Mark Shuttleworth interview (LXF 83)

The full text of the Mark Shuttleworth interview, some of which we couldn't fit in the magazine...

LXF: Is the way that Ubuntu looks important to you?

MS: Yes, 'pretty' is a feature, especially as we try to bring Free Software to a broader community of people. We have to make it something that people are excited to show their friends, and excited to show their parents. It's hard to get your grandparents excited about a super-reliable kernel - they assume that computers are reliable, or aren't reliable, and that's not something they can get involved with.

But it's much easier to get them excited about something that looks and feels unique and distinct, and works the way they want. It's interesting how we have these two different desktop initiatives, Gnome and KDE; many people think that's a weakness, but the reality is that they're both focused on different things, and the Linux desktop has benefited as a result. So the KDE guys are focused more on look, shine and bling, and while the pendulum's swinging back for them, they probably would retain a head-start on that front.

The Gnome guys are focused more on usability and interaction. This came home to me the other day when I was trying to format a drive under Windows - admittedly it was Windows running under VMware under Ubuntu! But I had to format a partition under it, and for the life of me I couldn't figure out where to go in order to make Windows recognise a new drive and format a partition on it. It just turns out to be so much easier to do that sort of thing under Gnome than under Windows.

It gave me real confidence in the Free Software process, not just for designing kernels but for designing desktop software you can give your grandmother.

LXF: Should they still have to worry about what a partition is, or whether it's formatted?

MS: That's an interesting question; I do think it'd be bloody dangerous to have computers automatically go round formatting and partitioning. But if a user decides, for whatever reason, that they want to format a partition, it should be clean, clear and they should know what they're doing. Interestingly, we're getting to the point where it's a lot easier to manipulate that kind of thing in the Free Software world than it is in the proprietary software world.

It's always been easier to do it at the command line, behind the scenes in Linux, but we've now started to develop a set of desktop metaphors which are really rich and powerful. That's quite exciting.

LXF: With Ubuntu, have you gone out of your way to appeal to people who'd ordinarily be Windows users? Perhaps the geeks in us weren't all that bothered about how something looks, when we could always go down to the command line...

MS: I think it's really dangerous to say that how something looks is all there is to it. There've been a couple of versions of Linux who've defined themselves solely in terms of making things look shiny, and they never took off because they never gave developers anything to get excited about. They didn't really invest in the underlying technology, and they didn't do a good job of making the most of the amazing work that is being done outside of the bling, in the rest of the Free Software sector.

So you have to reach a balance of all of those things. I think we recognise that our art team, and the art direction that they get, is just as important as the kernel team: important to different people for different reasons. I never let the kernel team crack jokes at the art team's success. They need to recognise that each of these community components plays an important role. That's exciting.

LXF: I think Ubuntu is the first distribution to have come along with that attitude...

MS: We won't be the last, because I assume that Red Hat, Novell and others will pick up on that lead, and match us in that regard. But that's how I think it should be done. I waited a long time before starting Ubuntu because I was convinced one of the major distros would do it themselves, and the fact that they didn't eventually got me to the point where I was frustrated and wanted to see if we could make it work.

So hopefully us driving that community philosophy back into the commercial distro space, is a good thing for the Free Software community as a whole.

LXF: How far can the spin-offs like Kubuntu, Edubuntu, Xubuntu etc. go? Does it dilute the Ubuntu brand?

MS: We will definitely see a multimedia version of Ubuntu kick in, and we'll probably see an embedded version. At that point we'll pretty much span the gamut from the server, through the very lightweight client through Xubuntu, and to the embedded client, with Ubuntu and Kubuntu in the middle. Beyond that, I think we'll start to see more fragmentation around vertical markets. Edubuntu is a leading indicator of that -- an aggregation of stuff from both Ubuntu and Kubuntu, but optimised for education.

I can imagine that we'll see one of those potentially for healthcare, or one for any number of verticals, once we have a strong enough community to do the work that it takes to aggregate, produce ISO images and so on. Hopefully, we're pushing the edge of limits in terms of the codebase that we need to support all those different variations and permutations. The embedded thing is going to be tricky because you have to do some real hardcore engineering across the board to make that work, but I think it's nice to have people focusing on the really lightweight environments, because their patches tend to make your and my desktop go faster.

So in summary, I think we've basically covered all the desktops, with some interesting new edge-cases to address, and then we'll see permutations and combinations to address vertical markets.

