The future is here

CEO TIM O’REILLY and the publishing company he founded, O’Reilly & Associates, are in the business of identifying emerging technologies, exploring their implications, and charting new directions for computing. InfoWorld Test Center Director Steve Gillmor met with O’Reilly at the recent JavaOne conference in San Francisco to discuss O’Reilly’s roles as publisher, futurist, and activist.

InfoWorld: As part of your mission at O’Reilly & Associates, you were mentioning you felt a need to step out in front and take a stance — an activist stance.

O’Reilly: Right. A lot of what we do at O’Reilly, first of all, is we watch what the hackers are doing and try to capture their knowledge in books, conferences, or whatever. But periodically we feel like we need to do something activist. And the activism is often when we see the stories in the media diverging from the reality. We did that back with open source — people were missing a big story. They were talking about Linux but they weren’t realizing that there were all these other programs that were providing a big part of the Internet infrastructure. In this environment right now, [after] some initial excitement about Web services, a lot of misinformation [is] starting to be spread around. We hear news like Microsoft just figured out they don’t know how they’re going to make money, so they’re going to have to go slow. And meanwhile, the hackers are in fact progressing apace. And so I thought, well, I’ve got to start letting people know that [Web services are] happening and are going to happen, and you can either get with it or you can wake up a year from now and go, darn, I missed that market.

What I see happening is a combination of factors that are just starting to come together. And the fundamental change is that connectivity is no longer an exception. [When] there’s a paradigm shift that happens over a period of time, a sort of signal event [will occur and] all of a sudden you’ll look back and go, oh, it was obvious, it was actually happening for quite some time, but now we really know it. A lot of people are still effectively thinking PC; they’re mired in the PC era. And it’s like the way the PC was off the radar [when] a lot of the big computer companies in the mainframe era saw it as a toy. They didn’t realize that it was the future. That’s why in [“Inventing the Future,”] I started off with this great quote from William Gibson: “The future is here, it’s just not evenly distributed yet.” So you start looking around and you say, well, what are these signs of the future? And they’re oddball little signs, these little clues.

For example, we host [developer] summits from time to time and one of the guys, Kevin, works on speech synthesis, an open-source project called Festvox. Another guy was working on the Freenet project, which is a peer-to-peer file-sharing system. And the Freenet guy says [to Kevin], “You sound familiar.” And Kevin says, “Yeah, well I do Festvox.” And the guy goes, “Oh, that’s it. I listen to you all the time — I pipe IRC to Festvox so I can listen to [chat] in the background while I’m coding.” So these hackers [are] already living in this world where the idea of routinely converting some text flow to speech [is real]; he’s living in a multimodal application future. He’s getting this stuff off the Net, and he’s converting it to something else. That’s not related to this Web services thing, but it is one of those little signs. The hackers tend to be out front; they’re using the tools that are available now in ways that other people are going to be using them routinely three or four years from now.

InfoWorld: Similarly, what about your point regarding Weblogs, that Weblogs are the tip of the iceberg in the content management space but also they’re an extension of the kinds of communities that have evolved on the Internet?

O’Reilly: [It] has to do with watching the grassroots. It’s very easy to get into watching the technology press, which is often influenced by the technology marketing of big companies. [But] there’s a lot of really interesting information that comes when you watch what early adopters are doing. They might be entrepreneurs, and what they’re working on is going to be next year’s product announcement. Or maybe they’re open-source hackers who are doing interesting things because they’re really good at what they do and it seems obvious to them that [something] ought to be possible. Coming back to the idea of Web services, I see a lot of people who are saying that we’re not quite sure how this market is going to take off. And I’m convinced it’s going to take off. The hackers are already treating the Internet as this global data resource and they’re building Web services however they have to. And right now, a lot of the time that means somebody doesn’t provide a Web services interface to their data. But they have a big database-backed Web site. And somebody says, “Well, with all this great data there, I’m going to go get it.” And so they’re doing screen scraping. They’ll download the page, figure out what data they want, and throw away the rest — sort of unauthorized, brute-force Web services.

