Attempts to Count Linux Users Remain Pointless

An anonymous reader writes "A great deal of attention is paid to numbers, but rarely does one actually ask what these numbers mean. One problem that many people have been trying to tackle is gauging the extent of use of Free software, including Linux. Questionnaires are not a solution here and neither are statistics, which are usually derived from the wrong data. The following article looks at the various challenges at hand and concludes that the growth rate of Linux is likely to remain an enigma."

words from microsoft:

"Do not attempt to count the number of Linux users, thats impossible, instead, try to realize the truth... there is no Linux" :-P

hmm.

well I skimmed TFA and conclude we can now expect in these comments:-

(1) a lot of foaming at the mouth rants and statistics from Linux evangelists
(2) some distie bashing thrown in for good measure
(3) the inevitable vista comments and hints about massive marketing campaigns
(4) maybe some mention of PCs shipped with Linux pre-installed
(5) if we are really lucky maybe the odd referenced fact

.. and nobody being better informed at then end of it.

Re:hmm.

You're way off base. All of the people in my department at work run Linux, so clearly Linux is already dominating the desktop. The fact that my department is made up of entirely Linux sysadmins should not take anything away from this single statistically relevant sample. Of course, we all run Red Hat because Gentoo is for masochists and Ubuntu has a stupid name. Sure, Microsoft's brainwashing^Wmarketing may lead you to believe that Vista is all the rage, but everyone knows it's a memory hog that barely runs on most supercomputers. Especially now with Linux being pre-installed on so many desktops, Microsoft is bound to go bankrupt any day now.

Also, did you know that the longest recorded frog jump was 33 feet 5.5 inches [exploratorium.edu]? Amazing!

Re:hmm.

Speaking as a person who has to go out on site as a consultant and recommend the best product for the job (regardless of personal bias) - and yes, that includes the places that are MS-only where i find myself saying "yeah, sql server 2005 is briliant", or "yeah, you should really get ms systems server op's center". I do believe things are changing somewhat. It started about 6-9months ago, i'd go out to a traditional windows company and they'd be running a samba server with apache where there used to be IIS and win 2k3.

Then a friends company (not a small one mind you) went off to do a linux-on-the-desktop study as alot of their windows agreements were about to become eol so to speak. At first I thought this was a bargaining tool to get cheaper software, but I was surprised to find that not only was it about replacing the desktop but also the server side functionality. It turned out they'd started looking at linux desktops because they'd managed to gain some linux servers to replace most costly machines (some windows, but alot were aix or solaris) - interestingly, alot of the now-linux server hardware are sun x86'ers running centos. As a result they took on some linux types to administer them, and it grew - they replaced a few non-essential file servers. changed a few mail gateways to linux. Moved proxies to squid. As their CTO put it "i was suddenly surrounded by linux and didn't realise it until i looked at the balance sheets, all we are paying for is hardware and alot of the things we are using linux for are internally grown and maintained. I started to think we weren't paying for licenses were we should be". One of the things that did take him by supprise is that half his IT department by this time had switched to a linux desktop and used mail thru imap or some such (some were using windows still thru vmware player or from a terminal server running outlook). Apparently if you pxe boot off alot of the networks, you'll get a pxelinux menu that allows you to boot various things like dsl or install a customized ubuntu (though i didn't see that myself). I know they're also running some systems with RHEL too because they "feel good" to know they have support.

To sum it up, i was quite shocked. 12 months ago I was feeling "unix was coming to an end" and feeling quite disappointed by that, but I feel quite elated by what i've seen lately - Especially so in Australia where linux has had a really tough time of it.

Having said all that, i think the author wasn't just referring to linux users but also the users of FOSS replacements for commercial applications (like open office, gimp, etc). I can't say i've seen a tonne of that myself, but its not uncommon to see things like gaim, firefox, jedit, eclipse - smaller things really.

It will be very interesting to see what the next 12months brings us.

It depends on your definition.

A "Linux user" could be anything from a hardcore Gentoo-compiling mad man of a Linux user to somebody who uses a phone or other device which has embedded Linux. I for one dual boot so for purposes of this attempt at a survey am I half of a linux user? I use several devices with embedded Linux distros so am I 80% Linux user? Does the device need to be capable of browsing to a webpage or (as is cliche on /.) does it just have to run Linux?

Re:It depends on your definition.

A "Linux user" could be anything from a hardcore Gentoo-compiling mad man of a Linux user to somebody who uses a phone or other device which has embedded Linux.

A point that is not actually made in TFA. I was talking with my father-in-law the other day, and we were discussing my software-engineering job, and that I use Linux preferentially simply because it's so much more reliable and "commercial grade" despite it's being free.

He announces to me that "Well, that's all fine and dandy, but I'm never going to bother learning that...". So I pointed to the Dish DVR under his TV and the Linksys router next to his Windows PC, and indicated that he was already using it more than he was using Windows!

This is a point that TFA didn't cover at all. The desktop is losing its dominant position.

