The father of Berkeley Unix shares his thoughts on Linux, Open Source licensing, and the history of vi.
|ALL PHOTOS © GARY WAGNER|
As one of the creators of Berkeley Unix, Bill Joy knows a thing or two about developing and marketing a free operating system. Sun Microsystems’ chief scientist has survived the Unix wars and has watched both his company and its chief competitor, Microsoft, grow from tiny start-ups to industry giants. Though he has had a major hand in the development of such important Unix technologies as NFS (Sun’s Network File System), the Berkeley Unix TCP/IP stack, and the vi text editor, Joy’s current obsession is trying to build a thriving development community around Sun’s Jini distributed computing technology and its not-quite-Open Source software licensing model. Joy recently accepted Linux Magazine’s invitation to dinner, where he gave Publisher Adam Goodman, Executive Editor Robert McMillan, and Associate Editor Eugene Kim the lowdown on what Sun thinks of Linux and Open Source.
Linux Magazine: One of the reasons we wanted to talk to you was that you have a long history with and a broad perspective on Unix and free software. What do you think of Linux? A lot of people talk about it as more than just an operating system.
Bill Joy: It’s actually less. It’s just a kernel if you want to be technical about it. It’s politically incorrect to conflate Linux with the applications. At least one person will get upset. So to be quite precise, it’s just the kernel of the OS. When we did Berkeley Unix, we were doing the operating system and all of the applications.
In a lot of ways, the Berkeley Systems Distribution (BSD) was on the road to being free with source available and many of the things that Linux is. But it got hung up in this legal fight between the University of California and Unix Systems Labs.
Those are the accidents of history. Now with Linux, we have this new version of Unix written with similar kinds of values that BSD had. One of the great strengths of Unix is that it’s been rewritten and reimplemented several times. Applications with similar names and similar functions are widely understood, which allows this healthy kind of invention and reinvention to occur.
LM: So if it weren’t for the lawyers, we’d be called FreeBSD Magazine?
BJ: If BSD had been free, there would have been no reason to rewrite it. The new thing that happened with Linux was cultural. The Internet is now coupling people together in ways that probably couldn’t have happened before. How else would the developers have found each other?
I did my work in the era of the magnetic tape. We sent Unix in source form to thousands of people; they sent us a few hundred dollars, because I had to pay for the postage and for the printing of the manuals, and that was our network. It was a postal-age speed thing. It was not very convenient.
LM:Were licensing issues as important back then as they seem to be now?
BJ: No. I knew I needed a license for BSD because at some point Berkeley was going to discover it. So I just took a license from the University of Toronto and modified it a little bit and started using that. I figured if I sent people a tape, and there was nothing for them to sign, they wouldn’t take it seriously.
When you give things away for free, often people think that’s what it’s worth: Nothing. So charging them a small amount and giving them a license to sign actually created a perception of value. I’m not saying the tape didn’t have value, but an awful lot of stuff comes across your desk that you just throw away.
LM: So, what did your license actually say?
BJ: I don’t remember. It was a one-page thing. I didn’t have any lawyers look at it and I’m not a lawyer. I just made it up as I went along.
What happened was that at some point we were getting to be big enough that we were sending out hundreds of these [Unix tapes] a year and charging hundreds of dollars for them. A quarter of a million dollars in revenue is a great deal of money for a graduate student. Scott McNealy likes to say: “To ask permission is to seek denial.” And we were operating with that philosophy.
But there were huge amounts of money involved and we were becoming pretty visible. So eventually we decided to send AT&T a letter asking them: “Is this okay what we’re doing?” And 18 months later they sent a letter back: “We take no position.” We won’t answer your question. So that’s what it was like to deal with a regulated monopoly of lawyers. That same sort of legal structure is what caused [AT&T] to license the transistor for nothing.
So we couldn’t actually get an answer from them and it was only years later that this whole fracas erupted around who owns the code. It turned out their code was as tainted with Berkeley stuff as ours was with theirs, so they eventually came to a truce. That’s what I’ve heard second hand or just drinking wine with people. So there’s a very tortured and funny history to all this code.
LM: Have you ever contemplated what it would have been like if you’d released your code under the GNU Public License (GPL) or something similar?