InfoWorld: They’re authorized by implication. That’s why the material is being presented on the Web for free.

O’Reilly: But there are issues. For example, there is a tradition of the robots.txt file that says don’t search this file, don’t search this Web space. And so people might say don’t spider me. But I don’t think that’s stopping [it].

When you look at the advanced hackers, what they see is this data out there. Web services are really about data, in a lot of cases. I’m not saying that that’s not an interesting market — it’ll develop over time and there’ll be a lot of money [in it]. But what’s interesting is the development of a market in which … I always like to talk about the architecture of unintended consequences. The original Internet made it possible for people to build independent services without knowing each other, without having to enter into a contract. They just built the service, somebody else figured out how to use it. Similarly, in the original Unix architecture, the whole idea of pipes and filters was that I’ll write a program and because everybody knows what the output is going to be — it’s going to be an ASCII stream written in this particular way – [and] they can build some other program that will take it and transform it and use it. So Unix was built as a Tinker Toy or Lego kind of environment. As we look at Web services, I’m interested in seeing that same kind of market development.

InfoWorld: Is this close to Dave Winer’s bootstrapping metaphor?

O’Reilly: In a lot of ways. To me, the thing that’s interesting in the technology market is that some architectures encourage participation by developers and others are command-and-control architectures. A lot of the focus in open-source discussions on licensing misses the point. Open source is a lot about architecture: Does this system architecture let someone hack it easily, or is hacking hard? If hacking is easy, then you get a lot of people who are experimenting. And a lot of those experiments later get replaced by technologies that take those hacks and make them accessible to more people. [For example], the original Web was for putting out pages with links. [Then] somebody came along with the idea of allowing dynamic content. So we got CGI. A lot of the early CGI stuff was pretty crude, but before long that little loophole, as it exploded with people building all kinds of dynamic sites, is where the Web got really juicy. Then you had the introduction of ASP, ColdFusion, a whole bunch of languages that made building dynamic sites much simpler for ordinary developers. And in a similar way we’re at this stage right now where the advanced developers, the hackers, are effectively building Web services that mix data from different Internet sites. The next step, of course, is for people to come along and say, “We’ll make this easier, we’ll make it possible for your garden variety developer [or] power user to start doing these things.”

InfoWorld: Isn’t that the impetus behind the commercial vendors going to the portal space?

O’Reilly: Absolutely. A lot of Web services aren’t about technology, but about data. The screen scraping is a clue to what people are trying to do with that data. I know one guy who built a carpooling planning service. Because MapQuest can give addresses and routes, he built something where you take the roster for the company [and find] carpool companions. It would look up people within a certain distance from you. Other things are in MapQuest data [that] you could imagine, for example, being supported as a feature in a program like Excel. When I fill out my expense report and I want to put in the mileage from one location to another, I’d go to MapQuest. I don’t want the map, I don’t want the address, I just want the distance, [and] here’s this engine that could produce that directly. Similarly, there are all kinds of other things [Amazon] is good for. You could easily imagine building a service that, when you’re doing that term paper and you get to the bibliography, [takes your] list of titles and authors and fills in all that other stuff.

InfoWorld: By spidering Amazon?

O’Reilly: We’ll just call it from a big public database of books.

InfoWorld: And that leads to your discussion about the business models, the relationships, the contracts that will emerge?

O’Reilly: What I worry about is that companies are going to start out saying, “We don’t know how we’re going to make money, so we’d better not do anything.” And what I encourage people to think about is creating at least some low-volume options. It might be anonymous. But it could be authenticated to some extent. You sign up, and you get an ID, which you provide [when you access the service]. That would allow somebody who has the data to say, “Wow, somebody’s using me a lot. I’d better talk to them.” The overhead of setting up the service is the barrier to entry of the market — if in fact you have to negotiate with someone. I want to see Web services APIs where somebody says, “OK, I have data, you want data. You can sign up pretty easily and we’ll give you a data feed. But we’ll only give you, say, 1,000 queries a day. If you’re using more than that, we want to talk to you.” That might be one way to do it. There may be others.

