Torvalds Q&A: SCO Lawsuit Has Nothing To Do With Linux

Linus Torvalds, the creator of Linux, spoke with Editor Heather Clancy and Editor/News Steven Burke last month at CA World about the SCO lawsuit and the future of Linux.

CRN: How has the SCO-IBM lawsuit affected Linux?

Torvalds: The biggest effect by far has just been a lot of time wasted on discussion. Obviously there have been a lot of people worried. But it hasn't actually affected [Linux] in any real sense. Part of the reason is that it hasn't affected it in any real sense is the way we have done development, because it has been so open, there has always been a very real electronic trail of exactly how everything came into the kernel from which source and stuff like that.

So we actually have a very good notion of where the code came from and what the [intellectual property] rights are. ... It seems to me that what SCO really minds is not so much the code itself. It is the contract lawsuit with IBM. I don't know what the IBM contracts are. It is kind of ironic, because especially when it comes to the stuff that IBM has given Linux, we have been very, very careful about how we accept them. The one thing SCO has mentioned has been the Read Copy Update code that IBM gave us, and that wasn't accepted for the longest time into the kernel exactly because we knew the patents were owned by IBM. [But] we said we couldn't take it until you [IBM] said very explicitly that you also license the patents.

CRN: Do you have an explicit IP protection or due diligence process for Linux development?

Torvalds: We don't have an explicit one. It is kind of strange because the open-source community is regarded as being fairly laissez faire. But at the same time, the people who actually do the work take copyrights very seriously. Copyrights are what we use ourselves to kind of do our work. I know way too much about copyright law. I should not know as much as I do. The way things are organized we don't have a process like you would have in a company usually.

CRN: With the current situation, are you changing that at all?

Torvalds: I am personally convinced that exactly because we are so open we can follow the code through any time. If something bad happens, you have the trail, you can see who did it, what happened, how did it get here, which is actually not that common in proprietary systems. It is actually much harder, usually, to see that in other systems just because you can't go through the main list archives. That in itself says if something bad happens we can stop it. We can go and look at what was going on.

CRN: What is your advice for solution providers who may be concerned about the suit as they are building business solutions with Linux?

Torvalds: One of the issues is the suit really isn't about copyrights or IP at all. SCO and Darl [McBride, SCO CEO], dear Darl let's call him, have been talking a lot about IP, but in fact the suit is about a contract dispute between SCO and IBM. And I don't care about contract disputes between SCO and IBM. I think IBM has the lawyers to take care of that. I also think whatever happens, happens. The good thing about it being a contract dispute is that is purely between IBM and SCO. It has nothing to do with Linux. It has nothing to do with any users. Obviously SCO is trying to kind of push that notion that IBM violated their contracts and now IBM lost their license to AIX, which, let's face it, nobody really believes that. But it has nothing to do with Linux at all.

CRN: It's awfully weird timing, considering that Linux is really starting to build a business following.

Torvalds: But that's the thing that makes it not so very weird timing. I can't say that I expected SCO to sue IBM. But I mean it was clear that in the U.S.-business kind of climate [that] once enough money is involved, lawsuits will happen. This is not an 'if,' this is a 'when' question. And most lawsuits are resolved. This one has gotten a lot of press because Linux finally got big enough that people decided we can make money more easily by suing somebody than by using Linux. In the end, SCO is not a very surprising [company to bring a lawsuit]. Their business was zero and it was shrinking.

CRN: Are you playing a role to try to resolve this thing?

Torvalds: Not really. I want to have as little as possible to do with lawsuits. I am in the situation that maybe I will end up being a witness to one or the other side, most likely it will be IBM. But I am not involved in any way and I don't really want to be.

CRN: What kind of feedback are you seeing from solution providers?

Torvalds: I am not working with those people. Everything I hear is basically saying nobody cares. ... The people I work with are my technical people. They are worried about the lawsuit just because they want to make sure that we didn't do anything wrong.

CRN: Recently Microsoft lowered its price by as much as 35 percent to try to win the City of Munich's desktop upgrade business. How are Microsoft's selling strategies changing?

Torvalds: You are talking to the wrong person. You should talk to the CA person. I am really, really happy that I have never been involved with the salespeople, because that is never what I was interested in. So I don't see Microsoft.