BJ: I don’t see what the advantage to it is. The important thing is that people have the source code. I actually think it’s fine that people can take BSD and make improvements to it and reap software profit.
I don’t think, given where we were and what we were trying to do, that the license made that much of a difference. At Berkeley, we had the model that software is the result of your research. The university tradition is that when you do research, you publish. So not giving people the source code for software meant that you weren’t publishing your research. A fundamental model of BSD was: Software is a result of our research. We’ll publish it and other people will use it if they choose. If someone commercializes it, I don’t particularly care, because if you publish research in a university, people can commercialize it. That’s just the way it is.
The important thing in my mind is that people share stuff. We’ve done something at Sun — Community Source Licensing — which is another spin on this. But the fundamental principle in my mind is that people get to see the results of other people’s work in a way that they can stand on shoulders rather than on toes. The details can vary; there can be many approaches and they work in different contexts.
I think the GPL is fine. I just don’t necessarily agree that it will achieve everything that Richard Stallman thinks it will. I’m not as religious about this as other people are.
LM: Just what was your involvement with Sun’s Community Source License?
BJ: I was the instigator of it.
LM: Did you at any point evaluate the GPL for Sun’s projects?
BJ: I can’t license all of Sun’s intellectual property under the GPL, because it just won’t work. I don’t see any reason why I should give somebody who’s doing commercial reuse unfettered access to stuff that cost me millions of dollars to do. We’re spending over a billion dollars a year in research. I can’t just throw it all on the street. Not only because it’s worth something, but because I’m not convinced people will respect its values — the values I would want to see expressed in the way people used it.
If I make code available under the GPL, I’ll lose control of it. The Europeans have this notion of artistic rights, and it seems to me an artist — the person who creates something — has some right over the ultimate use of what they do. Artists’ rights also allow an artist to get paid on resale of their stuff later. My view is that programmers are like artists. I think there’s got to be some economic reward back to the people who do the creative work that turns out to matter.
The GPL just doesn’t solve my business problem at Sun. I would like all of our intellectual property to be available in source form, but I can’t economically do that under the GPL.
In the object-oriented world [of programming], binaries are almost as usable as source because they have clean interfaces and boundaries. This whole thing about open source makes much less sense once you start talking about [object-oriented programming languages like] Java, except to the extent that people don’t get the boundaries right.
LM: What about your original implementation of TCP/IP for BSD, which was freely available and which became the basis for a lot of the other implementations that are out there? It seems that from a compatibility standpoint, Java, for example, would have benefited from freely available source code in the same way TCP/IP did.
BJ: The top predator now is Microsoft. We didn’t have a top predator back when I did TCP/IP. When you have a person with unlimited funds who is clearly focused on destroying the value proposition of what you’re doing, you’d be a fool not to account for them in the strategy that you adopted.
LM: Do you feel that Microsoft might actually try to create Microsoft Linux in an attempt to fragment the Linux community?
BJ: The enemy in terms of fragmentation is usually yourself — the people who know the most about making the software better. It’s likely to be two separate groups that both decide that they’re right and they’re both going to make it better and just diverge. You’ve seen the history of the family tree of Unix. It’s all over the map. It’s certain to happen to the Linux tree at some point.
LM: Then why hasn’t it happened already?
BJ: It has. Depending on what we’d say Linux is — the kernel hasn’t fragmented, but the distributions have. People’s systems aren’t the same.
LM: Do you think that the GPL discourages incompatibility by requiring people to make their source code freely available?
BJ: I don’t see that it really prevents incompatibility. The only thing I know that prevents incompatibility is requiring people to be compatible. The GPL permits compatibility. It does not encourage it.
LM: Can you explain Sun’s position toward Linux on Sparc? Sun seems to be supporting the development effort somewhat.
BJ: Right. Well, the customer’s always right. If the customer wants Linux, that’s great — then we should give it to them. Sparc is the hardware that we make, and we’re supportive of and very glad that people in the Linux community have done the hard work that they need to do.