InfoWorld: One of the things that we’ve been seeing is Web services beginning to emerge from the lab or from the hacker environment as you’re suggesting — as a second or middle tier of aggregators who are, for example, acting as trusted intermediaries between small companies. Does that fit into this vision?

O’Reilly: In some sense aggregation is a fundamental Web service. You can see this even in aggregators like search engines. If I want to find certain types of data, I’d rather spider Google than spider all the original sites because [Google has] already done some chunk of the work for me. You see how various types of aggregators have become destinations. And this is, again, where people are missing how many Web services are already there; they’re just not standardized. Look at IMDb [Internet Movie Database]. Or look at CDDB [CD database]. CDDB is a great example. It was originally built as an aggregation of a whole lot of data users were contributing, and they eventually realized they had a pretty valuable database. Now when you go to burn a CD and [the software] says, “would you like me to go look up the track IDs?” Well, that’s a Web service. It’s an application that’s getting some of my data routinely over the Net.

InfoWorld: I know this is a little off track, but that example starts to collide with the issue of hijacking ports — the peer-to-peer ports on the network. Do you see that as being an impediment to your low-cost freeware idea?

O’Reilly: Yes and no. First of all, there is a lot of data people think is worthless, and then somebody starts to gather it, and it becomes valuable. And sometimes there is effective hijacking. Network Solutions is a great example, [and] now VeriSign, [which] managed to, if you like, hijack a very valuable collective resource that had been built in the free days. Again, people often talk about music being pirated. Well, there are also things that have been built collectively, in the free spirit of the Net, that have been hijacked by commercial interests. CDDB was built by a collective community, [but] the people who were in control said, “Wow, we might be able to make some money off this,” and effectively changed the rules. So hijacking goes on in both directions. Is that necessarily a problem? It is a collection of individual problems, but it’s through solving those problems that we move the industry forward. When you look back at the evolution of the software industry, open-source and proprietary software both came out of the same mix, as people solved a set of problems in different ways. Bill Gates said, “My gosh, people aren’t valuing software; I want to get people to see it as something very valuable.” Other people said, “No, we really like this free-sharing culture, we’re going to take it in that direction.” Let the different models run. The market is pretty good at figuring out what works and what doesn’t work, and [there’s] room for different models.

InfoWorld: My concern about this vision of an open architecture is that when the implications of that kind of connectivity become apparent to the government, for example, they — or the shadow government in terms of the record companies — can step in and close down access, essentially staving off this evolution until they can figure out a business model.

O’Reilly: That’s why I’ve started to do a little bit of activism about this, [because] people are making the wrong choice. And the wrong choice is to say we’re afraid of the future, so we’d better stop it from happening. First, it does not work. Second, look at copy protection with software — it’s astonishing to me that the music industry is acting like this has never happened before. There have been copyable [software] products available on pirate sites, pirate networks, for 20-plus years. It’s not a new problem, and it hasn’t kept Microsoft from becoming one of the richest and most successful companies in history. So to say the Internet is going to change everything and piracy will destroy all intellectual property is either uninformed or real misrepresentation.