LXF: What will the multimedia Ubuntu do?

MS: Have a heavier emphasis on real-time responsiveness of the kernel: for example, audio recording, video recording, playback and things like that...

LXF: So stuff like JACK will work?

MS: Stuff like JACK, Audigy, Cinelarra...

LXF: I guess there'll still be issues with codecs?

MS: There will be -- obviously that's not a problem for Norwegian people, but what we can do is at least make it easier for people who have legal access to functionality where there aren't patent issues. We can make it easier to get that out-the-box on Ubuntu: click-click trivial. Things like EasyUbuntu are great, because they tell you 'Ooh, there's some talent in the community over here, these are guys who know what they're doing, they're excited and they're contributing work.' You can imagine something like that folded into the distro, or something new that addresses that same space.

But with the proviso that we have to respect the legal regime that is pervasive in any particular country. So the core distro has to aim to the lowest common denominator, and then we make available and enable functionality for people who say they're from a juristiction where those patents aren't an issue.

LXF: FreeSpire, the Linspire project, offers a version that's '99%' open source but includes proprietary extras, and a fully free version like Ubuntu is today. Have you considered doing something similar?

MS: I guess it depends on what you think is really important. What we think is really important for everybody is the Free Software angle. The only place where we compromise that is on the hardware enablement front -- making available proprietary drivers in the distro where we have to. But we don't put any proprietary applications in. That's because we're really confident in the Free Software stack! We know you can't pop a DVD into a virgin Ubuntu installation and it'll work, that's an issue, but there are solutions for that.

We're also confident that, in time, the Free Software community will deliver solutions to all of those edge cases. To a certain extent, the fact that we limit ourselves to Free Software applications provides a strong incentive for people to step up and write Free Software applications for the pieces that are missing. It also creates an opportunity for distros like Mepis, which are taking Ubuntu and adding commercial software to it, to give people people something that works out-the-box who're willing to pay for it.

That gives them a solid platform in Ubuntu, and a well-defined market opportunity outside of Ubuntu, that Ubuntu won't go after effectively. It creates an ecosystem, with us playing with one another's strengths. We hope in a future release of Ubuntu, perhaps the next release of Ubuntu, you'll be able to enable those commercial things and confirm that you have the legal right to do that. That'll allow us to let people do what they want to do, respecting whatever laws prevail. The moment the MP3 patent expires, we can do that globally!

LXF: It's very different approach to SUSE and Novell, where they've explicitly denied you to install LAME -- you have to recompile Xine to get it to work, rather than just dropping a library in, like in Ubuntu.

MS: It's interesting, because they also include proprietary software. It's fairly easy for them to get a patent license -- it wouldn't cost them anything to get an MP3 player -- so I'm surprised that they take that view. One of the things I really hope we'll see is the emergence of open DRM-free protocols and codecs which are genuinely better from a quality point of view. If you look at what's happening with the BBC's Direc, it's an extremely compelling codec that they've put together, and if they can get it out there... It could be world-beating, trend-setting stuff.

I'd like us to get to the point where if you buy an MP3 player, it plays Oggs. If you buy an iPod, it plays Oggs. And when you buy a recorder, it records in Ogg. Then all of these questions start to disappear, and I think it's worth retaining a little bit of pressure to make that happen.

LXF: So is Edgy Eft going to be released in October, or has the delay of Dapper pushed it back?

MS: There's still some discussion about the release date. We pushed back the release of Dapper because it's a long-term supported release, and we wanted to get that extra bit of quality. So the debate is now: do we make Edgy a full six-month release, which will put it permanently two months behind Gnome, or do we catch up one month, or do we catch up the whole two months?

It seems like there's a lot of support for catching up, making Edgy a very short bleeding-edge release. Then the question comes: are we going to offer commercial suppor for Edgy, in the same way we offered for Warty, Hoary and Breezy? There's an argument that says because Dapper is there, fresh and supported for a very long time, we can get away with not offering support for Edgy, but supporting Edgy+1. So I don't want us to get to the point where support is only offered for the long-term releases - that smacks of Fedora and RHEL. We probably want to offer security updates for Edgy, but not commercial support unless it's under special terms.

That sort of discussion is going on at the moment, and basically we're trying to find a balance between how bleeding-edge we can get, given that we have a very short time if we want to catch up fully...

