Interview with Linus Torvalds -- "On Being Linus"
Interview with Linus Torvalds
Linus TorvaldsThe creator of Linux and uberpenguin, talks of his work at Transmeta, adjusting to life in Silicon Valley, some thoughts on Linux and Open Source, and the Tao of being Linus.INTRODUCTION:
The month of April, 2001 marks the thirtieth anniversary of Nikkei Electronics. As part of the plans for our celebratory 30th anniversary issue, we asked our readers, mainly engineers working for large Japanese companies, who they would want to see interviewed. One overwhelming answer was Linus Torvalds, the Finnish creator of Linux, the wildly successful Open Source operating system which is credited with giving Bill Gates and Steve Ballmer at Microsoft, at the very least, a few nervous moments. While a student at Helsinki University, Linus's decided to release the source code to the operating system over the Internet, a decision which will certainly be looked upon as a defining moment in computing history.
Linus and his family, which now includes his wife and three daughters, have now abandoned the northern clime of his native Finland for the warmer environment of California's Silicon Valley. What brought him here was a job at Transmeta, a Silicon Valley startup developing a new CPU (Central Processing Unit) which is now giving Craig Barrett and Andy Grove at Intel a competitive prod in the ribs.
This led to Naoki Asami, the editor in chief of Nikkei Electronics, Rocky Eda, Silicon Valley correspondent for the same magazine, and myself showing up in the scantily decorated, nondescript one of a thousand in Silicon Valley lobbies of a Transmeta building, mikes in hand and cameraperson in tow.
Transmeta's CPU is innovative in that it uses software to help it emulate Intel's x86 architecture, the most popular personal computer CPU architecture. Linus, a former student of CPU architecture, is on the Transmeta software development. We talked with him of his experiences in working at Transmeta, life in Silicon Valley, musings on Linux and the Open Source movement, and not least, the philosophy of being Linus.
Nikkei Electronics: How did you get your job at Transmeta?
Linus Torvalds: I knew an engineer here at Transmeta who was originally from Sweden, and he'd been a long-time Linux user. He went back to Sweden on a personal trip, and, once you get to Sweden, it's very easy to come over to Finland where I lived at the time. He actually came over just to talk about Linux, and we hooked up. He told me that I can't tell you what I'm working on, but if you're interested, I can talk to some people. So they flew me out here. This was the end of '96. Linux wasn't that well-known. It was well-known in the technical community, but it wasn't a huge deal. Your average person on the street had never heard about Linux. So I had to sign an NDA (Non Disclosure Agreement). Everybody had to sign an NDA before they came into Transmeta. Transmeta flew me over from Finland without me actually knowing what they did. I was here for just two days I think, and the first day basically was interesting. But at the same time, I felt it was rather crazy. In '96 at Transmeta, I think we had five software engineers.
Q: Only five?
A: Only five people when I joined. And we didn't have any actual hardware. So everything was done in simulations. Simulations at that time booted Windows 3.1. This was '96, '97, so Windows 3.1 was kind of the normal Windows operating system at the time. And the demonstration at the time was to run Solitaire. That was pretty much the only thing you could do. And so after the first day I felt, okay, these people are crazy. But when I came back the second day I said, hey, it might just work. I mean it did work, but back then it wasn't that obvious. But I felt that the fact that people were a bit crazy was kind of a good sign, that it mean that people were doing something new and unexpected and on the second day, I was actually fairly excited about the position.
Q: What were some of the things that excited you?
A: I was just thinking about some of the optimizations I could see us doing. It takes awhile to kind of understand the architecture, and realizing that the chip is actually kind of a small database engine. And it's a very limited database. From an OS designer viewpoint, it's kind of interesting what we actually do under the covers. So that was probably what made me the most excited, just thinking about what you can do with an architecture like ours.
Q: Why did you decide to join a hardware company and not a software company like Oracle or Microsoft?
