Reports of Ubuntu’s death are greatly exaggerated

(Update 1: There seems to be some confusion about what I’m saying in this post, so let me be absolutely clear from the start: I am not questioning or criticizing Distrowatch’s data.  Their data is, as far as I know, 100% accurate.  What I’m questioning is whether or not this data is a measure of the “popularity” of any given distro, as so many news stories are claiming it is.)

(Update 2: If anybody else wants to run a story about this post, please contact me before making it sound like my blog article is somehow an official Canonical response.  I’m more than happy to have a conversation with you for the sake of accuracy.)

It seems that the tech blogs both inside and outside the Linux sphere have picked up on a graph supposedly showing a decline in the popularity of Ubuntu based on statistics from Distrowatch.  I’m not going to point out all of the flaws in these reports, or the basis of the graph in general, that has been done already here and here (especially in their comment threads).

Instead I want to take a step back for a moment and examine what the statistics are actually counting, and what that actually means for both Ubuntu and LinuxMint.

Hits per day

The numbers themselves come from the number of page views per day on Distrowatch’s page for each distro.  So if you go to http://distrowatch.com/arch you’ve added to the count for Arch.  Now the first thing this tells us is that the statistic is in no way tied to the actual number of users a distro has, just the number of people looking at that distro’s page on Distrowatch.  Now there are three possible reasons why a user might visit one of these pages:

  1. Curiosity about a distro
  2. Following a link from somewhere else
  3. Attempting to boost the hits-per-day count for a distro

I’m going to disregard #3, because I don’t believe that anybody involved with Mint is doing anything underhanded to boost these numbers.  But an examination of the other two will shed some light on what exactly is happening.

Follow that link

Even though a Distrowatch ranking isn’t connected to number of users, it’s still exciting to see your distro rise in the list, and it’s natural to want to tell people about it.  Mint does it, Ubuntu does it, lots of distros do it.  There’s nothing wrong with this, and if enough people are reading your announcement to impact the ranking, then it most likely deserves to be impacted.

But something all together different happens when 3rd party sources start sending people to your distrowatch page because of your rank.  When the Register and PC World run articles about you being on top, their readers will naturally visit your Distrowatch page, further increasing your rank, which will in turn prompt more stories about it, sending more people to your page, etc.

While I have no doubt that Mint deserves the top spot (more on that below), I think the amount of its increase has been affected by this positive feedback loop.  This cyclic reaction will likely continue for a few weeks until people finally get bored with the story, at which point I expect Mint’s numbers to fall back down into the 2500-3000 range, comfortably at #1, but well below the 7728 it’s at as I’m writing this.

The Buzz

All of which brings us back to the first reason for visiting a distrowatch page: Curiosity.  Distrowatch is a great resource for finding out about a distro, and it’s how a lot of young distros get attention.  When Qimo got a mention there, we saw a huge traffic increase, and we also rose pretty sharply in the ranking (nowhere near #1, but still something I was proud of).

But there comes a point, when a distro has become established, where the vast majority of those curious people will be going directly to the distro’s own website, rather than Distrowatch.  Nobody would deny that Red Hat is one of the most used Linux distros, but it currently ranks at #42 on Distowatch.  Suse ranks at #64.  Even the free-as-in-beer CentOS, which we all know is widely used, is only at #9.  As a general rule then, we can assume that as a distro becomes more established and gains more market and mind-share, fewer people will be going to Distrowatch to learn about it.

So what does that leave us?  I like to call it “Young Buzz”, a large amount of excitement about a relatively new (in terms of mindshare) distro.  This is something that absolutely describes LinuxMint.  As the seemingly anti-Unity distro of choice, it has been getting a lot of talk and attention and, while I disagree with the anti-Unity sentiment, Mint is certainly deserving of attention.  Its user base is growing rapidly and I hope its community is too.  They are doing some interesting work with both Gnome-Shell and the Mate, the Gnome 2 fork.  Every other distro will be keeping an eye on them, seeing what gains traction and what doesn’t, and I expect some of that to make its way upstream and into other distros as well.

What does it all mean?

Is LinutMint more popular that Ubuntu?  No, not by any measure I have seen.  Will it become more popular than Ubuntu?  I don’t know, but my gut says not anytime soon.  But Mint certainly has the most momentum, at least for the moment, and will continue to grow at a faster rate for at least the near future.

But they aren’t the first to be in the position, PCLinuxOS had much the same buzz a few years ago, but wasn’t able to maintain it.  The challenge for Mint is to keep this momentum going, and to do that they’re going to need a strong, open, supporting community that gives new users somewhere to belong in the way the Ubuntu community does.

This entry was posted in OpenSource, Work and tagged , , , , . Bookmark the permalink.

