ACADEMIA
State of the Technology Union
By Network World Staff, Network World -- Six research executives, together controlling billions of development dollars, examine the changing face of R&D and tell us how the network industry has become more innovative than ever. 
Mario Mazzola
Chief development officer, Cisco Number of researchers in the organization: More than 860
R&D budget: $3.3 billion for 2003
Number of patents issued in 2002: Unspecified number issued, but Cisco says it files approximately 650 patent applications yearly and it has more than 2,300 pending applications. Past "A few years ago, the R&D path was more defined. We had the opportunity not only to combine and repackage technology in a more convenient way, but also to introduce functionality that could create a great deal of value for our customers. There was a clear path involving the development of our hubs and concentrators to switches to provide more bandwidth, dedicated bandwidth and superior routing capability. This technology shift provided many new characteristics - much higher speeds, much higher density and much lower costs." Present "We are looking at a new innovation wave, which is more fundamental in some respects than previously. Some interesting aspects have to do with support for different types of media and traffic, which has implications in terms of universally supporting quality of service and multicast, for instance. Other aspects have to do with blending security, embedding firewall, VPN capabilities, strong intrusion detection and antivirus prevention. The intention is making it easier to integrate different applications, and to that extent, network intelligence is increasing. Businesses will transform themselves and their information with these intelligent networks. "The most relevant investments for customers are those taking advantage of the benefits relating to Moore's Law - meaning the rapid evolution of the technology - in a way that also ensures preservation of customer investments. The difference now is a move from point products and a move toward global systems and solutions. It's an exciting time." Future "We continuously look at new and disruptive technologies, but keeping in mind the important translation into capabilities and characteristics that are relevant to the customer. We focus a great deal now on how to increase productivity - how to make not only the capital equipment, but also the operational cost, more convenient for all of our customers. "We have defined several areas as new growth opportunity. A major one is in IP telephony, which we started to make investments in a few years ago, and storage and security. We have technologies here and there on the very edge of innovation that don't take the approach of doing basic research like in a Bell Labs. But even though I'm an engineer at heart, I don't believe in force-fitting technology for the sake of technology. I'm more interested in understanding the ways we can bring as much real value as possible to customers." - Phil Hochmuth Richard Lampman
Senior vice president of research, HP Labs Number of researchers in the organization: 750
R&D budget: Undisclosed
Number of patents issued in 2002: 1,385, including those obtained from work done at Compaq. Past "In the past, many corporate research labs had indirect visions. They pursued technology work without understanding what the result would be. When competitive pressure increased, people realized that labs had to be a real contributor to the business. This forced many research labs into painful transitions in the past 10 to 12 years. "At HP, there have been three disruptive technologies. The first was ink-jet printing, taken for granted now. It started here and was an uphill struggle for awhile. But ultimately, business people got excited about the technology, and it created this huge success for us. The second was the program that got HP into Reduced Instruction Set Computing processors - the discontinuity that moved HP into the first tier as a computer company. That technology was controversial, too, because we had many existing products that had to change. The third is the Itanium processor technology. It is emerging from the 'Is it going to happen?' stage to 'It is starting to happen' phase." Present "Itanium, which has the potential to shake the whole computer industry, is a strategy we've pursued for the last couple of years. When you look at the potential for building chips with hundreds of millions or billions of transistors on it, the best way to use that silicon shifts. That's a lot of what drove our early theory about what ultimately became Itanium. "We always have existed to look beyond the current product horizon, to develop technologies that give HP competitive benefit and give our customers an advantage. Chip technologies, such as silicon-on-insulator or copper metallurgy of a couple of years ago, had pretty long development cycles. There's still a lot of variation, but across the board, development cycles have gotten tighter. We have a range of things from some with short cycles to some that are quite long. Architectural issues, such as a new CPU, are in the middle, and software has the shortest cycle. All continue to accelerate in terms of the velocity at which we work." Future "Look at the use of virtual resources in the data center, a hot topic. That has been a rapid development cycle for us, and we have an extremely strong position in it as the result of focused work over a few years and in the labs over a short cycle. You can win in either [development] game and build value. It's about getting the right tool for the job. Even on our longest-range programs, those people have a clear vision of 'If we can do this, it would create an opportunity for HP.' It is motivational for people to imagine how their work could be used." - Deni Connor Alfred Spector