A: Why didn't I join Microsoft? [LAUGHTER] We are a hardware company, but we do have a fairly strong software side, and I've never been on the hardware side of the company. It's been kind of interesting to be at a company where hardware is so important because it gives a different view to some of the problems. So in that sense, I did join a software company. The thing that made it interesting was I still think that Transmeta is one of those companies that are doing things that nobody else is doing.There are parallels to Linux in the sense that one of the reasons, actually maybe the major reason, why I started doing Linux in the first place was actually to learn about the x86 architecture, and through that, I had a lot of knowledge about the details of how an x86 works. There are a lot of details which, even if you're a programmer using PCs, you don't need to know. You don't need to know how paging works, what the bids and the segment tables are. So I had actually the right kind of background for Transmeta and Transmeta was actually doing something that was very interesting to me because it was one of the reasons I'd gotten into Linux in the first place, namely looking at x86 from a software viewpoint. This meant that I could actually do something interesting. Actually it was a very natural fit for me at the time.
Q: Did you interview with any other companies?
A: Not really. To me even the Transmeta thing didn't feel like an interview. I came over to listen to what crazy things people were doing. I didn't have to get a job. I took it more that, okay, maybe these people are doing something really interesting and maybe I'd like to be part of that. So I didn't actually have a curriculum vitae or anything like that. When the word spread that I was taking a job with Transmeta, there were a few companies who said, okay, if you're going away from the university, we'll take you. And I never actually interviewed of any of them because none of them really offered anything interesting. Some of them basically said come here and do whatever you want to do, right? Which a lot of people think is the dream job, right? But what was interesting, what made Transmeta much more interesting than that was the fact that, hey, I was already doing what I wanted to do. What Transmeta offered was a very specific goal and something different that I hadn't ever done before.
Q: If Transmeta didn't allow you to work on Linux while working here, would you have joined them?
A: That was one of my requirements, that I'd get to continue doing Linux. I could have done it on my own time. It just helps a lot, if the company allows me to do it at work time and I don't have to worry about having to use a different Internet account for my Linux work and my Work work because it becomes fairly nasty.
Q: Since Transmeta is both a hardware and software company, have you found the two to be different cultures? I would think a project like this would bring together two cultures that normally don't work together so well.
A: One of the things we've certainly noticed is that software people are less controlled. Because if you have a bug, you can fix it, and it's fixed in the next version. In hardware, if you have a bug, it takes half a year until the next tapeout (initial chip - editor). I think hardware people tend to be much more careful, [LAUGHTER] while software people tend to be more let's try this. I think it works fairly well. But there certainly have been cases where software people say, hey, this is really hard to do in software. Couldn't you just redesign your hardware and do it this way and we'd be really happy and the hardware people say, mmm, that sounds dangerous. I think software people also, especially the people who have been most involved in defining the architecture, have become more careful. But that's been fun sometimes.
On Silicon Valley
Q: So what's you're here, what do you see as your next challenge, the next crazy task for you?
Linus: Well, that's kind of what the advanced development group does. And that's one of the reasons I'm in that group is because we can think of slightly crazy things that may not actually be useful in practice. I like moving to Silicon Valley, but at the same time, I have to say I almost expected more of Silicon Valley in some sense because I haven't seen any crazy projects. I mean last year you saw tons of crazy Internet startups. However, they were crazy in the wrong way. They weren't interesting technology crazy. They were just the market is crazy crazy. And I still think that from what I've seen at Transmeta, it is the most interesting project out here. Maybe crazy is the wrong word. It's just that in order to be really interesting, it has to be something that people think is slightly off the wall. It's not something every day boring, right? So when I say crazy, I mean that in a good way. Don't get me wrong.
Q: You mentioned that there's not as many crazy technology projects going on right now. Do you think that's a because of the times?
Linus: No. I mean times were really good two years ago. You could get money for anything. So in that sense it's calmed down a bit. I think it's really mainly the fact that the only company I came to look at here was Transmeta. I kind of got this notion that the whole Valley is 10,000 companies just like Transmeta and it turned out that it isn't. There's a lot of interesting stuff going on here, but at the same time, a lot of it is just engineering. Which is good, but it's not really exciting.
MALE: How about living here in Silicon Valley? Is this an area you need to be to do what you want to do now?
Linus: I really like it here. It started out as being a safe place to go because when you move to another country, you really don't know how you'll adapt to a new work environment and things like that. Silicon Valley seemed to be a safe bet in that, okay, if Transmeta doesn't work out, there's a lot of other stuff going on in this area. So it started out that way, knowing that a lot of the technology was done here. Then we got used to the weather which was a big factor. And there's actually a lot of things you can do here. There's a surprising amount of nature not that far away. You can go windsurfing, you can do stuff like that, which I never do because I'm too busy.
