Upside: How would you contrast the open source approach under which you developed Linux to Microsoft's proprietary approach with Windows and Windows NT?

Torvalds: With Linux, the people who use the system get to [affect the way] the system [develops]. It's democracy in the sense that you don't surrender control. Anybody can do anything. It boils down to [the fact that] you must be [competent], but that's a good way of separating the people who do the work. And even the [people who] don't make changes can make suggestions and can do testing and things like that.

I think that as a development model the open source model is superior--especially [compared with] the traditional commercial proprietary setup. It [comes] down to motivating people to [create] certain things.

But there's a lot of work that doesn't fill the [criterion] of being interesting, and when something doesn't fill that [criterion], you [must] have some other motivational factors. The obvious one [is] money. It's easier to make money to some degree in a proprietary setup. A proprietary model takes advantage of the fact that you limit competition by not giving everybody access to all the pieces of the puzzle. And limiting competition is nice for a company.

Upside: Do you have an analogy that illustrates the differences between the open source and proprietary approaches?

Torvalds: I'd say something like this--that you'd have a printing press where the presses own not the language but the letters of the language. So, you could have an English printing press that owns the English letters. Users can read or write whatever they want, but the person who controls the printing press controls what is made widely available. You can read or write by hand, but when you want to make improvements, you'd better make them yourself or find favor with that printing press.

Upside: On the other side--the open source version of the analogy--one printer wouldn't control everything.

Torvalds: Then you could have any number of printers, and they can share the same letters.

Upside: That's an interesting analogy. Tell me why you think Linux and the open source model has been so successful.

Torvalds: One reason Linux has been so successful is [that it's] the underdog. To some degree, the market forces didn't matter to a lot of the Linux developers. People developed [Linux] because their interest wasn't getting market share, and they didn't have a board of directors [to whom they had] to explain every quarter's results.

You also change the market by changing the rules in ways that Microsoft isn't willing to fail on. So, making source [code] available is a selling point. Suddenly you have a new rule; not everybody cares about that rule, but those who care about [it] dismiss Microsoft out of hand. It's like in war: You don't let the enemy select the battlefield, you select [the] battlefield.

Upside: Do you think Linux may lead Microsoft and other big players to alter their strategies or the battlefield?

Torvalds: To some degree I hope it will. I'll use an analogy I think is accurate. Let's pick five big software companies in the [United States] right now and compare them to [the major] car manufacturers in the [United States] in the 1950s, where [manufacturers compete] on adding things to the car, and every year you have a new model.

Does this sound familiar? Every year you have a new version of Windows. And you make people aware of [which] model year they're driving. So, there's a big difference between the '51 Cadillac and the '52 Cadillac. People are very aware of the year. Why? Because the car manufacturers want them to be aware of the year or the version they've been running--[the manufacturers are] making a lot of money selling the next year's model. Even though the previous year's model works fine, [they] still want to sell next year's model.

Think of me and Linux as the Japanese car industry. Yeah, we had an oil crisis and other issues, but basically what you have is a car industry that is not oriented [toward] quality but is oriented toward new features and a new release every year. Then in come the Japanese carmakers, and they aren't known for quality at first. But it eventually dawns on [consumers] that, "Hey, the Japanese cars are better made. They may not have the fins, but on the other hand, why am I changing cars every year?"

I think that's an accurate analogy. If Microsoft is going to change [its] practices [the way] the U.S. car industry did, I'll be happy. If Microsoft can change and compete on quality, I've won.

Upside: Microsoft President Steve Ballmer recently broached the idea of open sourcing Windows. How do you view such talk?

Torvalds: I've done all my development on the Internet, and the first thing you learn on the Internet is that talk is cheap. I don't believe in anything until I see the results. It's true of anything, and it's true of Ballmer spouting off. Show me the money!

Upside: On the subject of Ballmer and true intentions, do you think Microsoft is truly afraid of Linux, or has it been using the publicity as a legal argument in its DoJ case?

Torvalds: I think it started as a legal argument, especially last fall when Linux started to get some attention. Nobody inside Microsoft took [Linux] that seriously, but they saw a great [opportunity to] use it as a legal argument. [But that] backfired because it got a lot more journalists interested, saying, "What is this Linux thing?"

Upside: Do you think Linux could ever become more widely used than NT?

Torvalds: Between Linux and NT, the answer is a clear "Yes." Windows NT and Linux have [nearly] the same size market share. [Editor's note: According to International Data Corp., at the end of 1998, Windows NT had a 38 percent share of the server market, compared with 16 percent for Linux, 19 percent for other Unix servers and 23 percent for Novell Inc.'s NetWare.]