We treat each of our divisions as entirely separate businesses, and I don’t necessarily know what’s going to be important in the long run. People have now figured out that companies that are a little more chaotic in this way actually are better adapters to environmental changes, and I think it’s one of the reasons Sun has done so well. We don’t try to get everybody signed up to one credo. We do not have one ironclad set of rules. We allow this kind of diversity internally.
LM: Sun is providing machines for Linux developers. What else is it doing to support Linux?
BJ: I don’t actually know. I’m more involved in Java and Jini. The company’s very large — we have like 30,000 people — and I probably get involved with about half of the R&D. The Solaris stuff I have the least to do with.
Sun wins if somebody has a Linux machine with Java because that improves the Java community. Sun wins if it’s a Sparc. That’s even better. To be honest, if it was Solaris, Sparc, and Java, that would be even better. But we’re not infinitely greedy here.
The old Macy’s model was if they didn’t have what you wanted, they’d send you to the store that did, even if it was a competitor. If you come to us, we don’t expect that we’re going to solve all of your problems. You may want Apache on Linux on x86, and we’ll do the best to operate in that environment because there may be some reason that’s beyond our ability to affect that that’s the right answer for you.
So to be customer-driven is to accept that and to contribute what you can. We just did this big deal with Apache to put more Java stuff in Apache. So we’re coming at it from all directions.
LM: You’re referring to the Jakarta project, where Sun agreed to donate its JavaServer Pages (JSP) and other Java Web server-related source code to the Apache project and have it released under the Apache license.
BJ: That was a local business decision; I’m glad that they did that.
LM: But you still have the same concerns about standardization, compatibility, and so forth?
BJ: Which I think the Apache community should have also. To the extent that Apache is a platform, and you want to have it be a healthy platform, you want the platform to be stable. But, I’m not opposed to limited cases depending on other licensing mechanisms.
I would always rather have a legal thing to fall back on to enforce compatibility. Think of contractual enforcement as sort of the right-wing approach, and community licensing as more left-wing. We’ve taken a step to the center. We expect in most cases the community to enforce compatibility, but in the limited case of a rogue, I want the ability to enforce a legal contract, because I don’t see any reason why I shouldn’t have that ability. In the left wing, amongst ourselves, we can argue about these things, but in reality, most of the commercial guys are so far to the right that we already seem radical by being in the middle.
It’s innovation and sharing versus centralized control. It’s basically the Romans versus the Greeks. That’s what it comes down to. Microsoft is the Roman model, and the other people are basically the Greek model. That’s the real root of it.
LM: Isn’t that the same situation with Java? With Jini?
BJ: No. Because the Java source code has always been widely available. That’s never been the issue. You can download it yourself. Even under the old license, we basically had a clickable license to download all this stuff.
Basically, we think that it’s much better to work together than to not work together. That’s not a very complicated value. Microsoft thinks: “anything you do, you compete with us.” I think that if you don’t necessarily like what we do, we’ll find some other way to work together. There are not enough of us IT professionals anyway.
LM: Speaking of Microsoft, have you ever meet Bill Gates?
BJ: Oh yeah. Mostly in the eighties. I met him in the early 1980s.
LM: So would you consider him someone you know fairly well personally?
BJ: No. That would be a stale evaluation of him.
LM: So you believe he’s changed?
BJ: I believe it’s possible that he has so I can’t speak to his current state.
I haven’t seen him since — the last time I talked to him was probably five years ago.
LM: Is Linux the major force pushing against Microsoft?
BJ: I think Java is probably the major force pushing against Microsoft right now. I think Linux is a threat but Java’s a bigger threat.
LM: Do you see Linux as a threat to Sun at all?
BJ: No. More Unix is better. Anything that isn’t Microsoft is better. Anybody who buys a Linux machine has a lot better chance of buying a Solaris machine as their next machine or buying a Sparc machine running Linux or buying Java. The probabilities are greater for all those cases.
If I look at the graph of what percentage of customer dollars I’m likely to get next, it’s much higher if they start with Linux than if they start with Windows. So in all cases, I’d rather win and get Sparc/Solaris/Java as the solution. But Linux/Sparc/Java would be my second choice.
LM: Do you know of a company named VA Linux Systems?
BJ: I met somebody who said they were working for them. I don’t track the Linux community, though.
