It has been an incredibly interesting week in free desktop-land, in that kind of “interesting like a soap opera” kind of way. I guess it’s not news that different participants have different recollections of the same series of events, but it is a bit sad to see it writ so large on a public stage.

Timing-wise, it’s quite co-incidental, but it’s enlightening (I think) to read Mark Shuttleworth’s latest “Internal competition is healthy, but depends on strong and mature leadership” alongside Mark Wilcox’s “What happened to Nokia?” of a month ago. I’m quite clearly going to side with latter-Mark on this one: internal competition is generally not healthy; in fact, in my experience, it can be of the most damaging things you can do to a group of people. That’s not to say that it’s always a bad thing – to a large extent, it works for the Linux kernel (who I think are a special case in this regard) – but in a community telling someone their contribution isn’t wanted is a hurtful thing. You can see the hurt if you read what Mark S. is saying, it’s both implicit and explicit. Internal competition isn’t a solution to this, though, of course – it’s the equivalent of taking the disagreement outside and settling it mano-a-mano, swapping one hurt for another. It’s a red meat solution, a particularly macho form of solving problems.

For me, from the outside, Gnome 3 has been an example of a particularly successful collaborative project. If you go to gnome3.org and “Try it out”, you’re not downloading a copy of Red Hat / Fedora there – it’s OpenSuse underneath, built on their rather wonderful Open Build Service. All of the design has gone on in public (Hylke’s list of designers was interesting), and as a long-term gnome-shell user (I’ve been using it regularly since Fedora 12 on my Netbook) it’s easy for me to appreciate just how much work has really gone into this system.

Of course, Gnome 3 is not going to be for everyone. That’s ok, there’s KDE (and others). This is another example of where competition isn’t really: sure, you can run the same apps in both desktop environments, but generally users of one are not going to be immensely happy in the other environment (particularly power users). They don’t compete head-to-head in that sense. In the same sense, I think that’s the same thing that happens with distros. Yes, Fedora 15 will almost certainly lose some users because of the default setup. The inclusion of Gnome 3 will irk some, the inclusion of systemd will irk others, and to a large extent it was always thus (pulseaudio, networkmanager, etc.). Again, there are other distros, the Debians, OpenSuses, and even Ubuntus of this world, and to a great extent they really don’t directly compete with each other. Sure, some people move from one to another like they’re changing underwear. I think this is why Fedora can afford to be an adventurous distro, why Debian can’t really afford to put out bad releases, etc. – each to their own.

So, how much competition is too much? Where does the line lay? I don’t think it’s easy to tell. What is clear is that the amount of drama on this issue way, way exceeds the amount it deserves. Owen Taylor has said that including “appindicators” in gnome-shell is still on the table – so in that sense, there is a bit of a fuss about nothing (of course it’s arguable, and hypothetical, that his opinion has changed on this subject).

What is really needed, though, is a much clearer vision of where the desktop ought to be going. Mark S. has said that the Gnome 3 “trajectory” is wrong and has already failed. So where should it go? Where is Unity trying to go? Mark S. has already given up on Gnome, but talks about having Unity and KDE co-ordinate closely via freedesktop.org.

I think what is sad about this is the focus on the differences by the participants. Looking from the outside, gnome-shell and Unity are incredibly similar, and have been since release. Anyone looking at screenshots can see that; they obviously have the same influences. The Unity 2D system is even implemented in QML, which is another variant of Javascript just like gnome-shell’s gjs underpinnings. I find it difficult to believe that one could not be modified without a lot of work to look/behave much like the other. This isn’t a technical dispute.

As predicted, Microsoft and Nokia are tying a knot of sorts, and all sorts of people are extremely disappointed by this news. I’m an Android user right now, but I’m particularly disappointed because Android just isn’t the free platform it claims to be.

A lot of people are blaming Microsoft and dreaming up “entryism” conspiracy theories. These people are entirely wrong; the decision to go MS was signalled a long while ago by Nokia’s board. Nokia are a $40B business: decision making doesn’t work like that. What is true, though, is that occasionally the free software community gets the benefit of large corporations putting resources into developing software, and occasionally those corporations change their mind later. We celebrate the former and mourn the latter, it’s only natural – I’m a big GNOME fan, but it seems that GNOME Mobile, MeeGo, and the various related stacks are basically dead in the water at this point.

This match-up makes a huge amount of sense for Nokia, but sadly it is going to alienate some of their current user base. I liken this to Bob Dylan’s move to electric guitar: his fan base called him a sell-out, and never ever forgave him. Fundamentally, his music changed beyond all recognition. Whether this is right or wrong, of course, lies in the eye (or ear) of the beholder.

Amusingly, this also means that of all the development platforms for native mobile apps, Mono is now exceptionally well-placed. It can compile native code and make full use of native APIs, and comes in an Apple flavour already, with Android along the way. I guess this is an additional sting in the tail for some, particularly since Qt could have also played that role exceptionally well, but we must acknowledge that the free software mobile development stack is actually in quite good shape right now. We don’t have the right development environment for HTML5 apps yet, though.