Vice president of services and software, IBM Research Number of researchers in the organization: 3,000
R&D budget: $5 billion for 2003
Number of patents issued in 2002: 3,288 Past "In this industry, there was less emphasis on commercial value in the past. People have done interesting things, but now we would like the needs of business and society to drive those interesting things. During the bubble, capital in start-ups was so available that the economic system might not have been efficient at allocating society's resources to the most important problems. Society is probably somewhat more efficient today and more conservative in capital allocation, which means the most important projects are the most likely to get the necessary funding. That's a good thing." Present "Our research goals have shifted toward service-related software. Now our research lab has a clearer focus on work that's driven by real problems. The work might be equally innovative, but it's driven by problems that are valuable to our society. "We're doing a lot of work in natural language, focused on conversational voice systems for call centers, translation of Web sites into multiple languages and categorization of documents so you can find them quickly. Those are great issues that are every bit as interesting as some of the early artificial intelligence work in natural-language processing that wasn't focused on business problems. "Our goal these days is to do 'factorizable' research. We want things that have long-term goals, but we want to achieve intermediate benefits, or factors, along the way. One example of factorizable research is our work in superhuman speech recognition. Our goal is to build systems that recognize human speech as well as humans do, or even better. Our idea of the way to get there is to make an improvement in quality every year. That has commercial benefits. With compounded benefits, it can deliver a really incredible result. It's marketable as you go, but the result is something that makes you say, 'Gee whiz.' "An example of something nearer term is our work around optimization in our on-demand innovation services unit within the research division. We've created a unit of researchers that spend half their time on traditional research and half their time working in the field alongside IBM business consultants. They find important projects that need the innovation a Ph.D. or a researcher can provide." Future "We're going to have considerably more direct interaction with our customers through technologies like on-demand innovation services. We see that as profoundly valuable - connecting the minds of our thousands of research division employees with the problems that our customers have to catalyze rapid innovation. So many of the interesting problems associated with networking and IT are now not about the technology but about industry problems that the technology can solve. "One area that I know is important is secure, pervasive networking. We like our wireless services, but we want them seamless and secure, and with enough privacy that we can trust our systems to have significant amounts of information about consumers and businesses. There's a lot of good work to be done there. "Research is going through a transition as it grows into more industry domains. In biotechnology, computer technology is exceedingly important as it relates to genomics and custom drug treatment, for example. All sorts of things will need to be done in areas of business process integration and self-managed systems. That might be where the great innovation is. And you want to be working alongside customers in other domains, not just with other computer scientists, to do that work. "We think businesses will procure services essentially through transactions across a network using technologies like Web services. To make that work properly requires autonomic computing and service-level agreement management. It requires strong security and high availability. There will be significant growth in these fundamental networking areas as customers realize the huge cost advantages they can get by performing some activities on utilities across a network." - Ann Bednarz Sanjay Subhedar
General partner and co-founder, Storm Ventures R&D spending: Oversees a $310 million fund focused on seed and early-stage communications infrastructure companies.
