GNOME Online Desktop: "We will have to include Windows"


    Red Hats Havoc Pennington in an interview about opening up new possibilities, the crucial advantage over Microsoft and why he doesn't just want to "make a desktop"

    During this years GUADEC Red Hat developer Havoc Pennington proposed his idea of an "Online Desktop" to the developers of the GNOME project. Through deep integration with web services and "zero-maintenance" the Open Source client aims to get the "perfect window to the Internet". During GUADEC Andreas Proschofsky had the chance to talk to Pennington about advantages and possible problems of the Online Desktop concept, the necessity of Windows-support and about Red Hats "return to the desktop".

    This interview is also available in a german translation. Judging from the reactions you got to your keynote about the GNOME Online Desktop, do you think there is a good chance to get this rolling?

    Havoc Pennington: I hope so, it's hard to tell tough. It seemed very positive over all. There are still a lot of things to figure out, but it's not harder than when we originally got GNOME going in the first place. And you don't have to have everyone agreeing, you just need a few people who want to work hard on something. As Novell seemed to agree on your vision, does that mean, that both companies together just can go forward anyway? Do you need the community to join at all?

    Havoc Pennington: I think so. People really overestimate both how much of open source gets written by companies and also the ability of companies to kind of do things on their own without having community buy-in. I don't really think it is workable without community support. That doesn't mean that we need community unanimity, it's fine if just some segment of the community is interested in the project. But neither Red Hat nor Novell is a large enough company to just go at it alone. In the past there have been a bunch of other endeavours in trying to get a better integration between web services and the desktop - for instance the "social" browser Flock - none of them really successful until now. So why do you think you can do better here?

    Havoc Pennington: Whenever you have a general category description, whether it is cell phone or software that connects to online services there's going to be a few things that succeed and many more things that don't. It generally comes down to specifics, how the business aspects are done, how the community aspects are done or how the user experience is. Flock is actually - as I understand - not really aimed at a mass audience, so that might be part of it. Also it's just in the browser, they are not addressing the desktop as a whole, so it's a different thing, more like Firefox plus extensions. So you don't think that people might just be happy with what they have now?

    Havoc Pennington: You know, that's always possible (laughter). But to me when I use a desktop today I feel like there are major negatives to it - there are all kinds of maintenance and backups to worry about. Just cleaning up the desktop background, cause it's cluttered full of junk to not having calendar notifications, having recent documents be the wrong documents and not the online documents that I'm actually using.

    One way to think about it is, say you're a small business and buy Google Apps for your domain it keeps you from having to maintain your own servers, you know things like an IMAP-server. But you still have to maintain your Windows desktops. If we could offer a desktop, that is also self-maintaining, then you don't really have to do anything custom, you don't have to pay someone to maintain it for you, it's just completely off-the-shelf and just works. So I think that might be really compelling to lots of people, that's sort of the business kind of scenario. That would be one way to use it, are there other use cases you are aiming for?

    Havoc Pennington: I think this could also be useful for a nice simple Internet appliance for your home. So that's one direction. Another I think is just making Linux distributions and Live-CD type stuff better. It makes Fedora / Ubuntu / whatever Live-CDs a lot more interesting, if you just drop it in and have access to all your private stuff, it's kind of a toy otherwise.

    And then there's the mobile stuff,of course. So having the same configuration on different sorts of devices is supposed to be one of the big things?

    Havoc Pennington: I think so, yeah. Potentially we will have to include Windows in that. So for instance: Firefox on Windows, why shouldn't it know about the same things than Firefox on Linux. If you have services like social networks, you always have to support multiple platforms, so that all your friends can join in on your conversations and other stuff. How will that differ from the Linux implementation, are you going to have things like BigBoard (the panel that integrates with online services, ed.) on Windows too ?

    Havoc Pennington: Well, I think that's open to decide. I think it's potentially interesting, obviously some parts of it would make sense on Windows - like the list of your friends and whether they are online - others don't, like the Linux-specific software installation thing. So why should people still use Linux, when they can have it on Windows too?

    Havoc Pennington: They are not going to have everything on Windows, there's more we can do on Linux, cause we can actually change the whole thing and we can really get rid of all the maintenance headaches and all the little pieces of local state.

