Debunking the 1% Myth

By Caitlyn Martin
September 7, 2010 | Comments: 152

There's nothing so absurd that if you repeat it often enough, people will believe it." --William James
It seems like almost every day someone in the tech press or someone commenting in a technical forum will claim that Linux adoption on the desktop (including laptops) is insignificant. The number that is thrown around is 1%. These claims are even repeated by some who advocate for Linux adoption. Both the idea that Linux market share on the desktop is insignificant and the 1% figure are simply false and have been for many years.

Linux market share is not tiny. Linux and UNIX have held a majority share of the server room for over a decade. Linux is very competitive in embedded devices. It is also making great strides on the consumer and business desktop, which includes laptops, notebooks and netbooks.

Let's start with netbooks, the area where Linux has made the biggest inroads. According to ABI Research Linux regained 32% of the netbook market in 2009 despite being next to impossible to find in brick and mortar stores. That number did not include systems sold in dual boot configurations with both Windows and Linux. On such systems Windows is still considered to be the default operating system.

Dell also reported that nearly a third of their netbook sales in 2009 were systems preloaded with Ubuntu. Recent reports that there was no longer demand for Linux on netbooks and that Dell was dropping Linux proved to be false. Indeed, as of today Dell is offering laptop and desktop models preloaded with Ubuntu in addition to the Inspiron Mini 10n netbook.

What do the netbook numbers mean in terms of overall desktop and laptop sales? According to Forrester Research netbooks were 18% of total desktop/laptop sales last year. If we do the math we find that due to netbooks alone Linux captured nearly 6% of the desktop market in 2009. In order to reach a total number we need to add larger laptops and desktops both from companies like Dell, HP (their business line) as well as smaller boutique vendors.

Additional confirmation of the growth in Linux desktop market share last year came from an unlikely source: Microsoft CEO Steve Ballmer. Using a slide to visualize OS market share Ballmer had Linux desktop market share as a slightly larger slice of the pie than MacOS. Nobody considers Apple insignificant on the desktop and neither is Linux. Here is, in part, what Mr. Ballmer had to say about Linux on the desktop and the competition for Windows:

Linux, you could see on the slide, and Apple has certainly increased its share somewhat.


I think depending on how you look at it, Apple has probably increased its market share over the last year or so by a point or more. And a point of market share on a number that's about 300 million is interesting. It's an interesting amount of market share, while not necessarily being as dramatic as people would think, but we're very focused in on both Apple as a competitor, and Linux as a competitor."

Does anyone believe that Microsoft would see Linux as a serious competitor is Linux had captured just 1% of the market? That doesn't seem very likely, does it? All the figures I have quoted so far represent sales of systems preloaded with a given operating system: Windows, MacOS or Linux. They do not represent actual usage. If you go down to the local brick and mortar computer shop or big box retailer, buy a system with Windows, wipe the hard drive and install Linux that still counts as a Windows system, not a Linux system.

Where does the 1% number come from? There are two sources: very old data and web counters. The problem with using web counters to try and ascertain market share is that they generally only include websites that have paid to be counted. That pretty much guarantees that Windows will be overcounted. ars technica recently demonstrated how dramatic the error can be in an article on browser market share. Overall they found the Internet Explorer has just over 60% of the market, Firefox just under 23%, and Chrome a little under 8%. The percentages for ars technica's own website were entirely different with Firefox at just under 38%, Chrome at almost 22% and Internet Explorer a distant fourth at 16.63%. The reason for the discrepancy is fairly obvious: ars technica caters to highly technically literate readers who, in turn, are far more likely to be aware of security issues in IE and far more likely to use Linux or MacOS. Similarly, most Linux and technical websites do not pay to be counted by the web counter companies, which in turn skews numbers wildly in favor of Windows.

So what is Linux real market share on the desktop? The best estimate for present sales is around 8%, which puts Linux just a little behind or perhaps just about even with MacOS. 8% translates to 24 million systems per year sold with Linux preloaded. Windows represents at least 80% of the market and is a de facto monopoly still. However, there has been a steady erosion of that monopoly status.

If we talk about actual usage there really is no way to get an accurate measure. Educated guesswork probably puts Linux at close to 10%, just about even with MacOS. That is a far cry from 1% and is in no way insignificant.

You might also be interested in:


I'd like to note that as an Ubuntu user, it's getting easier to find .deb installation files. The ecosphere is expanding.

Even if netbooks are sold with Linux, that doesn't mean they are used with Linux.

How do you prove what is used?
(hint - web counters)

What nonsense! You have to go out of your way to buy a Linux system. You don save any money when you do. Why would anyone special order a Linux system and load something else on it? Web counters are an accurate way of counting visits to specific pages, nothing else.

In my country this is very common. My first computer came preinstalled with Linux, I installed a pirated Windows on it. The reason is very simple, Linux is cheaper (usually free).

People from developing countries don't want to pay for Windows and in many countries laws forbid selling a computer without an OS, so it's either Linux or FreeDOS.

Well, your country (Romania if I remember correctly from DistroWatch) is very unusual if you have Linux or FreeDOS systems available in shops. Most of the world does not at this point. In most of the world there is no cost savings in buying Linux rather than Windows either.

In general Windows systems are readily available and others require special effort. It is far more common for a Windows system to be wiped and for Linux to be loaded than the other way around.

Dell is a great example. 100% of their Linux sales are special order and they account for nearly a third of the netbook sales. They also do not give a price break for Linux.

If anything, the figures I came up with are too low, not too high for precisely the reasons I list above.

well in our country, if you order from Dell, you pay more for a linux system because, when you order a Windows system, Dell gets all sorts of kickbacks from including crapware.

if anything it's the other way around. i know many linux users who buy the windows based dell because it's cheaper, then they just wipe the drive.

One other point: FreeDOS doesn't show up in the numbers at all. If that is common in your country then sales in your country either are not being counted or are a very tiny part of the picture.

In Romania is not very usual to buy a branded system (yes you can find DELL,HP etc at super/mega/hyper-market) but most of the people i know custom build their own PCs (the trend now is order your PC - with your custom components from online stores)
My dell laptop(notebook) came with a linux version (red linux if i remember -chinese lol) and i've wiped it out and installed Ubuntu.

I live in Indonesia. In my country, many laptops and netbooks have two options: linux (or Freedos) or windows.
Why? Because the linux version is cheaper, usually by about $80-100
What happened then?
The non-technical people will by the Windows version.
The technical people, most will buy the Linux version, and then wipe it and install Windows pirated! Some stores will even help non-technical people to do this. All to save that margin.

I am a Ubuntu user, and I would love linux to succeed, but 10% sounds optimistic to me. Maybe 5%? Noone really knows for sure, but I agree that 1% is surely not correct.
Well, Canonical has been advancing the idea of a census.
Of course, some think that's spyware. I would agree to a census. But then, there are still the many other distros.

I think, in developing countries, the main competitor to Linux is not Windows, but Windows pirated! If only Microsoft has a way to enforce their copyrights, I'm sure there will be many more linux users here.

Actually, Mr. Ballmer addressed "Windows Pirated" as a separate category, and indeed claimed it was second to legal Windows. "They have a heck of a product and a great price" were his exact words, if I remember correctly.

And yet... Mr. Ballmer showed Linux to be third in market share, at around 5%, ahead of Apple at around 4%, in 2009. That's *after* accounting for illegal Windows installs (it's unclear if installs of Linux over Windows were included).

Caitlyn's netbook calculation independently puts the floor at 6% in 2009, however - that's just netbooks. That's a new angle to me, but consistent with other measures I've seen.

I'm more familiar with Canonical's measure of 12 million IP addresses regularly checking the repositories for updates. It's hard to estimate the average number of Ubuntu computers behind Network Address Translation (NAT) switches (each representing "an IP address"), or what percentage of the desktop Linux market is held by Ubuntu. If you assume "2" (it's a technically savvy user base) and "1/3" (based on DistroWatch figures), and given roughly a billion desktops worldwide (Steve Ballmer's estimate), then Linux install share is around 7.2%.

That web browsing is a poor measure of installed base is made most clear by the smartphone market. iPhone swamps most browser stat reports, yet we know from sales figures that Symbian holds over 40% of that market worldwide, RIM dominates the US market, and Android Linux is now outselling iPhones and growing at over 500% per quarter. Apple / AT&T's requirement of an unlimited web browsing plan for every iPhone is perhaps responsible for the ridiculously biased stats.

Two lessons: Browser measures don't mean squat. And Linux is kicking butt and taking names across the entire computing landscape.

I have even problems with this method because my brother has worked in south america helping set up small computer labs (server and thin clients) in poor localities, he was in Venezuela just last year.

Not only does he say that he used the iso's he had on his computer and burned many, many, many times (it counts as only one download: his original one) but that updating he says is a huge problem because many localities dont have high speed. Heck, just having a phone line to run a 56k modem is a small miracle in some places.
So you have many, many computers installed from the same download iso but many, many more that dont/cant update the way we do in western DSL laden countries.
We take high speed for granted.

That's why I think the number could be even higher.

Good point - I hadn't thought of that. Darned, it makes the number even more difficult to calculate - even the "census" idea would fail with a lot of computers without reliable Internet access.


"Why? Because the linux version is cheaper, usually by about $80-100"
Well, in Fouquets'stan, things go differently:
the only linux box which keeps being sold in a small provincial french town is the ordissimo
It is a desktop (there are laptops, too), meant for very old people, it comes with a very tedious manual, and very little application (mail, IT browsing, text processing, photos/video viewing), with a huge screen and a small minitower.
The maintenance is most easy: if something breaks, one justs renetinstall (there are no other solutions). One cannot add other softs (I asked if one could install a gcc and a {K|g|c}onsole, and was laughed at....). It is impossible to get less choice .
It is very easy to know the difference of price :
the same HardWare with XP or W7 installed is 200 E$ -ca 250US$-
, which is explained by a great advertising effort in TV and in newspapers dedicated to the 70-170 years old segment. I will wait 20(0) years before testing it further.

As I pointed out in the article, Steve Ballmer had Linux ahead of MacOS when MacOS had 10% share so I do not think 10% is overly optimistic. In Asia in particular a lot of low end systems with ARM and MIPS processors are sold and those netbooks cannot run Windows at all. The Dell numbers are largely from the developed world. Finally, in the developed world most people who run Linux buy a Windows machine, wipe it, and install Linux. I strongly believe that far outstrips what you and Xtyn are describing.