Investments include: Airespace (wireless LAN infrastructure); Ample Communications (high-bandwidth optical connectivity systems); and Fourelle Systems (Web acceleration for WANs, including cable, frame relay, satellite and wireless). Past "In the old way of doing things, customers felt comfortable buying end-to-end solutions from one company. When you bought a new IBM computer, IBM architected the hardware, the operating system, the applications and everything around the computer. Today, this is disaggregated. Computer manufacturers put pieces together and offer distribution channels. "For this reason, centralized corporate R&D labs don't make sense. There are too many technologies to track and improve. Today, in anything you do, you have to share with other companies. You can't be proprietary. "Initially at Storm Ventures, I felt that the demand for bandwidth would continue unabated. What I learned, after about a year, was that at the core, there is in fact not a bandwidth problem, that we'd overbuilt capacity. And we had not solved problems in the access and aggregation layers. "We had the rise of the competitive local exchange carriers. Then the CLECs collapsed, and in that collapse we killed broadband access. We've fallen behind countries such as Korea and Japan in terms of broadband access. "At the aggregation layer, there continues to be an alphabet soup. Should it be ATM or IP or [Multi-protocol Label Switching]? What do we need in quality of service, and how do we achieve it? The market was trying to build what we called 'God boxes' -'I'll put everything into this product because I don't know what customers will bet on.' But when you don't target products to particular, known problems, you make the solutions very expensive." Present "Part of innovation is not just showing that something works in a lab, but being able to produce hundreds of thousands, even millions, of products at an affordable cost. "The old R&D cycles have compressed significantly. We have disaggregated many different elements. I can get the same advantages of scale and speed that Cisco has by going to Solectron, a $12 billion vendor of manufacturing services, including new product design, manufacturing and product warranty support. "The process of converting ideas into products in six or eight or more years has compressed to 18 to 24 months. And this same process is pushing you to create global products. It's not as expensive anymore to have a global presence. This is all positive, because it lets me solve the problem of 'How can I get this technology in many people's hands quickly?' It can put a DVD or digital camera into the hands of 50 million people, and that no longer takes years to do. But this positive process also places a huge burden on a company's ability to execute its business plan. You have to be right in every decision, and you have to be No. 1, 2 or 3 from the start or you're irrelevant." Future "Venture capitalists have to recognize this compressed process. We need people who can identify solutions that are global, who can scale those solutions, and who can use outsourced facilities and create virtual organizations to make all this work. "Instead of centralized R&D models, I would say a better model is working with academia. Academics tend to collaborate rather than compete. And collaboration is a huge benefit when you're doing research. And academics are oriented toward sharing information, so research tends to be less costly. "Also, there needs to be a fairly good dialog between users and customers on the one hand, and the technology researchers and suppliers on the other. We tell our companies to find a teaching customer, one who can tell you that if you're trying to make a better product or a new one, and then you have to meet this specific price, and interoperate with these existing systems and so on. Otherwise, you're innovating with no goal in sight." - John Cox Dan Ling
Vice president, Microsoft Research Number of researchers in the organization: 700 across five labs
R&D budget: $4.8 billion for 2003
Number of patents issued in 2002: 511 Past "In the past, research labs tended to be in remote locations from the parent company. The classical example is Xerox Park, located in California when most of Xerox's operations were in upstate New York. But now more than ever, people are aware that getting technology from the research group into products is a difficult process - we always say that technology transfer is a contact sport. It requires the researchers knowing the product development groups well, both at technical and personal levels. It's important to have personal ties to individuals in the product groups so when technical opportunities arise those connections are in place. It has been fortunate for us that we decided to locate our research lab in the middle of the Microsoft campus." Present "Getting feedback from customers - what the technology is good for, how they are going to use it, what they like and don't like - into the R&D cycle fairly quickly is a positive thing. The research group here gets involved in a lot of company meetings. Obviously, one still needs to take a long-term approach, but along the way you learn a lot by having people using those products. "One can do that with a shortened R&D cycle on top of longer-term research. A good example is our work in the natural-language area. Clearly, for computers to understand any human language is still five to 10 years off. But this is an area where we have made many contributions to Office since the inclusion of the grammar checker in Word 97. "Innovation today is still healthy, despite the downturn in the economy. There is exciting work going on from raw bandwidth availability through wavelength division multiplexing and on the wireless front. And there is tremendous opportunity for new products and services in the software area. Storage is improving at a fast rate with special-purpose processors such as the graphics processor unit and high-bandwidth connectivity, all of which leads to tremendous opportunities for digital media and communications. The computer can help users manage their communications by routing calls to the most convenient device, suggesting use of the most convenient medium to reach someone and screening calls." Future "Technology is still on an amazing growth curve. It is important for companies to invest in longer-term work - say, five or more years out. Also, it is important for the federal government to invest in research activities in the IT market so the work can go on in the leading research universities. A lot of innovation has come from the work at these universities and the funding for a lot of that comes from Defense Advanced Research Projects Agency or National Science Foundation. It continues to be important for the federal government to be involved in that activity. "Microsoft realized long ago that it was important to invest for the long run and establish research activity. That is the philosophy that we continue to hold strongly, and certainly other companies, such as IBM, believe in research strongly." - John Fontana Hossein Eslambolchi
President, AT&T Labs Number of researchers in the organization: Approximately 300
R&D budget: Undisclosed for 2003; about $300 million for 2002.