LM: What do you think about the business models being built around Linux?
BJ: I understand that people think they’re going to build a business on the service model, but the truth is customers don’t want to pay for that, so I don’t get it. I don’t know how it’s going to work. People don’t like to pay for service.
The whole proposition with Linux is that nobody can control the operating system. Some invisible hand controls it; a community controls it. Any individual company can’t affect where it goes. How is everybody going to use this in a sense? The Linux companies are hobbled by it because if you say they can add value, then I say it’s going to fragment Linux.
If you accept the proposition that they can’t fragment it, then you also are saying that they can’t really differentiate themselves. Because other than tuning it up a little bit, to differentiate would cause fragmentation.
I would argue that for most people the performance is going to be more than they’re going to need anyway. I’m not sure performance differentiation is going to be that significant. So I’m not exactly sure how these companies will differentiate themselves technically.
LM: Have you ever considered making the Solaris source code more freely accessible?
BJ: Yeah. The difficulty is that it’s got a lot of third-party stuff that’s licensed under funny terms. So I think it will be really healthy for both the Solaris and Linux communities to work more closely together.
LM: Think that will ever happen?
BJ: It already is. We run a lot of Linux binaries, and we’re trying to find ways to work together. Merging isn’t a goal. I think Linux and Solaris have different goals. Linux is not worried about providing MVS class or VM370 or whatever IBM-class services for corporate data centers. That’s not the center of the Linux community.
LM: But there are certainly areas of overlap.
BJ: That’s okay. It gives people a choice, and that’s not a bad thing, right? I still prefer to win. I’m not saying we’re not competitive, but I’d still rather have it be Linux than NT. If there’s two Unix choices and one Microsoft, that improves our chances.
LM: Do you think it’s likely that parts of the Solaris operating system will be individually released as open source software?
BJ: I think that would be a good thing. There are logistic issues. You have to spend money to do that and it’s hard work. In return, you get the value that the source code’s available so the customer can become more self-reliant. I think self-reliance is a good thing.
LM: Were you in favor of Sun’s decision to move to AT&T Unix with Solaris?
BJ: It was hard to do a deal with AT&T and it was hard to work with them. It was a very close call and I went into Scott’s office and I said to him: This is a really close call and I can make the deal happen if you want. There are pluses and minuses. Personally I think it’s a plus because I think a unified Unix community is better than one that’s not, and I’m concerned about this. But I also think it would be okay if we decide to go our own way. It’s your call. It’s a CEO call.
LM: Do you see similarities between the development community and the cultural community that’s surrounding Linux right now and the community that surrounded BSD when you were developing it?
BJ: No. Our community was so small. It was Robert Elz and the people at Berkeley and the people at Bell Labs. There was one guy in Austria and one in Australia. No one else contributed much of substance that I recall.
LM: Do you believe that the BSD guys in general have a different philosophy toward software development then the Linux guys do?
BJ: No. I think that if you exclude device drivers, you’d find that there’s a bit of a myth operating here; that a whole lot of people wrote the system. It was actually a small number of people.
LM: In BSD or in Linux?
BJ: Both cases. We have this myth that distributed development works, but it’s a slight bit of a lie in that a small number of bright people can create an operating system. It does take a lot of people to write all of the device drivers. That’s true. But that doesn’t necessarily mean we can coordinate the programming of hundreds of people writing C code. I don’t know if that’s true or not, and I personally don’t think Linux proves it. I don’t think Apache proves that. That’s the myth that people have propagated. Maybe it’s true, but if you called me as an expert witness, I would testify that it has not been true in my experience.
LM: Is there something to the notion that the people working on BSD are more exclusive than the Linux community?
BJ: That’s an us-versus-them thing.These things just get amplified. I don’t think these people vary from each other by much. They just identify with some group, and that’s a human-nature thing.
BSD is older. It doesn’t need as much hacking. So if you’re a new person learning how to hack, BSD was not as good a place to go. It didn’t need as much work. Linux grew up with the Internet. By the time the Net came along, BSD didn’t need the same level of work and wasn’t as amenable to getting people interested in it.