NE: How about your family? Does your family like it here?
Linu: I have three daughters now. The oldest one was born in Finland. She was 10 weeks when we moved. She doesn't even realize the difference. My wife, it took her maybe half a year to adjust. But we're adjusted.
NE: Do you want to stay here forever?
Linus: Forever is a long time. [LAUGHTER] But the only thing I really don't like is that coming from Finland, there's a lot of space. Finland is not very densely populated mostly because nobody sane wants to live there. That's unfair. But it's a bit too cold. So the only thing that really disturbs me about this area is that it's too crowded. House prices are ridiculous and things like that. But at the same time, there's so many things that make me like this area despite all the power outages.
On Linux and Open Source
Q: When you working on Linux in Finland and you decided to make it an Open Source project, what were your expectations and reasoning at that time?
A: Looking back it looks like a huge decision and something I would have thought about a lot, but it wasn't. I was basically communicating a lot on the Internet with technical groups and people. Back then, in '91, '90, most Internet people were at universities which meant that the whole philosophy on the Internet was fairly university-minded. The best reason for making Linux available at that time was probably just because it fit the culture.
I didn't have very high expectations. I wanted people to give some feedback, but from a market share standpoint, I didn't expect it to make all that much of a difference. It was more that I know some technical people who would be interested in operating systems. And I wanted to get feedback from them.
Q: So now that 10 years have gone by, what were the things that were most surprising, that you never really expected to happen after you made it Open Source?
A: Very few things were really that surprising at the time. It's not that I expected this to happen because I didn't. But all the steps were fairly small. For example, just the fact that Oracle ported its database to Linux was a huge thing. That was something that suddenly made people realize that this thing actually can work in major commercial places. And yet the rumors about Oracle porting had been going around for two years. So it wasn't a huge surprise to me that it happened. Looking back, it's really surprising that everything worked so well, right? But none of the details were that strange.
Q: None of that was surprising. As it happened, you said, oh, this is natural?
A: Actually, the biggest surprises were probably fairly early on, in early '92 when I'd only released it maybe three, four months before. I released it expecting some of the people I'd already talked to to take a look. Then at some point, I just noticed that there were a lot of people actually using this and that the people using it were people I didn't even know. That was probably psychologically the biggest surprise. But at the same time, looking back it was maybe 50 or 100 people who were using it. Now that sounds really, really small. But at the time, it felt like a lot. Basically it's a big jump to go from one to 100. But then from 100 to 10 million, that's big numbers. [LAUGHTER]
Q: Looking at your decision, when you decided to release Linux as an Open Source project, you were a student. That tends to be the most carefree period of most people's lives. Now you've got a family that you have to support and if you were in the same situation, if you were to come up with a new project like Linux would you have released it as Open Source?
A: I would actually. I can see where you're going, but actually when I released and did most of the early Linux work, I was a student, but I was also a TA (Teacher Assistant-editor) at the university and the most important part maybe was the fact that I wasn't nervous about the fact that I would get paid. I'm not trying to make money per se, but at the same time, I wasn't having trouble actually being fed. Whatever students eat, drink, pizza and beer, was plentiful. And even now I don't have this feeling that okay, I need to think about money all the time. I still do computers because they're fun. It so happens that they're also my work. But I still do my work, get paid for that, and if I did something on the side, which I would do just because it's fun, I'd do that Open Source. Considering that Linux ended up paying for my house, I have made more money out of Linux than out of my real work, even though it wasn't really meant to be that way. It just happened.
Q: How did Linux pay for your house?
A: Linux companies basically gave me friends and family stock, and I was the most friendly of the lot, [LAUGHTER] and it was like no ties attached. They said, hey, thanks, and here's a few options, and maybe they'll be worth something some day. And everybody was laughing about that and not actually believing it, and then -- The IPO's came around. Uhm-hum.
Q: How much of your stock have you sold?
A: Yes, I've sold the stocks. I'm not stupid. I sold when the market was crazy. I said I'd rather have a house than worry about where the stock market is going.
Q: When you look at Linux as it is now, obviously it's a major part of the infrastructure for companies. There's an industry that's grown up around it. Do you feel that it's still your baby? Or do you feel more like that you're the father of the bride who has just given his daughter away?