Number of patents issued in 2002: 300 Past "We used to have a not-invented-here attitude; we would have developed technology ourselves even if we could buy it off the shelf. Our methodology has changed, and we've reversed the order. Today, our first priority is to reuse what we have, second is to buy off-the-shelf products, and third is to develop in-house. I've built a database of all our software capabilities that we search before any new development gets under way. This change was driven primarily by the need to reduce cycle time." Present "There are five major thrusts around corporate research today: Get your research to extract value for customers; improve the enterprise customer experience; automate all processes and lower unit costs; introduce new capabilities and feature functionality; and scale and grow the network. These are predominant factors for success. If corporate research is focusing on a technology that can be useful in 15 to 20 years, that's good, but I don't believe it's affordable on a corporate level or that it meets the needs of customers today. "About 80% of our investments before 2002 focused on fundamental or academic research. Think about quantum physics or delivering next-generation TV via Gigabit Ethernet. These are good programs that have to get done, but are of limited value to AT&T. I reshaped the focus at AT&T to put 80% of investments into direct research, which focuses on this rule of five. The other 20% is academic research, with more than a three-year horizon. "As a result, we've generated more patents this past year, and 40% to 50% of those patents are being implemented into AT&T's infrastructure. In the past, 90% of patents developed in a year were based on academic research. Somebody might pull one of these patents off the shelf in the next 17 years. "Our four strategic thrusts for direct research are: the simplification of our processes and network to eliminate legacy systems (one operating, ordering and billing system); driving to zero defects and zero cycle time; predictive networks, so if a customer hits a certain capacity threshold, additional bandwidth automatically is provisioned; building an infrastructure that would attack a network defect before a customer is affected." Future "We'll see a lot of innovation at the edges of the network. We'll see multiservice, multinetwork access. This is critical for us. Also, we'll see the building of an intelligent optical core with IP - via Multi-protocol Label Switching - and better network management tools, which are lacking when you get to these large complex infrastructures. We need a much better, much more robust network that includes advanced security. "Also expect to see innovation in the area of services over IP - not just voice over IP, but video, instant messaging, chat and data over IP - and network portability, which [would] let users access data from anywhere, at any time. This ties into storage innovation. Next-generation storage systems will be the success of global infrastructures driven by data." - Denise Pappalardo
Mario Mazzola
Chief development officer, Cisco Number of researchers in the organization: More than 860
R&D budget: $3.3 billion for 2003
Number of patents issued in 2002: Unspecified number issued, but Cisco says it files approximately 650 patent applications yearly and it has more than 2,300 pending applications. Past "A few years ago, the R&D path was more defined. We had the opportunity not only to combine and repackage technology in a more convenient way, but also to introduce functionality that could create a great deal of value for our customers. There was a clear path involving the development of our hubs and concentrators to switches to provide more bandwidth, dedicated bandwidth and superior routing capability. This technology shift provided many new characteristics - much higher speeds, much higher density and much lower costs." Present "We are looking at a new innovation wave, which is more fundamental in some respects than previously. Some interesting aspects have to do with support for different types of media and traffic, which has implications in terms of universally supporting quality of service and multicast, for instance. Other aspects have to do with blending security, embedding firewall, VPN capabilities, strong intrusion detection and antivirus prevention. The intention is making it easier to integrate different applications, and to that extent, network intelligence is increasing. Businesses will transform themselves and their information with these intelligent networks. "The most relevant investments for customers are those taking advantage of the benefits relating to Moore's Law - meaning the rapid evolution of the technology - in a way that also ensures preservation of customer investments. The difference now is a move from point products and a move toward global systems and solutions. It's an exciting time." Future "We continuously look at new and disruptive technologies, but keeping in mind the important translation into capabilities and characteristics that are relevant to the customer. We focus a great deal now on how to increase productivity - how to make not only the capital equipment, but also the operational cost, more convenient for all of our customers. "We have defined several areas as new growth opportunity. A major one is in IP telephony, which we started to make investments in a few years ago, and storage and security. We have technologies here and there on the very edge of innovation that don't take the approach of doing basic research like in a Bell Labs. But even though I'm an engineer at heart, I don't believe in force-fitting technology for the sake of technology. I'm more interested in understanding the ways we can bring as much real value as possible to customers." - Phil Hochmuth Richard Lampman
Senior vice president of research, HP Labs Number of researchers in the organization: 750
R&D budget: Undisclosed