Its funny that you mention your Indonesian experience because I remember two years during a June computer show in Taiwan when the netbook push really started, there were articles how the netbooks with Linux were for housewives while XP was for the geeks. Different strokes for different folks.

If you google 'taiwan linux housewives', you will see those artcles.
if youre lazy, here is one: but there are others which also quote this article:

That's about the time I first put my parents (mid-70's) on PCLinuxOS. Dad was a Windows user and mom never touched a computer so I totally believe the housewives part.

I was surprised this summer when I worked in Germany for two months at how prevalent Linux was. Im not sure their numbers but I travel about twice a year for work and Germany really stood out when compared to others.

In my country (Philippines) it is similar. Computers are sold with either Windows (more expensive), Linux, or FreeDOS.

As soon as people buy computers with Linux or FreeDOS, they format it and install pirated Windows XP.

However, the good hardware specs always come with Windows. So in other cases it's the opposite. Some people buy the good computers with Windows and then wipe it to install Linux. Hence they benefit of having a high end computer and Linux.

Excellent point that machines sold with X-OS do not necessarily continue to run X-OS. I have 6 physical machines in my home. Only 1 physical machine has an MS-Windows OS still. The rest run Linux and many, many Linux VMs

I run a blog and track usage statistics on it. So far this month, Linux leads all other OSes. BTW, I do not count internal access in these stats (my machines do not count at all).

Linux - 60.2%
Windows - 32.7%
Mac - 5.1%
Unknown - 1.4%
Solaris - 0.3%
BSD - 0.1%

What does that mean in terms of market share? My blog is fairly low volume, with only 11K pages hits this month (so far).

>Even if netbooks are sold with Linux, that doesn't mean they are used with Linux.

Same argument for Windows. Non of my four computer is running windows anymore...

I don't think it's unusual for developing countries. Look here:
Look at "Sistem de operare" (operating system) on every laptop. The page is sorted by "most sold" (cele mai vandute). In the first pages you will not see Windows.

You keep ignoring everything from the Dell numbers to Steve Ballmer's chart and keep repeating yourself. I have made clear that what you are describing is more than outweighed by Windows systems being wiped and having Linux installed throughout the developed world. You have no facts or figures to represent, just repetition of your opinion with nothing to back it up. I get it. You think Linux has no market share. Fine, you can believe that if you like. People believe in Santa Claus and the tooth fairy too.

I would love to know the number of legal photoshop owners :)

If the proprietary sofware vendors manage to crack down on piracy, it will only be for the benefit of Linux.

As for the number of Linux users vs other OS's, consider my case, I have 3 pc's at home, all came preloaded with windows, but I wiped them all, and only use Linux (OK I have a partition with a legal windows, that I boot on every month or so to do some gaming). But I have been counted 3 times as a windows user. On the other hand, I have downloaded and installed maybe 30-40 images of various nix distros in the last 3 years, does that mean I am counted 30 times as a Linux user?

Counted by whom?


Web-browser stats companies with unclear methodology and hidden biases and incentives?

Market-survey organizations that measure by dollar sales amounts (perhaps even including hardware, antivirus, maintenance services, office suite and other applications in the figures), and frequently only looking at brick-and-mortar stores (who furthermore are often adverse to non-Windows sales, even when customers seek alternatives)?

Top-tier on-line companies, who charge the same for Linux and Windows systems, and only allow certain upgrades (eg. better battery, more RAM) for Windows versions of the particular system you would like.

In other words:
No, they aren't counting -- not even your Linux-only machines.

People are ignoring a major factor, here, in this discussion -- namely the fact that the desktop OS market is not a "free" market: Microsoft bribes and punishes system manufacturers and sellers to insure that Windows is favoured. Many large distributors decide that Linux isn't profitable enough for them, even if there are customers clamouring for it.

This leads to spectacles such as:

* major netbook manufacturers publicly apologizing for demoing a Linux-powered netbook at a trade fair,

* leading manufacturers claiming that Linux-seekers are out of luck because their product is made with the OSs loaded in certain proportions, so if the Linux models are sold-out -- well, too bad (and then they get rid of their Linux product-line on the excuse that it's not selling well),

* employees of certain "big-box" consumer outlets revealing in their blogs that the higher-ups actively oppose non-Windows systems, because these don't lead to high-profit "add-ons" such as antivirus, bi-annual malware cleanings and "tune-ups" (and service packages to provide the same), additional software sales, etc. This can go to the extent of lying to customers about product availability, if the customer reveals that s/he plans to install Linux on the system,

* "Big Box" vendors training their employees using blatantly biased training materials and programs supplied by Microsoft, that disparage Linux and it's capabilities,

* Software houses make their desktop product available only in Windows (and perhaps Mac OS X), even if they already make Linux versions available to large enterprise as a server-based product.

Every debate about the numbers (and what the numbers mean) is incomplete without acknowledgement of this underlying bias. Every desktop Linux "sale" or installation is in fact made in the face of a pervasively hostile environment in which vested interests actively hinder, discourage and misinform end-users about Linux as a credible option. Linux users have to fight the system, go to extra effort, even pay higher prices (despite Linux OS itself being free). The facts and figures so readily bandied about generally reflect this ongoing pervasive bias.

Yet somehow Linux usage on the desktop still makes steady gains -- and here we are, arguing about just how quickly or slowly it is gaining.

How true! I work for a very large company who has a sweetheart deal with Microsoft so we receive every PC with Windoze loaded. Our group does in-house web app and database development and support and we have two Windoze PC's which let us see our offerings on the company's standard op. environment with all its silly little quirks we need to work around. But all our development PCs and production servers run linux and were never booted into Windoze - in fact, their first boot invariably happens with a linux install DVD.
Our company was seriously going to switch to linux, lock stock and barrel - and we're talking something like 50,000 PCs, but Microsoft started throwing money at our CEOs and accountants and that was that. Que sera sera :)

In China you can buy lots of computers now with no operating system. For instance I just bought an MSI and put Ubuntu on it. No Windows tax.

Those systems do not count as Windows systems or as Linux systems when numbers for sales are collected. They simply do not show up in the numbers at all.

Here in Spain it is almost imposible to buy a laptop or netbook without Windows (Dell do not sell the ubuntu ones here) so all the Linux users have to buy a Windows one and replace it by a linux. And all these laptops counts for Windows and not for linux (double error!)

Well, that is ironical: last year, the only way I knew to buy with cash Linux laptops was to travel to ...Spain and it made a 2*1000 km travel ... so I forgot the adress!!!

Can you remember the city at least? The cities that i know are Madrid and Valencia and I did not find anything...

Finally a sugestion, i find a bit stupid to tralvel 2000km for a linux computer, you will save money and time if you buy one with windows and replace yoursel de SO... or ask the shop to send you by postal ;)

"Can you remember the city at least? "

I remember Barcelona (which is compatible with the distance from my North french provincial town )

"Finally a sugestion, i find a bit stupid to tralvel 2000km for a linux computer"
So did I , else I would not have forgotten the address!
"you will save money and time if you buy one with windows and replace yoursel de SO."
I did not even replace : with unetbootin I can have nice live CDs (Victor, Scientific, Sabayon ) on USB sticks, with convenient keymaps; Mandriva can be installed -making his package manager working, unlike live CDs images- very easily -Fedora too I believe)
Is recent (BTW, it is the nearest place I know from my Northern French town where I could find laptops/netbooks -there is no gap- with linux preinstalled ; Romania is another 1000 kms away, and Ordissimo s seem very boring)

I think we can safely conclude that buying and using are not the same thing. The question is: do we want to estimate what we buy or what we use?

If we want to estimate what we use, we need to find a way to do that. I believe the best way is web counts. If only Google would share its results with the world...

Until then, I think we can assume that Wikimedia is representative, Linux visitors having 1.88%.

"do we want to estimate what we buy or what we use?"

I suppose the Microsoft tax has some economical value; even if I has a dual-boot, what interest Microsoft (and my "porte monnaie") is whether they are paid or not at a world scale...

"I believe the best way is web counts"
Web counts might be biased: I have "net""books" and no direct connection (therefore, I download *isos, *rpm , *src*rpm or sources from Internet cafe, under XP -it is unlikely I get viruses into them and they have high speeds-) When my neighbor asked me if I would download from his home, I asked him : what is your download speed? and it was 10 times slower than walking to and back an Internet Café, and downloading!).
This situation, where IT connections are too slow to be worth buying IT se(r)vices, are more common than commonly thought: I noticed cygwin, since last century, offered the opportunity of having a local mirror. In 2008, Gentoo (in Linux Identity Kits, bases in Newark, USA and available in good french railways stations) offered the same possibilty of having a local mirror -in total contradiction with the rolling release concept) Debian's *deb are sold about 19$ in Distroquatsch ads -the whole repo -a normal distro is ca 2$-, and,
in GNUlinux Magazine, France, Hors Série no 48, pp15-17
, tools (apt-rdepends and apt-zip) to manage packages (know what one has to download, with every dependency*) when not connected are discussed -and I miss them for *rpms -could make parts of them for R's *tar.gz-)
I know that this situation of having unconnected PC is what Kiki Novack calls "ordinosaurs", but their proportion might be huge, and cannot be objectively known ( at any *logical* way, not by web statistics!).
Have a nice year...

If you need to run Windows Programs or Games on Linux, FreeBSD or Mac you can use software like Bordeaux. Then you have the best of both worlds.. Linux security and that favorite Windows application.Bordeaux for Linux FreeBSD or Mac

After reading your article I looked at Dell (Belgium) for a notebook with Ubuntu on it. No luck. Then I had a chat with a Dell agenda:

Basically, he tells me that there are no Dell ubuntus in Europe and that I have to go to the States for that. Also no Dell ubuntu notebooks are planned.


Not a problem - try They ship internationally.

One major problem, you do not have to pay to use web counters, and many sites use them for free, including my own blog. I simply copy-n-pasted the html code into my blog, and now I get a web counter of hits and if I click on the counter, I get taken to a nice page that shows break down by OS, referral site, time, location, etc.

This article was pretty lame overall, another thing is "MS considers linux to be a competitor so it can't have 1%/insignificant market share" - now we're basing facts off of guesses about the motivations of Steve Balmer? Maybe he wants people to leave him alone with the anti-trust issue, so he makes a big deal out of linux and mac, who knows. All I know is every unbiased web counter, including free ones, including wikipedia, show Mac OS X at about 5-6% and linux around 1%. I'll go off that, and leave your assumptions and gut-feelings where I found them.