A: Some of both. For example when I go to a big Linux conference, say LinuxWorld or something, I don't feel like, hey, I did this. [LAUGHTER] Quite frankly when you walk on the floor it's so far off from anything I did. At that point I feel like hmm, okay, that was my baby and it grew up. But at the same time when it comes to the technology, the kernel itself, that part is still mine. Not all of it because I've always tried to not get in the position where people get irritated at me for being in the way. I control some parts of the kernel and for other parts there are other people who are better at it than I am, then they're maintainers of that part. But certainly the core functionality of the kernel itself is still my baby.
Q: One impression I have, even though you have all these journalists coming by to interview you, is that you really like to value your privacy. Have you ever regretted giving the name Linux to the operating system?
A: No. That was a good thing. That may not have been planned either. I had a plan for calling it Freax. As in "freaks" but also in "free." And I thought I'd call it Freax just because I thought Linux was a bit too egotistical. However, the name Linux stuck. I have a few people I can blame on that. [LAUGHTER] But I'm really happy it stuck because it's a good name. It's a lot better name than Freax which would have been hard to accept in a commercial setting. I think it made things easier from a PR standpoint, especially two years ago when people were talking about Linux versus Microsoft. It made it so much easier to personify Linux as me, Microsoft as Bill Gates. Good against Evil kind of thing. Which made it much easier to write stories. Which shouldn't really matter, but it does, for journalists. The only part I've ever disliked about Linux, is I'm not very good at giving talks. That's the only part that I'm kind of actively trying to get out of. It's not that I try to keep my privacy. It's more that I try to do the things that I find interesting and most of them tend to be staring at the computer for 10 hours a day which kind of implies that I don't want to be that public a person.
Q: Now that Open Source has really become an accepted software development style, looking at the next five to 10 years, how do you see the future of it?
A: The future won't be all Open Source. That's clear. I think it's going to be one accepted way of doing things, and it will take over in certain niches, especially in niche markets where you don't have a lot of economics of scale. I mean developing software is really expensive. It is only cheap to copy. So if you want to make money on software, you have to sell millions and millions. That's obviously what Microsoft does. And so in small projects, in niche markets, I think Open Source has the best leg-up, and then what tends to happen is that once you've found one niche, it kind of tends to spread just because it can. And I think that's going to be the pattern in most areas. I personally think that the only really hard market to break into is the desktop. And that's partly because of Microsoft and being such a good position from the marketing standpoint. But I think equally because the desktop inherently is a very hard market to enter. Because you have a lot of people who don't even like computers and they've already learned this one system and they don't want to change. They may hate the system and the computer, but they would hate it even more to have to learn something else. So that's one reason why the desktop market is always the hardest. Apple has shown it and IBM has shown with OS2 that it's usually easier to enter just about any other market. And that's true of Open Source too. I think that the big advantage that Open Source has and the thing that makes it different from OS2 and Apple, is basically that competition happens on a different level. When you try to go face-to-face with Microsoft on a commercial battleground, you're going to lose. There is no question about that. So you have to change the rules and do something different in order to not lose. And Open Source does that. I'm not saying that Open Source will win. I think it will actually, but it's going to take a long time, just because the market is hard.
Q: Looking at the factors that have made Open Source successful for developing software, do you think it could apply to other industries, for example, hardware?
A: I don't think hardware is necessarily the best industry to apply it to. I think Open Source is really an issue of information. And information is only useful in areas where the information is the most important thing. In many hardware areas, information is often very expensive. But it's equally expensive sometimes to produce the chips or produce the cards or something like that, so you have a noticeable amount of the expense coming from the actual raw materials.
Q: I was thinking perhaps medicine?
A: I suspect that's one area where it might be very interesting. If you look at where the cost is, it is almost never in making the drug. Eventually -- we haven't reached that point yet -- but it might be that if you look at genetics and see who is trying to de-code and understand what it's all about, there's a number of commercial companies in there, too, but there's a number of universities. And depending on what happens to the intellectual property rights in that area, it might well be that a lot of intellectual property might be more of an Open Source kind of property, if it's developed at a university, and I think that would be a good thing. But I'm biased.