CRN: Are you being called in by vendors such as CA and systems integrators to help win over some of these big Linux deals?

Torvalds: No. I never go to customer meetings. I don't like customers (laughing).

CRN: Is that why you decided to go to Open Source Development Lab and not a commercial vendor?

Torvalds: To me, the most important thing has always been that people be able to trust me. That doesn't mean that they agree with me. It just means that they know what my motivation is. And then it is very important not to be at a company where people start wondering, 'So is he motivated by the company?' I'd much rather be in a situation where people know that I make my decisions on my own personal grounds. Even when people don't agree with me, they are a lot happier about that than me being part of a commercial company and maybe making my technical decisions because of something their competition is doing.

CRN: So really the only thing you have is the Linux trademark?

Torvalds: In the U.S. legal copyright system, I do have one special right as the copyright holder of the kind of Linux kernel itself. I don't have copyrights to all the pieces. But I am the original copyright holder, and that gives me some legal rights. But then I give those away, so I don't actually have any other real rights or recourses. I own the trademark, which I try not to have anything at all to do with. I was forced to take it. This was back in 1995 or so. The trademark had actually been bought by another person, and it [basically involved] going to a judge and saying, 'We want this trademark because this other person shouldn't have the rights to it.' The lawyers getting the trademark felt that the person who needed to be the other party was me. Not because I wanted the trademark. But because I was the only other party where it was clear that I should own the trademark. So since then I have had it.

CRN: How has owning the trademark changed your life in terms of the financial compensation you get?

Torvalds: The trademark doesn't bring in any money at all. It sometimes covers some of the legal costs of just owning it. But that is the minority of the cases. So the trademark has really been much more of a bother than anything else. Especially as trademark law actually requires you to maintain your trademark. Unlike copyrights, you own the copyrights when you write something, you don't have to do anything. Trademarks you have to go through the motions. This is why companies send cease and desist letters when you use their trademarks. Even kernel.org is maintained by a kernel.org foundation. The foundation doesn't do much. It's really a legal entity to allow like companies to donate hardware and things like that. Actually, the network management system is the more important one.

CRN: Could you talk about the more formal role you are taking at OSDL and what it will mean for the Linux community?

Torvalds: It doesn't really change a lot of things. What OSDL allows me to do is a. continue to do Linux full time, and b. for me, more importantly, do it on a neutral ground.

CRN: Will it accelerate the Linux development cycle?

Torvalds: I am seeing it more as a way to avoid burnout, frankly. I spent a lot of time at Transmeta doing Linux. It came and went. Sometimes Transmeta stuff took precedence. Lately, for 2.6 releases, Linux stuff took precedent. So in practical terms, I spent most of my time for the last six months anyway on Linux. So moving to OSDL will not change that. I will still spend most of my time on Linux. What it does mean is it is my job now. So there is less distraction. There is less feeling of guilt about taking over my time. So no, it won't accelerate Linux.

CRN: Is the Linux 2.6 kernel complete?

Torvalds: Being a test kernel, it doesn't mean it is complete. We are not touching it anymore. The features I know. Not all of them are actually in my tree because for 2.6 right now Andrew Morton will be the maintainer of it. I don't like dealing with stable stuff. I like doing the development stuff, the exciting new stuff. I have never maintained a stable kernel. ... Now we are just testing. Development is fun. This is the other side of development, having to make sure it is all done.

CRN: What are the exciting things that have your attention now?

Torvalds: Next week is going to be the Linux kernel summit up in Ottawa. That is going to be one place where we talk about what is going to happen in 2.7, the next development kernel.

CRN: What is the time frame for 2.7 and 3.0?

Torvalds: I don't know whether the next one will be called 3.0 or something. There are always two simultaneous tracks: there is the stable kernel, which for now is 2.4, and there is the development kernel, which has the odd number, so that is 2.5. Now we are getting ready to make the stable kernel move from 2.4 to 2.6, which is based on 2.5. That takes a few months. At that point 2.4 will still be maintained. The same way, actually, there are people maintaining 2.2 and 2.0 because some sites, especially in the commercial world, something works you don't touch it right. So people are still writing kernels that are five years old and they are maintained. But it is like a very low level of maintenance. It still works. Security issues are fixed. Not a lot. So 2.4 will eventually move to that. 2.6 will be the stable kernel. For a while we will concentrate everything on 2.6. Everybody will work on that. That takes usually like half the year or so. Then we open up the next development tree especially during the half year when we concentrate on 2.6 a lot of the developers get antsy and you have people who have ideas that they want to start working on. So 2.7 will have a lot of new things.