But we’re kind of getting off the topic. For me, the topic that’s really interesting is the way that the Net is morphing into its next generation — into the third phase. The first phase was the various command line tools, the comments, the FTPs, and the original TCP/IP utilities. The second phase was this wonderful Web phase. And the next phase … here we are at JavaOne. To use Sun’s phrase, “The network is the computer.” Microsoft clearly gets this as well with .Net. We are moving to a world in which connectivity is the platform. And people are looking to build applications that can simply assume connectivity. And when you make that shift, you think differently. The real significance of Napster, for example, wasn’t that it was the end of the music industry or that content would be free. What was interesting was that Shawn Fanning was a college kid who grew up in the age of the Internet, and he thought differently about the problem. The MP3.com people said, “Yeah, we’re going to put all the songs in one place.” And here we have 3,000 servers. And along comes this kid and he says, “Wait, why do I have to have all the songs? My friends have all the songs. All I have to do is point to them.” And that’s that shift from the old client/server thinking of my computer is local to my computer is ubiquitous. And that shift to network-centric thinking is cropping up all over, whether it’s in file-sharing, Web services, the idea that I could connect my digital camera to my cell phone and send a photograph along with my instant message over my e-mail, or somebody [using] Jabber [to] control SAP from a cell phone. We really are getting into a world where the network is becoming the platform, and that’s a really interesting shift. The last shift of this magnitude was in the early 1980s, which was the devise of the widely available personal computer, the move out of the glass house of centralized computing. And people think that the Internet of the 1990s was a big change, but it was really just an add-on to the personal computer era, it was something that you could dial up, it was this cool application. It was a fascinating new killer app. But I think that this next generation of network computing is a fundamental shift at the level of the PC, where ubiquitous networking is the key fact. And [we’ll] start thinking about devices very differently. Going back to this William Gibson quote, “The future is here, it’s just not evenly distributed yet” … look at the iPod. Very, very significant class of device now. People think of it simply [as] another music player. But if you look at what the hackers are doing, they’ll say, “Hey, can I borrow your computer?” They connect their iPod to it and they have all their files, all their programs. So they’re effectively carrying around access devices to their computer, which is the network.

InfoWorld: IBM did a test device that essentially is another version of that idea — the I/O is separate from the little storage device and it plugs into various things. But it’s your unintended consequences comment I think is most relevant to the iPod. It’s the kind of convergence described by Nicholas Negroponte’s saying, “Everything is bits.”

O’Reilly: [And] even though I was talking here about the network being the central fact, storage is getting cheaper faster than bandwidth or CPUs or any of these other gaiting factors, and there are enormous changes [brewing] in the fact that you can have massive storage that you can carry around with you on a key chain.

InfoWorld: There are also implications with Mark Lucovsky’s work on Hailstorm, or .Net My Services, regarding how the cloud interacts with that data.

O’Reilly: That’s right. You have your personal data, but you also [have] massive databases that are going to be Net connected. And the reason you have to have Web services, even though storage is getting cheap, [is that] transporting all that data from one place to another isn’t. You have to have some way of drilling down. A good example of that [came out] at our bioinformatics conference. Ewan Birney, who’s the developer of an open-source bioinformatics toolkit called Ensembl, talked about a support call he had from a biologist who called up and said, “I’m using Ensemble, and something seems to be broken. It’s been running for two days and it hasn’t done anything.” And Ewan said, “What are you trying to do?” And the caller said, “I’m using it to run GenBank against Swiss-Prot.” These are two big bioinformatics databases. And Ewan answers, “That’s a 50-year project.” So although this guy had the Web services, you need to be able to call [them] programmatically. The databases are so big that you can’t have them; even with cheap storage, you don’t want to have your own copy. But you also need to be able to get at them in small atomic calls. You don’t want to do the 50-year comparison of this one to that one. You want to be able to specify some programmatic queries and get into that data. And so what we’re moving into is a world where you imagine tens of thousands of very, very large Internet-facing databases, and a programming environment in which people say, “OK, I need a piece of data from over here, I need a piece of data from over there. I’m going to mix them, I’m going to run some operation on them. I’m going to get some other data from over there.”

