Unscientific Linux Popularity Contest

Introduction

Have you ever wondered which Linux distro is the most popular? Many people will point to the statistics at DistroWatch.com. These statistics are generated by the number of "hits" for each distro page at DistroWatch. Unfortunately, this allows the possibility of "ballot-stuffing" by making multiple visits to distro page that you like.

Therefore this data cannot be considered scientific, and it is has a high probability of being inaccurate. Even so, it still has some value. We can tell which distro has the most (real visitors + "ballet-stuffers"), which will roughly correlate to the actual popularity of the distro.

Here are some interesting long-term trends:

2002: Top 5 Distros
  1. Mandrake
  2. Red Hat
  3. Gentoo
  4. Debian
  5. Sorcerer
  6. Suse
Comments: I have never even heard of Sorcerer Linux! Year 2002 statistics could be very inaccurate since DistroWatch.com was not well-known at the time. Also, I included Suse in position #6 so that we can see how it fares as the years go by.

2003: Top 5 Distros
  1. Mandrake
  2. Red Hat
  3. Knoppix
  4. Gentoo
  5. Debian
7. Suse

Comments: Knoppix makes a huge splash as it enters the scene at #3. Knoppix was the first distro to introduce the concept of a Live-CD, which is still very popular today (used in the Ubuntu install CD). Many people saw Linux for the first time when they booted up Knoppix.

2004: Top 5 Distros
  1. Mandrake
  2. Fedora (was Red Hat)
  3. Knoppix
  4. Suse
  5. Debian
13. Ubuntu

Comments: This is the year that Fedora replaced Red Hat. This also marks the third year in a row that Mandrake has held the coveted #1 rank on this list. Note that a brand new operating system which was introduced in Q4 of 2004 still managed to achieve a rank of #13 for the year.

2005: Top 5 Distros
  1. Ubuntu
  2. Mandriva (was Mandrake)
  3. Suse
  4. Fedora
  5. Mepis
Comments: Ubuntu shines in all its glory as it manages to beat Mandrake (now known as Mandriva) and take the #1 position. This is also the first year that Suse has beaten Fedora or Red Hat.

2006: Top 5 Distros
  1. Ubuntu
  2. OpenSuse (was Suse)
  3. Fedora
  4. Mepis
  5. Mandriva
Comments: Suse was renamed to OpenSuse, and it still manages to keep its lead over Fedora. Mandriva is dropping fast as it is beat by the new-comer Mepis. The success of Mepis was probably helped by the fact that it is based on the number #1 distro, Ubuntu.

Statistics from Google Trends

There is another source of statistics that we can use to determine the popularity of Linux distros. Google Trends allows us to see the relative popularity of search terms, based on a sample of total searches. Here are some fun ones:




Statistics from Alexa

You can also play around with statistics from Alexa. The data from this site comes from users who have installed the Alexa toolbar. We can now find the relative popularity of websites by querying this data. Here is a query that I did:


Summary

I hope I have demonstrated something that is useful to you. These tools can be used to determine the popularity of almost any topic. Just keep in mind that they cannot be considered accurate, but they do show relative trends. Have fun!

Comments

  1. Its too bad you don't give distrowatch.com the credit they rightfully deserve in your post. Yes the numbers on the ticker can be fudged but I can remember going to distrowatch right after its first publishing and it just blowing up in the following months, it is an incredible site that still today remains quite unbiased towards the different distros on the market. I still read distrowatch weekly regularly and at the time it was one of the only sites that had quality news for open source os's compared to today when linux has every poser organization claims they are the news site for the masses. At my university (EWU, very small school) we used it almost daily for tracking our currently installed distribution's default packages and new up and coming distributions onto the market like sorcerer. Also on a side note sorcerer linux was a very popular alternative distro back at the beginning of this decade with an amazing source based package management system that unfortunitly didn't catch on outside of the core distro. I used sorcerer at home, work and school for about two years alongside freebsd and debian. Anyone who compiles their operating system should try sorcerer atleast once.

    ReplyDelete
  2. Your post is interesting, but has no real significance regarding the popularity of the linux distros, neither does distrowatch, and that's the real problem.

    The "bursts in popularity" of Knoppix, or Ubuntu then, and then OpenSuSe, were just due to increased marketing all over the web at that time.

    We really need a poll that take into account IP connection correlated with one unique login and password, and ask people to vote, and verify that they only vote once.
    I'm pretty sure we'd be surprised to see the results...

    This is the only way how this discripancy can be solved.

    (and not like the pseudo-poll we saw on tuxmachine recently, where PCLinuxOS was considered as the most favored distro..... what a pity some believe in that kind of sh.t).

    ReplyDelete
  3. if you check out distrowatch's faq, you'll see that they do, in fact, take into account continuous page reloads by the same ip.

    /kubuntu user

    ReplyDelete
  4. "if you check out distrowatch's faq, you'll see that they do, in fact, take into account continuous page reloads by the same ip."

    Still, it doesn't mean anything, like I said, since their method is based on "how many people visit a given distro web site"....
    which doesn't mean anything in terms of users preference for a distro in terms of usage, but rely mainly on marketing and ads on the web.

    /kubuntu and Mandriva user too ;-)

    ReplyDelete
  5. I think that Distrowatch mainly shows what distributions people are _interested_ in, not what OS they _use_. Does anyone ever visit Distrowatch after becoming a happy user of a particular distro?

    ReplyDelete
  6. Someone just pointed me to this:
    http://www.google.com/trends?q=+windows+xp%2C+virus&ctab=0&geo=all&date=all
    Isn't the correlation interesting?

    Now wait until you see this:
    http://www.google.com/trends?q=ubuntu%2C+virus&ctab=0&geo=all&date=all
    Is this proof that more interest for "Ubuntu" actually means less interest for "virus"? ;-)

    ReplyDelete
  7. @JanC
    Your form of reasoning really fascinated me. You are actually comparing apples with oranges!

    ReplyDelete

Post a Comment

Popular posts from this blog

Using the Cisco console in Linux

Linux NIC teaming recommendations

What it takes to make Ubuntu ready for use