Every machine in my home has Linux. Yet, my firefox (or opera) browsers are configured to identify themselves in many cases as IE on Windows. That is because some web pages refuse to load just because of what browser and/or OS it *thinks* you are running.

Oops, I guess that kind of puts a kink in the accuracy of your fancy web counters.

Nope. You are within a margin of error.

So what is the margin of error? Care to state it? 1%, eh ?

"All I know is every unbiased web counter, including free ones, including wikipedia, show Mac OS X at about 5-6% and linux around 1%."

web counters count PCs *if* connected to the net. Many PCs are not connected (ex. laptops in a train/plane : some are used only if you are professionaly travelling; some boxes at work are denied net access, for safety reasons; even in developped countries, the net may be too slow to hire a connection), or cheat to make believe you have iexplore (some French IT se(r)vice providers assume you have iexplore.exe).
Can *your* experience compete with a US/world wide experience?
Can an ***unbiased*** web counter count unconnected PCs?

OTOH knowing how many PCs are sold, and how are sold with a Windows tax is quite easy for M$, and they might lose some shareholders if they unduly claimed they are threatened by linux...

Any web counter, even if one is truly unbiased, will under count Linux precisely for the reason you describe. Many people will change their user agent to report IE and Windows to get around websites from companies that only want to support Windows. Not all Linux distributions are configured to have their browser report the OS accurately. The numbers will always be skewed downwards for Linux.

"Not all Linux distributions are configured to have their browser report the OS accurately" -- can you back that up with any evidence?

I've used Linux on my desktop since 1998, and MSIE-only sites used to be a significant problem. I have changed my browser string in the past, but I haven't done so in years because most modern sites support multiple browsers.

I find it hard to believe that anyone would bother changing their browser string these days.

As I mentioned I visit the Israel Broadcast Authority website regularly and I have to report as IE to get their streaming media. I have a client that uses Outlook and I have to report as IE for the OWA (Outlook Web Agent) to work properly. It is still a very common problem.

Slackware and many of it's derivatives to do not report as Slackware or Linux AFAIK. What evidence would you like for me to show you in that area?

Does the web counter that you are using pool the info from your site with a lot of other sites? Is it the same web counter as any of the ones that are normally pointed to? Your statement about your web counter is just another anecdote and doesn't carry any more weight than what you are criticizing.

Also, Wikimedia puts Linux hits at closer to 2% than 1% (1.88%). Eighty eight percent more than what people have been claiming is actually a significant difference, especially if the number is on the rise.

Your web counter on your little website is not what is being quoted when 1% is claimed. They are paid, professional web counting organizations and their numbers are skewed.

You call my article "lame" despite the fact that I presented links from research organizations and computer vendors to back up my numbers. You presented absolutely nothing, no data, no links, nothing verifiable to back up your assertions. So... if you don't believe it then it isn't true? Is that it?

It's very easy to throw stones. It is not nearly as easy to bring evidence to refute the article. You certainly have failed to do so.

"All I know is every unbiased web counter, including free ones, including wikipedia, show Mac OS X at about 5-6% and linux around 1%."

Ooh, ooh, let me. Enter "OS statistics" in Google. FIRST HIT is this:

2010 Win7 Vista Win2003 WinXP W2000 Linux Mac
August 22.3% 10.5% 1.3% 53.1% 0.4% 4.9% 6.7%

That would be Linux just shy of 5%. It's just one, and I don't believe theirs any more than I believe the others, but since you mentioned it, it is the very first hit on Google. (I trust that you don't define "unbiased" as "agrees with my pre-conceived notions". ;-)

Dafydd: last time I checked, UK and Germany are part of Europe, so your Dell representative in Belgium was simply ignorant, just read the following link:

I agree that the sales of OS loaded computers are a poor guide of what is finally used. But what is more interesting I find, is why a Ubuntu PC from Dell with the same specifications as a Windows PC often used to be either MORE expensive, or had no price differential. I had queried this with Dell before, but never acquired a definitive answer...instead, on their website it has now become nearly impossible to compare the two side by side, like for like. My guess is that the cost of the window tax is offset by commission earned when people pay to update their free virus killer, or other time limited "free" software e.g. Office. So, I guess it is possible that MS may in some cases be effectively subsidising my laptop purchase, which I will subsequently load with Linux.
The other thing that can not be measured is the number of people that use Live USB/CDs for maintainance or portability. So many IT guys I know keep Linux on a stick handy just in case a Windows (or linux) PC is having problems

So, I guess it is possible that MS may in some cases be effectively subsidizing my laptop purchase, which I will subsequently load with Linux.

You pay MORE for the same hardware and you think Microsoft is subsidizing your purchase by donating part of the purchase price of Office or some other demoware?

Actually, the hardware that comes with Linux pre-installed has fewer options and peripherals available, so you are paying MORE for less.

DELL and the other PC OEMs are PAID indirectly by Microsoft in the form of ad rebates to pre-install its OS. The PC market had long ago reached the commodity state and now profit margins are so slim that ad rebates are the difference between profit and loss. IF Microsoft cuts off their ad rebates they become unprofitable.

We've recently learned that between 2001 and 2006 DELL took bribes (euphemistically called "credit memos") from Intel to NOT install AMD video drivers. The bribes amounted to over 70% of their income. When DELL installed AMD on some of their PCs Intel cut off their bribes and DELL nearly went bankrupt.
It's hard not to conclude that the PC processor market is now a monopoly cast in granite - and that the SEC settlement has come too late to introduce any meaningful competition.

The Intel-DELL agreement was pure FRAUD. Why some real "corpus" didn't get jail time shows how disparate the legal system has become. One wonders when the SEC will investigate the ad rebate fraud that Microsoft is using control the PC OEMs.

You don't get my point. By buying a windows PC I get a cheaper better specified machine. If you want to use linux therefore it is more cost effective to buy a windows machine and then discarding windows and putting Linux on it. Effectively MS subsidizes my linux machine. QED

Regardless of any other opinions or statistics, we recommend our clients and customers purchase hardware with Windows pre-installed.

There is only one reason for this and that is the immediate resale value benefit.

Try selling a used machine on the open market. Most folks just want a machine with a valid Windows COA.

No point in arguing or wasting time fretting over it. It's just a decision based solely and wholly on secondary market values.

Tending Wyatt's Torch,
John and Dagny Galt
Atlas Shrugged, Owners Manual For The Universe!(tm)


Its ok to recommend, it is better to offer the choice transparently. By the time the user comes to resell his computer the odds are that his version of Windows will be out of date, replaced no doubt by Windows 8. The reason manufacturers always put "XX recommends Windows etc" is presumably, that by doing so they get a discount from MS on their OEMs.


I am an GNU/Linux user for several years and I try to spread it as much as possible. I have a web page about histology (in Spanish) and the users mostly come from Spanish speaking countries. From 2008 till now (nearly two millions of visits) Gnu/Linux scores 0.75 %, Apple 1.20 % and Windows 97.86 % (Google analytics). I know that this results are biased to these countries and to the histology subject but I think it means something: young people is not so enthusiastic with GNU/Linux as one may think. In my opinion there is a lot of work to do yet.


"Similarly, most Linux and technical websites do not pay to be counted by the web counter companies, which in turn skews numbers wildly in favor of Windows."

That doesn't really hold up, unless you posit that people who run Linux *only* visit geeky technical websites and never visit general interest sites (or that they visit general interest sites vastly less often than people who use Windows). I find this unlikely. It's the numbers for geeky technical sites - and especially Linux-related sites - which would be unrepresentative; they would be skewed *in favour* of Linux users. Also note that even very geeky sites which publish numbers have very low numbers for Linux. w3schools has it under 5% - Wikimedia has it under 2% - - and that's including Android. Do you really want to argue that Wikimedia's numbers will be unfairly biased against Linux users? Linux users don't visit Wikipedia as much as Windows ones do?

2009 netbook sales numbers are not a particularly reliable indicator of current trends. See my article at and the extensive commentary discussion for details. In addition to what's mentioned in that post, remember the iPad has blown a rather significant hole in netbook sales.

So where did Steve Balmer get his numbers from? Your same source? I doubt it.

Your dispute is with Steve Balmer the CEO of Microsoft, not the author.

You are posting as a fanboi with scared beliefs that Linux is ...

You omitted ( by mistake I'm sure ) one chart that debunks the/your 1% myth.

PS. If you want someone to believe a number, put a percent sign behind it.

Adam Williamson is NOT a Microsoft fan. He works for Red Hat and is a highly respected member of the Linux community. Please remember that educated and well informed people can legitimately disagree. I may disagree with his analysis of the netbook market but I certainly respect Mr. Williamson and his work.

Web counter numbers are NEVER accurate. Many Linux users have their user agent string reporting Windows and IE to get around restrictions placed by web designers and companies who only want to support Windows. That includes my computers since I go to an Israeli news site (Israel Broadcast Authority) at least once a day that is setup for IE/WMA only.

Netbook sales are still up even with the tablet iPad sales on the rise. In addition, there are a number of new tablets running Android, which is Linux.

Regarding Dell, their page isn't well hidden and the sales numbers I used came directly from Dell. You also dispute the ABI Research numbers in your article with an "I think" but provide no alternative data or evidence from elsewhere in your article. You talk about Best Buy and CDW. Best Buy dumped Linux a long time ago and never had much. CDW had nearly none at all. You are taking mainstream U.S. retailers and projecting for the rest of the world. You can't do that and get realistic results.

In Asia in particular low end machines with ARM and MIPS chips that run Linux or Windows CE (ARM only) only are still being sold, and an awful lot of them are sold with Linux. Changing the OS is non-trivial and in the case of the MIPS systems not even Windows CE will run. In addition the Chinese government is backing Lemote which does not sell systems in the U.S. Those have Longsoon 2F processors (64-bit MIPS technology) and come preloaded with either Red Flag Linux or Debian. Don't expect to see these at Best Buy anytime soon.

You are looking for 2010 figures. 2010 is still in progress. We won't have good figures until after the end of the year. Your article makes a good argument that the 32% figure from 2009 for netbooks won't go up in 2010 and we won't reach majority share in 2012 on netbooks as some projected. I agree with you on that. However, as more tablets with Android and Meego come out the percentage of Linux on the desktop will go up overall.