29 Responses to Reports of Ubuntu’s death are greatly exaggerated

  1. Galland says:

    I’ve been using Ubuntu for years, happily until 11.10.
    My reasons for having downloaded and being testing other distros (Fedora, openSUSE, Debian, …) are:
    - Unity is a giant step back
    - Gnome 3 is a step back

    • Michael Hall says:

      Unity and Gnome-Shell aren’t steps back, they are brand new paths and we’re just as the beginning of them. One of the biggest complaints I hear about both is a lack of comparability, but neither one has a stated goal of being unconfigurable. So while it may feel like a step back, and while some functionality has been lost in the switch, don’t expect it to stay that way. Both Unity and Gnome-Shell will continue to improve, add features and functionality, and fix the annoyances you currently feel. The Unity in 11.10 is already significantly improved over what was in 11.04, and Gnome-Shell in Gnome 3.2 is much improved over what was in 3.0.

      • Anonymous says:

        Gnome has always had at least an unstated goal of unconfigurability.

      • Bill says:

        “while some functionality has been lost in the switch”

        That’s what everyone means by a step back!

        As of today with GNOME Shell, almost all of that lost functionality is available through Extensions. Unfortunately, Unity does not have that, and is such a big step backwards in functionality.

        • Michael Hall says:

          But with both you also gained quite a bit of new functionality or improved usability, which is why I called it a sideways move. You took one step back, but then two steps forward on a new path.

    • Anon says:

      More of a step sideways. It just takes time for some to adapt, and others may not like it. I always found KDE unbearable, but I did not go around saying that it must be removed. All to themselves, and unless you’ve got a real opinion on the subject, you should not speak up. I like Unity how it is, I find it to suit my workflow the best out of all the choices. So that’s it.

  2. Jorge Encarnação says:

    Nice post. I’ve read the post on omgubuntu before, and although I agreed with the author, the way he wrote (the way I would probably talk about it as well) was probably not the best way to do so.

    You shown a more careful analysis of the data of websites like distrowatch, so kudos for that.

    I also agree with your reply to the previous comment. People keep complaining about both Unity and Gnome shell, but personally I think that Unity is the step that Ubuntu needed to get a spark of attention from the mainstream. Both interfaces look beautiful and I look at them as a step forward towards a more modern interface. Granted that, specially for Unity, that step wasn’t done the correct way.

    Unity was way too buggy on 11.04 and still is on Oneiric, I learned the hard way that I need to wait for the LTS for work machines, or to recommend it to non tech navvy person. I still use it, both at home and at work, but I can’t honestly say I have no issues with it, so I’ve never been as eager for a LTS release.

    I think that the LTS release will be crucial for Unity, things need to be fixed, access to apps menu must be improved etc..

    Sorry for the long deviation, keep on postin :)

  3. cgb says:

    Of course it’s a feedback loop. But that’s how projects become popular. Ubuntu started with a feedback loop as well.

    • Michael Hall says:

      There’s a difference between interest sparking more interest, and page hits sparking more page hits. Mint’s current hits-per-day on Distrowatch are higher than they should be based solely on the interest people have in Mint because everybody is talking about how high their hits-per-day is on Distrowatch.

  4. Jef Spaleta says:

    I look forward to _Canonical_ writing up their _methodology_ on how they _estimate_ Ubuntu userbase. The only entity on record actually making any claims with regard to Ubuntu is _Canonical_. Everything else that is being done is being done in response to _Canonical_ making userbase growth an important part of their laypress talking points.

    Canonical executives and managers have been making userbase size estimate claims annually to the laypress as far back as 2006. It’s well past time for _Canonical_ to explain the userbase estimates they throw around so glibbly.

    As a new _Canonical_ employee who feels compelled to talk about relative popularity you now share the responsibility to come clean with regard to estimation strategy being used by your employer. You can’t stand on the soapbox and be critical of the flaws in other _documented_ estimation techniques when your own employer has _refused_ to document the methodology being used to feed glowing userbase growth numbers to the laypress year after year after year.

    Earn your place on the soapbox, and present a methodology statement for Canonical’s public userbase estimates for public scrutiny. Be thankful that distrowatch feels an ethical duty to describe their methodology for ranking so that you can judge its validity for yourself. Its really unfortunate that Canonical has never felt the same ethical compulsion with regard to their userbase estimates and instead requires everyone to take it on faith that its a valid methodology.

    -jef

    • Michael Hall says:

      My position in Canonical doesn’t give me any more insight or information about user base than you have. It may surprise you, but doing infrastructure web development doesn’t give me many privileged insights. I would also note that this is my personal blog, not a Canonical publication. It was that way before I started working for them, and it’s that way now.

      From what I’ve heard, and this is public knowledge, the best way Canonical has of gauging the number of current installs is by counting the number of unique IP addresses checking the repos for package updates. There was a “census” app at one point that was designed to give an anonymized way of counting individual installs, but it was never included in the default install as far as I’m aware. Neither of these methods is entirely accurate, for obvious reasons, but I don’t know of a better way that at the same time preserves user anonymity.