CRN: So what type of new things will be in 2.7?

Torvalds: Well one of the things that we still don't have is cluster file system stuff. That has been talked about. There is no firm plans. ... I am actually very happy at where 2.6 is. We did everything I wanted to have done. ... I still need to get the stability phase done. But I don't have any personal things I want to push until 2.7. So I am OK with that. So this is why you have events like the kernel summit. They start griping to me about what they have and felt was left out. I know some of the things they feel was left out. But you just need to start talking about this. A lot of it will happen during 2.7 development, quite honestly.

CRN: What influence will the on-demand industry dialogue have on the summit next week?

Torvalds: It probably won't impact the kernel all that much. The kernel is largely not affected very much. There are small pieces. Actually, the cluster file system does actually end up playing a big role, but most of it ends up being the tools to administer the thing and things like that. Apart from maybe the file system issues, it is not going to be any of the major focus for the kernel summit.

CRN: Will there be any desktop issues discussed at the Summit?

Torvalds: This is purely a kernel summit. It turns out that there are more issues on the desktop that are kernel-related than often in the server space. The obvious example ends up being device drivers. On the server space, five device drivers are fine: you like need a SCSI controller, you need simple graphics, you don't need 3-D graphics. ... But on the desktop you just have tens of thousands of devices and things like that, and those all impact the kernel in some way. So the desktop tends to be a lot more interesting from a kernel perspective, right? Things like doing 3-D graphics for games--that is what drives 3-D these days. The desktop has been driving a lot of these things for a long time, which people didn't expect ten years ago.

CRN: What kind of technical improvements might we see in the kernel that are going to make Linux more attractive on the desktop in the future?

Torvalds: I think the real attraction is going to end up being all the programs outside the kernel. I said for the last five years that the kernel isn't the most important part. The most important part are things like Mozilla, OpenOffice, KDE, Gnome--all of which depend on the kernel, but they depend on the kernel in the sense they don't want to care. To them, the kernel really just hides all the ugly hardware details. They just want to have the infrastructure, but they don't want to really worry about it. That is really what drives most of development. There are some things. For 2.6, one of the things that is new is threading.

CRN: What is your goal to get accomplished at the Linux summit?

Torvalds: I don't have a goal of having a five-year plan when we come out of there. That is actually against all my goals. The only real goal, as far as I am concerned--and I suspect most of the other developers too--is more to get a feeling for what other developers and what companies kind of are expecting for the next generation. It is more of a get people together, start the communications channels going.

CRN: Are you putting in place anything to make it simpler for collaboration to occur in the Linux community?

Torvalds: We do some of that. A lot of it just evolves from how people work. And then sometimes quantifying those things. Some of that can be very simple, like just having certain rules: When you send a patch to me, you actually put in the subject line patch. Not everybody does it, but most people do. It is like very simple social mores.

CRN: What applications do you see having the biggest impact on Linux?

Torvalds: I think the one a lot of people are talking about is obviously the Exchange Server replacement things. There are a number of them. There are some of them out there commercial already: I know SuSE sells one. ... There is also [Mitch] Kapor's application. So ... for the businesspeople, that is one of the things going on. The thing that has had more impact has been the infrastructure to make it easier to do one off applications. KDE as a desktop is wonderful. It is very important to have a nice desktop that people, especially those coming from Windows, are not that confused. Also, Gnome does the same thing. But what is even more important there is all of the infrastructure, all these libraries, all the tools to build these applications that would be completely impossible to build from scratch. But if you have enough infrastructure, you just put the infrastructure together, you add some glue and you have a nice application. Suddenly something that could have taken you 15 years to build, you actually have 15 years of infrastructure to build on and you just add your piece. And that has been much more important, I think, than any single application in many ways. And that takes a long time. We are in much better shape. For Microsoft the one application that made a huge difference was Visual Basic. And I am talking about this kind of infrastructure. Visual Basic itself was just the tool to build all these things. And that took the world by storm. And now, just within the last year or so, all the Gnome and KDE have all that infrastructure in place so it is much easier to do these SQL applications. That has been very important. We didn't have those four or five years ago. And that was like a complete vacuum. These days they are mostly there. Even like OpenOffice. The OpenOffice people don't just make OpenWriter and OpenCalc and all these applications, they had to build the infrastructure to do so. So part of the OpenOffice stuff is again all the infrastructure to, like, embed a word processor into bigger things. Sometimes you lose sight of the infrastructure when you only concentrate on one particular application. And I think that OpenOffice has been hugely important for having OpenWriter and basically having a word clone. But in the big picture, it is probably even more important that there is now the infrastructure to build on.