    Also you couldn't do stuff like the LiveCDs yourself with Windows, as the code isn't open source. Speaking of Microsoft, you cited Bill Gates in your keynote, saying that Windows will go into a similar direction, why do you think you can do better than them?

    Havoc Pennington: Mostly cause we won't have any kind of fixed connection to one one service in our desktop, like they will have to "Windows Live". We could have all the best-in-class services and basically everything you can find on the Internet integrated with the local desktop, while Microsoft will restrict it to their own stuff. In the Q&A-session after your keynote quite a bunch of security and privacy concerns were raised by the audience. How are you going to approach that topic?

    Havoc Pennington: The first thing here is you just have to be careful, trying to not have security bugs, trying to use encryption, where encryption is needed, privacy policies and all that sorts of things. Some people just have the objection, that they never want there data on any server in any scenario, I think that's just a very small minority kind of view. Most people use online banks, like several people pointed out to me after the keynote.

    Havoc Pennington: So I think the reasonable concerns are things that can be addressed just by being very careful. And if you decide it makes sense you can always keep your private key and just send encrypted stuff to the server, and as all of this is going to be implemented in open source code you can take a look yourself and make sure that no data ever leaves your machine unencrypted.

    Another possibility would be to just go on keeping all your stuff locally, I mean it's not like we are taking away capabilities, people can always do whatever they like. And even if - and I don't think we will do this - we require the desktop to run a specific service at one point, you could still run that locally on your own machine. What happens to an Online Desktop, when there is no net connection?

    Havoc Pennington: I think for some applications it's just ok to just say "doesn't work when I'm not online", I mean a lot of people are already using web mail, and you get the same problem there, so I think the offline problem can be overstated. But at the same time that's something that is possible to address technically, you can do different kinds of offline modes. Google Gears just came out, which is an open source browser plugin that allows you to do that.

    It's also something that I think is increasingly going to be less of a concern as WiFi and mobile broadband gets more widespread. This will get slowly better over many years and making something like Online Desktop really good is probably a two year process at least, anyways. Still, something like the Online Desktop will use a lot of bandwidth, which might not be a problem in Europe or North America, but in other parts of the world this could be a show-stopper.

    Havoc Pennington: Well I'd say, obviously it's going to be feasible in some places before it's feasible in others, but that's true of most technologies. Even the Linux desktop historically tended to use too many resources for the computers some people have.

    As far as bandwidth: Everyone is building networks at the moment to handle video streaming and all that sorts of stuff. Everything we'll do is going to be tiny compared to video streams. Are online applications going to replace desktop applications?

    Havoc Pennington: I think people are switching from the desktop to the online applications and in a lot of cases it's because the online applications are really different, they are just not the same thing. I mean Google Documents and OpenOffice Writer to me aren't the same thing. OpenOffice has much better printing controls and is like 400 times more powerful in areas like document formatting. But the Google Documents have the ability to get to them anywhere and share them with other people. For me: I don't really use the OpenOffice word processor, I never had a reason to, but I use Google documents.

    So what I think that we will see is that people that are heavy users of traditional applications are going to keep using it, cause they were looking for things like those powerful word processing feature. But there will be new people using the online application because it has different benefits. I think we will just see co-existence, but with a trend towards the online stuff, cause the features of the online apps are more important to people than the power of the traditional apps.

    Also there will be some increased blurring of lines between those both worlds with the online apps supporting offline storage and more complicated kinds of user interfaces, and on the other hand the traditional apps starting to support online services. But you are not actively pushing in a direction to replace traditional software with online applications?

    Havoc Pennington: I don't think it would make sense to tell people "You can't use this anymore". If people want to launch a local app, then let them launch the local app. The issue is, what the defaults are going to be. I think there should at least be a mode that kind of encourages you to use the online applications.

    And also for things like The Gimp or Photoshop, I don't think there is going to be a replacement for that in the online world which suits the needs of an artist for quite a while. Even for word processing, I don't think there are lot of professional novelists that are going to use Google Documents. Will people have to get a in the future to use the desktop, or will all of this be optional?