When you already have several million lines of code, it’s not as much fun to work on. Linux was a great thing because it allowed a lot of people to get involved in learning about operating systems by helping to finish this system. That process of creating something is the process of creating a community.
So Linux came along at the great, perfect time in a perfect, incomplete state for lots of people to participate in. It was still small enough that people could read the code. On the other hand, BSD was already mature, and the things that needed to be done to it were hard enough that it made it difficult for any person to come and participate.
So BSD wasn’t as amenable to parallel innovation because the bar to participating was pretty high and the code base was too large. When I started on Unix, the source code could be listed in ten or twenty thousand lines as a 50-page or 100-page book.
If I came in today and wanted to do something with Solaris, I’d be overwhelmed. I can’t have the kind of impact I had on Unix with Solaris. The second-generation people coming into the Linux community are going to have the same problem.
LM:: What inspired you to write vi?
BJ: What happened is that Ken Thompson came to Berkeley and brought this broken Pascal system, and we got this summer job to fix it. While we were fixing it, we got frustrated with the editor we were using which was named ed. ed is certainly frustrating.
We got this code from a guy named George Coulouris at University College in London called em — Editor for Mortals — since only immortals could use ed to do anything. By the way, before that summer, we could only type in uppercase. That summer we got lowercase ROMs for our terminals. It was really exciting to finally use lowercase.
LM: What year was that?
BJ: ’76 or ’77. It was the summer Carter was president. So we modified em and created en. I don’t know if there was an eo or an ep but finally there was ex. [laughter] I remember en but I don’t know how it got to ex. So I had a terminal at home and a 300 baud modem so the cursor could move around and I just stayed up all night for a few months and wrote vi.
LM: So you didn’t really write vi in one weekend like everybody says?
BJ: No. It took a long time. It was really hard to do because you’ve got to remember that I was trying to make it usable over a 300 baud modem. That’s also the reason you have all these funny commands. It just barely worked to use a screen editor over a modem. It was just barely fast enough. A 1200 baud modem was an upgrade. 1200 baud now is pretty slow.
9600 baud is faster than you can read. 1200 baud is way slower. So the editor was optimized so that you could edit and feel productive when it was painting slower than you could think. Now that computers are so much faster than you can think, nobody understands this anymore.
The people doing Emacs were sitting in labs at MIT with what were essentially fibre-channel links to the host, in contemporary terms. They were working on a PDP-10, which was a huge machine by comparison, with infinitely fast screens.
So they could have funny commands with the screen shimmering and all that, and meanwhile, I’m
sitting at home in sort of World War II surplus housing at Berkeley with
a modem and a terminal that can just barely get the cursor off the bottom line.
It was a world that is now extinct. People don’t know that vi was written for a world that doesn’t exist anymore — unless you decide to get a satellite phone and use it to connect to the Net at 2400 baud, in which case you’ll realize that the Net is not usable at 2400 baud. It used to be perfectly usable at 1200 baud. But these days you can’t use the Web at 2400 baud because the ads are 24 kilobytes.
LM: Do you still use vi?
BJ: No, because I mostly use Netscape.
LM: To write code?
BJ: I mostly do e-mail. The last code I wrote of any substance, I wrote in vi.
LM: Did you have a sense back in those days — even in the furthest region of your mind — that you were working on something that would eventually build an industry or change the world?
LM: At what point did it occur to you? At what point did you look around and say: Whoa?
BJ: I probably constantly under-estimated it.
LM: You must have realized that it was happening at some point in your career.
BJ: I think the Web was a “wow” for me because my dad was using it. [laughter]
LM: You said earlier that Sun would like to work more with the Linux community. Do you have any thoughts on how something like that might happen?
BJ: We have Linux mode on Solaris and there’s Solaris mode on Linux. We’ve done analysis on both sets of APIs and what commonality they have. If the Linux community believes it’s okay for there to be other choices, then that’s kind of a prerequisite to working with somebody who’s different. It’s okay that there’s another version of Unix out there and total world domination is not our goal.
LM: Are there opportunities to do things with Linux that Sun was never able to do as a company? For example, Unix never got the desktop. Or at least, Sun was never really able to bring Unix to the desktop.
BJ: That was our whole business for years.