LXF: So does that mean things like Compiz, XGL and AIGLX?

MS: I hope so! The planning process for Edgy is different to the others. For Warty, the process was 'see if we can do it'. We didn't commit to any specific feature goals, other than shipping the latest release of Gnome on the day of its release. For Hoary, Breezy and Dapper, we set from the top-down goals for the developers, based on the ability to do something like Dapper. So I've had my say, as it were - shot all my bolts. Now I want the pendulum to swing back, and I want the developers to say 'this is your release - play!'

LXF: You've said you're going to have 'almost no direct input' in Edgy. What is the 'almost'?

MS: The 'almost' might be in a case, for example, where we have an ongoing partnership commitment, where Canonical is under the legal obligation to update packages so that we can build the latest Oracle, DB2 or something like that. So I would go in and say: 'Hey guys, are we going to have this version?' And they'll say yeah, and that's it. I'm not going to impose any significant feature goals.

So I would really like to see AIGLX, and Compiz, working together really nicely. But I'm not going to go in and say, 'Please will you do this for me!'. I can cheerleader, I can talk about things people really want, but this really is the developers' release. They can do what they want, as long as it's aligned with where we want to go, and is a constructive use of their talent and time. It's not a holiday, but a self-driven release.

Something you said earlier triggered... I made a note to say something. Just like some distros try to define themselves by being shiny above all else, some have also defined themselves by trying to be just like Windows. I feel very strongly that that's a mistake. Linux is its own beast, it has its own character, its own strengths, its own style. We shouldn't be ashamed or shy of being distinctive.

And that's why I'm so pleased that - partly by accident, partly by design - the Ubuntu 'look' has emerged as being so clear and distinct. People recognise it, and they can rally behind it. It becomes a distinctive community and a distinctive skillset.

LXF: What do you think are the main strengths of Dapper?

MS: On the server side, it saw a lot of engineering and collaborative work between ourselves and the big server vendors, making sure that the kernels went through robust testing. The big server hardware vendors have big stress-testing procedures, and they like to kick the shit out of kernels. So we put our kernel through that.

On the desktop side, there's quite a lot more unique polish, uniquely developed features and functionality. A lot more quality assurance and documentation. The doc team did a fantastic job during the Dapper cycle, I think because they had a longer freeze period, a longer time when things are stable.

LXF: What motivates them [the doc team]? Is it Linux in general, or the Ubuntu approach?

MS: I think to a certain extent we all have selfish motives - we want something we can share, we want something we can be proud of, and have our name on. These guys have built something that they will give their friends, that they will talk about, that becomes part of what they do. It's an awesome thing to be part of. I saw some community burnout: some guys who threw themselves into the community to the exclusion of many other things in their lives. That's dangerous - unless it's a job it shouldn't be a job.

What I love to see is guys who love being part of something, and who make great contributions but keep their sanity and pace themselves as they do it. Many of the teams are starting to mature to the point where that is the case. The other thing that is happening a lot, is that we have some guys who've been around long enough that they're becoming natural leaders of the different teams. And you can clearly see the difference between a community team where that natural leadership has emerged, and one where it hasn't. We're learning how to spot and encourage that leadership.

LXF: So where do you see Ubuntu in 10 years time?

MS: I hope that Ubuntu will be the standard bearer, the torch bearer for Free Software for non computer specialists. At the same time, I say that without wanting to compromise the idea of Ubuntu as a standard bearer for Free Software for developers and professionals. I don't ever see Ubuntu as the 'dumbed-down' version of Linux - but along with being cutting-edge and developer-oriented, I hope Ubuntu will drive Free Software into new areas where it currently hasn't penetrated.

In 10 years time, I would expect that any significant consumer hardware manufacturer is publishing free software drivers for Ubuntu. Our strong emphasis on free software is starting to bring that message home. I would expect that people graduation from university will be completely comfortable with the environments, and businesses will recognise the skills that people have, if they have them on Ubuntu.

So in 10 years time, I would fully expect Ubuntu to be - if not an equal partner to - at least in people's minds on a par with Mac OS and Windows, as being a universally recognised computing platform, with its strengths and weaknesses, and admirers and detractors.

LXF: And Goobuntu [Google's in-house variant of Ubuntu]?

MS: I don't know - I haven't heard from the Google guys for a long time. For us, it's awesome to have this group of really smart guys kicking the tires. Every now and then we get a bug report and think, 'Only some really smart guys would have spotted that!' They're very good free software citizens.

