OneStat.com today reported that Microsoft’s Windows dominates the operating system market with a global usage share of more than 97 percent. The leading operating system on the web is Microsoft’s Windows XP with a global usage share of 38.48 percent. Microsoft’s Windows 98 has a global usage share of 22.56 percent and is the second most popular OS on the web. The global usage share of Apple’s Macintosh operating system is 1.49 percent and the global usage share of Linux is 0.51% (that’s double since last year‘s 0.26%) on the web. Google also has some stats on the same subject.
OneStat: Microsoft’s Windows Dominates the OS Market on the Web
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I know quite a few people that have their Web Browsers “pose” as IE. Is it _really_ checking to see what browser and OS is being run? Or just what “appears” to be running.
I feel the Linux die hards comming…
Okay, my friends, what is wrong with these stats?
Just kidding, of course.
To make it simple, let’s assume these stats are right… I would’ve thought that 97% would be something between 90-95% by now. Too bad…
For arguments sake, let’s say these numbers are fairly accurate.
For one thing, I’m actually surprised that WinXP has passed up Win98 already.
So WinNT, Win95, Win2000 make up the other 37% of windows operating systems(WinXP and Win98 only add up to about 60%)?
I believe this was already posted a few weeks ago.
Why do we need two flame wars?
My problem is that web usage isnt an accurate measure. How many *nix thin clients,servers,etc are out there? More than home usage, I can tell you that.
Yeah, 97% does seem pretty high.
And yes, you’ll probably see posts saying that it’s a nefariour plot sponspored by MS to keep linux’s numbers down.
There are several ways in which their sample could have been biased. Which site did they survey? That is the real question.
Personally, I think that Google’s estimate is a more accurate picture, and even then it is biased because Linux browsers often masquerade as IE in order to be compatible with the site.
I looked at the onestat web site and my impression is that it only monitors sites that use its software. How many linux sites use third party monitoring tools? Do you think slashdot.org, apple.com, or even the osnews.com web sites are taken into consideration? Just my two cents…
Maybe linux users know where they’re going and dont have to ask for directions?
So the “other” OS’s total up to 5%? Which is more than Linux and Mac combined. What exactly are these “other” os’?
They cannot be very accurate for a very simple reason: They use some sort of client-side script (javascript) to track visitors. Now this has at least two problems:
1. The script may not work on all platform/browser combination because of the browser.
2. The users might have turned off scripting comletely.
There are other problems (like some alternatives pose as IE).
The ones I would have an easier time believing are the Google.com Zeitgeist stats, at http://www.google.com/press/zeitgeist.html . Google’s page doesn’t use scripting, and so the results are gained directly.
The two critical differences are (drum rolls): Mac and Linux stats! In both cases, the OnsStat press relase cites approximately half the percentage as the Google page!
Given also that most people are more likely to visit google than a onestat tracked site, I would say that the onestat results need to be taken with a bit of salt.
Better yet, how about a poll, not here but on the street to find out what percentage of people that use/own a computer can name more than two Non-Microsoft Operating Systems? Oh heck, I’d make it easy and cut that down to 1, or even a multiple guess, but put obscure auto names in with the OS names.
The only thing those stats show is that the vast majority of computer users are content to understand as little about their software and hardware as possible in order to “just” (or just barely) use it.
>I believe this was already posted a few weeks ago.
No, it hasn’t. This is a new press release. You are probably refering to this:
http://www.osnews.com/story.php?news_id=4138
Get back to work!!
Just kidding. I would think Linux users would search more, due to the “do it yourself” mentality often prevalent through out the Linux community.
>What exactly are these “other” os’?
That is all other OSes, but *ALSO* Windows ME, Windows 2003 Server and Windows 3.1.
No, this:
http://www.osnews.com/story.php?news_id=1706
That was posted a year ago though, so my memory is WAY off.
Sorry.
Almost the exact same press release though,very odd.
>Almost the exact same press release though,very odd.
Maybe nothing has changed in the world.
Actually, it has. Linux went up to 0,51% from 0,26%. That’s double.
Score one for linux users!! All 7 of us. 😀
the over 90% usage is the pirated chinese copies.
People have already pointed out the issues with the client-side stuff, so I’ll leave the OneStat stuff alone. The Google figures are more interesting, however.
In all fairness, the browser masquerading argument is limited, in that I believe only maybe 30% of people change their browser identity to IE for Mozzy, Konqy, etc.
The point I would impress, is that we seem to get listings for non-networked systems. I’m assuming proxies/caches quite often have an effect on this, especially at the business level (not to mention companies where employees don’t have net access from their desktops).
In effect, we have a guage of home use for single-connected machines. That would explain the windows XP penetration (which, in my experience is nowhere near 37% (from google) of current machines).
It would be interesting to find more stats on these figures, such as the ratios of ISPs and general locations, % of broadband users (most especially per OS). With this information, I think we could have a truly fascinating view of the market and development, which could be very meaningful in the long term.
