Thoughts about Kubuntu's Status, Canonical, and your distribution's sponsors

Yesterday I woke up to the news that Canonical are no longer going to fund Riddell to work on Kubuntu. I've trying to figure out what that means for KDE and for community Linux generally.

Disclaimer: I work in the same role as Jonathan at SUSE, a competing Linux company that sponsors the openSUSE project. This is my personal opinion, not that of the openSUSE Board or SUSE Linux GmbH.

I'm sad for Jonathan personally. He has put a lot of his lifeblood into Kubuntu over the years, at no little cost to himself, and to be pulled off one's favourite project hurts. The same thing could happen to me if the powers that be decide, so I can easily empathise with him.

In the bigger picture, I have to say that this doesn't surprise me at all. For Canonical, Kubuntu fulfilled its purpose a few years ago already. Kubuntu, and the other official Ubuntu derivatives, have always been a spoiler move to tie up community contributors who believed in the early community-centric image of Ubuntu, but who didn't agree with the main Ubuntu's direction. Otherwise, there was the risk that Ubuntu design decisions would polarize the Linux community and send people towards Ubuntu's competitors. With the derivatives, they are safely occupied under the big tent of the Ubuntu brand.

If we look back at the Ubuntu game plan as history neatly lays it out for us, we have

1) Establish the Ubuntu brand amongst early adopters (check, by about 2005)
2) Expand it to the wider Linux user base (check, by about 2007)
3) Make Ubuntu the default Linux for non-technical users (2009)
4) Tie up a paying market. Initial targets have been enterprise desktop Linux (maybe next year ;)) or consumers in the massmarket netbook segment (but that was squashed by tablets and Microsoft rounding up the manufacturer back to the XP prison), and now they are aiming at embedding into consumer electronics (TVs) and will probably snare a tablet OEM as a cloud OS (hell, if KDE can do it...) or a bookseller or someone who wants a platform to digitally sell something else off of.
5) Profit
6) Buy more spaceflight (Probably. For some, 5) is enough)

Somewhere after 1), the massive demand for KDE on Ubuntu in KDE's main territories (Germany, via the ubuntu.de forums, which IIRC threatened an unofficial fork) caused Canonical to realise that it was better to control a large dissenting minority with some token gestures than to have them really doing their own thing. So Jonathan, at that point a KDE packager at Debian, was hired, and Mark Shuttleworth did his salesman job at a couple of KDE events making some insubstantial promises (If I had a dollar for every KDE eV board member at the time who told me "But Mark has promised to install and use Kubuntu on his workstation" multiplied by every Ubuntu developer overheard chuckling that "But they don't know that Mark *never* uses his workstation, he's always on a notebook"...), a few community people got flights to events, and Kubuntu was born, and legitimised by the then-leaders of the KDE community.

Once 2) was consolidated, Kubuntu was redundant to Canonical, but on the average professional Linux hacker's salary, Jonathan was an affordable luxury. Now, I suspect that with the trend at Canonical to develop more and more in-house to chase 4) rather than just distribute what the FLOSS community provides, putting paid man-hours on a mature product is no longer a good way to spend engineering budget.

By cheaply tying up competitors' resources, Kubuntu has hindered KDE's overall growth via other distributions and balkanized the KDE community. It can be argued that Kubuntu has brought users and contributors to KDE as part of the rapid initial growth of Ubuntu, and Kubuntu has been a success in focussing their developers on improving KDE, but this came at the price of cementing KDE in the role of a second class environment in the eyes of everyone who came to Linux via Ubuntu. I suspect that the GNOME community, which previously surfed the wave of Ubuntu's growth, will feel the pinch of necessity as Canonical moves towards its endgame, and having already been displaced as the default desktop for an inhouse development, will move further towards just being an anonymous organ donor to Unity and subsequent productisable UIs.

Why am I writing this? I don't want to be so crass as to just say 'come to my project instead'. I'd like to take this opportunity to suggest that you should have no illusions about what your community Linux distribution means to the businesses that sponsor it.

For openSUSE, it's some engineering contribution to and testing of SUSE enterprise products' codebase, and supporting the enterprise brand via a halo effect from the community brand. In setting up the openSUSE project, SUSE has been militant in giving the community complete control of the project and the distribution that comes out of it. Call it an insurance policy or a lifeboat, but by opening and freeing all the tools that create openSUSE (as well as the source code), we assure that the results of 20 years of work are indefinitely available. SUSE is secure enough in its business and believes strongly enough in free software to do this with the rootstock of its enterprise products, because the modular, federated Open Build Service allows SUSE to derive enterprise products from openSUSE without having to steer it.


I think the decision is rather consistent with all the actions done lately. They do not like to build up on desktop upstream-projects like GNOME or KDE but get something own done. Probably to apply own different visions[1], to distinguish from competition[2] and/or for other reasons. Future will tell if they are able to build up a strong user *and* developer contributor base around that while keeping the control necessary to apply own visions. In any case I think that is a perfect valid way to go. I think we should take the actions done now as what they are: A focus on there strategy and vision.

