Interview: Bill Joy, Sun's vision guy

Java co-author lets Jini out of the bottle




As chief scientist of Sun Microsystems Inc., co-founder Bill Joy has spent a quarter of a century designing Unix software.

While a graduate student, Joy was principal inventor of Berkeley Unix. He went on to design Sun's Network File System and codesign the Sparc processor architecture. After joining Sun in 1982, he co-authored the Java language and, more recently, conceived Jini distributed computing technology.

He received a bachelor of science degree from the University of Michigan and a master's degree in electrical engineering from the University of California at Berkeley. He is co-chairman of the Presidential Information Technology Advisory Committee.

Joy will deliver a keynote address on Java and Jini at the Software Development Conference this week in Washington. GCN chief technology editor Susan M. Menke interviewed Joy by telephone.




GCN:'Has anyone alive invented more currently used software than yourself?

JOY: No one comes to mind.



GCN:'When you look back on the last 10 years, do you feel Java has taken the world by storm?

JOY: Yes; look at the Internet. It's becoming clear that the nature of the Java language makes it much easier to produce reliable software. That was a specific area I was involved in, and I'm very pleased about that. For the first time since about 1960, Java is making it possible to do scientific, system and symbolic programming in a single language. Historically, they were done in Fortran, C and Lisp or Smalltalk.

I see strong movement toward Java to get the benefits of simplicity and object-oriented programming in a modern language.



GCN:'Java is maddeningly slow to start up in Web browsers. Is it ever going to get any faster?

JOY: The alternative is that you download software every time you visit a Web site. You're looking at some artifact of the way Java is used in a particular program today, whereas I'm taking a longer-term view. There's no reason why Java couldn't start in background once you launch the browser. That's just a matter of engineering it better.

Of course, Microsoft Corp. has no business interest in making your life easier. The Netscape Communications Corp. browser has been in limbo for a while. Hopefully you'll see that fixed in the next version of Communicator.

The truth is, if you want to author Web content around the client, there are not that many choices. That's why the cutting-edge interactive stuff uses Java.



GCN:'When is Jini coming?

JOY: If you look around your office, you see a lot of things using electricity, but they don't tend to have software in them. When those things have software and get networked, they can work together. If you have a cell phone in your car and want to make a hands-free call, there's already a speaker in your car radio. But there's no software protocol level at which the phone can talk through the radio.

By putting software and networking into electrical devices, we want to create an architecture in which individual devices appear as objects. They can come together to do things in ways that weren't necessarily preplanned but are simple.

Objects replace protocols in the sense that you can send objects between machines or devices. The objects themselves implement protocols. In a world with agents, you don't need protocols anymore. The protocol that moves the agents is the protocol to end all protocols.



GCN:'How will this arrive on the market?

JOY: The breakthrough place is going to be in cellular handsets. The first one is going to be the iMode phone in Japan, which is the first country to have high-speed Internet access from handsets. We showed Jini in prototype form in some handsets at JavaOne last summer.

The first generation mostly will be for wide-area use. Our expectation is that it will start in Japan and go worldwide. The next generation will give the handsets a kind of proximity wireless connection, either with Bluetooth short-range radio technology or with wireless Ethernet. This is all rolling out in the next three to five years. The numbers are staggering.

By 2010, the industry expects to have a billion of these handsets in use. There's a 30 percent to 40 percent replacement rate every year. Once the stuff is deployed, it will be adopted quickly'in the same way we've seen digital wireless technology take off in the United States.

We think that these will become the PCs of the future for most people. No other consumer appliance is on a trajectory in the next decade to ship anywhere near that number of devices with wireless connectivity. It's a staggeringly large number. More handsets than cars will be built, more handsets than DVD drives or videocassette recorders or televisions. It's off the scale until we have electronic Post-it notes or paper clips, which probably will happen at some point, too'a computer in every piece of paper is possible in the next 20 years.

In the 10-year time frame, the center point for most people's computing is going to be a pocket computer that looks like a cell phone.



GCN:'The battery problem will go away?

JOY: It isn't like a notebook computer. It'll have the battery life characteristics of a phone. The best battery standby time on a digital network is almost a week now. My Palm VII [from Palm Computing Inc. of Santa Clara, Calif.] that uses similar power management techniques lasts on two alkaline AA batteries for a week to two weeks. My two-way pager lasts a week.

We're not talking about running Microsoft Windows CE or the big power pigs. We're running the handset code we already have, maybe a small browser and small Java applets. The Java won't have any start-up time because it's in the read-only memory. The wireless network is probably faster than an Ethernet connection unless you have digital subscriber line or a cable modem. The Palm and pager are instant-on'the batteries last a long time.



GCN:'Will this play into some telecommunications carriers' plans to give people a single, portable phone number that they carry around in a handset and plug in anywhere?

JOY: I'd be loath to predict people will have only one phone number. There are good reasons to have more than one. The trend seems to be the other way. Some phone numbers represent places, some represent people.

I think it's more likely your handset will have several phone numbers you give to people to cause different things to happen, or at least Caller ID will be so ubiquitous that you effectively have that behavior.



GCN:'How soon will Sun put up a Wireless Application Protocol site on the Web for such devices?

JOY: That's a complicated question that the Sun iPlanet people are working on. Another question is, what are the plans for WAP on Solaris? And another one is, are we going to put up our own site accessible by that protocol? I don't know the answers.



GCN:'Will the Sun Ray do any better than the network computer did?

JOY: I think the NC will do fine eventually, and the Sun Ray will do fine. It's not something individuals will buy. It's something companies buy.

The advantage, even in a small office, is a high-availability server and no single point of failure. If the Sun Ray on my desk breaks, I can go to another one. If the server breaks, it hot-swaps over to another one.

When hundreds of people do something against a database, they can't take orders or whatever when their PCs are down. This is a much better architecture for high availability to a lot of seats. Most people do e-mail and the Web and maybe a couple of applications. Those things are becoming services on the Web. It doesn't matter which device you use, it's a question of being at a browser. Sun Ray has a browser, local access to a database you need to do your job, and e-mail.

Citrix Systems Inc. of Fort Lauderdale, Fla., has tried to do similar things with

WinFrame, but Windows NT isn't reliable enough and doesn't scale enough to capture the bulk of this market.



GCN:'How many computers and cell phones do you use?

JOY: I have a screen at the office, a screen at home, a notebook computer, a Palm VII, a pager, a digital cell phone for the United States and one for Europe.



GCN:'How bad do you think year 2000 will be?

JOY: As long as the power stays up, we'll be fine. Most systems aren't completely automated. And people are pretty clever. It's a cautionary tale'we shouldn't take people out of the loop on a lot of systems. The problem is when a lot of things break at the same time. There'll probably be a few things that are hard to fix, a cascade of effects.

What's more


  • Age: 44
  • Last book read: For Common Things by Jedediah Purdy
  • Favorite Web site: www.nytimes.com
  • Leisure activities: Skiing, hiking and playing with his kids
  • Worst job: 'Digging sewers as a summer job.'
  • Best job: 'Working for Scott McNealy.'

inside gcn

  • high performance computing (Gorodenkoff/Shutterstock.com)

    Does AI require high-end infrastructure?

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above