A Talk with Tim O'Reilly
At the first Linux World Expo after the turn of the millennium (January 31-February 3, 2000), I had a brief talk with Tim O'Reilly that I recorded and then lost. Recently, when that tape reappeared, I gave it a listen and found that it was as relevant as ever—perhaps even more so—because Tim is one of those rare businesspeople who not only takes the longest and broadest possible view but acts constantly on Yogi Berra's most impossible advice: “When you see a fork in the road, take it.” Looking back, Tim seems to have taken every fork along the way and sometimes several at once. Yet, there also seems to be something consistently broad and encompassing about every one of those decisions.
In retrospect, the vision Tim shared in that talk seems exceptionally knowing and prescient, especially since I was mostly interested in just one question: Why didn't he sell out? Even though publishing companies are not held in especially high regard by venture capitalists (who tend to value businesses with potential valuations in the billions over ones with actual revenues in the mere millions), O'Reilly has a long track record of growth and success. More significantly, it is one of the most influential companies in the software world.
Yet O'Reilly expressed almost no corporate vanity—at least not compared to other companies in its communities that were either already public or headed that way. Relatively speaking, O'Reilly & Associates never seems to care about selling itself, beyond whatever it takes to publicize its various products and services. It has always been, in the oddly discreet vernacular of finance, private. And Tim has told me more than once that he intended to keep it that way.
I wondered why. For freshly public Linux companies, Y2K was a heady time, as all of them were still surfing a tsunami of fashion investing. Red Hat, Cobalt Networks, Andover and VA Linux Systems had all gone public in late 1999 to enormous run-ups in valuation. At the time I spoke to Tim, VA Linux had just bought Andover.net (now the OSDN division of VA Linux) for more than $1 billion in stock. Less than two months earlier, VA went public to the largest first-day run-up in history, closing at $239 after trading as high as $320. Red Hat had gone public only four months earlier and peaked at close to $270 around the time of VA's debut. When I spoke to Tim, VA was trading in the $120s and Red Hat at around $100. Even though those numbers were way off their early peaks, both companies were very, very hot, and Linux along with them.
Today the Linux investment tsunami is flat as a pond. VA Linux shares now (on November 20, 2000) sell for around $13 and Red Hat's for around $10. Meanwhile O'Reilly & Associates continues to truck along in its merry, myriad ways, looking far the wiser for not selling at any price.
Here's the conversation with Tim. The date was February 1, 2000.
Doc: One of my pals here just suggested that more money was spent on booths at this show than all the Linux companies in the world combined made in revenues last year. That make sense to you?
Tim: [Laughing] No, but they have a job to do and a lot of money to spend.
Doc: Linux is hot right now. You're a major player in the category. It seems to me that you could easily sell out if you wanted to. Yet you haven't. Why?
Tim: This is really the second big explosion that we've been an interesting part of. The first was the Web. The second was Linux and open source. There were others before the Web, actually. And there will be others after Linux. We've been around long enough to know that every trend is more like a wave than a line that goes up toward the sky. One wave builds on another. When you identify with any one of them, you get a short term advantage out of it, but then you get locked into that stage in history. We could have been a UNIX company, or a PC company, or a groupware company, or an on-line company, or an Internet company, or a web company or, now, a Linux company. Trust me: another wave is going to come after this one. Linux will still be here, but it won't be the hot thing it is right now. Nor should it be.
Doc: You just don't want to get locked in, then.
Tim: Right. It's just never interested me very much. We've just been around too long. We're good at what we do, and we have a role to play; we'll just keep on playing it.
Doc: I used to have a partner who said, “Don't say no, say how much.” That's another way of saying everything has a price. If that's so, and you're still saying no—and companies like Andover, which had a small fraction of O'Reilly's revenues last year, are selling for upward of a billion dollars—then you're telling me that you value O'Reilly & Associates somewhere in the billions.
Tim: [Laughing] I suppose so.
Doc: Okay, so you like to take the long view. Give me some background on that.
Tim: Like a lot of people, I got started doing programming. And, like a lot of people who do programming, I saw a need for manuals. So, the company got started in 1978 as a technical writing consultancy specializing in manuals. In the early eighties, open systems were getting hot, and we started licensing our manuals to various UNIX vendors. Gradually our consulting business gave way to a publishing business.
Doc: But you've been more than a publishing business. You've been highly involved with, rather than just selling to, a market or a category.
Tim: We do like being part of the industry and what makes the industry change over time. I think we probably have a better sense of where things are going and the pulse of technology, than most companies do, mostly because we've been around a long time, and we cover a lot of topics.
Doc: You're not tied down to any one category, or title or whatever.
Tim: No, we're not. Whatever programmers and other technologists need, we're there publishing a book about it.
Doc: Or running a conference.
Tim: Or starting something like GNN, which we sold to AOL.
Doc: I had forgotten about that.
Tim: There are a few things I'd like to forget.
