A Conversation with Bill Joyby Tim O'Reilly
Fortune Magazine calls Joy, Sun Microsystems' chief scientist, "The Edison of the Internet." Joy is a cofounder of Sun and a member of the Executive Committee. His work on BSD Unix and Berkeley networking qualifies his as one of the founding fathers of both Unix and the Internet; work springing from his research group at Sun led to Java, Jini, and various networking technologies yet to be announced.
Joy keynotes the O'Reilly Peer-to-Peer Conference Thursday, February 15, "From UNIX to Java to XML to Peer-to-Peer."
Tim O'Reilly: You've been talking about peer-to-peer for a long time, and I know you're still in stealth mode on some new projects you're working on, but I'm wondering whether you have anything that you're ready to unveil?
Bill Joy: Well, Mike Clary and I have been working with the idea of connected computers since the early '90s. In 1992, I gave a talk at Esther Dyson's conference that looked forward 10 years - which isn't very far in the future now - about having a quarter billion connected mobile professionals each carrying several devices with them. So we've long been concerned that software be architected in a way that would allow you to live with those devices. Having that many devices creates a nightmare because they all have to be managed and found and addressed and named and so on. We just couldn't imagine that the devices would be as complicated as conventional systems. We also felt that they had to be more reliable, so we took some early steps with Java to make things simpler and more modular. I think we've had good success with that. We also brought out the Jini technology with the idea that devices need to find each other dynamically.
Now we believe that the kinds of things people are calling peer-to-peer are part of a set of technologies to deal with these same problems. Of course, there's the problem of coming up with any sort of business model, especially in the current market environment, but we believe it's an appropriate area to invest in at this point. Sun's always had a very long-term approach.
O'Reilly: I think peer-to-peer is part of a much bigger movement in the world of networking, one that really caught on with all the buzz around Napster. Napster woke everybody up to the idea that the issue wasn't just a matter of centralized vs. distributed computing, but it was somewhere in between. What do you think of Napster?
Joy: I'm a shareholder, but as I wrote in Time magazine, I don't understand how Napster isn't infringing on the rights of the artists when their music is taken without compensation. I don't know whether you saw today's Dilbert cartoon, but Dilbert says, "I created software that makes all copyrighted work on the Net available for free." The second panel says, "Wouldn't that destroy all forms of creativity and throw us into a depression?" The third panel says, "Yes, but it's very neat."
I think the big appeal of Napster was that people could get music easily and that it was free. I think we need the "easily," but I don't see how people's work can get taken without compensation. I understand it's very popular, but we do have to have a better policy. I think the Napster people recognize that. I'm going to be working with Senator [Orrin] Hatch [R-Utah] on the Senate Judiciary Committee to try and look at some of these issues and come up with better public policy so that we can encourage innovation but also find a happy medium here. It's not going to be easy, but not to try is not acceptable.
O'Reilly: Can you say more about possible solutions?
Joy: I think that the copyright laws need to be enforced, and maybe they need to be changed. We need an enforceable digital-rights management scheme or else we will do severe damage to the book industry, which is the one that I'm most concerned about. Books are the best bargain in the world. And perhaps only if we go through taking an industry apart by having widespread copying for free without compensation, we'll realize in the end what we've thrown away. Hopefully that won't happen. The reading devices fortunately are bad enough that hopefully the newer ones can have reasonable rights management software in them.
O'Reilly: What do you think about Larry Lessig's concern that when we go down the path of strong rights management that we will end up in a situation where we lose some rights that we've taken for granted. For example, the first sale right on books or the ability to redistribute portions.
Joy: Exactly. The reason I'm working with the people at the Senate is that you clearly would like to be able to put the e-books in libraries and have people check them out, and give the book you have to your friend, and we have to find workable mechanisms within the context of the rights management for socially acceptable solutions to that - I mean which should account for the fact that the consequence can't be that everybody gets everything for free, because then it will destroy the ecosystem. You can't say, "This is really neat," but you just destroy the ecosystem. That's too much of a nerd view of the world. What I said in Time is, "Just because it's possible doesn't mean it's OK." Yeah, it's easy to steal stuff, but it's not OK, because it destroys stuff without creating any value in return. I mean the fact that we can easily get access to music is of no use if the music industry falls apart.