LXF: What are they doing with it?

MS: I only see it very distantly; I have no behind-the-scenes access. Which means that I can speak quite openly about what I know, and it's simply a platform for the developers. So they have to have a Linux desktop - it could be any Linux desktop, and they chose Ubuntu. I don't think there's any masterplan, as far as I'm aware, to take over the world with an Ubuntu derivative. I'd be thrilled if they would, because then I could go back to space!

LXF: In the last issue of our magazine, DistroWatch.com maintainer Ladislav wrote about feedback he's seen around the net for the last few distro releases. He said he's seen more problem reports for Dapper than the latest Fedora or SUSE versions...

MS: It's interesting. One of the things we've noticed is that the size of our community doubles from release to release - which means we get bug reports about things we never used to get bug reports about, because people are now using Ubuntu in ways they never used to use it. So that continually pushes the edges of the envelope. We're going to have to grow our development community to keep up with that.

If you're getting bugs from twice as many sources, about twice as many different things, you have to have enough developers to 'context switch' across all those different areas. That balance becomes extremely important. I've no worries about the quality of Dapper: I think it's the best release we've ever put out by a long way. But I can accept that it's not perfect, and I can accept we're going to uncover corner cases we've never found with previous releases.

LXF: One of the things that annoys is is that SUSE, Fedora and Ubuntu all have functionally identical update tools - yet they're all completely different applications. Is there any way distros can work together to make a single interface?

MS: We do have a single interface at one level of the stack, but what you really want is better interfacing at other layers of the stack. So the stack to me is: there's the source code level, the binary package level, and the user interface level. We have pretty common functionality at the user interface level, and we can almost certainly converge that down so that an RPM-based distro and a .deb-based distro use the same applications and user interface.

At the binary package level, there was a huge pissing contest about binary compatibility. There's a huge difference between 'it'll probably work', and 'we've certified that it'll work in every case'. Unfortunately, if you go with the 'we have to certify the work in every case' kind of approach, you completely stall your development, because you spend you entire life trying to chase down corner-cases at the periphery.

You get into making horrible decisions like 'Do we fix this bug, and therefore break compatibility with other people who have the bug'. That's at the binary package level. So at the binary package level, I don't think those incompatibilities are such an issues - that's where you get the ability for guys to make decisions on when they're going to move forward, and when they're not. I don't think you should ever expect to take a package from SUSE, install it on Ubuntu, and then move it to Debian. A binary package is built for an environment.

LXF: But that's what, say, our parents would expect - you download a program and run setup.exe...

MS: But in a free software world, your interface to software is through your software repository, and every application you can imagine should be there. In a free software world, why would you need to download a piece of software somewhere else? If Photoshop was free software, you don't think it'd be in the Ubuntu repositories?

Let's dig even deeper and look at the source stack, and that's where I think we can make huge leaps and improvements. We do all this work adding value to Debian, adding things that are important to our userbase. We want Debian to take that, but there's no common language for us to speak, in moving code between ourselves and Debian. They use the same package format! So right now we have an automated process, which does its very best to strip out the changes we've made from the changes they've made, and say 'Hey Debian guys, here's a couple of changes you might want to consider taking - you don't have to take them, but in the easiest possible automatic format we can give you, here they are.'

That format's not ideal, because the Debian changes are a monolothic diff, and our changes are a diff on top of that. It's hard to say what each change is for and why it's there. Contrast that with if we wanted to exchange code with SUSE, for example. They're using an entirely different package format, and that's entirely unnecessary. So one of the areas where we'll be investing over time is trying to get commonality between ourselves and SUSE and Red Hat. I'm sure Red Hat will resist this until the cows come home, but it's in their interests I think to tap into the broader community. To do it properly, it has to be a two-way flow.

LXF: Are there any Easter Eggs in Dapper Drake?

MS: There's some fun bits and pieces - the video from Nelson Mandela, for one! Also, you know how some of the screensavers want photographs, and if you don't point them at photographs they use those test screen images? We stuck a couple of photos in there that you'd like - pictures from space! And for the rest? You'll have to dig around... [Laughs] We have a good April Fool's idea for people part of the development process of Edgy+1... I can't wait for April Fool's?

LXF: Let me guess - it's going to be blue? TWM as the default window manager?

MS: [Laughs] TWM with Compiz!