Again, going back to this Unix pipe-and-filter philosophy, how do we build data sources, data syncs, and connectors for those things in an easy-to-use way that allows people to build applications that rely on this current Internet operating system? What an operating system does is provide a layer of consistency that allows applications to access resources. In the old days, those resources [allowed users] to access the keyboard, the disk, the monitor, the tape drive. Somebody would write an application and request those services from the operating system. [Now] we’re heading towards a world in which the services that need to be accessed are various kinds of Internet-connected databases. So the Internet operating system, whether it comes from what Sun’s doing with Java or out of .Net or out of some grassroots set of hacker options or some new logic, whatever, what we’re going to see is the development of an infrastructure that keeps track of what are the key resources out there on the Net. And [these are] going to be things like identity, location, various kinds of specialized types of data, whether it’s mapping data in MapQuest, bibliographic data in a site like Amazon, or information about music tracks in CDDB. There are thousands and thousands of databases that are going to be turned into programmable components, and those components [will] need to be accessible from a variety of devices. And somebody’s going to put all that together into a system that makes that kind of development easy.

InfoWorld: And with a healthy dash of caching to deliver those services with enough speed.

O’Reilly: That’s right, and the opportunity for the software vendors is to [improve] this stuff that’s now pretty crude. Again, in the early days of any new industry, there’s a lot of cheap hacks that don’t work all that well. It’s very much like Craig Christensen’s book, The Innovator’s Dilemma . [The new] stuff doesn’t work all that well, but it gets better faster than the existing stuff.

InfoWorld: And when the critical mass of users is reached, then there’s money in it.

O’Reilly: Right. So anyway, we’re doing this conference in mid-May, the O’Reilly Emerging Technology Conference , which has the theme of building an Internet operating system. It focuses a lot on Web services, but also on peer-to-peer file-sharing and things like software that’s adaptive and learns and also broader work going on in biological models. I know people who are looking at the way the immune system functions, for example, and [asking,] how do you have certain types of automatic responses? This is a lot more complex than the software you’re dealing with now, so you have to have software that can learn.

InfoWorld: I’m not going to hold my breath for that.

O’Reilly: Yes, but there’s a lot of interesting work going on in that way, and technologies that [used to be] pretty abstruse are starting to crop up in [interesting] contexts. Something like simulated annealing, which was originally used in figuring out routing on circuit boards, people are all of a sudden using it to figure out other types of predictive problems. [Or] hidden mark-up models. This is something [that gets] a lot of use in speech recognition, but it’s also cropping up in bioinformatics in gene research. [Or] machine learning, [which] was sort of identified with AI — it’s cropping up in a lot of different contexts. As we get to the critical mass of the Net, fresh ways of looking at problems [will] take the solutions to another level. A good example of this is Google. Google realized that there was more information in the Net than people were using. Google said, “Well, we can actually follow who links to [whom] and use that to figure out what’s more important. And so our search rankings will be better.” And that’s just the tip of the iceberg because in fact there’s so much information that’s out there now from the way people are using the Net.

InfoWorld: Google is also a good thesaurus and spellchecker. When you mention somebody whose name I don’t recognize, I just type as close as I can get to it in Google and I get 20 hits and I’m there.

O’Reilly: That’s the kind of thing, again, that [shows] people just don’t realize how much the world has changed. We take certain things for granted, and we take them for granted partly because we’re still thinking in the old frames. People think they know what software is, something that you package up and sell — shrink-wrapped software. And [there’s] the enterprise software market where you’re paying some big price and buying service to boot. And yet some of the key services that have been introduced that are really world-changing — that in some previous model would have been recognized as the killer apps — they’re just Web sites. And because they have an e-commerce model like Amazon or an advertising-based model like MapQuest, people don’t think of them as software products. But MapQuest is just as revolutionary a killer app as the original spreadsheet.

InfoWorld: It’s the value proposition of this transition to the network is the computer because the operating system, the UI, is in fact the network. In other words, software has made that move from being stored on the disk to interacting with the cloud.