Number of patents issued in 2002: 1,385, including those obtained from work done at Compaq. Past "In the past, many corporate research labs had indirect visions. They pursued technology work without understanding what the result would be. When competitive pressure increased, people realized that labs had to be a real contributor to the business. This forced many research labs into painful transitions in the past 10 to 12 years. "At HP, there have been three disruptive technologies. The first was ink-jet printing, taken for granted now. It started here and was an uphill struggle for awhile. But ultimately, business people got excited about the technology, and it created this huge success for us. The second was the program that got HP into Reduced Instruction Set Computing processors - the discontinuity that moved HP into the first tier as a computer company. That technology was controversial, too, because we had many existing products that had to change. The third is the Itanium processor technology. It is emerging from the 'Is it going to happen?' stage to 'It is starting to happen' phase." Present "Itanium, which has the potential to shake the whole computer industry, is a strategy we've pursued for the last couple of years. When you look at the potential for building chips with hundreds of millions or billions of transistors on it, the best way to use that silicon shifts. That's a lot of what drove our early theory about what ultimately became Itanium. "We always have existed to look beyond the current product horizon, to develop technologies that give HP competitive benefit and give our customers an advantage. Chip technologies, such as silicon-on-insulator or copper metallurgy of a couple of years ago, had pretty long development cycles. There's still a lot of variation, but across the board, development cycles have gotten tighter. We have a range of things from some with short cycles to some that are quite long. Architectural issues, such as a new CPU, are in the middle, and software has the shortest cycle. All continue to accelerate in terms of the velocity at which we work." Future "Look at the use of virtual resources in the data center, a hot topic. That has been a rapid development cycle for us, and we have an extremely strong position in it as the result of focused work over a few years and in the labs over a short cycle. You can win in either [development] game and build value. It's about getting the right tool for the job. Even on our longest-range programs, those people have a clear vision of 'If we can do this, it would create an opportunity for HP.' It is motivational for people to imagine how their work could be used." - Deni Connor Alfred Spector
Vice president of services and software, IBM Research Number of researchers in the organization: 3,000
R&D budget: $5 billion for 2003
Number of patents issued in 2002: 3,288 Past "In this industry, there was less emphasis on commercial value in the past. People have done interesting things, but now we would like the needs of business and society to drive those interesting things. During the bubble, capital in start-ups was so available that the economic system might not have been efficient at allocating society's resources to the most important problems. Society is probably somewhat more efficient today and more conservative in capital allocation, which means the most important projects are the most likely to get the necessary funding. That's a good thing." Present "Our research goals have shifted toward service-related software. Now our research lab has a clearer focus on work that's driven by real problems. The work might be equally innovative, but it's driven by problems that are valuable to our society. "We're doing a lot of work in natural language, focused on conversational voice systems for call centers, translation of Web sites into multiple languages and categorization of documents so you can find them quickly. Those are great issues that are every bit as interesting as some of the early artificial intelligence work in natural-language processing that wasn't focused on business problems. "Our goal these days is to do 'factorizable' research. We want things that have long-term goals, but we want to achieve intermediate benefits, or factors, along the way. One example of factorizable research is our work in superhuman speech recognition. Our goal is to build systems that recognize human speech as well as humans do, or even better. Our idea of the way to get there is to make an improvement in quality every year. That has commercial benefits. With compounded benefits, it can deliver a really incredible result. It's marketable as you go, but the result is something that makes you say, 'Gee whiz.' "An example of something nearer term is our work around optimization in our on-demand innovation services unit within the research division. We've created a unit of researchers that spend half their time on traditional research and half their time working in the field alongside IBM business consultants. They find important projects that need the innovation a Ph.D. or a researcher can provide." Future "We're going to have considerably more direct interaction with our customers through technologies like on-demand innovation services. We see that as profoundly valuable - connecting the minds of our thousands of research division employees with the problems that our customers have to catalyze rapid innovation. So many of the interesting problems associated with networking and IT are now not about the technology but about industry problems that the technology can solve. "One area that I know is important is secure, pervasive networking. We like our wireless services, but we want them seamless and secure, and with enough privacy that we can trust our systems to have significant amounts of information about consumers and businesses. There's a lot of good work to be done there. "Research is going through a transition as it grows into more industry domains. In biotechnology, computer technology is exceedingly important as it relates to genomics and custom drug treatment, for example. All sorts of things will need to be done in areas of business process integration and self-managed systems. That might be where the great innovation is. And you want to be working alongside customers in other domains, not just with other computer scientists, to do that work. "We think businesses will procure services essentially through transactions across a network using technologies like Web services. To make that work properly requires autonomic computing and service-level agreement management. It requires strong security and high availability. There will be significant growth in these fundamental networking areas as customers realize the huge cost advantages they can get by performing some activities on utilities across a network." - Ann Bednarz Sanjay Subhedar
General partner and co-founder, Storm Ventures R&D spending: Oversees a $310 million fund focused on seed and early-stage communications infrastructure companies.
