Slashdot Mirror
Is Linux's "Overall Market Share" Statistic Meaningful?
ruphus13 writes "Linux recently achieved 1% market share of the overall operating system market. But, does that statistic really mean anything useful? This article makes the case that it doesn't. It states, 'Framed in the "overall market share" terminology, the information (or how it was gathered and calculated) isn't necessarily questionable, it's more that it's meaningless. It's nebulous, even when one looks at several months worth of data. [How] Linux is used in various business settings answers an actual question — and the answer can be used to ask further questions, form opinions — and maybe one day even explain to some degree what 1% of the market share really means. ... Operating systems aren't immortal beings, and by rights, there can't be (there shouldn't be) only one. ... No one system can be everything to everyone, and no one system (however powerful, or stable) can do everything perfectly that just one person might require of it in the course of a day. While observing trends and measuring market share are important, the results (good or bad) shouldn't be any platform's measure of self-worth or validation. It's a data point to build on (we're weak in this area, strong in this area, our platform is being used a lot more this quarter, where did all of our users go?) in order to improve and stay relevant.'"
What? This directly contradicts the widely-known fact that Linux is The Highlander of operating systems.
To the developers, at least, marketshare is absolutely irrelevant to their efforts. With some exceptions, the GNU/Linux systems is largely built to benefit the developers themselves, and if other people find it useful, good for them.
both reveal some interesting things but may hide the essential."
These statistics seem to a be a bit flawed. Windows has 90% of the market, Mac OS X has 9%, and Linux has 1%. However, Linux is heavily used in servers, handhelds, and other devices. Not to mention, the fact that there is no way reliable way to track Linux installs (100s of dstributions with users installing everywhere and no phoning home to report it).
I don't think this statistic is meaningful. I think Linux should keep chugging along and show the world that freedom, volunteers, and good will can equal money. Something to tip the scales...
The quote from TFA misses the point entirely. It's not about there "being only one," it's about there being enough users to make Linux (or any OS that isn't from Microsoft) a viable alternative to Windows. If a particular OS has 0.0001% or 0.01% or even 0.1% market share, very few developers are going to develop for that OS. You won't be able to connect your machine running that OS to anyone's network, even if it's technically capable of making the connection, because IT will be paranoid about this unknown platform. Etc. But if you reach 1% or more, that's kind of a magic number. You may still be seen as kind of weird for not following the crowd, but you'll be able to use your computer for the same tasks for which everyone else uses theirs.
I'd say 1% is about what any non-Windows OS needs, as long as the aggregate of "alternative" OSes stays above 5% or so, as is currently the case with Linux + OS X. When the number gets significantly below that, as it did in the days before Linux took off and when you couldn't say "Apple" without first saying "beleaguered," things are pretty rough for anyone who's not running Windows on the desktop, using IE for the web, and writing everything in Word.
The correlation between ignorance of statistics and using "correlation is not causation" as an argument is close to 1.
I mean, how is that measured? I mean it certainly must be way more. Do they measure commercial sales of distributions? Well that's certainly misleading. For example I have a laptop which came with a Windows XP license, now it runs Ubuntu. Few Linux users actually buy their distribution and the amount of them has decreased over the years. That would also explain why the market share of Macs seems to be so large. There they simply could count the sold machines.
Measuring the user-agent strings of web-browsers also isn't verry precise as different sites tend to attract different kinds of users.
I don't know, I don't think Linux has a huge marketshare on the desktop, it's probably a few percent in corporate desktops and less than one percent for home desktops. That being said Linux is probably has more total OS installs that WIndows, all the virtual hosts and ubiquitous embedded devices that have been moving over to Linux in droves add up to a ton of actual usage.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
Other sources estimated the number to be 5% or even 6%. Which just goes to show how statistics can easily be used in ways that are misleading or distorted.
But this also bears examining: 1% (or 5% or 6%) of what OS market? Linux is sure as hell a lot higher than that in the server market, and if you are talking about internet servers, higher still.
So, maybe it doesn't have wonderful desktop penetration yet. But I bet it's higher than those statistics say! My bet is that Linux is the secondary OS for an awful lot of people, often via dual-booting. Just as "one and one only" voting has been shown to be inferior to "instant runoff" and other voting methods, saying that people have only "this or that" OS does not present an accurate picture of the landscape.
The overwhelming mainstream demand of Linux is that it become as much a clone of Windows as possible.
No no no!!! Please, if anyone gets anything from this let it be that Linux cannot be just a Windows clone, it has to be something better! Why would anyone go through the trouble of installing a completely different operating system that is exactly the same as the one they have? There HAS to be something extra, even for the average user.
Really, it shouldn't be too hard. Look at what Apple has been doing: they make little applications that draw people in, like photobooth. It is totally silly, and mostly useless, and really easy to make, but I've seen teenagers in the Apple store after school just taking pictures of themselves in photobooth. It's easy to get to and addictive.
Another example is time machine. It is simple, straightforward, and fun to use. It makes you WANT to go buy a second hard drive, just so you can look at the cool animation. Never mind that you've seen way cooler animations in made-for-TV movies, that animation is seductive.
The dock was the same way when it first came out, it bounced when you put your mouse by it. It was fun to play with. It drew you in. Linux needs to draw you in.
And it can. Linux has Compiz, which is graphically the most impressive of any desktop. KDE has some great artists. Now they just need the focus to make Linux sticky, make it draw you in, make you feel happy when you look at the screen. That's what Linux needs to do. Be better than Windows.
Qxe4
Seriously, computing is about more than just desktop users.
How do various hobbyists, and I.T. professionals use Linux? It would be easier to count the niches that Linux is not filling. According to Netcraft, Apache still had over 50% web server market share, while IIS only had 30% in April 2009. I am sure there are some people running Apache on Windows, but I would venture a guess that it is not the majority.