It’s exceptionally sad that a really free mobile OS hasn’t come to fruition. OpenMoko took a long time to come to market and wasn’t developing quickly enough, that same verdict has now been given on MeeGo. Android is close, but is not developed in an open fashion and in the matter it is delivered is not a free OS. The “commoditization” argument has been shown to be wrong.

What the move to WP7 does signal is strong integration into Windows and, I guess, Exchange and Sharepoint. People aren’t going to care about the OS in a couple of years. For free software to matter in this space, the focus has to be on integration and apps. It doesn’t matter, after all, what’s running the hardware: what matters is what you can do with it. By struggling for freedom at the hardware and OS level, it’s very easy to lose sight of the bigger picture – and with it, strive for things which are totally irrelevant to 99.99% of phone users.

Today was Open Source Expo day. While I had been asked by one of the organisers whether or not I could propose a .Org to exhibit there, I decided against it for a couple of reasons: mainly, because I hadn’t heard very much about the exhibition, was a bit worried about the timing, and questioned whether or not it would be a good use of time for me or anyone else involved in an open source project to attend. To be clear, this event is held over two working days, and is in the middle of London: not the end of the world for me, I could take time off work, but others I know are consultants and would be literally losing money by going. I have to say, I’m glad I didn’t spend much longer there than my lunch break today allowed.

Now, Open Source Expo this year was “co-located” with Cloud Expo Europe. This immediately sent up red flags; while all of the interesting Cloud stuff is of course being driven by free software projects, the cloud expo speaker list is primarily composed of execs from relatively large corporations attempting to flash their with-it credentials to the kids of today, giving rise to various vacuous talk titles such as “Ahead in the Cloud”, “Networking the Cloud: Is it the Journey or the Destination?” (network and cloud? surely you’re kidding!!), and “SaaS 2.0 Open Cloud Computing”. I apologise to the speakers whose talk titles I have used; it may be that your talk was exceptionally entertaining and at the cutting edge of cloud technology, and certainly there are speakers such as Michael Meeks who I would pay to be informed by, but I’m afraid it was a deeply uninspiring schedule.

However, if the “Cloud” side looked a little bit dated, the “Open Source” sister website was even worse. Even as I write – on the 2nd February as the event is actually taking place – there are still significant proportions of the site missing. In particular:

• “The conference programme will be available here soon.” – well, no. In fact, there doesn’t seem to be an open source conference at all. There are five or six vaguely open-source talks and speakers, no idea if you have to pay for the conference but I’m certainly not hanging around for two entire days just to see a couple of open source talks.
• “This year’s speakers list is to be announced shortly, please check back soon.” – again, no. I’m not sure the difference between the “speakers list” and the “conference programme”, anyway – perhaps it’s just the same list sorted alphabetically and then chronologically.
• [Sponsors] “LinuxIT is seeking Open Source projects to include in the .org village, this exciting area of Cloud Expo” – no list of .orgs, and there are serious problems with the village (see below).

So what part of this conference is really open source? It’s difficult to tell. Aside from the .org village, there is no obvious “open source” area, and it doesn’t look to me like there’s an open source conference happening at all. If you’re interested in open source technology, and not cloud per se, there really isn’t much here for you at all.

But here’s the rub. Ok, there are not many open source companies there. But to be honest, there’s not much here at all. On a single level at the Barbican, it’s a very small and claustrophobia-inducing space, and it took me literally ten minutes to walk around the entire show. Large parts of floor space are taken over by the speaking areas, the meeting rooms and the cafe, and there are only 35 exhibitors. It’s difficult to get excited about this, and don’t get me wrong: I want to be interested in the corporate area. I can name four exhibitors of whom I’m a customer. But what wonderful new stuff is there? There’s only so many hosting companies or people offering SaaS applications I can stomach; I want to come to Expo to see what’s coming, not what’s been around for the past ten years.

Back to the village. I repeat what I said just now, “there’s not much here at all”. The .Org area, which is one of the most interesting and cutting-edge areas of the Expo in years gone by, has been decimated. I kid you not, this is a picture of the entire .Org village:

Yes, I can hear what you’re saying from the other side of the internet. “That can’t be it! That’s just a stand”. No, no it isn’t. Front left you have a PXE project (I’m sorry guys, I forgot your name – maybe it was gPXE – but you had hardware and it looked a polished presentation), and behind them back-left is Ubuntu UK. In the centre is Drupal, and back-right is Debian. LPI are front-right: I’m not totally sure what LPI are doing there, and since their stand was unattended for the 40mn I was there I didn’t get chance to ask, but it seems like LPI took over sponsorship of the .Org area from LinuxIT. How/why is not to speculate, but it’s an extremely odd match in my opinion.