      • Jef Spaleta says:

        That entire last paragraph is speculation trying to stand up as fact.

        If you don’t know how canonical actually gets its X million of user estimate as presented to the laypress, then just say that…don’t speculate…don’t make it worse.

        And if you want to play the this is my personal blog card we can do that..for now. As an Ubuntu supporter and contributor you don’t you feel that the Ubuntu leadership (a certain Mr. Sbuttleworth) has an ethical responsibility to transparently document and communicate the methodology as part of making any public statements concerning Ubuntu usage growth?

        How would you feel if Distrowatch didn’t provide a documented methodology for their rankings? Hmm? Would you just accept their rankings as credible even if their methodology was undocumented?

        You clearly care about estimation accuracy, or else you wouldn’t have written this blog entry and yet I haven’t seen you call out Ubuntu leadership for a lack of disclosure and transparency concerning public estimation. I guess when the numbers look good, you don’t really care how they are calculated because they say what you want them to say. But when the numbers look bad, totally different story right? Reap what you sow. Being selective in your skepticism is not going to help create a laypress culture that is rigorous in applying theirs.

        -jef

        • Michael Hall says:

          My last paragraph was about what Mint needs to do to keep it’s momentum going. That is my personal assessment of their situation, I wasn’t trying to pass it off as fact.

          Again, I believe it’s been publicly stated in the past that Canonical uses unique hits to our archives to estimate current install base. It’s an inaccurate measurement to be sure, but I don’t think anybody has been anything but transparent about that.

          Distrowatch is showing a very objective statistic: page hits. Canonical is sharing an estimate based a limit amount of concrete data. Again, I’m not entirely sure what information you think Canonical has that they’re not sharing.

          My blog entry wasn’t about the accuracy of estimations, my entry was about the mis-application of the numbers. If people were saying that Mint has trounced Ubuntu in Distrowatch.com page hits, there wouldn’t be any need for this entry.

      • Manish Sinha says:

        That census package was meant only for OEM as far as I know not for non-OEM installs. I also heard that it won’t work if you installed it manually on a non-OEM install as it checks for a file which is present only in OEM installs

    • Michael Hall says:

      I’d also note that I wasn’t being critical of the way Distrowatch gathers their statistics, nor was I discussing relative popularity specifically. Rather I was making clear what exactly the statistics represented, and what that means for Ubuntu and LinutMint. Distrowatch is very clear what their numbers mean, but people who are reporting on them are taking them to mean something altogether different, and therefore drawing the wrong conclusions (and sensational headlines) from them.

    • Josh says:

      _Dear_ _God_ _Jef_,

      _Your_ _usual_ _criticism_ _would_ _be_ _so_ _much_ _easier_ _to_ _read_ _without_ _all_ _of_ _the_ _bloody_ _underscores_.

      _Can’t_ _you_ _bold_ _or_ _italicise_ _or_, _well_, _anything_ _else_?

    • Anonymous says:

      From what I heard it was based off of the Ubuntu NTP server traffic.

  5. Josh says:

    As I just posted elsewhere, also according to Distrowatch “stats”, by the same bloody metric, the hobbyist CommodoreOS is more popular than: FreeBSD, Mandriva, SLACKWARE, Solaris, Gentoo, Kubuntu, Xubuntu, OpenBSD, Mepis, Red Hat, SUSE, and Archbang.

    So there’s for the Distrowatch stats. I’m by no means an expert, but I have serious doubts that the users of the above distros have suddenly abandoned those projects for CommodoreOS…

  6. jake says:

    I have to say unity is not really cup of tea – yet. It is maturing and improving. I didn’t care much for gnome 3 either. I tried gnome 3 in “classic” mode and it’s just not the same as gnome 2.

    Surprisingly, my curiosity about other choices led me to try out kubuntu – I say surprisingly because I’d been a very happy kde 3 user who was driven away by the bad experience that was kde 4.0.

    But kde 4.7, part of kubuntu 11.10, is the first version of kde 4 that I felt was as good as the classic 3.5 series. The LTS release of 12.04 should bring performance improvements and even more polish. Kubuntu is bidding fair to become a very popular distro.

    • Michael Hall says:

      Unity and Gnome-Shell are both at about where KDE was at the start of the 4.x path. And just as KDE 4.7 is not only as good as 3.5, but’s also put itself on a more solid footing for future development and innovation. Unity and Gnome-Shell will catch up to Gnome 2 in terms of functionality and usability soon enough, and like KDE 4 they will be in a much better position for future development when they do.