Even webserver market share does not represent the whole server market share. Approximately 40% of all hardware in the server room where I work run Linux in some form, only 25% of all the servers run Linux. There are more than a dozen third party network appliances in this room. Third party examples I can think of are load balancers, spam firewalls, content servers, and NAS filers. I cannot think of one third party Windows Server based appliance in our server room, aside from servers. I am sure there are Windows appliances out there, just not in our server room. If it is part of Microsoft's mission to lock customers in to commodity desktop and server hardware, that is not something that really scales for vendors designing and selling specialized appliances and hardware.
How much Internet infrastructure runs on Linux? I wonder what the percentage of postfix/sendmail servers on Linux versus Exchange servers on Windows is? What is the number of external BIND DNS servers on Linux, versus external Windows DNS servers. What is the market share of Linux iptables/tc routers, load balancers, VPN gateways, or 3rd party appliance running Linux) versus Windows RRAS routers used in small and midsize offices? How many companies are using Asterisk versus the number of companies using Microsoft Office Communicator Suite (Not sure OCS qualifies as a PBX, though)? How many companies are virtualizing their data centers with VMWare ESX, Xen, or KVM, all running on Linux versus Microsoft HyperV?
How many consumer electronics devices have popped up with Linux on them, versus Windows? I can probably name 20 devices with an ARM processor, and some version of Linux running on it. Here is a short list: Linksys Wireless routers, webcams, Tivo, Roku, Netgear ReadyNas, Sony flatscreen televisions, POS terminals, etc. Windows mobile has notably made its way onto mobile phones and Wasp barcode scanners.
How about high-performance computing? How many Rocks clusters, and render farms are built on Linux versus Windows HPC servers?
Seriously who cares if Linux isn't prevalent on the desktop. Linux has filled every other niche, besides the desktop computer, six ways to Sunday. While Microsoft and Apple are laughing at a 1% desktop share, Linux is taking over every other niche which it is able to quickly evolve and adapt. World domination fast, indeed.
/^([Ss]ame [Bb]at (time, |channel.)){2}$/
I agree completely that you cannot place much trust in the percentage, for all of the reasons that get mentioned whenever we talk about OS or browser market share.
The trend, however, is much more interesting because it cancels out much of the systematic bias that will be present in any given series of results.
In this particular case Linux shows a fairly steady increase from 0.43% to 1.02% over the last two years, a compound annual growth rate of about 50% (albeit from a low starting point). I think that's good news.
(In fact the actual figure may be even better than that, because there was a suspicious 25% decline in October 2008. It could be that they changed methodology in some way, perhaps by reclassifying one of the embedded Linux-based platforms, because that month's change stands out as being very unusual.)
It would be interesting to come up with a metric that evaluated "real" work done under each platform. The numbers might be surprising.
"To those who are overly cautious, everything is impossible. "
Absolutely. I want to be able to walk into a store, buy some random piece of hardware and be absolutely sure that it will work under Linux. I don't care how many people use Linux, I just want to make my own personal choice to use it easier.
The question is, what market share is required to achieve this? I'm betting it's fairly low, I mean, even at 1% we are starting to see some traction. Boxes with Linux pre-installed are available from major manufacturers (albeit in a limited and hidden manner), more and more hardware makers are starting to produce drivers or release specs so the community can (I'm looking at ATI here).
If we are getting all this at 1%, then surely full-support can't need a huge amount more, I'd guess at 10% we should be good. How long it will take to get there is another question.
I claimed, and continue to, that linux can be a fine desktop for people who know how to set it up well enough. I personally don't want to invest the time to do that.
Most long time users don't want to fiddle with their machines any more. Been there done that... wrote X11 conf files and modelines, compiled kernels that would actually run their hardware (after getting the missing drivers), wrote window manager rc files... now most of the "old timers" I know just want their stuff to work. Hence the popularity of "ready to use" desktop distributions such as Mandrake, ?Ubuntu, SuSE or any of the less vocal ones. Even with experienced people (not to mention the newcomers of course).
It's really exceptional nowadays that you have to do anything more complicated than add a repository when you need some exotic software. I think I haven't even compiled anything in ages. It just works. And when it doesn't, it's a regular system that's (usually) easy to fix. So I can just do my stuff, process my images, talk to my servers, in a comfortable environment. Works for me at least. To each his own of course.
May contain traces of nut.
Made from the freshest electrons.
Mod parent down.
What's tested here isn't a hypthesis, hell, there's even nothing being tested! This is a measurement of the installed base of different OSes. Now, you can argue about the the way it's measured, but expressing a marketshare as a ratio of the total sample has nothing to do with statistcal significance.
The question is, what market share is required to achieve this?
I doubt that it has much to do with market share, it seems more to be an issue with Linux being pretty much incompatible with how hardware manufacturers like to ship their drivers. Most drivers in the Windows world are not just drivers, they come bundled with a whole bunch of software and stuff that is tied to the specific piece of hardware (i.e. standard Windows Logitech mouse driver is 50MB instead of a few KB). A clean separation between the code that makes your hardware work on all that other additional software doesn't really exit, because the supplied software plays a big part in the marketing and feature lists you find on the box.
I think to get proper Linux support hardware vendors would first need to learn that their job is to produce hardware, not software. Once thats done they might have less problems with releasing specs, but I somewhat doubt that this is going to happen anytime soon because of Linux. The best thing for Linux hardware support in the end are really the open standards. Any USB HID or storage device works on Linux out of the box, not because the hardware vendor cared about Linux, but because he implemented the spec. The more specs we have for common hardware, the better the Linux support will be.