LM: Right, but Microsoft owns the desktop right now. It’s not meant as a cut against Sun. It’s just a fact that Unix is basically a server operating system.
BJ: We haven’t given up. We’re doing Java clients now.
LM: Why do you think that Unix was never successful on the desktop?
BJ: Because Microsoft had a person who was very greedy and who was very brutal in his business dealings and was handed a monopoly by IBM due to ineptness. They had several opportunities to rein this guy in and the management blew it. So the IBM monopoly got transferred basically due to blunders. Microsoft is a direct successor to the IBM mainframe monopoly. The corporate guys coalesced around the PC standard because it came from IBM. Not because it was any good.
LM: But does something like Linux offer Sun an opportunity to rectify past mistakes? For example, there are windowing systems being developed for Linux that need …
BJ: We’ve had windowing systems, several of them, for many years. The presence or absence of a windowing system didn’t win or lose the war. We have had CDE and Open Windows and X-Windows and NeWS.
We’ve had applications too. We’ve had all of these things. I suggest that the desktop war was not won based on technical merit, but on business decisions. Microsoft came along and took over the apps base with Office. Office locked people in to the point where corporation don’t feel they can change their desktops, not because they’re locked to Windows but to Microsoft Office.
Now what’s happening is the wind is blowing hard for the companies to put everything on the Net. Make the browser the access point for the desktops. So the desktop is really becoming a browser. But the people — the companies — still have these Microsoft Office hairballs that nobody likes. Have you seen a good review of Office 2000? Everybody hates it. But Office is what locks up the desktop.
LM: There is a lot of talk about Linux possibly conquering the desktop.
BJ: It’s easier to talk about than to do. The Macintosh is easy to use, and it even has Office. What’s the difference between Linux and a Macintosh? If Linux with future apps is going to be good enough, why aren’t more people buying Macs?
LM: You think this is a fight that’s already been fought, basically?
BJ: No, I’m not saying you can’t find a way to win. It’s just that I haven’t heard what it is. Given a sufficient number of people who care and an ability to be flexible about the way you achieve your objective, Linux might get there. You have to find a way around some of these things that are preventing people from switching.
These guys at Microsoft are very aggressive business people, and they have been very successful as aggressive business people. I don’t think they’ve been very successful at building good products. I think history will judge their products to be the lowest-quality consumer products ever built and manufactured in any scale.
It’s similar to how Detroit got itself to where they manufactured incredibly low-quality cars, which coincided in history with GM’s maximum market share. What happens is monopolists don’t tend to value product quality. Very high market share is what they value.
As GM’s market share was declining, it always talked about getting back market share. Why didn’t they talk about making products that people wanted to buy that were high quality? That was the problem. I heard Steve Ballmer [President of Microsoft]say this in a speech: “Our number one goal is maximizing our market share.” Excuse me. Market share should be a consequence of your goal. That can’t be the goal.
The goal is to build great products. I have an infinite respect for Steve Jobs because whatever else you say about him, his passion is to build a great product. Good things come from that. Bad things come from a focus on market share. So the Linux community should have as its goal to build the greatest operating system. Its goal should not be, “Beat Microsoft.” Because that’s a market-share goal. That’s a very, very destructive, counterproductive goal.
LM: Do you think there are any other goals the Linux community should pursue?
BJ: Make it the best product it can be. Figure out who you want it to be for and build it to serve that community — if it’s for yourself, that’s okay. Make it the best hobbyist — in the best sense of the word — operating system. Linux to me is like amateur radio was to radio. Amateur radio developed all the radio technology. Linux is developing some good technology, and these people are hobbyists. Probably some Latin root of the word “hobbyist” means people who love something and care about it. So it’s a sense of love and caring for reasons that are noneconomic.
It’s like amateur astronomers. In essence, it’s amateur in the highest sense of the word, having the highest affinity to caring that it’s always the best. And tinkering and all this kind of stuff, that’s a very positive value.
Eugene Eric Kim writes, programs, and consults on a freelance basis. He is the author of CGI Developer’s Guide(Sams.net 1996), and is currently writing a book on the history of free software. He can be reached at firstname.lastname@example.org.