How many Linux users are there?

[NedSlider]

Those who know me know that I'm interested in stats. Like how many Linux users are there, how many CentOS users etc...

There are no easy answers to these questions. Even the mirror maintainers at CentOS can't tell us how many users we have. Part of the reason for this is the freedom under which Linux may be distributed - users aren't restricted to getting their updates from centralised resources, they can use any number of mirror sites over a large distributed network, so we have no way to count the number of IPs or machines hitting the update servers.

So I was reading this article earlier:

http://www.linuxplanet.com/linuxplanet/reports/6671/1/

and it mentions the statix project - an open source project to report back the number of Linux users:

http://sourceforge.net/projects/statix

It's an opt-in type deal and reports no personal info.

So, the question is, if the major distros all included some independent (non-distro specific) mechanism to report the number of users would you opt in or not? Would you want to be counted?

[AlanBartlett]

[quote]So, the question is, if the major distros all included some independent (non-distro specific) mechanism to report the number of users would you opt in or not? Would you want to be counted?[/quote]
First answer -- [i]no[/i].

Second answer -- [i]no, not by that method[/i].

But then you know me, so you would have already predicted my answers. :lol:

Now what was that quote? -- [i]There are lies, damn lies and statistics.[/i] :-P (Apologies to whoever I've mis-quoted.)

[ot]I used to wind-up colleagues in the sales department with a modification of the above -- [i]There are lies, damn lies and sales talk.[/i][/ot]

[NedSlider]

[quote]
AlanJBartlett wrote:

First answer -- [i]no[/i].

Second answer -- [i]no, not by that method[/i].
[/quote]

Would you care to expand on that a little?

[gerald_clark]

In my testing, I load 100s of copies of different versions of Linux each year.
These are not permanent installations, and should not be counted.

[NedSlider]

[quote]
gerald_clark wrote:
In my testing, I load 100s of copies of different versions of Linux each year.
These are not permanent installations, and should not be counted.[/quote]

Yes, hence why presumably you would just not opt in on those installs to not have them counted.

[NedSlider]

BTW, some very approximate (and some guesswork) figures I've seen/come up with:

Estimated total Linux users/installs: 30-50 million

Red Hat has around 2.5 million paid subscriptions
CentOS - estimate around 3 million users/installs

Fedora (all versions) estimates around ~9-10 million from unique IPs hitting their update servers.
Ubuntu also estimates ~9-10 million

If the big two have ~18-20 million between them, then totals in the region of 30-50 million seem reasonable.

Another figure I read around the time of the release of Firefox 3 in the browser wars is that the browser market was an estimated 700 million (extrapolated from an estimated 140 million Firefox users with a 20% market share) and growing (China being the fastest growing nation with an estimated count of 500 million alone by 2012). So if we take 30 million Linux users in a total market of 700 million we see a 4.3% market share. If Linux can break into the rapidly expanding Chinese market then there's the very real potential to break into 10% market share territory and beyond as this is by far the fastest growing market segment (China has already overtaken the US for the most number of Internet users).

[MatsK]

2% [url=http://www.w3counter.com/globalstats.php]http://www.w3counter.com/globalstats.php[/url]

0.8% [url=http://marketshare.hitslink.com/operating-system-market-share.aspx?qprid=8]http://marketshare.hitslink.com/operating-system-market-share.aspx?qprid=8[/url]


/Mats