Investments include: Airespace (wireless LAN infrastructure); Ample Communications (high-bandwidth optical connectivity systems); and Fourelle Systems (Web acceleration for WANs, including cable, frame relay, satellite and wireless). Past "In the old way of doing things, customers felt comfortable buying end-to-end solutions from one company. When you bought a new IBM computer, IBM architected the hardware, the operating system, the applications and everything around the computer. Today, this is disaggregated. Computer manufacturers put pieces together and offer distribution channels. "For this reason, centralized corporate R&D labs don't make sense. There are too many technologies to track and improve. Today, in anything you do, you have to share with other companies. You can't be proprietary. "Initially at Storm Ventures, I felt that the demand for bandwidth would continue unabated. What I learned, after about a year, was that at the core, there is in fact not a bandwidth problem, that we'd overbuilt capacity. And we had not solved problems in the access and aggregation layers. "We had the rise of the competitive local exchange carriers. Then the CLECs collapsed, and in that collapse we killed broadband access. We've fallen behind countries such as Korea and Japan in terms of broadband access. "At the aggregation layer, there continues to be an alphabet soup. Should it be ATM or IP or [Multi-protocol Label Switching]? What do we need in quality of service, and how do we achieve it? The market was trying to build what we called 'God boxes' -'I'll put everything into this product because I don't know what customers will bet on.' But when you don't target products to particular, known problems, you make the solutions very expensive." Present "Part of innovation is not just showing that something works in a lab, but being able to produce hundreds of thousands, even millions, of products at an affordable cost. "The old R&D cycles have compressed significantly. We have disaggregated many different elements. I can get the same advantages of scale and speed that Cisco has by going to Solectron, a $12 billion vendor of manufacturing services, including new product design, manufacturing and product warranty support. "The process of converting ideas into products in six or eight or more years has compressed to 18 to 24 months. And this same process is pushing you to create global products. It's not as expensive anymore to have a global presence. This is all positive, because it lets me solve the problem of 'How can I get this technology in many people's hands quickly?' It can put a DVD or digital camera into the hands of 50 million people, and that no longer takes years to do. But this positive process also places a huge burden on a company's ability to execute its business plan. You have to be right in every decision, and you have to be No. 1, 2 or 3 from the start or you're irrelevant." Future "Venture capitalists have to recognize this compressed process. We need people who can identify solutions that are global, who can scale those solutions, and who can use outsourced facilities and create virtual organizations to make all this work. "Instead of centralized R&D models, I would say a better model is working with academia. Academics tend to collaborate rather than compete. And collaboration is a huge benefit when you're doing research. And academics are oriented toward sharing information, so research tends to be less costly. "Also, there needs to be a fairly good dialog between users and customers on the one hand, and the technology researchers and suppliers on the other. We tell our companies to find a teaching customer, one who can tell you that if you're trying to make a better product or a new one, and then you have to meet this specific price, and interoperate with these existing systems and so on. Otherwise, you're innovating with no goal in sight." - John Cox Dan Ling
Vice president, Microsoft Research Number of researchers in the organization: 700 across five labs
R&D budget: $4.8 billion for 2003
Number of patents issued in 2002: 511 Past "In the past, research labs tended to be in remote locations from the parent company. The classical example is Xerox Park, located in California when most of Xerox's operations were in upstate New York. But now more than ever, people are aware that getting technology from the research group into products is a difficult process - we always say that technology transfer is a contact sport. It requires the researchers knowing the product development groups well, both at technical and personal levels. It's important to have personal ties to individuals in the product groups so when technical opportunities arise those connections are in place. It has been fortunate for us that we decided to locate our research lab in the middle of the Microsoft campus." Present "Getting feedback from customers - what the technology is good for, how they are going to use it, what they like and don't like - into the R&D cycle fairly quickly is a positive thing. The research group here gets involved in a lot of company meetings. Obviously, one still needs to take a long-term approach, but