I cited numbers for Canada and the UK as well as the U.S., and quoted numbers from 'geeky' retailers - NCIX and NewEgg - as well as Best Buy. And like it or not, mass market retailers are where people buy computers. You say Best Buy and CDW don't carry Linux, I say 'exactly'. That's the *problem*. :)

"Web counter numbers are NEVER accurate. Many Linux users have their user agent string reporting Windows and IE to get around restrictions placed by web designers and companies who only want to support Windows. That includes my computers since I go to an Israeli news site (Israel Broadcast Authority) at least once a day that is setup for IE/WMA only. "

I dealt with that excuse in my articles. Frankly, I don't give it much credit. I see the excuse cited a lot but you're the first person I've seen who actually says that they set a UA to spoof Windows. I don't, and have never had any need to. I don't know anyone else who does. And as I mentioned in my post, honestly, do you really believe more than 5% of people who use computers actually know *how* to spoof a browser user agent string?

Of course user agent derived statistics aren't perfectly accurate. Few statistics are. They're probably the best we have, though, and I'd trust them more than tenuous extrapolations from old sales figures of a single market sector.

Android is, more or less, technically Linux, yeah. I don't particularly rejoice about Android-using devices though. If you just want to count 'Linux' numbers and go 'hooray!', yeah, Android's fine. But if you dig into it much it's a pretty icky project. For a start, it's completely different from anything else; unlike Nokia, who built a mobile operating system mostly out of existing bits of F/OSS infrastructure by co-operating with many existing projects, Google took the Linux kernel and pretty much wrote their own userspace on top of it. You can hardly leverage (ew) any of the existing typical F/OSS infrastructure on Android, it's its own little universe. And it's very much owned by Google, no matter how technically 'open source' it is. It's very hard to get contributions into Android, Google retains complete and very tight control over its overall direction, and in practice it's quite difficult to meaningfully exercise your F/OSS rights on Android platforms as they're almost all quite tightly locked down. Just look at the Android world - the Marketplace, which reads exactly like a more chaotic version of the iPhone app store, and the actual OS hacking scene, which much more closely resembles the world of Windows Mobile flashing than it does the rest of the F/OSS world - and you can see what I mean. Android is a different world and in many ways one which isn't much better than closed source projects.

Adam, you know I have a great deal of respect for you and both your professional Linux work and your work in the Linux community. I'm going to give you the same answer I gave fewt: Will have to agree to disagree, at least about web counters. I believe they have absolutely zero value and are an accurate gauge of nothing except traffic to specific websites. I find sales data more convincing. You don't and that's fine.

Where I agree with you 100% is in your analysis of Google and Android. I trust Google about as far as I can throw my house and I believe their approach is to use FOSS software in as proprietary a way as the license allows. The fact remains that Android is a very significant force in the embedded market and that it is nothing more than a Linux distribution customized for that market so it has to be included in any honest counting. My hope is that Meego, with the support of Intel and Nokia, can make inroads and provide a serious challenge to both Android and proprietary alternatives.

Agreeing to disagree isn't a bad thing, I think it's much better than continuing to disagree when neither thinks they are wrong (because I'm not! /funny).

I'm fine with it if for no other reason than the level of respect that I have for you.


Does anyone believe that Microsoft would see Linux as a serious competitor is (s/b "if") Linux had captured just 1% of the market?

Nice analysis, excellent article!

Although Linux comes pre-installed on 3% of DELL's laptop offerings and on only 6% of their server offerings, the hardware and associated peripherals on which Linux is offered by DELL isn't as complete as those on which Windows is offered. Shoppers are left paying nearly equal prices for substandard hardware. So, lots of Linux users like myself shop the Windows offerings simply to get hardware with the specific options we wish to purchase. The VISTA Home Premium that came pre-installed on this Sony VAIO VGN-FW140E was totally scrubbed and replaced by Kubuntu. The retail channel count does not reflect the reality that the number of Windows installations has decreased by one and the number of Linux installations has increased by one.

Also, while Windows gets one count for each laptop sold with Windows preinstalled, that copy of Windows cannot be moved to another machine, EVEN IF it were legal to do so. But, MOST installations of Linux were achieved by the user downloading their favorite Linux ISO file, burning it to a CD, and then using the CD to install Linux in either a dual boot mode or, increasingly, to replace Windows altogether. THEN, that SAME CD is used to install Linux on one or more other computers. My CD of Kubuntu has been used to replace Windows on over a dozen laptops of friends who got tired of paying for a subscription to keep their Windows installation free of malware only to have to pay to have it cleaned again when the latest vaccine updates failed to do their job. The most common problem I found was keyboard logger Trojans stored in the MBR and the unnamed partition where the restoration copy of Windows is stored. Do a recovery or restoration of Windows and the keyboard logger is also restored. None of the replaced Windows copies had their retail channel count deducted.

In my experience about 5-10% of those who try Linux return to Windows. Usually it is because they can't find a replacement in Linux for a program they find indispensable or a hardware component isn't well supported. But, about two weeks ago I installed Puppy 4.3.1 on a 12 year old laptop running a 486 CPU with 256MB RAM. The finger button in the center of the keyboard worked beautifully as a mouse, and Puppy was several times faster than the Win95 that was clogging its RAM.

Bottom line: The retail channels report a LOT of Windows machines that are now Linux machines.

@Xtyn - I understand what you're saying but I think you have to distinguish between FreeDos & Linux. FreeDos is just there so you can boot the machine - no-one would ever use the machine as is after buying it, however that is not the case with Linux

I see a lot of people have a real need to continue to present FUD in the form of the 1% number as truth without any evidence to back it up. One country, whether it is Spain or Belgium or Indonesia or Romania is not representative of the world as a whole. One website counter or any number is certainly not accurate. Some of the proof presented has already been debunked by conflicting comments. Some simply ignore the numbers and research I have presented but not one comment has provided a source of conflicting information which is reliable or verifiable.

Clearly some have a vested interest in seeing Linux presented as insignificant. That's fine. Keep your heads buried in the sand if you wish and Linux will continue to slowly and steadily grow.

The facts of the matter are that you cannot purchase a Linux netbook or laptop or desktop in the retail stores. You talk of Dell saying 30% of its netbooks were Linux, but Dell sells very few of the world's netbooks. ASUS, Acer, and HP don't offer one at all. So, at best, you are looking at 30% of the Dell production.

Even there, a quick check of shows that all of their mini systems are advertised with Win7. Maybe they are selling 30% Linux, but they sure are not promoting it anymore. Isn't that a sort of clue?

I don't think that it matters one whit as to how many people are using Linux, be it 1% or 8%. In either case, the OEMs are not offering it in the mainstream and that gives Linux essentially a 0% market share of things that people pay money for and that is all that matters to people in the business of selling things.

It may very well be that Linux is highly popular in some corners of the world, but those people don't seem to spend much money and that is why the suppliers are ignoring them.

I stated clearly that Linux cannot be purchased in retail stores in North America in the article. Almost Linux systems are purchased online at present, which is what makes the growth of Linux market share all the more compelling.

HP does offer Linux in the business line, particularly SUSE Linux Enterprise Desktop (SLED). They don't offer it in their consumer line. ASUS does offer Linux on the EeeBox series nettops despite your claim to the contrary:

Dell does have Linux on their website and they are promoting it, again, despite your claim to the contrary:

Your claim that Linux has no market share for people willing to spend money is absurd. Would you like link after link after link of Linux systems for sale? I can provide them.

You have provided another ridiculous comment with no supporting data and nothing to back it up. Don't you Microsoft supporters have anything better to do than try and claim that Linux is insignificant? Why does Linux threat you so very much?

From your link:

"Asus has pretty much stopped offering its netbooks and nettops with Linux preloaded. Instead, most of the company’s products come with Windows XP or Windows 7. But retailer Newegg is selling a first generation Asus Eee Box EB2020 nettop with Red Flag Linux, a Chinese Linux distribution that’s designed to look an awful lot like Windows XP."

So what Asus is offering is not a netbook and it is not so certain that Asus is offering it at all. It would seem like NewEgg is loading it with the XP look-alike from China.

Dell isn't advertising Linux, my dear.

Doubtless you can still find some way of ordering it, but it is not very likely that it is 30% of their shipments.

As to link after link of Linux systems for sale, does that really say anything useful? Post some financial figures filed by someone selling Linux systems showing that there is a viable business there and you will have achieved something worthwhile. Linux has been around for some 25 years now and it is still largely unknown.

What are you Windows people so afraid of? This article is not about netbooks. It is about the desktop as a whole so the ASUS offering is a legitimate part of it. Dell says that their netbook shipments are 30% Linux even if you don't believe them. Why would they lie about that? Do you really believe that Dell would keep selling anything for years that isn't making money for them? Do you think all the little Linux boutique vendors would still be in business if they weren't making money? Yes, the link says something useful. It says there is a viable business there or else Dell wouldn't bother supporting another OS.

Linux is "unknown". Linux and UNIX dominate server space. Nobody making money on Linux? Look at Red Hat's profits for the last few years which are most certainly published financial data.

Nice try but it's pure FUD backed with no data and nothing to refute the numbers I have linked to and posted. I ask again. Why do you Windows FUDmeisters feel so threatened by Linux? Why is my article so important that you take time to comment if Linux is unknown and insignificant?

The article was about Linux use, true, but your refutation of my comment that ASUS and Acer and HP didn't offer Linux netbooks was the sub-issue certainly and netbooks are 75% of your claim of Linux being an 8% player. So the dearth of anyone even making Linux netbooks is a clear indicator that you are wrong in your assessment.

Why do you suggest that we "Windows people" are afraid of Linux anyway? The corollary is that Linux people seem deathly afraid that they are not being recognized and all is for naught. Are you afraid that the Linux developers will give up the fight and cut you off from your source of free software?

Nobody makes Linux netbooks? ROTFLMAO! If you wanted to prove that your statements are ridiculous you've done it. Take a look at: The same Chinese manufacturers that build systems for the large OEMs build them for a large selection of smaller Linux vendors. The better known ones, like ZaReason and System 76, have been in business for many years. They all offer netbooks specifically designed for Linux use. Sorry to burst your bubble.

Linux people aren't afraid of anything. Linux and UNIX the dominant players in the server room where the real money is made so Linux isn't going anywhere anytime soon. On the desktop there has been steady growth as well despite Microsoft's best efforts and strong arm tactics to stop it. Linux developers won't "give up" because they make nice salaries thanks to companies like Red Hat, IBM, Novell, Intel, HP and many other large corporations who are heavily invested in Linux and employ those developers.

Look, if you want to strengthen my article by making wild and ridiculous statements, by all means, be my guest. Why do I think Windows people are afraid of Linux? How about all the effort they make to try and dominate any discussion of any article that features Linux in a positive light? How about all the marketing the Microsoft does that specifically targets Linux?