Linux in the future will pass [NT]. I'm not going to be [too] surprised. The real goal--it's not going to happen in a year or two, but it [may] start happening in three, four, five--is the desktop.

Upside: You think Linux may challenge on the desktop in three or four years.

Torvalds: In three or four years, I hope Linux will be there as an alternative for [nontechnical computer users].

Upside: What would that do to the industry and to Microsoft?

Torvalds: It doesn't necessarily mean that much. I think [the market is] in a sick position where people don't have choice.

Upside: You call it sick?

Torvalds: Yeah, it's sick. You have a huge homogeneous market, and you have one entity that controls a large portion of that market. [But] eventually monopolies do not work.

In emerging markets, often one company holds [enormous power]. That happened to the oil barons, it happened to cars, and it happened to computer hardware, where IBM [Corp.] was seen as unbeatable.

It always results in a more stable market when you have five or six big companies. That's why I think the current market is sick.

Upside: What about the rivalry between the freeware and the commercial advocates? Some purists don't like Red Hat Inc. or other companies that sell commercial versions of Linux; they think Linux should forever be free.

Torvalds: I try to avoid things [that are] black and white, and my opinion is that with Linux you can avoid the commercial. You can download everything you want [from] the Internet, and you don't have to pay a cent. And frankly, I don't want to be in that position anymore. I'm happy that these days when I get a new machine, I just insert a Red Hat CD or SuSE Inc. CD, and I install it that way and then add my own special things on top of [it]. I don't have to pay for that, but even if I had to pay I would. They're providing a real service. It's a choice. And if you [think in terms of] black and white and think Linux shouldn't make money, then you have that choice, but I think you're limiting yourself.

Upside: It sounds as if you're not concerned that Red Hat and companies like it may co-opt Linux.

Torvalds: It's [because of] experience that a company like Red Hat doesn't want to take over kernel development--they know it's expensive. They want a lot of kernel developers on staff because when a problem comes up, they want to have people they can consider theirs.

But they don't want to get involved [in kernel development]. A commercial company, if it's any good, wants to do market research to see what the hell people are doing, where [it] could give extra benefits and value to users, makes sure [the product is] easy to install and [addresses the] important details. It's about finishing touches, and it's marketing and logistics.

Upside: What do you want to see Linux become in the next five years?

Torvalds: To me, the most interesting part is always [the many] different uses. I'm interested in the embedded market because that has some specialized uses you don't find anywhere else. Supercomputers are sexy from a technical standpoint, but they're also interesting because they have requirements you don't find anywhere else.

I think the most interesting market is the desktop because the desktop isn't specialized. And that, in turn, means there's never one perfect solution because people are using so many different things. That's a hard problem technically to solve, and that's why I've been most interested in the desktop environment. I hope that in five years you'll see Linux being a desktop choice. Maybe not displacing Microsoft, but at least being a real choice so [that] when people go into computer stores and buy their computers in 2004, they'll have to [stop and ask themselves whether they] want Linux installed, or Apple, or Microsoft.

Upside: The Linux, open source business model seems to be working. Are there any other emerging business models you find interesting?

Torvalds: The one I find intriguing but don't believe is the "Internet phrase" business model. The Internet phrase business model [consists of] adding e- before your name and increasing your stock [valuation] by an order of magnitude. I find it intriguing, but I don't think it [will] be successful in the long term.

Upside: What do you think will happen?

Torvalds: I think you'll find that the old-fashioned industries [will] be very aware of the Internet. A few companies [will] get a big name [from] the initial rush, and those few [will] do OK; the rest [will] be one in a dozen. Maybe Yahoo [Inc.] will survive because it [has gotten] such a big name that [it] can make money [from] just the name--name recognition is important. But they aren't going to make money on their business models, and they probably know that.

Upside: One last question. Where do you think the cost of software is headed?

Torvalds: PC [companies] and Microsoft got into the market by making cheap software at a time when software was expensive. You had these small, aggressive companies--Borland [International Inc., now Inprise Corp.], for example, was well-known for making compilers, and Microsoft started doing Basic. These weren't expensive [software packages]; that's how they entered the market. The sad fact is that now that Microsoft [has] such a strong hold on the market, prices [have started] shooting up again. [Inprise] made a big splash by making one of the first compilers for [less than] $100, but that big splash is long-forgotten. We're getting back to the point where software is [becoming] expensive because you have big companies that again have the clout to say, "You will pay this much."

I'm hoping we'll have the big splash again. Now there's an economic model that makes it possible to sell software at a high price, and I'm convinced it's going away. That makes people think I'm a communist, [but] I'm not--I'm a big believer in competition. I just think competition will win, and prices will come [down].