Paul Horn was put in charge of IBM Research in January 1996, just as the Internet was entering the popular lexicon.
More than a decade later, Horn's stepping down to the applause of his colleagues who credit him with transforming Big Blue's research and development efforts--including speeding up its corporate metabolism to bring more tech innovations to market. CNET News.com recently spoke to Horn, who is looking back on a 28-year career and is looking forward to a new career as a scholar at New York University.
Q: You're becoming a Distinguished Scientist in Residence at NYU. What exactly does a distinguished scientist do?
Horn: Anything he wants to do. (Laughing). Basically, it's a fancy title. I have no formal responsibilities, but what I'm hoping to do is find synergies and perhaps help them commercialize some of the technology...So basically, the job is "Come down and figure out what you can do to help change the university."
Part of your charge, as you suggest, will be to get things a bit more market-oriented. The modus operandi at IBM Research also has been to try to get its scientists more involved in customer engagements and be more market-oriented.
Horn: That's true. More and more of the value is higher up the innovation stack. That requires not just a new device (but also) to really understand what new business models can fundamentally change companies within an industry. The only way you get that information and help build those changes is to work with the customer.
It's not so much a question of size but a question of culture.
What's been the learning curve? Has it been hard or easy to get them to follow your lead on that?
Horn: Well, some of our scientists--we just love to keep them down in the basement. We don't unlock the doors that often but, to be honest, you have a lot who, when they interact with the customers, they actually love it. It's important that they get to interact with them on really challenging problems.
Does that come with the expense of theoretical experimentation?
Horn: I don't think so. If you don't do this, you really run the risk of being like the old Bell Laboratory: very inventive, but not really good at innovation because innovation really requires understanding the market. So the more of these channels you have to the marketplace, the more positive feedback the scientists get because they get to do things to change the world.
I recall touring an IBM Comdex booth in the mid- or late-1980s and seeing the digital mouse for the first time. But it took at least another year or two before IBM got around to incorporating the device into its laptop line. How does a big company avoid that kind of bureaucratic pitfall when there is a cool product under development in the research labs but because of the size of the company, it's going to take an awfully long time before it hits the market?
Horn: It's not so much a question of size but a question of culture. The way we used to be is that our researchers really didn't understand our businesses that well. So they didn't understand the barriers and what it takes to get new innovations into our products. You have to understand the product stream; you have to understand how to build code that doesn't have to be rewritten; and you have to understand how to be able to integrate a new material with an existing process.
The barriers you're talking about, I don't believe are so much associated with size and bureaucracy. It's more a question of culture and trust.
You've been at the helm since 1996?
What's been the most important development to come out of IBM Research during your tenure?
Horn: Oh boy! I knew somebody would ask me that question, Charlie.
Come on. It's a softball.
Horn: This will sound crazy because it wasn't a product. It was the time when Deep Blue beat Gary Kasparov. It changed the culture. If you looked at the newspapers or journal articles about IBM before that time, every one of them started off, "IBM, the troubled computer maker"...because we were still coming out of the near-death experience we had in the early '90s. After that, you never saw that again.
What was the impact internally?
Horn: It changed the culture, because people started to feel good about what they can do here again. I'm pretty proud of that accomplishment even though it wasn't something that instantaneously led to a lot of revenue.
What's been the biggest surprise for you over the course of that era?
Horn: Probably the biggest surprise was how rapidly the Internet went from some cute thing to a fundamental way to rethink every aspect of a business, the way it's been able to transform business.
My view of history may be somewhat slanted by the company I work for and how we think of the world. But we had a very simple-to-manage-and-operate world in the old mainframe era. But a line of businesspeople rebelled against that model and at having the CIO build all their applications.
Because they could grab a PC and build a line of business applications trivially in a matter of days or weeks rather than having the CIO shop...that created the whole client server era. The distributed computing world has got great advantages, but the cost of managing distributed systems is bringing distributed computing close to the breaking point. The people-management costs are going through the roof, and what we're seeing is now some level of reconsolidation. With Web 2.0 technology, there are new forms and new ways to build applications trivially, and the interesting thing about Web 2.0 is whether that model of mashing up new applications is going to be another disruptor--like client server--or whether we'll be able as an industry to manage it better than we did.
It seems that the really interesting innovations, at least as far as the consumer market goes, has been coming more from independents and small developers rather than the big guys. Is that because new interesting technologies that don't really have an immediate business application don't garner serious attention until some smart bulb figures out how to make money off of it?
Horn: No, I don't think that at all. Today you can build an application or a service by matching up other existing services that are perfectly good lines of business applications. And then all the kids, the hundreds of thousands of programmers who are growing up, they're going into the industry and they're going to know how simple it is to build the applications on a LAMP stack. So, the traditional IT suppliers like IBM and others have got to help corporations to be able to incorporate and utilize all that innovation and those novel ways to build software. If we don't, then it's client server all over again and there'll be two different worlds. There'll be the sort of corporate J2EE world and then there'll be the LAMP world, and applications won't work back between the two of them.
The cost of managing distributed systems is bringing distributed computing close to the breaking point.
I wanted to get your thinking about IP and licensing; IBM makes these big gestures to the open-source community. But how do you decide what's too vital to give away and what it can afford to contribute?
Horn: Part of my job was to make sure that we look out into the future and we don't miss future client server revolutions or don't miss, you know, a PC revolution that's going to disrupt the minicomputer business. So that we embrace disruptive technologies in the Clay Christensen sense, from the Harvard Business School.
Technologies that upset old businesses and change the world.
Horn: And one of the things that can change the world is the proliferation of Web services and Internet standards...They flatten the world in a Tom Friedman sense, and they allow work and business processes to go on around the world and be reintegrated fundamentally better in something that tends to be called services-oriented architecture.
What all that means is that companies are evolving into much more distributed and collaborative ways of doing business and innovation...and all of that is disruptive for the traditional old-fashioned way. Think about intellectual property. The traditional old-fashioned way you think about it is you invent something, and you keep control of the intellectual property. Now what we're saying is you can fundamentally speed up innovation by collaborating in a more open, interactive way.
And that's expressed how?
Horn: We're giving away patents in areas where we believe we can help speed up innovation by allowing for more collaboration and, hence, rapid development. A good example is Linux. IBM embraced Linux a number of years ago. Any operating system really requires about a half billion dollars a year to develop and maintain. The community is spending about a billion dollars a year on Linux, and IBM spends around $100 million to customize Linux for our own customer needs and makes sure all our software works on it and that our systems will utilize it.
One of the things that can change the world is the proliferation of Web services and Internet standards.
So for $100 million instead of $500 million, we get a fully customized operating system, and we can take that $400 million and spend it higher up the value chain. We get a lot of money from IP, but if we can replicate that open style of innovation, we get much more value than we get by controlling intellectual property we have on an operating system, and that's why we've given that operating system IP to the Linux community.
There has been a lot of hand-wringing about the quality of math and science education in this country. What's your view?
Horn: We've got major problems. But you know, it's still the case that our graduate programs and our graduate schools are where everyone in the world wants to come. So, we've got a large gap and we have a major problem...but as far as advanced education, I think the U.S. has got the best in the world.
You know, it's not an accident that eBay, Amazon, Google, YouTube and MySpace grew up in the U.S. Now, there may be other countries that will have companies that, you know, follow that, but we have the most innovative economy, and our economy is driven by innovation, I think more than any other.