> Linux has been around for some 25 years now and it is still largely unknown.

For the record, Linus T made the first public release of the Linux kernel in August 1991, about 19 years ago. Linux became a contender on the desktop arguably when the X Window system ran on it for the first time in 1993. The first release of KDE was in 1998 and that of GNOME in 1999. StarOffice was open-sourced in 1999 as OpenOffice, the first version of the proprietary product for Linux having been released a few years earlier. I'd say desktop Linux has "been around" as we currently understand it for maybe twelve years, not 25.

Thank you. That was another point I could have tackled but you have done it beautifully and your timeline is correct.

The sad truth is that Windows "as we know it", had only about 4 years of head start, with Windows 95. Was that enough to flood the industry and maintain its dominance? I think it was the lack of marketing that mostly left Linux behind in terms of rate of adoption, in those early days.

Take the smartphone business for example, which is now in more or less the same state as the computer industry was in the late 80s. Windows Mobile, RIM and Symbian where the only players for a while. Apple came along and flipped everything around. Later on Google entered the game and become yet another game changer. And we haven't seen the last from Microsoft yet, who will re-enter the game with Windows Phone 7, not to mention the other newcomers.

On a different point, as we move into cloud services, where the OS plays a less dominant role, coupled with the proliferation of ARM based system (ARM moving into the desktop and server domain with their Cortex A15 etc), Linux will probably become more relevant in the consumer/desktop space.

Going back to the subject, Canonical has done some impressive job to bring linux to the mere mortals (without trying to diminish the efforts of the other distro providers) and I agree with you that quoting 1% linux market share is a FUD. I also agree that measuring the actual figure is very difficult, especially if we take into account machines that are not connected. But instead of trying to figure it out, I invite you to write an article on why Linux is not dominating the market. Microsoft's anti-competitive tactics? Microsoft Office's dominance? Lack of professional-grade software? Lack of hardware support? User friendliness? Marketing? Need a quantitative approach.

A Linux user since early 2000 (solely since 2005)

I've already written some of the articles you want, as in ones about Microsoft strong-arm tactics or the lack of preloaded systems in big box stores (another article on that soon). Those are the real reasons, not a lack of applications or hardware support. After all, Linux supports far more devices out of the box than any version of Windows. (Before someone disputes me on this, the minute you have to add drivers from a source other than the installation media it doesn't work out of the box.)

However, to claim that I shouldn't write this article to explain why the lie is, in fact, a lie, is a bit presumptuous, don't you think? Has O'Reilly hired you as an editor? Seriously, I think debunking pernicious myths like this one is every bit as important as the stuff you want to see written.

Here's a suggestion: if you don't like what Linux advocates are writing then perhaps it's time you start writing articles that stress the issues you believe are most important.

Certainly, I agree with you on all those points and I apologise if I sounded harsh towards your article. On the contrary, I'm glad someone did point out the fallacy of those metrics/methods used to measure the Linux market share. I will seek out your other articles you mention as well.

I consider myself a linux advocate as well and I truly believe in the power of community driven software development and its positive-feedback effect (though, I do recognise its shortfalls as well).

As for your advice, why not I guess. I'm sure there is room for yet another Linux blog out there.

"you cannot purchase a Linux netbook or laptop or desktop in the retail stores. "

In developped countries, may be (in *my* sowhat developped country, I am almost sure : I saw two other linux boxes than mine -1 at a conference {went to tens of conferences}, 1 in the train- within 4 yrs!) .
In emerging countries (where people are going to get rich, and buy their 1rst computer) it is unlikely; if Internet facilities are inexistant/ lousy (this year, I read in DWW a Brasilian having a download speed 100 times lower than what I am accustomed to in cybercafés) , they are not part of browser statistics, even if they were reliable. A number of {1- (Maxs + windows sold)/ (number of PCs sold) } seems a more reliable mean of measuring the growing speed of linux.

And these "corners of the world" have a fastly growing economical importance.....
And I spend much more money with linux, buying things like GNUlinux Mag France, Linux Identity kits and books ("Linux aux petits oignons" was very appreciated by my nephew) than with the Windows tax... and know many linux users do the same thing.

I still see lots of sites that discriminate, requiring me to use false
Browser ID strings. Ars Technica, for instance, doesn't work with KDE's Konqueror any more. Its links return an empty tab labeled "Dart Frame". So I don't trust their statistics either.

No way is Linux desktop as popular as Macs. Look around. I just do not believe it.

IMHO, the best way to estimate this number would be to look at browser figures for websites like Wikipedia and Google. Everyone uses them and I doubt that there's too much platform bias.

Since you don't believe it then it can't be true. Is that it? That is not a very convincing argument at all.

Caitlyn, what you are doing is called wishful thinking. I would like Linux to have a bigger market share but I'm not a wishful thinker.

Nothing is perfect, not web counters, not sales but I think web counters are the best way we have right now to estimate the market share.

If you like the thought of being part of a bigger herd, don't let me stop you from deluding yourself.

More nonsense! I provide solid statistical evidence and you dismiss it as wishful thinking. You provide no evidence whatsoever and expect me and everyone else to accept your view. Go back to the William James quote which started my argument: a lie has been repeated over and over again. It's absolutely absurd but because it is repeated so often it is believed by you and by many others.

You have done nothing to refute the provided data. Nobody has provided any valid, verifiable data to refute what I have written. That is the challenge I have thrown out there repeatedly and nobody has taken it up. Why? Could it be that there is no such data?

"Solid statistical evidence"? That is rather audacious. You take Dell's claim of 30%, which is very dubious, and multiply that by the claim that netbooks are 18% of the market to get 5.4% that you just round up to 6% and then gratuitously add 2% handwaving about regular laptops and desktops accounting for that number.

Somebody has a reading deficit. First, why would Dell lie about their own sales? What makes their claim "dubious"? If you don't like a number then it is dubious, is that is.

ABI Research put Linux market share at 32%, not 30% so your 5.4% number is too low. Dell is one vendor. I look at the market as a whole.

The 2% is not handwaving as much as you would like to dismiss it. Something has to be added for other systems that are clearly selling. The Microsoft CEO gives Linux a larger share than MacOS when plenty of sales numbers put Mac at just about 10% of the market so maybe my 8% number was conservative.

You have provided absolutely no contradictory evidence and just dismiss out of hand and you call me audacious? What about Linux is so very threatening to you and the other Microsoft fans here today? Why, if my numbers are wrong, can you provide none of your own?

1. Those DELL figures are from Feb 2009 - not all of 2009 as you claim.

2. And it isn't all DELL netbooks - just the Inspiron Mini 9s

3. ABI forecasted 11 million Linux netbooks out of 308 million PCs sold in 2009 - which comes out closer to 3.5%

I never said Dell figures were for 2009. I said that in 2009 they claimed nearly one third. That statement is accurate and it reflected all the netbooks they were selling at the time.

ABI forecast 39 million netbooks, not 11 million, for 2009 and revised the final figure to 35 million. Source: That makes 32% not 3.5%. Once again my information is accurate and well sourced. Yours is anything but.

I went to and the second sentence I read was
"Information about operating systems share is difficult to obtain. In most of the categories below, there is no reliable primary source or methodology for its collection."
Then I went to 'Desktop and laptops' and read :
"There is little published information on the usage share of desktop and laptop computers. Web client information (see below) is often used as a proxy for this, but many such computers are not used for web surfing"
About browser stats , they say :
"This is an inexact science for a variety of reasons. For a discussion on the shortcomings see Usage share of web browsers."

Therefore, unless you can prove that accountant figures (which are pillars of capitalistic economy) are less reliable tant web "statistics", I am obliged to think you, Paul Topping, are a communist..

Please, no name-calling or red baiting here. Comments like this one add heat but shed no light.

Windows is all that matters. Everything else is niche.

I'm sure folks said the same thing about WordStar when it dominated the word processing market. Ditto WordPerfect. Where are they today? Lotus 1-2-3 was king of the spreadsheets. Now it's gone. Go ahead, be cocky, and say that Windows is all that matters. It's a rather self-delusional belief.

It might be pertinent to note that 1-2-3 and WordPerfect were market leaders in the DOS days and were late to market with Windows versions, eschewing Windows for OS/2, not wanting to dignify Microsoft with their presence. As it turned out, that was a mistake on their part and Windows was all that really mattered and they let their own markets slip away.

I don't think that WordStar ever dominated word processing, though. Back in those ancient days Lanier and Addressograph and even IBM dominated word processing with specific machines for the purpose.

Don't forget Wang

How many names and non-names are you going to post under? Wang never dominated any aspect of the computer software market. Another unrelated smokescreen post. Come back when you have real data and something significant to contribute.

Of course they didn't dominate computer software for word processing, silly. Personal computers were not being seriously used for word processing in the WordStar era. Word processing was done with word processor equipment and Wang was a major producer along with several others.

Nothing about Linux terrifies me. What is it about low brain share of Linux terrifies you to the point where you have to invent these fantasies. You have yet to offer any real statistics. Only third hand references and a lot of handwaving.

The only thing silly is your claim the computers were not used for word processing in the mid 1980s. I was in the transportation industry back then as an IT Director and they were most certainly in heavy use. You are repeating an irrelevant argument about a completely different technology. I have linked and sourced my numbers and you just keep repeating claims that I haven't and provide no data of your own.

Guess what? You are done repeating yourself and trying to dominate the conversation. Since you have no new points to make I have decided not to allow you to continue repeating old ones. You're done.

Pertinent that those were DOS leaders? Certainly. DOS dominated the PC operating system market and now it's all but gone. Inevitably Windows will also be an historical system of the past at some point.

WordStar did dominate the word processing market in the 1980s: You "don't think" so just means you don't know and didn't bother to look it up. We're discussing computer software markets here so throwing out other sorts of hardware is just a red herring and a smokescreen. Hey, if you don't have facts on your side I can understand why you'd want to confuse the issue.

Again: where is your data to refute my numbers? What is it about Linux that so terrifies Windows fans like you?

Caitlyn, I appreciate and enjoy some of your articles. But this statement:

QUOTE: The problem with using web counters to try and ascertain market share is that they generally only include websites that have paid to be counted. That pretty much guarantees that Windows will be overcounted. ENDQUOTE

... This statement does not make sense at all. You don't explain at all or give reference to WHY websites that pay to be counted guarantees that windows will be overcounted.

This is a weakness that I see in your (and others') writings. Specifically, unfounded statements. Your statement might be true, but damned if I or other readers can verify the truth since there is lack of an explanation or references.