Q: How about opening up Transmeta CMS (Code Morphing Software - editor) software as Open Source?
A: We discussed it at length inside Transmeta, especially among the engineers. One of the issues, this is the usual issue, in all cases, people were nervous about what would happen. So we open sourced part of what we do, but we haven't open sourced CMS itself. The question is maybe still on the table, but it's not actually discussed.
Q: You prefer to work on Open Source but you're not going to be dogmatic about it?
A: Yes, if I do something for fun, if I do something because it's my interest and if I don't have any other external issues, I'm going to Open Source it. But no, I'm not religious about it. I think that Open Source is the almost certainly the best way to get the best technology. It's not necessarily the best way to get a company where it wants to be, always. Sometimes the two go hand-in-hand, sometimes they don't. So you have to make that position. In my opinion, it's not my decision to make.
On Being Linus
Q: Especially on the Japanese side, many engineers feel pride about their work when they produce a successful product that is producing revenue for the company. How about yourself? What areas about your work that gives you the most satisfaction?
A: Some people just like to program. I think that's programming for its own sake. And to a large degree, to me most programs are in the end only as useful as people actually find them. I still program just for fun, but at the same time, a lot of the satisfaction, at least these days, comes from fixing a real problem for a real person who couldn't get his work done because of something or maybe we added a feature that somebody really needed. So in that sense, it's not about selling well or generating money. But it's the same kind of thing. But I still think that a lot of the pride comes from knowing that you did something the right way, even if people don't even notice. You know you could have worked around the problem some other way, but the pride is in knowing that you did the clever thing, you fixed it for good.
Q: We are also asking all of our interviewees for our anniversary issue this same question: what do you see as the most exciting things that have happened to the IT industry in the last three decades?
A: I've been barely alive for the last three decades. I would just say Moore's Law. It's not any specific feature. It's just the fact that things just improve so much and became so much cheaper. It's not a thing, maybe that's a bad answer. But I think it's something that the industry has had for the last three decades and probably will have for three more decades. Eventually, it will have to stop. So in that sense, I think it's something that has been interesting and is fairly unique to this particular timeframe.
Q: Can you give us the names of some people who you look up to? Who do you consider your heroes?
A: Very few people in the technology area.
Q: How about in areas other than technology?
A: I'd have to say scientists. Like Feineman or Einstein obviously or Bohr. These kinds of people.
Q: How about any engineers?
A: Maybe Steve Wozniak? People don't remember him as well as Steve Jobs because he was just a technical guy and he got out of Apple when it wasn't interesting to him anymore. But if you had to ask who in my industry I'd like to be like, I would probably say Steve Wozniak.
Q: For you, what is the definition of an engineer?
A: It's somebody who does it for the technology. So that why I count Steve Wozniak as an engineer. Part of the reason why I mentioned Steve is that I grew up in the timeframe when other computers were competitors to the Apple. So I was aware of all the strange things Apple was doing with their graphics and floppy chips and things like that. That was true engineering. Some of the things that Apple did in those early ages were things that people actually would not have thought possible. That was not what the chips were even meant to do. That's engineering, and that was Steve Wozniak. That was not Steve Jobs.
Q: Have you ever met Steve Wozniak?
A: Never met him. I know he's around, and he's teaching kids about computers.
Q: How about Bill Gates?
A: No. He's a business person.
Q: He was an engineer.
A: I don't think so. He was technical, but if you look where he went, he wasn't at MIT. He was at Harvard. I think math and law or something like that. Yes, he was technically knowledgeable, but he was not an engineer.
Q: Have you ever met him before?
A: No. We've been to some conferences.
Q: Would you like to meet him?
A: I don't know what I'd say to him. I'd probably like to meet him.
Q: Have you ever thought about what you would like to be remembered for? Something you would like someone to put on your gravestone? You still have a lot of time ahead of you.
A: I don't know. I've never been the kind of driven person that some people are. They have a big goal in life and they want to climb Mt. Everest or something. I've never had anything specific like that. And if anything, my gravestone says, Good Guy. I'm okay. Maybe that sounds a bit too goody-goody. It's not that I want to be Mother Theresa. It's more that I want to do something interesting and do it in a way that's respected.
Note to Steve Wozniak. E-mail Linus and invite him out for a beer. I think you two will have a lot to talk about.