So yes, we have stats here with a number of flaws in the reporting, but the flaws themselves will develop further data, if the companies take the time to report those too.
“Methodology: A global usage share of xx percent for OS Y means that xx percent of the visitors of Internet users arrived at sites that are using one of OneStat.com’s services by using OS Y. All numbers mentioned in the research are averages of last month and all measurements are normalised to the GMT timezone. Research is based on a sample of 2 million visitors divided into 20,000 visitors of 100 countries each day.”
What were the target markets? I can imagine that wininfo betanews and microsoft would yield numbers like that where sites like google ebay slashdot and cnn would provide a grossly different number. Until more info is provided one has to assume that the stats are skewed.
Nice to see Mac and Linux both making gains.
FCKGW-…
😉
(posted from moz/freebsd masquerading as redhat, emulating os-x)
97 percent of drivers are in Yugos and 3 percent are in Porsches. Sounds about right.
It also can’t be accurate because linux downloads off of FTP aren’t tracked, so the Linux number should be higher then it is.
First, some assumptions:
1) On the low end, we have 0.51% Linux users. On the higher end, we have 1.0% (Google’s number) users.
2) According to various statistics, there are anywhere from 450 to 600 million internet users.
Now, by the conservative measure, there are 2.3 million Linux users out there. By the more liberal measure, there are 6.0 million Linux users. Now, these are based on statistics about who is browsing to various websites from a particular OS. This means that we are measuring *desktop* users, not those who use Linux on a server. In total, that’s a whole lot of Linux users. Small percentage or not, several million people are several million people. Thanks to the internet, the geographic seperation of these people doesn’t matter. What is important to remember is that there is a very large potential market here for a moderately sized company to take advantage of. If such a company could really make a product that people wanted to buy, there is a nice bit of money to be made.
“Research is based on a sample of 2 million visitors divided into 20,000 visitors of 100 countries each day.”
Well, according to this, they rank each country equally, despite that obviously each country will have a different number of internet users in total. Which means that if one country has 90% MS and 5% Mac OS, 5% Linux, and has let’s say 50million people, and another country that’s 99% MS with 5million people, according to their statistics, 20,000 ppl from each country would be taken, and each of those would be worth the same amount of “points”.
Anyone else see the flaw in this approach?
Dude, you had me until you mentioned the money bit. Yep, 6 million at the high estimate is a heck of alot of Linux users, but as for the making money part…..ask Loki…Corel….Mandrake (until recently)…
The article said their technology powers 50,000 sites, now how many of those sites are tech related or even linux related? 50,000 is a pretty small sample set.
Although the google stats surprise me alot.
What about win4lin, wine, suse. rh, etc???
How many “normal” users actually go to anything Linux/Tech related.
I don’t think that they are that screwed, Google puts it a 1%, but is that rounded up/down? if so what is the actually number (if it is rounded up, it could be 0.54).
There is a very high bar here in Linux land. What Loki and friends learned is that Linux people don’t want to buy outdated, low-quality Windows ports many months late. The key is to offer value: Linux people don’t want to buy products they don’t need, and don’t want to buy products for which equally capable open source replacements are available. In this age of multi-billion dollar monopolies, people seem to have forgotten what makes a company successful — finding products your target market wants to buy and offering them at prices they can afford. Now, I don’t know exactly what products Linux people would want to buy — if I did, I’d be getting paid for it
I know what I’d be willing to pay for, though:
1) If a company went and added some GUI polish enhancements to KDE (like Ximian does for GNOME) I’d certainly be willing to shell out $50 for it.
2) If a company offered high quality theme packs (ala ObjectDesktop) I’d pony up $30-$40.
3) I’ve already paid for CrossOver plugin, and am considering buying CrossOver Office in the coming months.
4) I’d buy something like Flash Studio.
A lot of it is going to boil down to finding the niches were Linux is really used a lot (software development, workstation apps, scientific computing) and catering to those niches.
There are companies making money on the Linux desktop. theKompany, RedHat, and SuSE come to mind. Linux will probably never be able to support multi-billion dollar software companies. The competition is simply too intense and the rules are too different. However, there is money to be made here
we have influence. otherwise bill g. “money” and “dancing monkey boy” ballmer would not mention linux “as _the_ threat”…as often as they do.
hell…i don’t even want majority representation. 15% would be fine.
i just want to have a choice in 10 years.
an all MS world in 10 years, would be quite distasteful.
we half percenters are over represented in the technical skills area.
the other 99 1/2 percent are over represented by the “sheep” category.
From a statistical point of view these results are just a bit of crap.
There are two things to consider here:
1) Measuring visitors to sites running onestat software. This seems to be the purpose of the software, and I guess that in this regard it provides useful information to the individual sites. There are bound to be sources of bias, of course, but operating system is probably not the most interesting statistic the customer are after.
2) Generalizing the results from 1) to the whole Internet. This is a totaly different task, full of methodological problems and traps. In the field of statistical inference I’d guess that this belongs more to the area of research than to reliable application. I have not found *any* relevant information on the onestat site with regard to this.
