In his first public interview since taking on the responsibilities of Sun co-founder and chief scientist Bill Joy, Sun Microsystems Inc.s chief technology officer, Greg Papadopoulos, sat down with eWEEK Senior Writer Darryl K. Taft to pay homage to Joy and to give a glimpse at his views on the future of the computing industry and for Sun under his watch. Papadopoulos spoke to eWEEK at the Object Management Groups (OMG) Integrate 2003 conference in Boston this week.
eWEEK: What would you say is Bill Joys legacy?
Papadopoulos: His legacy is broader than Sun. Its really to the whole community of the Internet. So all of Bills contributions have been ones that made the way into the substrate of what we do—whether its things around the network file server, or adding TCP/IP to Unix and sockets was a pretty big step, to the network file systems to his evangelism of Java, and kicking the chalk lines on new programming paradigms.
I just think that hes been a tremendous person who helped get the Internet where it is.
The fact is that over the past several years his attention has really focused on what he considers to be the next wave of technology, which is not so much IT-based. Its a lot of stuff around nanotech, bioengineering, and thats where his heads been.
eWEEK: How much of that will you carry over?
Papadopoulos: Well, we do have programs that are working in nanotech in the company, but his areas of concern were really more about what are the social implications of that. And there isnt a particular reason for Sun to carry that social agenda.
eWEEK: What will you miss most about Bill Joy?
Papadopoulos: I will just miss him as one of the many brilliant people at Sun. It truly is a candy store full of folks. Each one of those people brings their own unique perspective. Bill has one that is very much a renaissance perspective. You know, he is a voracious reader and well informed in lots of topics. So he brings that kind of perspective to things. Generally speaking, though, when I want more depth in any of those particular domains I turn to other people in the company. So what Ill miss from him is sort of that “out there” breadth that hes had.
From an operational point of view … hes not had an operational role in the company for years and years.
eWEEK: Right. And that is what I am trying to lead into. Sun has officially stated that youll take on his responsibilities, what does that mean actually?
Papadopoulos: There are about two dozen people that have worked in a couple programs hed been working on and there are some facilities in Aspen and some stuff up in San Francisco that hes been working on that gets assimilated into basically the CTO organization in terms of Sun Labs—the labs and advanced development, and engineering and the Java Community Process, all that stuff has historically reported into me. So this is for me, maybe a 10 percent increase in the advanced development resources in my organization.
And Ive been sort of virtually doing that in any case because over the last few years Ive been funding that work and the people are dotted lined into me just from a management perspective.
Hes not the sort of person you replace. His responsibilities are really easy to assign, but you cant replace him as a person in the sense of, and I wouldnt pretend to.
Just like I wouldnt pretend to go and replace James Gosling or Rob Gingell or Ivan Sutherland. These are folks that are just greats in their fields.
eWEEK: Well, several people have said that with Bill Joy gone the vision is gone from Sun. How do you respond to that?
Papadopoulos: If you think that Sun doesnt have a vision now, criticize me, not his departure.
eWEEK: What do you perceive your role to be as CTO of Sun?
Papadopoulos: Well, that vision thing. At a very basic level Im the fiduciary for the technology of the company. And it is a high-tech company, so it is fundamentally where we go with our R&D portfolio. Thats fundamentally the thing that I worry about. Its spent at a rate of $500 million every quarter. And any project that youre working on is three years away from being completed, because engineers walk around with signs on theirs heads that say “give me something to do that Im going to show you in 18 months.” So you always have to have this lead on where that is going. Extrapolation in this business is really dangerous. If you sort of take what the market is asking for right now and what people think they need, thats what I should be investing in. That gets you into some serious trouble in computing. You sort of follow the conventional wisdom and you sort of end up three years from now having built the thing that people wanted three years before. Its a lot of leading that duck and getting there.
For me there are two sides of the role in the way that its built up within Sun. Theres the “get out and fundamentally be a liberal force” that makes the company uncomfortable at any moment in time with its R&D portfolio. Thats one side. Sort of this liberalizing view and the view of making the company in this perpetual state of discomfort with what its doing.
The other side is the processes around conducting the engineering. Which is all the top, for instance the chief engineer, who is Rob Gingell, reports into me. He is deeply concerned with the how is it that we go create whole systems out of the company because ultimately Sun delivers a single product: its network computing. And if youre going to keep $2 billion dollars or R&D coherent, you have to think really hard not just about whats the next great thing thats cool to go do, but what is our ability to actually deliver a system and execute on that. So I have things that are cross-company things, like Java Community Process. Ive got a group of a few dozen over-the-top physicists there who work on a lot of basic physical technology.
eWEEK: So what excites you about these next great things?
Papadopoulos: What Im really excited about is what does that new world look like, whats the consequence from what the software stack is and how is that going to affect systems, because its going to re-invent a lot. And were kind of sitting at this classic triple point or phase change in that you judge things with the current lens of whats efficient, whats right and that kind of builds up into this extreme kind of commoditization and innovations over, its the end of the industry. While this next thing is really starting to accelerate and people get confused as to whether youre talking about the existing thing or where youre going. And that confusion happens all the time.
eWEEK: I liked how you pulled out words like open and standard, free and commodity in your presentation.
Papadopoulos: And they are completely different things economically, to competitiveness. And I will get ill if I hear things like Intel is the open industry standard processor architecture and were proprietary. You can license Sparc for $99 and go have at it. Lots of people have. In fact, we just acquired a company that went off and did that. Fujitsu does that.
eWEEK: I know you guys have argued against Microsoft selling software as an integrated stack and yet with this Orion offering youre doing the same thing. Did your thinking change somewhere?
Papadopoulos: I think that our critique of Microsoft in that is not that you shouldnt look for value of integration, but that in so doing it all needs to be open and substitutable components—that you dont use it as a way to lock people into a set of contracts of your implementation. And we were using this word that sort of caused you to trip over it too many times, so we dropped it out of our marketing vocabulary. But we talk about Orion as being “integrateable” as well as integrated. That it is a component architecture where all of those interfaces are open. And for everything that we have in Orion theres an equivalent open source component. So the value that were doing is not that I made this big wad of software. In fact, its not a big wad; its a set of components. But that I have caused the synchronization of the release of those components so that I get a verifiable set of contracts there that take cost out.
If you look at an operating system, if you say Solaris or you say Windows or Linux, for that matter, I can give it a version number. I can say its Red hat 8.0 or its Windows 2003 or its Solaris 10. And you know exactly what Im talking about. You know that whole collection of contracts that are there. You can go look up at a web site, you can verify it, there are tools set against it. If you look at any one of those systems: Linux, to Windows to Solaris, its not a big monolith. Those are all component architectures underneath. But what weve done is weve given those components a name, weve released them, weve verified them, and we do patches against them. What were doing very simply is were just trying to do that for the next level of contract because people have moved up from the OS to here. So lets give these components names and do the verification and the synchronization of the release. And while were at it there are a lot of margin dollars that people are investing in that software. And my view is not to go in and claim those margin dollars by doing competitive software at that level of margin. My view is actually to go drive efficiencies into the software business at those layers. And do for software what Dell does to hardware. Too many customer dollars are going into pieces of that stack that rightly ought to be part of that standard contract.
Discuss this in the eWEEK forum.