O’Reilly: Right. And when you think about this from a business point of view, there’s often a hidden value that is being exploited, but people don’t realize it. For example, shrink-wrapped software was very visible, [and] that was the heart of the business model. [The packages] are out there competing on the store shelves. Meanwhile Microsoft was locking themselves in below the water line, in a totally different way, through the OEM contracts and so on. So the public perception of where the value was in software and the reality weren’t the same. In a similar way, I think we’ve had this sort of focus on e-commerce or advertising as the source of value in the Internet industry. In fact, the source of value may be the accumulation of a unique and defensible data set that people need access to.

InfoWorld: Accessible and unique.

O’Reilly: Look for example at what Amazon owns right now. OK, they have this interesting brand, they have this interesting business. But they also have a unique amount of data that would be difficult to duplicate. And you start thinking about what other applications could be built using that Amazon data set, besides the one that already exists.

InfoWorld: The flow of interests to events, for example, take two separate pieces — large stores of data — and then correlate those two. That’s very valuable business intelligence.

O’Reilly: Exactly. You could look at market research as a byproduct kind of business … from putting together data from the traffic of a lot of different Web sites. In the same way that Network Solutions took that domain name registry and said, “Wow, this is a valuable asset, we’re going to start to see companies that have delivered an application suddenly [realize they] have a valuable asset that can be exploited in other ways.” This is where I think AOL keeps missing the boat. Their instant messaging user identities are an incredible asset. Now what if they were able to turn those into a component in other people’s services? I think this is where Microsoft has really …

InfoWorld: They got it.

O’Reilly: They got it, because they know platform strategies. There’re so many people out there with an application strategy, and it’s so obvious that a platform strategy will win. You can have this great application, whether it’s AIM or ICQ or whatever, but unless you can get developers to start building on top of you — as an identity service or for presence management or for transport — you’re not really locked in. And the point I’m making is there are all these opportunities for companies that have built one kind of asset to suddenly see it in a new context when they say, “OK, if I start thinking of myself as an Internet operating system component, how do I think differently? Who else could use my data? Who else could use my services? How could developers rely on me?”

InfoWorld: And clearly the Web services stack and the XML revolution is driving this.

O’Reilly: Yes.

InfoWorld: I recently had a conversation with Sun CTO Greg Papadopoulos about Jxta, and I suggested to him, going back to your point about pipes, that Jxta’s dynamic piping was a fascinating technology — it’s a very powerful mechanism that offers an opportunity for the Internet operating system to talk not just server-to-server, but also client-to-server and client-to-client. And I got the distinct impression from him that the main factor there was politics, that Sun was concerned about opening that can of worms because of concerns about peer services being unregulated and the intellectual property question raising its head again.

O’Reilly: I think that’s a real mistake. Two years ago at JavaOne I gave a keynote in which I talked about the pipe-and-filtered model and how important this was. If you go to tim.oreilly.com, which is where I have my archive, there’s a transcript of the talk. And it was after [this keynote] that [Sun] brought me in as an adviser to the Jxta team, which was very excited about the pipe concept. Quite a few other people have been too. It’s a fairly fundamental idea, that we want to be able to take the data, transform it, and pass it on to someone else.

InfoWorld: It’s also that bootstrapping idea again that Unix evolved out of, so it’s sort of the next stage of it.

O’Reilly: In one sense, people get that. But there are concerns. There’s been a lot of fear mongering about intellectual property protection. I believe that most people do want to do the right things. I don’t know if piracy is the right word for it, but there’s often a freewheeling phase early in a new market where a lot of stuff is traded freely. But that doesn’t mean that there won’t be a commercial market. The real limiting factor is people’s attention, people’s ability to find things. The Web, for example, started out totally free, everybody was building sites. It was kind of the mythology that everybody is equal. It isn’t that dissimilar from peer-to-peer file sharing. Here’s this network and anybody can get to anybody else. What actually happened was that over a relatively small period of time we built up a pretty significant infrastructure of “publishers” — people who were in centralized positions of power, control, revenue generation, and [drawing] a disproportionate share of the attention. And watch what happened on Gnutella. People were doing file sharing. After awhile, certain sites [became] more well-connected than others. And then we had some of the legal pushback on all of the file sharing and things went in different directions.