CRN: Talk about your vision for Linux and what it is going to be five years from now vs. the Microsofts of the world?

Torvalds: This is the one thing I never wanted to have. Five-year plans are something the Soviet Union had. And I am convinced they just don't work. And I don't want to have a vision. I mean, I have a feeling for where I want to go right to some degree. But it is a fairly fuzzy thing. And what ends up happening is all the requests coming in, all the patches coming in--they get filtered by that feeling. Sometimes that means that I won't apply a patch because I just don't think this is where we want to go. Sometimes, more often, it ends up saying, 'I see where you want to go with this patch, but it is not really what I want, can you do it this way instead?' But I don't have a big vision or a plan.

CRN: What is the technical issue with regard to Linux that keeps you up at night?

Torvalds: The only times technical issues keep me up at night is if I have a problem and I just can't figure out how to solve it cleanly. That happens but it doesn't happen that often. ... There was this big DRM [digital rights management] discussion three months ago, or something like that. That is the kind of thing that keeps me up at night--not because it is about technology, but because it is about making sure people are on the same page. ... There were some flames [e-mails] that weren't shone in public when this was discussed that led up to the DRM issues and the announcement I made. That keeps me up at night. Not very often. I sleep very well. I tend to worry a lot more about [things] like people's feelings than about technology. Technology is easy to fix. It's like you can make the wrong decisions, but that is okay because you can fix it later on.

CRN: Do you go with your gut when trying to decide what to do, or do you consult the community?

Torvalds: Quite often, I go with my gut feeling about things which works really well, because when you have done something like this for 12 years, your gut feeling has been honed over time. Even more often I just trust the developers I work with, especially when it comes to certain subsystems. I work with them for five to 10 years. I trust them. I am not going to micromanage them. It would be stupid for me to try to make their decisions for them. The whole point of them helping me is that they help me enough that I don't have to get my hands into all of it. I don't personally scale, the way Linux development scales. Other people around me, they don't scale either, but there can be many people I trust. ... There is still a fair number of odd patches from just people. During 2.5, I did the statistics a few months ago, there has been something like 750 people involve-- unique people, not just different e-mail addresses, in doing all the patches. That is a lot of people. Obviously I only really know 10 to 15 of these that well. Some of the patches come through the people I know. Some of the patches just come from random people who only do one thing. Of that 750 people, there were several-hundred that only had one small change that they sent to me, and most of those are fairly obvious, some of them aren't.

CRN: There is a clear choice here between Linux and the open-source movement and Microsoft. What is your advice to solution providers who are weighing those two choices?

Torvalds: I don't see it as my place to care. ... Literally, the only thing I care about is making the best code as possible. Then everybody else just has to make their own choice. To most other people, it is not just about technology it is also about how much impact can I have on the future direction of the choice I make today. That ends up being often the most important part.

CRN: Is there stuff that you are going to bring to the table in the 3.0 kernel that will make solution providers' lives a lot easier to build solution vs. the Microsoft product mishegas approach?

Torvalds: Most of what I do ends up being really boring for most people. It is the day-to-day crap of just maintaining and integrating stuff. The most important thing that is going on these days ends up being outside the kernel. All the tools to make it easier to maintain and integrate and create new applications. And that is where the excitement is. I am kind of in the center. But at the same time, there is a lot of stuff all around. Think of me as the technical lead, not as the person who does most of it himself.

CRN: Are you going to be more visible now?

Torvalds: I don't think so. I have been always fairly high-profile.