I do not believe that there sponsoring of Kubuntu such a long time was driven by an evil(TM) strategy. There strategy just changed. We do not really know why but I think parts of that are related to experiences collected within the past (missing control of the direction a community-driven project is moving into and general problems to deal with communities and as consequence getting patches upstream and resources lost on maintaining larger patch-sets that never went upstream).

Personally I think the whole story gives us an insight how difficult it can be to deal with a community. Let me add that I am not aware of any other distributor that had such massive problems with the communities what indicates that it's not all our fault. From the beginning there where visible problems with projects they build up on like debian, Linux Kernel, GNOME what makes me believe that it's a systematic problem in how things are decided, applied and communicated within Canonical.

What I really hope for is that all the energy they put into there new strategy will benefit the whole ecosystem. That means I really hope Unity will be a success, that a community builds around, that other distributors can ship it too, that patches made by them for the underlying projects are adopted by the projects and that the cooperation with other projects to establish common across desktops used standards continues. We will see.

[1] What may mean it's way harder if not near impossible to contribute to Ubuntu/Unity, influence the direction or even have a voice/vote. The code or person is secondary. Marketing, plan and roadmap of Canonical may overrule anything an external contributor could do. Sounds for me like a day-job (that should be proper payed) rather then a community-project. But that's only my very own impression.

[2] From reading about the amount of problems other distributors have with packaging Unity there are only two options for them. Either not package or do the job to get all the patches not upstream yet upstream what can mean lotttt of work including completely rewriting patches, patching Unity, etc. pp. In any case other distributors will have no voice/vote either in the project but could be overruled by there competition (means Canonical) just any time. Not any longer a neutral place to equally meet-up with your competition to get a common goal done but one of the participants (your competition) has full control over everything now. Sounds like a rather bad ground to cooperate on but that's also only my very own impression.

By Sebastian Sauer at Thu, 02/09/2012 - 05:24

I agree with the latter part of your post, and don't think it contradicts my original post. It shows that community involvement at Ubuntu is subordinate to Canonical's direction and inhouse work.

However you disagree with my analysis of Canonical's sponsorship of Ubuntu. What strategy do you think they did have prior to this change?

By Will Stephenson at Thu, 02/09/2012 - 12:08

I don't think my reply contradicts your original post either with the exception of the sponsorship-part :-)

My believe is that the original strategy was to make a desktop-centric Linux-distribution, get most of the existing Linux desktop-users to switch to that, get the distribution good enough to make the Linux-desktop-year happen (for that desktop-centric distribution), play a significant role as 3th big desktop-OS after Windows and OSX and finally that would magically open opportunities to get revenue and make the investment pay out or at least reach break-even with Canonical.

The reason why Kubuntu was sponsored was that there was and is a demand for KDE on the Linux-desktop. Lot of users and with it potential customers asked for it. I agree here with your analysis that if they would not done it then somebody else would have done it bringing the whole strategy into danger. So, yes, it made absolute sense to put some minimal resources on it and at least offer something before somebody else does. The difference is that I do not believe that its a bad intention. At the end each user can decide by his own and if Kubuntu is not able to fulfill there demand then there are enough alternates that can. It's up to the user what is how it should be.

Now two things happened. First while Ubuntu may play/played a large role at desktop-Linux it never got beyond that. The Linux-desktop-year just did not happen and so Ubuntu stays more or less irrelevant compared to Windows and OSX desktops (bug #1 is far away from being marked fixed - https://launchpad.net/ubuntu/+bug/1 ). Second they realized that a lot of people using your product does not automatically make revenue happen. But then it maybe would have when not failing at bug #1. As desktop-Distribution Ubuntu is a success when compared to other (desktop-)distributions but commercial it's a failure in still not being able to survive without lot of Mark Shuttleworth's money. Building profitable services around failed, making lot of money with support failed, getting significant at the server-businesses failed and so did a few other attempts. Now ~7 years later Canonical is still not profitable. The strategy did not worked out, bug #1 is in the very same state where it was ~7 years ago, and so the strategy was changed.

By Sebastian Sauer at Fri, 02/10/2012 - 04:09

I would love to come over to Suse, but just like the Windows holdouts, the issue is one of available software. Specifically, I don't want to build a new Anki for every release, and other Linux software vendors also see supporting Ubuntu as "supporting the Linux community". I don't disagree with this, we are already too fragmented as it is. *buntu's power of consolidation will only help get the likes of Adobe to port software to Linux.

That is why I prefer to stay with a Linux distro that is binary-compatible with the leading Linux distro.

By dotancohen at Thu, 02/09/2012 - 14:52

Well, from my experience (I neither work for SUSE nor for any other distributor) Opensuse has *way* more up-to-date Software available then Ubuntu has thanks to the Opensuse Build Service. Also, and that is a killer, Opensuse even has the newest Software, e.g. KDE 4.8, available for older versions of there distribution (e.g. 11.4 where I am still on where 12.1 is the newest) saving me lot of time not forcing me every 6 months to reinstall everything. Only think really missing is apt (no, zypper is still no alternate).

By Sebastian Sauer at Fri, 02/10/2012 - 04:27

I agree that the Opensuse Build Service has much popular software. My two problems were with Anki and Zim not having up-to-date packages for Suse. I will take another look, though, next time I do another install.


By dotancohen at Fri, 02/10/2012 - 16:30