But take Usenet, originally totally free. Everyone was sharing bandwidth effectively. What happened? Bit by bit we saw the emergence of the ISP marketplace. UUNet, which is really the biggest ISP even today, started out [as] a collection of people who wrote a free backbone site. They realized that free didn’t scale. They said to people, “Hey, if you pay us such and such a month, you can be connected directly to us.” And people were glad to pay. We saw a transition from a free Internet to a paid Internet over a period of about two or three years. And people still [believe] the mythology that the Internet is free, when in fact almost everybody today pays for it. So [there’s] this idea somehow that when things are free [originally], they will always be free. You hear people like [Walt Disney CEO Michael] Eisner saying no one will produce intellectual property. It’s total and unadulterated hogwash.

InfoWorld: That’s the one side and then there’s the other side, which is the Recording Industry Association of America (RIAA) and their president and CEO Hilary Rosen and how their objectives seem to intersect with some of what Microsoft’s doing with the Windows Media Player.

O’Reilly: They’re working to build a controlled market that locks in existing players. And what I’m saying is, we don’t have to be afraid in a free market that we will demolish the system and [lose] the ability to make money. The point is that the nerves, the places at which you make money, may be different. What we see are the existing players trying to stop the new industry with fear mongering, saying no one will make any money at all. But in fact, when you get a new technology, you get new ways of making money. When video came out, there were opportunities for video stores, for video distributors, for video manufacturers. Yes, it may in fact be different people who become the aggregators than the current music publishers. So they’ve got to figure out how to become those aggregators before someone else does, rather than trying to stop it from happening altogether.

InfoWorld: That’s a difficult problem. Novell did not succeed in staving off competitors by buying companies or by monetizing their model.

O’Reilly: It is a difficult challenge, but nobody guarantees a company’s existence. Look at the transition from the mainframe world to the PC world — IBM managed to do it pretty well. Yes, they’re no longer a monopoly. But they’re still a successful company. There will be ways for publishers, whether they’re music publishers or print publishers, to find the opportunities in the next generation of technology. I would urge publishers to look in that direction rather than trying to stop the future. Publishing is a fundamental act that’s mandated by mathematics, so there will be publishers of some kind, there will be aggregators. Because no matter what technology exists, people are not good at doing really large numbers. I think they say that crows can count to seven. Well, people can count to about 20 comfortably. I remember [AOL Time Warner chairman] Steve Case saying, “The truth of the Web is that most people visit about 20 Web sites.” Then he says, “I’m going to exploit that fact.” Other people exploit [it] in different ways. It’s easier to remember Google than it is to remember 1,000 [Web addresses].

There was fear mongering among bookstores in the early days of the Web that publishers would sell everything direct. [But selling books] didn’t change all that much, because the fact is, do I want to buy an O’Reilly book from O’Reilly and an Addison-Wesley book from Addison-Wesley and a Wiley book from Wiley? No. I want to go to one place where I can get one of each. So that led to the opportunity for people like Amazon. And there were a lot of people who could have become Amazon if they had looked at the future instead of looking at the past. In a similar way, there are a lot of opportunities right now for people to look firmly to the future and [ask], “Where is it going?” And how do you find that out? You find that out by looking at the people who have been in the technology the longest. How did they use it? What do they take for granted? How do they think about it? And then you say, “OK, how do we meet up with that trend line?” Going back to our conference, we try to bring together the people who are out there on the edge. [To them,] the Internet is an operating system, and there are enormous opportunities for making money in that environment. And the people who try to keep that environment from happening are going to be the losers, not the winners.

InfoWorld: If it’s going to be true, it is true.

O’Reilly: That’s right.

Source: www.infoworld.com