Doc: I'd like to talk a bit about the issue of intellectual property, which I know you care pretty deeply about. Jeff Bezos did us the favor of putting the issue of software patents on the front burner when he obtained the patent for one-click shopping and promptly sued Barnes & Noble for infringing on it. You've been very vocal about the whole matter. I'd like to hear what IP (intellectual property) means to you.
Tim: I think of intellectual property as topsoil. Hard to make, easy to waste. It's a rich substrate of ideas, techniques and opportunities that anyone can use. That's a rich soil on which you can build business opportunity. We should all be mindful of that and contribute to it. But there are some business models, particularly business models that create hype and fund themselves out of the stock market rather than providing real value to customers, that don't contribute. These models leech nutrients out of the system. If you can convince customers that they should not have to pay for high-quality technical information, for example, nobody will produce it. What some of these companies have done is exploit some short-term conditions that allow them to take a whole bunch of money out of the pockets of potential shareholders while, at the same time, destroying the conditions that made their success possible. So, it's like hey, there's all this stuff that we could get for next to nothing. We'll pretend that we have a viable business model by selling it cheaper than the cost of production. This is just like what happens with resource exploitation in the physical world. Hey, air and water are free! Somebody once asked Ted Turner what he would build a business on if he were a young man, and he said “Water.” And he's buying up watersheds all over the place. He knows there is stuff that has real value that's undervalued at the current time.
Doc: And you think this is the case with software IP?
Tim: I just see a lot of danger in business models that are predicated on some resource that was not created or enriched by the people who exploit it. Again, this is one of the imperatives of the Linux community. So far it's doing better than the Internet community did. I joined the Internet Society Board back in '95 because I saw this happening with the Internet. I had hoped they could model themselves on the Sierra Club. I thought the environmental movement was a great model of people saying, “I have some kind of right in this common good. I can speak for it even if I don't own it.” There is this common heritage represented by the Internet culture, by Internet software, by open-source software, that is being exploited in the same way that natural resources have been exploited. Users need to have some voice that says “don't do that to us.”
Doc: So we'd say to Amazon, “Don't sue Barnes & Noble over your patent on a natural resource. Beat them some other way.”
Tim: Exactly. You can compete but don't compete in a way that's destructive to the substrate that you're building on. Again, I think that the Linux and Open Source communities have done a much better job than the Internet community did five years ago. Linus and various Linux companies have done a better job of giving back to the community that depends on the environment the community is creating. They are much more conscious of what we hold in common. But I don't think it's as far-reaching yet as it needs to be. We need people thinking about this over a very long period of time. I don't know if you've seen the book, The Clock of the Long Now, by Stewart Brand. I was at a lecture he gave with Brian Eno in which Eno talked about speaking to this woman who lived in a wonderful house in a lousy urban setting. When he asked her how she liked living here, she said it was terrific. But he realized she was talking about “the small here”, not “the large here”. He said he realized that a lot of our problems in society come from our definition of what we're going to care about, how local it is, and we often miss the larger picture. Or worse, we obscure it by painting a small one on top of it.
Doc: You see this as relevant to our current infatuation with the stock market rather than the larger market that surrounds it.
Tim: Yes. We have a lot of businesses optimized for the short term and for a narrow set of interests. They don't spend a lot of time asking, “What will the big benefits be for the culture as a whole?”
Doc: How far back does open-source consciousness go?
Tim: We've talked a lot about open-source software, but the IBM PC was just as big a revolution in its time—and still is. It was open-source hardware. That's what made it cloneable. Its openness freed up a tremendous wave of innovation. But what happened on top of that was the building of a new proprietary empire that was even more oppressive than IBM was in the hardware days. I fear that we're actually going to get there again in spite of Linux. I've been saying for years that just because one layer of software is free doesn't mean that the next layer will be free. We can expect to see closed software being built on free and open-source software, where the new applications will be fundamentally closed. You can look at Amazon as an application. Or Mapquest. Or E-Trade. These guys are not touched by open-source licenses. The GPL does not matter if you don't distribute software. E-Toys won't give away their e-commerce software because they don't have to—which may be fine. I have no problem with that. But I do note that these developments are not part of the ecosystem. They don't give back to the common soil that we all grow from.
Doc: Do you prefer the GPL to other licenses?
Tim: The one thing I don't like is the compulsive aspect of it. I mean, people can choose not to use it, but it also keeps some people from using it. But my point is that it's most critical to establish the culture in which the companies that build on the open platform realize it's in their interest to keep it going. Microsoft has more to gain from feeding the Open Source community than by competing with it.
Doc: I think it's inevitable that Microsoft will support open source in a variety of ways over time.