      • Alex Fischer says:

        Yes, but don’t they both really depart from the traditional dekstop metaphor? It’s bad enough that Windows is doing it- such mistakes can be expected from Microsoft. But GNOME? Hopefully KDE will stick to its roots…

  7. Pingback: Canonical questions Distrowatch share slide figures | Install Ubuntu

  8. skierpage says:

    > amount of it’s increase
    > It’s user base is growing rapidly
    > and I hope it’s community is too.
    > make it’s way upstream

    These are not a contraction of “It is” or “It has”, so no apostrophe (you don’t write “I like hi’s computer”). I couldn’t figure out your e-mail, feel free to delete this, cheers.

    • Michael Hall says:

      Fixed, thanks. I know the proper way to use them, but I almost always subconsciously do it wrong.

  9. Pingback: Links 30/11/2011: Kororaa 16 Beta, Firefox 11 Mentioned | Techrights

  10. koolhead17 says:

    Nice Summery!!
    Hope the FUD will cool down soon!! :D
    +1

  11. Pete says:

    Nice to see a little rationale thinking about all this Distrowatch Buzz. The only thing I think you can safely conclude about Mint’s Distrowatch ranking is that it has gotten a a lot of buzz driven by blogs and tech site’s noting its new de facto status as the linux distro of choice for Unity/Gnome3 refugees.

    What I’ve really found remarkable about the whole issue is that so many Linux users, probably a smarter-than-average bunch, have been so quick to draw conclusions from numbers that don’t say what people say they’re saying (if you follow me). It’s noteworthy, sure, but there’s no way Ubuntu is crashing and burning here. Mint is cool but you’re not going to see businesses and large organizations jump from Ubuntu. Mint has none of the structure and support that Canonical brings to Ubuntu. It’s not even a close comparison. (One thing never mentioned here, too, is that Ubuntu is so popular that users don’t even need Distrowatch to find them. Ubuntu’s Website and community forum are fantastic. )

    The other thing is that is surprising to me is the depth of passion about UI. Anyone smart enough to install Ubuntu is smart enough to grab XFCE or GNOME. They’re also smart enough to install Xubuntu or Kubuntu. The howling about choice is ridiculous. You’ve got so much choice, it’s almost paralyzing. I spent months test driving all sorts of desktops (started with Ubuntu, went to Mint, Mepis, Bodhi, Xubuntu and back to Ubuntu). Heck, even Mint (not LMDE) is Ubuntu. Just choose what works for you and relax. I can’t imagine what it taks to actually get that fired up over a UI that you can change anyway.

    There are a few things I dislike about Unity, but nothing that makes me crazy. In some ways, I think they weren’t radical enough. I’d like to see some things that are really cool about Gnome 3 incorporated into Unity (the hot corner, the desktop view, the easier access to all installed applications). I understand there are bumps in the road, but I really like where both projects are going. I’m happy they’re thinking forward and not sitting still on a desktop system that hasn’t changed in what seems like 15 years.

  12. Pingback: Is Linux Mint an Ubuntu-Killer? « Ivan Drakic

  13. White Lotus says:

    Well i have commented the same in some other blogs talking about similar topics. Here is the copy :)

    What the author says is right. Distrowatch is not an accurate representation. net-a-holics and distro-hopers often visit distrowatch. The website also explains that the figures only show the link to that distro.

    To accurately know the actual usage is not possible by seeing distrowatch ranking or facebook or twitter. I can have both facebook and twitter a/c. So I will give 2 votes. After a month, I try Mint 12 and give it a like on facebook. Not even the number of downloads or CDs sold or distributed can accurately give the number of users. Some may copy CDs and give it to friends. Some share single CD to install. Not every user register at forums. In short accurate figure cannot be obtained from Online resources.

    Best way to gain accurate statistics, IMO, is to have a registration from the OS. Like Ubuntu friendly. Just like windows programs and windows have an activation key. Since everything is free, so after every fresh install, if Ubuntu can automatically send statistics with all the data like version number and PC architect. This will give near accurate results, atleast much better than all these fuss. The point is will to Community accept this step, or is it too much privary intruding.

    To a certain point, I agree to the author and the negative remarks about Gnome 3 and Unity made me to visit distrowatch. So obviously, i will not look at Ubuntu, since I already have it installed. I am going to look for another distros and begin with *buntu derivatives. It is natural that Ubuntu’s rating will stay or decline and hits of all other distros will increase. These negative comments made me to look for an alternative. Had Gnome 3 and Unity did not received negative comments, I would not have bothered to go for another distro. I have 10.04 and plan to upgrade to 12.04. As an end user, I do not tweak much than changing wallpapers :)

    Regarding switching from Windows to Linux, I was prepared to learn new things. There is also a learning curve if you upgrade from XP to Win 7. IF Gnome 3 and Unity are not good DEs than better shift to another distro, mostly Mint.

    As per one comment, distro guys are aware of how many people are using their distro, when they issue updates, they know how many have downloaded.

Comments are closed.