Firefox

I'm going to offer the same solution I did for counting Firefox users:

1) Require a national ID number to download any Linux distro, and validation of ownership of this number through an in-person meeting with the local authorities.

2) Have the software "phone home" that it's actually being used, when it's used.

3) Close the source so that 2) can be facilitated.

4) Made the ID numbers and contact information in 1) publicly available so anyone can audit the official count of users.

There, done, you've got everyone counted. Wasn't that easy?

Re:Firefox

How about we just ask for a show of hands.

Not possible

As long as you can get linux from kernel.org compile and make your own distro, download from a myriad of distros, multiple installs both in hardware and in vm's, and people single people using multiple versions it's really not possible to get a valid number on how many computers are actually running linux.

Plus are you talking about just Server/desktop? If you count the millions of embedded devices that run gnu/linux I'm sure it would be considered the worlds most popular OS. It's all in how you want to swing the numbers.

Well, duh.

It might not be entirely pointless to try, but I'm reasonably convinced of two things: I don't care (and don't need to) about the exact numbers, and it's growing.

I don't care largely because the software meets *my* needs. That's the most important thing to me. An assurance that it will continue to do so is also nice, and there are clearly a lot of people developing for it. I'm not worried on that front. People who have a big investment in *other people* using Linux (especially when said other people aren't developers) confuse me. (Well, except when they're trying to sell Linux software / services.)

It's growing. I can't tell you how much, but I can offer the anecdotal evidence that the responses I get to "I run Linux" have changed over the past few years. It's not always "What's that?" anymore. It's not uncommon to get questions about it in response -- people want to know how well it works, whether it runs the same software as Windows, etc. I just answer their questions and am polite and friendly about it.

couldn't you just

Take a sample of 10000 people / companies.
Ask them if they use Linux of not
Extrapolate the results.

Seems to work when there counting all kinds of other things that don't have a direct method of counting them.

Ah, an Onion-esk headline

That's a good way to start a Monday :-).

Actually, it's not so much that they are pointless - just that they are useless. There is a point to knowing how many Linux boxes are out there (demographic studies, confidence in support longevity as a function of install base, etc.) But most known techniques for counting remain useless.

To be honest, this might be just as well. Any technology that COULD count successfully all the Linux boxes out there would be a bit scary - many people probably don't WANT anyone to be able to know what they are running. (OK so nmap can probably figure out anyway...)

Large scale counts like this are a difficult proposition - the only things that approaches being successful in this respect are probably automobile registration systems, census systems, and the tax system - in other words, massive systems with compulsary reporting for every existing component member.

Now, of more interest might be to work with the BSA for a while (or someone else who has the authority to open random IT doors at random) and do an anonymous study of deployment percentages at random under guise of a random license check or soemthing. Probably (hopefully!) not legal but it would be a way to get statistically meaningful results if the sample was chosen well.

For the Bogglers

For those boggling over WHY this matters, try and keep in mind that Microsoft, Apple, et al provide these figures regularly. Whether or not they're valid is a source of debate, but some kind of numbers are out there. This is how we get to say things like 'Windows is 90% of the market', etc.

Perhaps we need a 'BeCounted' daemon that merely tracks the stats of those that would like to be counted? It would still be a fraction, but if that number were out there we'd at least have some kind of data point to discuss. Perhaps FSF or GNU or some other party would host the servers that collect the data? You could even make the thing multi-platform, reporting on specific apps, and providing other useful data and pitch it to Google and company. Not that they're not already tracking this in their own apps, but this would be OSS. You could have all sorts of opt-in/opt-out toggles for it and it would be transparent as to what it tracked. You could also have it gather from different places and homogenize the data after it was submitted. The possibilities abound.

Maybe there already is such a creature? If we supporters of Free-with-a-capital-F want to be relevant moving forward, a detailed head-count could certainly be a step in the right direction.

I suggest we do it like the MPAA and RIAA

Take a leaf from the MPAA and RIAA and extrapolate the losses from Microsoft's profits.

nmap

duh! nmap -sO 0.0.0.0-255.255.255.255

Two

World domination is at hand!
 

Re:Start counting here

Anyone else? Or shall we approximate the linux userbase size as being "1"?

Re:A good slashdot poll

and we'll conclude 18% of the world uses the CowboyNeal Microthreaded Kernel fork of Planet Nine

Re:Start counting here

I, for one, use GNU/Linux and only F/OSS

Where shall we mail your trophy?

Re:Start counting here

That's the whole problem. If I count home and work, I have more than 20 linux "servers" of which one is a cluster containing twice that many machines. But I'm the only person who uses them (well, the only person who directly uses them, e.g. logs into them), so really, in terms of users that's just 1...You can't count everyone who goes to a webpage, or uses a bind, ntp, samba, squid, etc service to be a linux user.

That's why it's hard to count. Windows users are easy: it's almost all 1 to 1. I have 1 windows machine, so mark me down for 1 in the windows category as well. You can be even more specific and count windows licenses; this is misleading...My workplace has a great number of unused windows licenses...But it's a good number with documentation behind it, whereas linux can only count support contracts with big linux vendors.