    Havoc Pennington: I would say this is totally optional for the foreseeable future, at least until nobody cares anymore. But in the end it's all about the defaults and that's something for the distributions to decide. I mean nearly nobody is installing GNOME from source-tarballs, people are getting it from there distribution, and these will have defaults. And I guess at least on some of these the defaults might be to use the online stuff.

    But I think the way we have done it in the prototype so far is a really fine model, we have different sets of default config options and we have that little command line utility which flips the defaults and then you are in a different environment. Do you envision all the distributions to use or do you think they will just set up there own servers?

    Havoc Pennington: It's hard to say how it will work out in the end. But I think at least if you want to have some sort of social features there is a big advantage to having one server, so I think we should do whatever we can to be sure that everyone is comfortable with that server. Cause you will want to have all your friends on one server, and not people using Ubuntu on a server and people using Fedora on another. That would be suboptimal.

    But if you are just storing peoples private data, then it doesn't matter on which server you do that. It might even be better to have each distribution to have its own server for that, because then each distribution can pay for their users. The would then just be important to make sure that you don't have to buy your own server to do a distribution. Like for smaller distributions or developers who just want to compile it themselves an open community server would be a big advantage. Do you have some sort of timetable for the GNOME Online Desktop?

    Havoc Pennington: That really depends on how many people are interested in the idea and what it evolves into, so it's hard to say. But obviously there are already some prototypes and we'd like to do it as quickly as possible. Something usable could be done in 6 - 12 months, in practice that might take a bit longer, though as software always takes twice as long as you think. But if it takes us four years we will be just to late to make any difference. Are the users going to see some of the first parts of it in Fedora 8?

    Havoc Pennington: Yeah, we're going to have a demo-ware-type something in Fedora 8, but really more like "Here's something, it's not ready yet, you can mess with it". But it won't be there by default. It's just not fully baked at the moment, it's just too early.

    Still it's moving faster than GNOME originally did, I mean GNOME panel started as a grey rectangle and that was pretty useless. And in my opinion the entire GNOME 1.0, 1.2, 1.4 series was pretty bad, so it just takes time to make software really polished. A lot of people seem to have the impression that Red Hat is now coming back to the desktop, would you agree on that?

    Havoc Pennington: No, I think that is misleading. I don't really agree that Red Hat ever left the desktop, I understand that's a popular deception, but most people seem to base that on when we canceled the retail box set product. People are confusing retail product with desktop and they just have nothing to do with each other really, certainly not how we thought about it at Red Hat. The retail box vs. download is a business model thing and has nothing to do with what the product does which did not change at that time.

    So I don't think that perception was ever right, and the other thing is, we were basically the largest contributor to GTK+ and GNOME for many years and we continue to be so. We have a lot of people at Red Hat working at the desktop.

    I mean the way we've approached the Linux desktop has been fairly consistent, which is enterprise desktop and then kind of new angles on the desktop, things that weren't just to do whatever proprietary competitors have been doing. So for instance One Laptop per Child was a way that we wanted to approach the desktop.

    For me personally - not speaking for Red Hat - "make a desktop" is like the worst possible mission statement. It's a terrible way to understand a problem, it builds into your problem statement, that you are just going to copy things, that have already been done and that people already have and that you think that they don't want to change what they are using. It's just not very forward looking, it doesn't tell you anything about what your desktop is like. I mean if you look at the new features in every Mac OS X release, they could have never come up with this feature list, if their way to think about it would be "we are making a desktop" and just left it at that. The way they probably think about it is: "Let's allow people to create lots of new content" or maybe just "let's have some cool demos", I don't know, but it goes beyond "let's make a desktop", that doesn't say much. There have been lots of statements at this years GUADEC, criticizing the lack of a consistent vision in the GNOME project, would you agree on that?

    Havoc Pennington: Yeah, there's some truth to that. But I don't think that's something that is inherent in the community and can't change. The reality is not that many legitimate opportunities for things to do just like come along every day. So I'm proposing this Online Desktop-thing and there is One Laptop per Child, there is mobile devices, those are maybe three opportunities for the future. Thank you for the interview.

    (Andreas Proschofsky)

    Share if you care.