O'Reilly: Isn't that somewhat alarmist? The fact is there are certainly studies that show that people are buying more music as a result of free sampling.
Joy: I'm more concerned, as I said, with the book industry.
O'Reilly: Yeah, well I'm a publisher and I'm not concerned about that one.
Joy: I know you aren't. And people also weren't concerned that the market was going to go down when the Nasdaq was at 5000. People's ability to see forward is very limited. People generally assume that tomorrow will be like today, and it's just not the case. I mean this technology is transformative enough that we could have a big problem, and if we don't take steps before we have the problem, we won't be able to take steps when we have the problems.
O'Reilly: Well, that's certainly true, but what about software? I mean software has been freely copyable now from the beginning, and we don't have that problem.
Joy: Well, software is a safe harbor that's morphing into a service. People don't want to install and manage software on their machines. They want someone to do it for them. You can steal software on an individual basis. It's very hard for an institution to steal a service, or in fact to deploy software that implements a service on a large scale. So that largely solves the problem. The personal productivity stuff is mostly given away in the context of a subscription or a communication service. You know, mail isn't an application, mail is a service.
O'Reilly: No, I understand, so the point is that there are business models that end up routing around the business-free problem. The fact is Windows is installed on machines through OEM contracts. People sort of take for granted that they just get it at this point.
O'Reilly: But the point I'm making is that given that when we started down the path of software distribution over the Net, it wasn't clear that it wouldn't be freely copied. We had many copy protection wars in the '80s and it ended up not being the problem that the fear mongers articulated.
Joy: Well, it's clearly a problem with music.
O'Reilly: Is it? Aren't we at about the stage with music that we were with software?
Joy: No, it's a problem. It's that the rights of the artists aren't being respected that I think is the problem. And with CD writers and MP3s and Napster, the artists have lost control of the uses of their recordings. In a way, then, whose financial viability long-term is unclear? So I think we need to take a good hard look at how we are going to do this and come to a consensus. We're going to have a scheme that's workable.
We do the same thing with the broadcast spectrum. There are frequencies in the air and people don't make devices in general that can transmit and receive and just jam - or else as a society the whole communication system would break down. That's essentially a social contract and we agree that it's quite easy to build a radio that transmits and receives on any frequency, but in practice we don't do it and we enforce those laws and the society benefits. And I think the same thing can be true for music and books. The fact that they're distributed electronically doesn't change the fact that we need to have an agreement as to who owns what and how it's protected.
O'Reilly: Yeah, that's fair. Let me move on to this whole question of wireless and the wireless spectrum. It raises an interesting question of standards. One of the things that always has appealed to me about Unix is this idea that there were a few simple rules, and once you have those rules, people could write programs that work together. And it seems to me that part of what we're lacking in the peer-to-peer space so far is that sense of what are those minimum standards that a program needs to follow so that it will work with other programs? Do you have any thoughts on that topic?
Joy: It's very difficult to get people to agree.
O'Reilly: But we did it with Unix and we did it with the Internet.
Joy: Well, there weren't very many people.
O'Reilly: That's a good point.
Joy: People just generally like to disagree. One thing I would say, though, about peer-to-peer is that the real trend here is optical networking. It's going to provide us with almost unbelievable bandwidth. You know, to the extent that it's optically switched, the photons really do go in one end and drop out the other. It's not really even packet-switched; it's circuit-switched. And that provides you with what you might call cross-sectional bandwidth. You know, the sum of all the point-to-point bandwidth is just huge, and that provides an opportunity for connecting machines in a way that is much more powerful than they have in the past.
Gordon Bell used to say any two connected computers, if they're connected with a tin can, or two guys in Morse Code, that makes a parallel computer. But obviously the economics of the parallel connection affects the viability of the algorithms that you can run, and things like SETI@Home, which are embarrassingly parallel, work quite well, but with the kind of optical bandwidth we see, to the extent you can tolerate latency, you'll have as much bandwidth as you want.
You're not going to repeal the speed of light in the near future. You're going to have a latency issue, but you won't have any real bandwidth constraints. This has enormous implications: It helps things to become a service, and it makes it possible to cobble together machines to do certain problems and to replicate things real easily. It affects the whole economics of peer to peer. It's a strong wind at the back of this kind of distributed architecture.
Pages: 1, 2