Their aim is *not* to measure the global usage of different os:es, the tool they use is not developed for this and they don’t provide any information on statistical methodology.
With regard to 2) this is a prime example of crappy statistics, spewing forth so called “results” without any real substance. I am prepared to believe that their results indicates that windows have a very large global share of desktop os:es, yes, but that’s not news. To draw any conclusions about the exact percentage, say 97.34% or 95%, is plain nonsense.
But then, I guess that the purpose of this pressrelase is to draw attention to their site not to provide any useful statistcs…
Am I the only one who find those Google browser stats a bit fishy? It looks like people are converting FROM Mozilla/Opera/* TO MSIE6. I’ve seen people convert the other way around, but never ever from Mozilla to MSIE.
Weird.
But Linux doubled its market share. 0.51% is not so insignificant.
May be it will become the 2nd operating system in 2005.
If only Linux users had bought the nice commercial products available around for Linux operating system, may be Linux would ha ve 2% of Market share now.
Assuming the stats are accurate, if the trend continues Linux will have 130% market share in 8 years
Very small percentage of computer users actually know that there are alternatives for MS Windows. So, do your MS Windows using friends a favour and give them a GNU/Linux live CD. It’s quite OK if people use MS Windows because they think it suits their needs best, but it’s sad if people use it because they think that they have no choice.
Hum. No commercial games except fews.
No commercial desktop software
I am wondering if it is a good idea to force people in switching from Windows to Linux when there is no Adobe, Corel and so on softwares.
When I buy a new hardware for my PC running Linux, sometimes I have to upgrade my kernel or to compile it.
Why the fuck can’t we have the possibility to install proprietary drivers on Linux whithout informing us : “Kernel tainted” !!!
Why is it too difficult to develop a driver for 2.4.4 kernel compatible with the 2.4.19 for instance ?
Is the driver API on Linux so crappy ?
Instead of laughing on Windows we should improve the Linux APIs in order to use proprietary modules more properly.
Of course, these figures are estimates and there are all kinds of things you can say that they are not very precise, but that does not mean they are useless.
I think it is very reasonable that Linux’es desktop share is estimated at around 0.5 – 1%. It’s not zero and certainly less then 5%.
What I would like to know is when Linux will reach about 5%, because then it will be interesting for companies that are working with cross-platform toolkits like Qt, WxWindows or java to target Linux beside their Windows offerings. (See my comments in the discussion on: http://www.osnews.com/story.php?news_id=4575 )
_if_ Linux keeps on doubling every year, we might expect Linux to reach 5% by the year 2007 – 2008.
This means that ISV’s have just a few years to decide wether they continue on the MS path or if they will transform into a multi-platform shop.
This is a very important decision to make, because if Linux really takes off you will be faced with a growing competitive disadvantage if you’re not cross-platform and not able to reach your full market potential.
“Assuming the stats are accurate, if the trend continues Linux will have 130% market share in 8 years
“
This is exactly what I have been saying on this forum and others about the future of desktop Linux. It is currently in the takeoff zone of an exponential growth with a doubling rate of about one year – this what the IDG survey for HP on desktop Linux showed about eighteen months ago (the numbers showed it even if IDG’s conservative conclusions didn’t).
This is way Gates and Ballmer are so scared at the moment and in my opinion why we see the SCO FUD phenomenon now.
Why the fuck can’t we have the possibility to install proprietary drivers on Linux whithout informing us : “Kernel tainted” !!!
A more appropriate question would be: why should you care? The kernel doesn’t print “TAINTED!” on the screen in blinking red 96pt font while blowing a boat horn thru your PC speaker when you insert a proprietary module. It’s just an information for the developers saying that they shouldn’t bother to support you. See below for why they shouldn’t…
Instead of laughing on Windows we should improve the Linux APIs in order to use proprietary modules more properly.
Linus already said that he’ll never support proprietary modules because it makes debugging a really tedious task. It can be quite hard to know if a bug is in your or in their code. These modules are also usually compiled without debug symbols and/or frame pointer. That’s like asking a mechanic to repair your car without opening your hood (or assembling something from IKEA without a plan
).
They could freeze the kernel interface for each branch (like FreeBSD I believe), but developers would be forced to either make more branchs (2.4, 2.6, 2.8, 2.10…) or use workarounds for flaws, something that Linus don’t want.
>>Am I the only one who find those Google browser stats a bit >fishy? It looks like people are converting FROM >Mozilla/Opera/* TO MSIE6. I’ve seen people convert the other >way around, but never ever from Mozilla to MSIE.
>>Weird.
Do not question the great Google overlords!
Besides… maybe more Mozilla and Opera users are identifying themselves as IE in order to get to IE only websites. I for one refuse to go these terribly made websites unless I absolutely have to. I wish people made their websites more W3C friendly.
0.5% -> 1% -> 2% -> 4% -> 8% -> 16% -> 32% -> 64% -> 128%
That is when linux% double each year within 8 years. You need more math 😉