You seem to be suffering from a reading or comprehension deficit. I backed up my statement with the ars technica example. I could have used DistroWatch, where the owner, Ladislav Bodnar, posted a story a few months back about how, finally, more that 50% of his vistors are running Linux. Numbers from his site are equally invalid and skewed in favor of Linux for obvious reasons. For a web counter to be accurate it has to count across a large variety of sites throughout the world on all sorts of topics, not just a few paying commercial clients.

Not only does my statement make sense, I would think the logic is obvious. I actually see your comment as another smoke screen response seeking to dismiss rather than providing contradictory data of your own.

Ars Technica publishes OS stats for their site, in addition to browser stats. Their OS stats have Mac around 27%, Linux around 6%, so even at a site that you used to illustrate Linux does better at technical sites, it is a distant third.

There is an implicit assumption in your web counter argument that Linux users are less likely to visit general purpose sites than are Windows users. Do you have any support for that? The Linux desktop users I know use Linux when they want to buy a book online, or read a news story, or check social networks.

There is no implicit argument in what I wrote about web usage. The problem is that the numbers generally come from companies paid to generate web stats and produce results that are often predetermined either by the selection of sites or by the people paying the bills.

The most common source of the bogus 1% number is the NetApplications Survey. Bruce Byfield wrote an excellent piece on just how dubious that number is last year for Datamation. Please see:

The 6% number from ars technica is probably the very bottom of the range I find believable for Linux usage. Please remember again that many browsers, including my own, are set to deliberately misreport running Windows to allow access to websites that require it.

(I'm a Ubuntu user)

As much as I'd like to believe what you are saying, your argument is flawed.

Many (most?) free analytics companies publish their statistics. For example, statcounter has

The wikipedia stats previously posted are another point:

You've only posted one piece of evidence to try and defend your claims: that Dell sold 30% of netbooks with Linux on them. As others have pointed out, that was for one month only, and on it's own isn't enough to change the statistics significantly.

I think a retraction is in order.

A retraction is certainly in order. You have completely misrepresented the information in my article. I'll gladly accept your apology and retraction whenever you care to give it.

First, the primary piece of data I gave for netbook market share was from ABI Research, not Dell, and it covered a full year. It was 32%, not 30%. Second, the Dell data was not for a single month. Where on earth did you get that from? It is significant as it confirms ABI's findings. I also provided additional supporting data from Microsoft which you conveniently ignore.

If you bothered reading my replies to flawed comments like yours you'll have noted that I linked to other articles that quoted older numbers from ZDNet and IDC that also refute the 1% number by very well respected tech writers. I also explained at length why those web counters are an absolutely unreliable way to judge market share both in the article and in the comments.

So, yes, I'm ready for your retraction now.

Dear Caitlyn,
I very much enjoyed your analysis. However, your opening quote from William James seems to have been reorganised and a "not" left out in the process. It now says that repeating an absurdity will not get people to believe it, with the further connotation that somehow more absurdity should lead them to believe.
As to why Microsoft fans fear the acceptance and growth of Linux, it is that they do not want to lose MSWindows. The growth of Linux on the desktop means that one day MSWindows will be gone, for it is dependent on the vicious cycle "everyone uses Microsoft because everyone uses Microsoft". There is a day coming when "nobody uses Microsoft because nobody uses Microsoft". This phenomena is called the network effect, where the value of something is dependent on the number of users it has.

With all due respect I read the quote again and it's fine. The usage is not modern but the meaning is clear. Take Lenin's version if you prefer: "If you repeat a lie often enough it becomes the truth."

Hi Caitlyn! I would like to think that Linux use is above 1%, but I just don't see it. Using popcon for example ( the top packages only have roughly 1.6M installations. My my own statistics from watching source forge there are (roughly) 10 Ubuntu users for every 1 Fedora user.

If Balmer considers 1% to be 300,000,000 then I don't think we are close to cracking 2% much less 10.

I could be wrong though.

300,000,000 is 100%, not 1%. That would make 1% roughly 3 million. A total figure of 30 million (or 10%) seems very reasonable to me. Gartner, which is not exactly considered friendly to Linux, put Linux at 5% back in 2005, before the introduction of netbooks which boosted the numbers and led to wider adoption.

Also, Fedora claims to have a larger user base than Ubuntu. I don't necessarily believe that to be true but I find the 10:1 number equally hard to swallow.

Hmm, he says "And a point of market share on a number that's about 300 million is interesting" I guess I read it wrong the first time, and stand corrected on that. Still, 1% of 300 million is 3M which Canonical implies they have far surpassed but they haven't released any analysis to back up their assessment.

I think the opposite is true concerning Fedora and Ubuntu users though, Ubuntu seemingly has a significantly larger amount of users than Fedora, but that is a non-scientific estimate.

Look to the World of Goo stats from last year to prove that theory.

I don't seem to recall Gartner putting Linux at over 5%, do you have a source on that? I'd really like to give it a read.

By the way, it was an interesting article and it inspired a lot of thought on the subject, thanks! :D

In 2003 IDC put Linux desktop market share at 2.8%:

I couldn't find the later Gartner numbers with a quick search and unfortunately I no longer have that bookmarked. When I have more time I'll do some more digging. It is out there.

I wish I had time to do more digging now but here is Linux Today and Linux Planet editor Carla Schroder's post on the subject for last year: It seems ZDNet had Linux desktop market share at 4% in 2004 and it has only grown since then. Carla's estimate, at 8-10%, is in line with my own analysis.

I would be extremely cautious about anything that uses "GreyGeek" as a source. In my personal experience with this person I have found that he has no problem bending the truth, making up facts to support whatever argument he decides to make.

As for Mr. Asay, without source data which Canonical refuses to publish anything that comes out of their marketing machine can't be considered credible.

The only thing that I see here with potential is the ZDNet article, but even it doesn't cite any sources.

That said, I would like to see 1% become 10% but that won't happen any time soon. The end product of a Linux "desktop" just isn't mature enough yet.

Honestly, I don't think that we should care if it's 1% or 50% what we should really focus on is making a product that's worthy of 100%. There is still a lot of maturing to do before we get there though.

If you do happen to come across that source again in your internet "travels", email me. :D

Let's agree to disagree basically on everything. I don't see the Linux desktop as less mature or any way inferior to any other. I also think the sales numbers and research numbers I've provided are irrefutable. We haven't been at 1% in nearly a decade and we never will be again.

Fair enough.

> I find the 10:1 number equally hard to swallow.

Actually, the Wikimedia web user stats show exactly that. Even assuming that there is a bias against Linux in web server log statistics, do you think that it applies to Fedora somehow more than to Ubuntu?

Let me put it this way: I do not believe any web counter numbers are in any way even vaguely accurate.

So, what is your explanation for the Wikimedia web user agent stats? They put Linux at 1.88 %, including Android, and Mac OS X at 6.63 %, excluding iOS (iPhone, iPad, iPod). The stats are from June, so they should be current enough, and they've been collected from Wikimedia server logs without any money changing hands. As I see it, if the Linux installed base is on par with the Mac and there really are five percentage points of Linux users missing, the possible explanations are:

* Linux users use Wikipedia about one quarter as much as everyone else.

* Linux users use the web in general about one quarter as much as everyone else.

* Three quarters of the postulated Linux users are spoofing their user agent identification and identifying as Windows or Mac.

* Three quarters of the postulated Linux desktops are in some locked-down corporate setting that doesn't allow web access or at least blocks Wikipedia.

* Wikimedia's log analysis script is seriously borked and the results are bogus.

* Wikimedia's system administrators sold their souls to Microsoft's marketing department and falsified the results on purpose.

I find all of the explanations extremely difficult to swallow.

An estimate put forward a couple of years ago based on Ubuntu and Fedora update downloads and web server logs put the number of Linux desktop systems in the ballpark of 25 million. That actually corresponded reasonably well to an installed base of 1 - 2 % of all desktops at the time.

I've used Linux almost exclusively since 1998 and I know a large number of Linux users at different levels of technical experience through my work. I've yet to see widespread user agent spoofing. If that was to explain the discrepancy completely, I think it there would have to be a distro three times as popular as Ubuntu that had set their default browser to spoof as a Windows browser.

Obviously there is more than one factor at work but of all the explanations you listed the one that is most correct is:

"* Wikimedia's log analysis script is seriously borked and the results are bogus."

Put "web counter inaccurate" into Google and enjoy about a week of reading. I'll also recommend Bruce Byfield's article for Datamation from last year again:

The fact that you find the numbers "hard to swallow" does nothing to disprove ABI Research's results or Dell's claim or any other of the mathematical data that fairly well proves that 1-2% is simply wrong.

You do raise one issue that I think needs some study: usage patterns. Linux machines in the consumer market skew to the very low end, whether it's tablet or those 300-400MHz CPU ARM and MIPS machines that show up on eBay and are popular in Asia due to low price. I expect that with such a limited system or when using tablets the usage patters are significantly different.

> Put "web counter inaccurate" into Google and enjoy about a week of reading.

As far as I can tell, almost all of that stuff is criticism about counters embedded on web pages as JavaScript and depending on the user to download images and such. That's not what the Wikimedia stats are. They are server log analyses. The counter criticisms ever recommend using server logs instead.

There are problems with counting users from a stateless protocol such as HTTP, because there is no information about a session, just hits on different pages. Do you count hits or unique IP addresses, or collate requests from one address within, say, an hour, into one hit, or what? Whatever method Wikimedia chose, it's the same for all of the operating systems. You're asking us to believe that usage pattern differences and spoofing account for a 4x discrepancy. I still find that hard to believe.

What I meant by borked script is that there is a real bug, such as failing to classify a Linux user agent string as such.

> Bruce Byfield's article for Datamation

All of the actual statistics in there and the links point to 1 - 2 % share, and the rest is more or less wishful handwaving.

> nothing to disprove ABI Research's results or Dell's claim or any other of the mathematical data that fairly well proves that 1-2% is simply wrong.

We know that the netbooks get wiped and installed with a different OS in both directions. I've recently purchased a ThinkPad laptop with Win 7 for a family member, wiped it and installed Fedora. No doubt that was counted as a Win statistic. Where are the Dell users and those in the ABI study? These are supposed to be internet devices, but the users don't seem to be on the web, or the usage pattern is dramatically different from other laptop/netbook users. Or do these netbook distros set spoofing on by default? Ubuntu on Dell surely doesn't.

> I expect that with such a limited system or when using tablets the usage patters are significantly different.