along the way you learn a lot by having people using those products. "One can do that with a shortened R&D cycle on top of longer-term research. A good example is our work in the natural-language area. Clearly, for computers to understand any human language is still five to 10 years off. But this is an area where we have made many contributions to Office since the inclusion of the grammar checker in Word 97. "Innovation today is still healthy, despite the downturn in the economy. There is exciting work going on from raw bandwidth availability through wavelength division multiplexing and on the wireless front. And there is tremendous opportunity for new products and services in the software area. Storage is improving at a fast rate with special-purpose processors such as the graphics processor unit and high-bandwidth connectivity, all of which leads to tremendous opportunities for digital media and communications. The computer can help users manage their communications by routing calls to the most convenient device, suggesting use of the most convenient medium to reach someone and screening calls." Future "Technology is still on an amazing growth curve. It is important for companies to invest in longer-term work - say, five or more years out. Also, it is important for the federal government to invest in research activities in the IT market so the work can go on in the leading research universities. A lot of innovation has come from the work at these universities and the funding for a lot of that comes from Defense Advanced Research Projects Agency or National Science Foundation. It continues to be important for the federal government to be involved in that activity. "Microsoft realized long ago that it was important to invest for the long run and establish research activity. That is the philosophy that we continue to hold strongly, and certainly other companies, such as IBM, believe in research strongly." - John Fontana Hossein Eslambolchi
President, AT&T Labs Number of researchers in the organization: Approximately 300
R&D budget: Undisclosed for 2003; about $300 million for 2002.
Number of patents issued in 2002: 300 Past "We used to have a not-invented-here attitude; we would have developed technology ourselves even if we could buy it off the shelf. Our methodology has changed, and we've reversed the order. Today, our first priority is to reuse what we have, second is to buy off-the-shelf products, and third is to develop in-house. I've built a database of all our software capabilities that we search before any new development gets under way. This change was driven primarily by the need to reduce cycle time." Present "There are five major thrusts around corporate research today: Get your research to extract value for customers; improve the enterprise customer experience; automate all processes and lower unit costs; introduce new capabilities and feature functionality; and scale and grow the network. These are predominant factors for success. If corporate research is focusing on a technology that can be useful in 15 to 20 years, that's good, but I don't believe it's affordable on a corporate level or that it meets the needs of customers today. "About 80% of our investments before 2002 focused on fundamental or academic research. Think about quantum physics or delivering next-generation TV via Gigabit Ethernet. These are good programs that have to get done, but are of limited value to AT&T. I reshaped the focus at AT&T to put 80% of investments into direct research, which focuses on this rule of five. The other 20% is academic research, with more than a three-year horizon. "As a result, we've generated more patents this past year, and 40% to 50% of those patents are being implemented into AT&T's infrastructure. In the past, 90% of patents developed in a year were based on academic research. Somebody might pull one of these patents off the shelf in the next 17 years. "Our four strategic thrusts for direct research are: the simplification of our processes and network to eliminate legacy systems (one operating, ordering and billing system); driving to zero defects and zero cycle time; predictive networks, so if a customer hits a certain capacity threshold, additional bandwidth automatically is provisioned; building an infrastructure that would attack a network defect before a customer is affected." Future "We'll see a lot of innovation at the edges of the network. We'll see multiservice, multinetwork access. This is critical for us. Also, we'll see the building of an intelligent optical core with IP - via Multi-protocol Label Switching - and better network management tools, which are lacking when you get to these large complex infrastructures. We need a much better, much more robust network that includes advanced security. "Also expect to see innovation in the area of services over IP - not just voice over IP, but video, instant messaging, chat and data over IP - and network portability, which [would] let users access data from anywhere, at any time. This ties into storage innovation. Next-generation storage systems will be the success of global infrastructures driven by data." - Denise Pappalardo