Tim: Possibly. But there really are a lot of choke points here. To me, the fundamental insight that Microsoft had in the software world was that there are choke points, and if you can hold those passes, you can control the markets that depend on them. And believe me, many people in the e-commerce and the Web world are thinking the same way. And some of them are getting a pretty good stranglehold on some of those choke points, and they will inevitably have others dancing to their tune. I am very concerned that we are going to see in this new layer a number of players who try to dominate by controlling choke points. They may not be as dominant as Microsoft overall, but they will be just as dominant in their own categories.
Doc: What about the big media players—such as AOL/Time Warner?
Tim: I'm not too concerned about the big media conglomerates. Well, we do need to be concerned about the ways we get broadband. The sources matter, in terms of how fast broadband is deployed, what kind of access they provide, and who controls it. But the fact is, the active producer/passive consumer model doesn't work very well anymore. We're all too connected. Those of us on the leading edge really do think of markets as conversations.
Doc: We see supply and demand as symmetrical.
Tim: Right. And we like technology that works the same way. It should be of interest to readers of Linux Journal that when we did our web site, we deployed the first NT-based web server. Why did we, a UNIX company, do that? Well, we saw at the time that the Web was breaking down the fundamental model. The vast majority of people using browsers were using them on Windows, and there were no Windows servers. So I used to joke that the Web was becoming the world's largest read-only groupware system. That was in fact the model. What we tried to do was keep alive the vision of the Web as a peer-to-peer medium. In fact, we did.
Doc: How did it work out?
Tim: Not very well as a business, but it did get Microsoft into the web server market. For all the bad stuff one can say about it, the people using that platform can publish to the Web. Prior to that, we were hearing all this stuff about push technology from people who were trying to make the Web look like television instead of an interactive model.
Doc: They were trying to weld the Web to the rusty back end of television's history.
Tim: That's right. And this choice of ours to deploy NT as a web server was an example of trying to think in a bigger way. If we said, “Our primary allegiance is to Linux, or UNIX, and we wouldn't be caught dead using NT,” we wouldn't have done anybody any favors. We said, “Hey, the users are using Windows, and they're getting a bum deal here.” We need to give them the same tools as the culture that we want to build. And that's a key concept here. Why did we publish Windows 98 in a Nutshell? Because we want to teach people using that platform how to think in the way that the people who are over on UNIX think. I want them to be empowered and show them that they're the bosses. If you look at that book, you'll see it teaches, as much as it can, the UNIX mindset, on this crippled platform.
Doc: Do you bring a similar message to the Linux world from what lies outside?
Tim: A lot of what I try to get across to the Linux community is the importance of the Web. Take “the network is the computer” as the mantra. You've got to be a player there. What I worry about is that so much of the focus is on the last war, which was over operating systems. I kind of feel like the real question is how are we going to play in the world being built on top of those operating systems, regardless of what they are. I look at the web-based open-source projects as what we need to think about more, get involved in more. Apache. SendMail. QMail. Even something like MySQL. Anything web-related is really important. The big question should be, what does the computer of the future look like? I think it all looks like one big computer. So, how are we building the tools to participate in that? For example, what we're doing with Collab.net is figuring out and training people how to work in a collaborative way. So much of the Linux conversation has been about how the roots of the movement have been in the GNU project and free software. I'd like to talk to people about its roots in Usenet. Its roots in the way people shared information. If we think the conversation is about a particular style of software licensing, and that Microsoft ought to put out its software under open-source licenses, and that “we” somehow win if “you” ignorant mortals follow what the priesthood hands down without question, we've all lost. At the end of the day, it's about a culture of communication and innovation that comes from many points and not down from on high. In some sense, we have a chance to build a new world. I see too many people jumping ship from this new world to go back and work in the old one—or in the part of the old one that's building closed stuff on the open new world we're all building together. So, it kind of goes back to the beginning of this conversation when we were talking about the money. If you don't think long term, you play into the hands of the old order.
Doc: But there is a lot of financial incentive to think and behave short term.
Tim: That's right. And those of us who take advantage of those opportunities need to keep the long term in mind, to keep the common interests in mind. This is a world we're building together here.
Doc: And there's still money to be made there, too.
After I turned the tape recorder off, I suggested to Tim that he take a look at Jabber, the new open-source instant messaging system. By August, Tim was talking up Jabber at his Open Source conference; and not long after that he joined the Jabber.com board of directors. (I was, and still am, on the advisory board.) Meanwhile Tim's patent reform conversation with Jeff Bezos has morphed into Bounty Quest, a site that involves several communities in the same patent reform effort.
The man just keeps taking those forks.
Doc Searls is Senior Editor of Linux Journal
- Let's Go to Mars with Martian Lander
- Applied Expert Systems, Inc.'s CleverView for TCP/IP on Linux
- My Childhood in a Cigar Box
- Papa's Got a Brand New NAS
- Returning Values from Bash Functions
- Tech Tip: Really Simple HTTP Server with Python
- Rogue Wave Software's TotalView for HPC and CodeDynamics
- Panther MPC, Inc.'s Panther Alpha
- Debugging Democracy