Compare and contrast this photo with previous Expos. If you haven’t been, the Open Source Expo website helpfully provides a picture from 2009 on the front page. Or compare to this report from 2005. Or this picture of the .Org village from 2008. This is by far the smallest .Org village by some significant margin. The question has to be asked; was it really right to label the “Open Source Expo” as a. existing and b. co-located with the Cloud Expo? The honest answer has to be no; open source / free software is a fact of life and if we’re being honest the various web developer shows have twice the amount of “open source” as was available here.

All this said, I have two major complaints. The first is the missed opportunity. Everyone knows that some of the most cutting-edge “cloud” development is happening in the open source world, but this Expo didn’t give that impression – in fact, you’d get entirely the opposite impression. Here are some of the key cloud technologies that were entirely absent from this Expo:

• Drizzle, MariaDB, and the other lightweight SQL-ish clustered databases
• MongoDB, CouchDB, and the other lightweight no-SQL clustered databases
• GlusterFS, GridFS, or any number of the other shared internet storage systems
• node.js, nginx, or other innovative server systems
• jQuery, YUI, Dojo, or any other front-end UI development stuff
• Eucalyptus, libvirt, Cobbler, or any other virtual machine provisioning/hosting system
• Hadoop or any other map/reduce style processing system

… and that’s just off the top of my head. Also absent were any developers knowledgeable about the above, or any IT managers who have deployed any of the above, or any business people running enterprises on the above. Yes, there are not many people working with the above because the above technologies are quite new: but that is the entire point of Expo. I don’t want to go to Expo to talk about people deploying virtual machines or network storage. Woop woop! Boring alert. Expo is about stuff to come. New stuff, things people haven’t seen/done before. Sorry, but the cloud happened about five years ago.

Second major complaint: timing. Why on earth early February? Linux Expo was always traditionally on in October, which seemed to work reasonably well. Early February, though, is traditionally FOSDEM time, and FOSDEM starts in a couple of days. Are people really going to come to an Expo in London and then go to Brussels a few days later? No. Are people going to visit the London Expo in preference to FOSDEM? No. So what you’re left with is the people in/around London who can’t get to FOSDEM (like me) but who can come in for Expo.

This whole post is horribly negative and I apologise to people who’ve put time and effort into Open Source Expo / Cloud Expo to make it work. I apologise to people who’ve taken time off work or otherwise invested time to come to Expo to exhibit or to be a speaker at the conference; I don’t write this to belittle your contribution. The question has to be asked, though: how do we solve the problems which have beset this Expo, and make it better in future?

If you haven’t read it already, Bradley Kuhn’s take on where Canonical are aiming is deeply interesting. There is bound to be push-back against the article, because it does connect a few distant dots, but I found it particularly interesting because it’s apropos of a recent discussion on Surrey LUG’s mailing list – which has no public archive so therefore I cannot link, but the gist of the thread was a discussion on the various approaches Canonical takes to getting income.

I personally disagree that it seems like Canonical are attempting some kind of open core strategy: I think it’s arguable that they’re seeking to leave that avenue open, but for them to sell proprietary software products at any point in the next few years would destroy pretty much their entire brand image. In particular, Shuttleworth has made extremely strong statements in the past about never charging money for software – and it seems pointless to be proprietary if you’re giving the software away (well, for the most part).

However, there are a couple of things I find pretty depressing that Kuhn brings up, particularly the admission that Canonical is still not profitable and would take “some more time” to achieve that. By 2007, some three years into the Ubuntu project, it had already cost some $25 million, and of course we’re another three years on since then – given the growth of Canonical, a total spend of $50 million to this point would seem particularly conservative.

What I never understood to begin with, and still fail to understand, is why Shuttleworth has ham-strung the project from the start by promising there would never be a pay-for version a la Red Hat. My thinking in this regard is pretty simple: when you’re paying for all this development, you have to recoup those costs somehow. When you’re providing support or doing all these other activities, you leverage that development, but you incur further costs which then also need to be recouped.

Pretty much everyone else who commercially develops free software has some method of recoup in position. Red Hat, as an example, sell security updates, support, and the OS itself. Nokia invest in software to sell their phones; similarly the likes of Intel, Oracle, and all sorts of other businesses invest in the software to enable hardware sales. Google invest in the likes on Android because it gives them a huge amount of control over the mobile market. And of course, there are plenty of small players too, all of whom have some kind of strategy to ensure their development activity is not some loss-making exercise in the grander scheme of things.

If Canonical’s commercial model confuses me, though, there is one thing which is abundantly clear: for that company to fail would be incredibly damaging. It would be difficult to imagine a more worse-case scenario than seeing Canonical at ten years old, still burning millions of dollars each year and still not making a profit.

Of course, Canonical likely has that luxury; previous break-even targets have been set and then moved for whatever reason, and the vision still seems to be about growing hard, branching out and becoming some “cohesive whole” (which is avoiding the issue). So it’s not likely that they’re going to fold in the next five years. But in away, that’s almost worse: much like a Government-funded entity, there is this insulation from commercial reality which ensures that “the vision” will be followed. The must be an opportunity cost there, room for a new Ximian or a new Mandrake, and fresher more innovative visions that are currently squeezed out of the market.