Well, Android alone has almost as large a share of the web statistics as desktop Linux, so certainly these devices are being used to browse the net. Isn't that what the tablets and such are primarily designed for in the first place? If anything, they should be over-represented in the web user agent data. Actually, there I could buy the case for spoofing as an explanation, if all these inexpensive devices say they are Windows. Android obviously does not.

With all your repetitive posts it comes down to you trust a web counter and are willing to dismiss the sales numbers I linked and call it all "wishful thinking". It doesn't matter how many respected writers or how many analyses I link. I get it. You've made up your mind and are certain WikiPedia's log analysis is perfect and everything else must be flawed. Sorry, not convinced.

I have noticed that the majority of the objections to my article are a very limited number of people posting over and over and over again. I have to wonder what the agenda is here.

P.S.: You're done. I won't let you dominate the discussion and say the same things over and over again on my blog posts. Feel free to write your own.

Perhaps it's just a matter of Linux users identifying themselves to the internet as a different OS for compatibility. Or the logger fails at appropriately identifying OS's.

People are ignoring a major factor, here, in this discussion -- namely the fact that the desktop OS market is not a "free" market: Microsoft bribes and punishes system manufacturers and sellers to insure that Windows is favoured. Many large distributors decide that Linux isn't profitable enough for them, even if there are customers clamouring for it.

This leads to spectacles such as:

* major netbook manufacturers publicly apologizing for demoing a Linux-powered netbook at a trade fair,

* leading manufacturers claiming that Linux-seekers are out of luck because their product is made with the OSs loaded in certain proportions, so if the Linux models are sold-out -- well, too bad (and then they get rid of their Linux product-line on the excuse that it's not selling well),

* employees of certain "big-box" consumer outlets revealing in their blogs that the higher-ups actively oppose non-Windows systems, because these don't lead to high-profit "add-ons" such as antivirus, bi-annual malware cleanings and "tune-ups" (and service packages to provide the same), additional software sales, etc. This can go to the extent of lying to customers about product availability, if the customer reveals that s/he plans to install Linux on the system,

* "Big Box" vendors training their employees using blatantly biased training materials and programs supplied by Microsoft, that disparage Linux and it's capabilities,

* Software houses make their desktop product available only in Windows (and perhaps Mac OS X), even if they already make Linux versions available to large enterprise as a server-based product.

Every debate about the numbers (and what the numbers mean) is incomplete without acknowledgement of this underlying bias. Every desktop Linux "sale" or installation is in fact made in the face of a pervasively hostile environment in which vested interests actively hinder, discourage and misinform end-users about Linux as a credible option. Linux users have to fight the system, go to extra effort, even pay higher prices (despite Linux OS itself being free). The facts and figures so readily bandied about generally reflect this ongoing pervasive bias.

Yet somehow Linux usage on the desktop still makes steady gains -- and here we are, arguing about just how quickly or slowly it is gaining.

> Every desktop Linux "sale" or installation is in fact made in the face of a pervasively hostile environment

Yes, of course, but this is beside the point. The question is, what are the accurate market share and installed base numbers, not whether that market is fair and competitive.

Wait... the definitive answer to the how much Linux is out there comes from our favorite sweaty, chair throwing CEO?

Wow!! Dogs and cats together, pigs will fly and so on...
its the end of days!!!

As for web hits, Im really not a server guy but I remember one of our guys explaining how the varnish cache works to lighten the load on our server and that its impossible to know exact details because it doesnt reach our server.
I will re-ask next week to have him explain it again but
I do remember him sayint that most requests never touch our server running apache/psp/cms and instead theyre are served through a medium.

> I do remember him sayint that most requests never touch our server running apache/psp/cms and instead theyre are served through a medium.

That shouldn't matter so long as the requests that do hit the server are a representative sample, i.e. that there's nothing in the behaviour of Linux users that makes them more likely to be served by the cache. I find it difficult to imagine such a behaviour, but I suppose it could happen. Linux users are boring conformists that only look at pages that the trend-setting Mac hipsters have already requested and that are therefore in the cache? Maybe so.

Congratulations Caitlyn. You wound up and let loose a real opinion storm with this article!

I don't think we'll see any remotely [i]accurate[/i] figures for Linux for a long long time. There is just too much variability in the way Linux can be implemented---which is why there is such variability and caution in the statistics. Too many Linux machines slip through under the radar and have to be estimated. It certainly is heartening, though to see estimates as high as they are!

I must be in one of the smallest minorities: I run 10 machines: 8 run Linux, 1 runs OpenBSD and one runs FreeBSD. None of them would have been counted as a purchased computer:" they were all built from parts. I connect to the Internet through the OpenBSD machine which runs NAT, so it appears as one machine (and maybe as OpenBSD at that). All my browsers report themselves as something else: Internet Exploder, and Idiot Explorer are two names I commonly use :-).

I see from previous comments, many buy, wipe Windows and install Linux. Some buy pre-installed Linux machines. But I don't and many of my friends and colleagues don't buy, but build too---and then install Linux.

Estimates of Linux use will remain just that for a long time: estimates. But one day we'll be able to say categorically: We've won!

I've known for a long time that the 1% myth is just that, a myth. Either that or visitors to my web sites are just a lot smarter--and therefore use Linux--than visitors to other web sites. The volume of sales of Linux merchandise at my shops convinces me that there are a lot of Linux users out there, and they LOVE boasting that they're Linux users. Rightly so, of course, considering its superiority. I know *I* love wearing my pro-Linux, anti-Micro$oft t-shirts and having bumper stickers on my car.

One thing that bugs the hell out of me is that every time I buy a new computer--which comes with that windoze garbage preinstalled--it goes down in the record books as a windoze sale. Um, no, the MINUTE I get my hands on it I wipe the drive and install Linux.

There is a word for the phenomenon you describe.


A piece of unverified or inaccurate information that is presented in the press as factual, often as part of a publicity effort, and that is then accepted as true because of frequent repetition.

e.g.: The WMD justifications for the Iraq invasion were not fact, but factoid.


Ironically, factoid has been so misused to mean "trivia", that it has itself become a factoid. Talk about recursion!

"That shouldn't matter so long as the requests that do hit the server are a representative sample,"
a representative sample of what?

Of the unconnected PCs , for safety reasons or because, in a developping country -where linux desktops are interesting; saving 20$ is ca 2 weeks of average wage in Madagasikara- (or even in parts of developped countries , sometimes it is better *not* to hire an IT connection if you are more than 2 km from an IT node), connections are inexistant or desespairingly slow?

* What is your methodology for telling, telling and telling net statistics are representative of *every* PC?

* Does such methodoly exist?

* Where is it sourced?

OTOH, how can you explain statistics based on sales are not " accurate market share and installed base numbers"
(most of the people do not know how to install another system(s) than the one they bought -and money movements are real ones, not credos, unless sellors are crooks!- and do not like getting rid of what they bought -would you like burning a 100 US$ banknote-: this makes reinstallation quite a tiny, uninteresting part-

Have an happy new year...

Caitlyn, about the Ballmer article you mention. In the pie chart Linux is above MAC OS X (which you put at roughly 10%) so, Linux should have over 10%. The problem is that Linux + MAC OS X account for only ~10% of the chart (together).
Can you see a problem in your maths?

The pie chart is about "home and business PC's", I don't know if servers are included in the "business PC's" category.

"I don't know if servers are included in the "business PC's" category."
Unless Micro$oft is very anticonformist, business PC are ordinary PCs (not that expensive) for professional use. Servers and workstations do not interest M$, and are expensive hardware.

And often professional PCs are *not* connected to the web, which makes web counter statistics

Suppose the police were at the mercy of any script kiddie? of any virus?

FYI : for security reason, the French police and gendarmerie are likely (for gendarmerie, I am sure : they are military and obey) to work under GNU linux; that makes about 150 000 PCs (I supoose every secretary/officer has a PC, and there is a PC for 2 patrol men ; numbers are taken from and I suppose there are 30 E6 PCs in France: that makes
............0.5 %
of professional PCs which are hidden (I am sure, for the gendarmerie, the police is more messy) from the web...And, that, for ONE profession (people working for paying wages have two PC, one to surf limited parts of the web and to email and the other for their work).

There is no error with my maths and I never claimed Steve Ballmer claimed MacOS had 10%. Twist my words all you want the fact is that doesn't change the fact that Linux has been way over 100% for many, many years. Yes, Mr. Ballmer under represents the competition on his pie chart. I wouldn't expect it to be otherwise. What I said, if you bothered to read and stopped simply blurting out opposition soundbites, is that MacOS market share was reported to be about 10% at the time and Mr. Ballmer showed Linux with a larger slice than MacOS. Is that so hard to understand?

Caitlyn, about the Dell mini 9 - 30% Linux article. Dell mini 9 started shipping on september 16th 2008, the article was written on february 23rd 2009. I don't think that statistics for a few months have ANY relevance.

I'll just carry on debunking your debunking.

I've seen earlier in the posts you were saying that you would like Linux to have that much market share. Well, if you'd like that, maybe you'll stop criticizing articles that put Linux in a better light. There are enough people saying that Linux isn't good because free. We don't need others bashing the people who are trying to help it.

And if you don't see this article helping Linux, I'll tell you one thing: I wouldn't buy a product that has 1% market share but I might consider buying one that has 10% market share.

Why does it matter if the figures are skewed? All estimations can be wrong. That's why they're called estimations.

Have you ever noticed that when Apple shows the iPhone market share, they do that only for US? Of course, because in Europe they don't stand a chance. The big companies always show data that's in their favor. Why wouldn't Linux companies/fans do the same? Seems only right.

I don't see how your "debunking of the debunking" could benefit you or Linux. Seriously! So, for a change, why don't you start being more supportive towards the people that are trying to improve the image of Linux.

I'm from Romania too and I do think the estimation is good.

I don't understand people that bash free software.
I mean... IT'S FREE, why would you say bad stuff about it?

"I mean... IT'S FREE, why would you say bad stuff about it(i.e. free software)?"
Well, if it is buggy, why should not I feel FREE (sorry for the case) to protest? Unless you can prove that buggy software is superior to human beings...

"I wouldn't buy a product that has 1% market share but I might consider buying one that has 10% market share."

If I want to buy shoes, I do not look at their market share, as my feet have an unusual length...

