “Linux” devices.
    4. As Android phones have both “Linux” and “Android” in User-Agent string, it’s possible that they are being mis-counted.

    I don’t claim this is the definite reason for this sudden rise, but it is plausible, and in my opinion based on the stats, more plausible than a desktop rollout somewhere.

    “how do the retailers sell the entire stock in one day? That’s a pretty rare occurrence.”

    You don’t really buy popular or hyped electronic devices, do you? 🙂 Ever heard of the Nintendo Wii? Or perhaps the iPhone? Not to mention pre-orders or on-line buying?

    “Why is there a jump followed by a quiet period? That’s roll-out behaviour.”

    I’d disagree with that. A rollout would be a jump followed by a plateau. It’s installed and running, why whould it disappear again? And the fact that Linux use mostly rises at weekends points strongly towards this not being a corporate, educational, or government rollout.

    “Why is the surge located in USA/Canada only?”

    Differing release dates to the rest of the world? Not like that’s an uncommon practice, is it? The fact that North America is arguably the single biggest market for high-end mobile phones and tablets might help too.

    “I just have not seen crowds crushing retail establishments to buy HTC’s stuff. Where are the hyped-up ad-campaigns? I haven’t seen any.”

    HTC *AND* Samsung. No trying to diminish my point by leaving two-thirds of the recently released models out!

    You’re talking about not seeing evidence of things related to the United States (your headline and the stats you provide say so), while you – in the nicest possible way – live in a Canadian backwater. Just because you do not observe it, it does not mean that it did not or does not happen.

    And if you’ve not seen an ad for the Galaxy S4, Tab 8 or HTC One, here’s a tip – *turn on a television*. 🙂

    PS. The Statcounter numbers do not mention 3% for Linux anywhere. It’s 2.71% at highest. Your headline and the extrapolation on your graph could be construed as misleading. And StatCounter haven’t “finally admitted” anything – they’ve just released their results as normal.

  25. Ted wrote, of smart thingies masquerading as GNU/Linux thingies “it’s the simplest explanation for the jumps in “Linux” usage.”

    I don’t understand the explanation. If I make a million smart thingies that announce, “X11, Linux” etc. and deliver them to retailers, how do the retailers sell the entire stock in one day? That’s a pretty rare occurrence. If a device were so popular, the OEM would make more and the share would ramp up after a spike instead of holding steady. That’s the real puzzle. Why is there a jump followed by a quiet period? That’s roll-out behaviour. Why is the surge located in USA/Canada only? Does HTC have such localization? I just have not seen crowds crushing retail establishments to buy HTC’s stuff. Where are the hyped-up ad-campaigns? I haven’t seen any.

  26. Ted says:

    I’d be looking at the release dates of mobile devices rather than searching for Linux desktop rollouts in business or governments.

    The HTC One, the Samsung Galaxy S4 and the Galaxy Note 8 all fit into the window where “Linux” use rises.

    Unlike some, I don’t pretend to be an expert in numerical analysis (I’ll stick to administering servers and client computers) but it’s the simplest explanation for the jumps in “Linux” usage.

    “Android” has not risen particularly noticably, despite these much-hyped Android devices being released. Even accounting for new devices cannibalizing existing share, the overall share should be going up, if the vendors (and you) are to be believed about the take-up of Android. Are these new devices simply being counted as something else?

    Use of “Linux” actually increases at the weekend. So it’s not a business/government rollout, as that would show the opposite. The most likely cause is more personal devices (not necessarily computers) being used at weekends. The usage of iOS matches this trend. People would presumably use iPads and iPhones more at the weekend when they’re not at work or school. Same for their shiny new Galaxy or HTC One.

    Steam’s usage has been dropping according to Valve’s own numbers, so I very much doubt it’s a surge in Steam installs. I’ve said it before – there is no compelling reason to switch to Linux for Steam. The third-party support is just not there. You can crow about Linux having over a hundred games now, but it’s quality not quantity that matters and Valve’s own offerings and two or three old (take note of this – OLD) ports aside, the quality is not there. All the indie games and their add-ons that make up the Linux numbers are also available on Windows. There are still no Linux exclusives, and I doubt any development studio wants to release their suicide note in playable form.

  27. Kevin Lynch wrote, “Maybe it’s all the gamers switching now Steam is available for Linux and has better support on the Mac.”

    Hmmm… What was the date of that release? 15 February in Ubuntu GNU/Linux… Why did they wait until April to suddenly switch? Do gamers stay off the web most of the time but with GNU/Linux they suddenly figured out multi-tasking?

    We wait expectantly for new stats from Steam to see if the jump in May registers. Perhaps they are one of StatCounter’s sites…

  28. Kevin Lynch says:

    Maybe it’s all the gamers switching now Steam is available for Linux and has better support on the Mac. The only reason gamers use Windows is because that is the platform the games were written for. Now that is changing, we should expect to see massive changes in market share.

Leave a Reply