There are two issues:
* is the 1% maket share a reasonable figure? It is not linked with linux future, but with (absence of) methods to have reasonable estimates.
* will Xtyn "debunk the debunking" : that might take **years**, as Ms Martin is wise enough to write that "there really is no way to get an accurate measure." (of usage, at the world's scale -not at Romania, my office or US scales, the world's one-)

Nota : As some professions cannot be connected to the web for security/infrastructure reasons, measures based on IT stats might be the worse one can imagine: suppose your boss is very paranoiac about security and decides
- every box will be under GNU linux
- there will be no|very limited access to the web
this a bias (among others) "against" linux.

No you won't. You think you can ignore my comment about a handful of people repeating the same nonsense over and over again and trying to dominate the conversation. You can't. You're done. Further comments by you will be deleted.

Once again: the Dell numbers were not just about one model of netbook and have been repeated. If you followed my previous articles on the subject you would have found where Todd Finch of Dell repeated the fact later. Of course, since you are taking something I used as evidence that the ABI Research number made sense and ignoring both the ABI number, which was for a full year, and subsequent Dell statements, you are cherry picking and ignoring evidence that simply isn't convenient for you. It's disingenuous at best.

Like I said, you're done. If you want to repeat nonsense over and over again do it on your own blog.

If you look carefully at that picture of this site will notice that Ballmer & The Boyz were estimating Linux marketshare about 6-7% and Mac about 4-5%. Their study (made by who?) was 1½ years ago. My guess for Linux is about 5% and for Mac globally 4%. Note: Mac could be big name in US but not ouside The States.

Once again: I never claimed Steve Ballmer claimed MacOS had 10%. Twist my words all you want the fact is that doesn't change the fact that Linux has been way over 100% for many, many years. Yes, Mr. Ballmer under represents the competition on his pie chart. I wouldn't expect it to be otherwise.

You say that you "believe" that Linux is around 5% but provide no evidence to support that other than the Ballmer chart and what you believe it represents. I've posted numbers by a research firm backed by a computer manufacturer that pretty well prove that it can't be below 6% and is almost certainly considerably higher.

I have yet to see anyone refute my numbers other than by trying to post flawed web counters and posting their own beliefs. Just saying "it isn't so" doesn't mean it isn't so. Discounting one piece of my evidence doesn't allow you to ignore everything.

I'm a Linux user. Or am I a windows user?

My home PC(s) all have a Linux based OS installed on them and use Firefox. But the PC provided by my employer is a Windows PC and I have to use IE.

I pretty much browse the same web sites during the day at work as I browse at home - apart from those that my place of work bars - Hotmail, YouTube, Facebook etc.

So in web stats terms, what am I, Linux or Windows?

Obviously I am both. Does the fact I use both essentially cancel out my addition to the statistical data? To web counters I am two unique instances, but I am only a single individual. If I use my phone to browse the web I am something else again.

My personal preference would be Linux if my employer would provide that facility, which obviously it won't. No large corporate or even government organisation is going to risk moving to Linux on the desktop with all of the implications of staff training, support staff avaialability and cheapness, application availability, vendor support issues etc.

Linux on the desktop is very much ready, but corporations - the ones who spend the real money - are not ready for it. It's a shame, but that will be the case for a long time.

My point here is not to refute (with evidence) any of the numbers more to highlight that we really can't rely on any of the data as it can all be skewed in some fashion.

And what does it matter whether Linux adoption growth can be statistically proven or not? I know in my world it has been growing since 2007 when I jumped to it. To me that's what matters.

I care not about Windows any more. I care not who uses it. I care so little about it now that I don't have enough knowledge of it (anything beyond XP) to even criticise it.

So I do wonder... why does anyone else who has made the effort to move to Linux and realise its obvious benefits really care about its adoption rates? Or care what Windows OS users and the vast corporate entities that perpetuate Windows are propagandising about Windows?

Who cares?

While your comments are mostly good there are two points I must disagree with. First, when I consulted for Red Hat I learned first hand that there are large corporations who have made the move to Linux on the desktop. It isn't just a handful either. Large corporations may not want to publicize such a decision but there are times when the cost benefit analysis definitely comes out in favor of Linux or business needs demand Linux. It happens more than some might think.

Second, the constant shouting of 1% is FUD designed to make people think Linux is a risky or odd choice when it is nothing of the sort. As such, it does need to be debunked. Bruce Byfield and Carla Schroder did it long before I did, as have other writers. Sadly until this myth, or more precisely this lie, is no longer believed articles like this one will be necessary.

The stats don't lie...

So, speculate all you like, but according to who's using the Web, 1-2% of the users are GNU/Linux users, netbooks or not.

"TThere are three kinds of lies: lies, damned lies, and statistics." --Benjamin Disraeli

I have already explained at length why the web counters are wholly inaccurate, as are your numbers. Your numbers do not count who is using the web but rather count a subset of what is reportedly being used on selected sites. I say reportedly because my system reports Windows and Internet Explorer even though I am using Linux and Firefox. This, sadly, is necessary for me to access certain sites. The web counters often select in favor of Windows based on the sites they use.

So... you've ignored my article and all the comments and posted meaningless numbers that do in fact lie.

Unfortunately you have made some very critical mistakes in your analysis in how you are using your information sources.

The Dell quote you reference about the percentage of linux netbook sales is from Feb 2009. It is inappropriate to extrapolate that forward in time to be indicative of sales for all of 2009. That qoute is most assuredly about their performance through 2008. What's your reasoning for relying on a Feb 2009 quote to imply performance through all of 2009?

The ABI Research estimate of 6% you refer to does not include the last quarter of 2009 which saw the introduction of Windows7. I'm really not sure its appropriate to extrapolate estimates that do not factor in the market disruption that Win7 caused and the impact it had on the Christmas shopping season. Purchase volume has seasonal variability does it not? Historically how large is the sales volume in the last quarter of any year? If you don't have a data source that can factor the last quarter of 2009 including the introduction of Win7 (which is now the dominate netbook operating system of merit) its difficult to find your analysis as credible as you would desire.

At this point isn't there be real post-mortem market analysis available that includes data from the last quarter of 2009 so as to provide a comprehensive picture?


Actually, no Jeff, I have not found any reliable data for the whole of 2009 or for 2010. If you read the comments you will see that I do have data for 2003, 2004, 2005 and 2008 that strongly supports Linux desktop growth and the numbers. Unfortunately I can only use the best published data available. Unlike some of the people defending the 1% number I am unwilling to go with my feelings or beliefs or use numbers which can easily be debunked.

Anyone who does an honest bit of market evaluation will come to the same conclusions I did. Of course, there are huge forces out there that prefer to see Linux dismissed and derided. I don't think there is any flaw in my analysis. Like all those throwing stones at this article you have utterly failed to provide any convincing data to counter what I have.

Okay, doing a little more backlog reading to try to find any newer information sources and I've come across this.

Why are you recycling articles months later with new headlines?

You wrote an article based on the same information in Dec 2009. Now, several months later (I say several because I can't actually count on my fingers to get the right number of months while typing) you write essentially the _same_ article with the _same_ thesis using the _same_ information sources to back up your argument.

C'mon. At least try to bring newer information to the table.

With that said, I look forward to reading your next article in your 12 part series on 2009 linux netbook sales in Jan 2011.

The Windows FUDmeisters and all of those who believe the big lie publish articles day in and day out repeating the 1% number. An article every once in a while refuting that is worthwhile.

Despite your claim of recycling the December, 2009 article was about netbooks only. In this article I used netbooks as one piece in a much larger puzzle: the desktop as a whole. Are you saying that data, once published, can only be used once? I can't think of a more ridiculous assertion.

Despite your claim to the contrary I wrote an entirely different article using the 2009 data as one element. I could accuse you of lacking in reading comprehension but I think a more likely explanation for your comment is that you are someone who came looking to ridicule or criticize and decided to be highly selective in picking on just one piece of a much larger whole. I suspect most of my readership will see right through that. Nice try, though.

Refuting bad data with more bad data is not worthwhile.

That is precisely what you did with your miles long rant, so I took this comment as permission to delete the other one. Repeating yourself over and over and telling the world my data is bad or or my analysis faulty is something you can do on your own blog if you wish. Saying that my data isn't good enough does not refute anything.

I refuse to let four or five people post the same thing over and over again in an attempt to overwhelm or dominate the conversation.

The following is from a parallel discussion on at It was posted by golem and I do not have a verifiable source to confirm it:

"Going back to a presentation by Ballmer, he stated that Windows has a market share of 83% in the US. I believe that was on the occasion of the release of Windows 7. If as you say Linux and OS/X each have about 8%, that would be quite consistent with Ballmer's 83% figure for Windows."

83% market share is very consistent with the Linux market share of at least 8% and that MacOS has a similar number.

There's always some kind of dependency on windows, for any of you who like to play games or use adobe or office applications.

It's rare for where I live to find places that sell netbooks for a lower price because they have linux installed.

No doubt there are people who'll wipe the disk and install ubuntu on it as soon as possible.

Regardless, one can always dual boot.

Really? I know way too many businesses and individuals who have no dependence on Windows at all. I haven't found a need to run it for years. Office applications? What is missing for Linux? Oh yeah, the Microsoft branded ones? There are plenty of others out there, you know. Ditto alternatives for Adobe's offerings. There are also plenty of games for Linux. The dependence on Windows is only there if you want to be.

8%? Where did you get that number from?

Anyways your propaganda has already been debunked at binary playhouse:

If you read the article you know I got the number from well documented sales figures, Your blog post debunks nothing. It starts out by saying I'm "at it again" and provides no data other than web counters which are notoriously inaccurate to support the 1% fairy tale. Debunked? With no facts? Ignoring sales figures? Ignoring the meat of my article? I don't think so.

I looked at the blog. It's a Windows fanbois blog. It dismisses Linux on the server too. It belittles Steven J. Vaughn-Nichols. I see I am in very good company indeed.

Playhouse? Yep, the children are at play all right, spouting all the silliness children always come up with.

That is another meaningless web counter. Susan Linton over on Tuxmachines found one that puts Linux at 14% on an adventure sports site: I consider that meaningless as well. Unless you can tell me who they are counting, who they are not counting and what their methodology is it becomes yet another number that has little or no validity.

Sadly, I am going to close comments at this time. The comments I found this morning had devolved into name calling, foul language, and screaming "Linux zealot!" Clearly I allow opposing comments and allow everyone to enter civil discussion.

Some people want to add heat to the discussion without any light and now they are the only ones posting. The idea that a market share number so anger people and the fact that they get so emotional tells me that there really are people who find Linux threatening or who simply don't like their preconceived notions challenged.

News Topics

Recommended for You

Got a Question?