29-76.the Impact of Academic Research and Industrial Performance

THE NATIONAL ACADEMIES PRESS
This PDF is available at http://nap.edu/10805 SHARE

   
The Impact of Academic Research on Industrial Performance

(2003)
DETAILS
264 pages | 6 x 9 | HARDBACK

ISBN 978-0-309-38522-0 | DOI 10.17226/10805
CONTRIBUTORS
GET THIS BOOK Committee on the Impact of Academic Research on Industrial Performance; National
Academy of Engineering
FIND RELATED TITLES
SUGGESTED CITATION
National Academy of Engineering 2003. The Impact of Academic Research on

Industrial Performance. Washington, DC: The National Academies Press.
https://doi.org/10.17226/10805.

Visit the National Academies Press at NAP.edu and login or register to get:
– Access to free PDF downloads of thousands of scientiﬁc reports

– 10% off the price of print titles
– Email or social media notiﬁcations of new titles related to your interests
– Special offers and discounts

Distribution, posting, or copying of this PDF is strictly prohibited without written permission of the National Academies Press.
(Request Permission) Unless otherwise indicated, all materials in this PDF are copyrighted by the National Academy of Sciences.
Copyright © National Academy of Sciences. All rights reserved.

Report of the Panel on the Network

Systems and Communications Industry
The Panel on Network Systems and Communications, one of five panels

formed by the Committee on the Impact of Academic Research on Industrial
Performance, was asked to examine the impact of academic research on the
performance of the network systems and communications industry and recom-
mend ways based on trends in the industry and the research community to in-
crease this impact. The panel of six included three members of NAE (all from
industry), one other member from industry, and two from academia. Three of the
panel members were also members of the parent committee. The panel reviewed
the literature, developed several case studies, and sent a questionnaire to experts
in academia, the computer industry, the network systems and communications
industry, and government. The questionnaire was followed by a workshop at-
tended by approximately 30 senior individuals in the network systems and com-
munications sector (see Addendum).
The network systems and communications business sector flourished
throughout the 1990s, when the growth of the Internet, the technologies that
implement it, and the businesses and services that use it were unprecedented.
Telecommunications services—especially wireless digital telephones and paging
services—also grew rapidly. Much of this success was attributable to exponential
improvements in the performance-to-cost ratio of microelectronics over the past
three decades. Technical innovations emerging from within the industry and from
academic research have been essential. Some innovations were the culmination
of decades of research; some were short-term developments that entered the
29
Copyright National Academy of Sciences. All rights reserved.

30 THE IMPACT OF ACADEMIC RESEARCH ON INDUSTRIAL PERFORMANCE
market via start-up companies; and some were incremental improvements to

existing products.
In the last 30 years, digital technologies have transformed the telephone
network from an analog system to a computer-controlled system with digital
switching and transmission. The process of digitalization has changed the indus-
try from two distinct businesses—computers and communications—to one busi-
ness in which computers and communications are intermingled in products and
services. This convergence was accelerated by advances in microelectronics and
increases in the bandwidth available for communications (Messerschmitt, 1996).
The result is increasingly pervasive data networking, based largely on the packet-
switching technologies that emerged from academic and industrial research to
spawn the Internet.
The network systems and communications industry has a large and expand-
ing services component. For the telecommunications industry, which has always
been a service provider, the challenge is to invent and offer customers new,
valuable services that generate new sources of revenue. For the computer portion
of the industry, high-performance communications are making a wide range of
new services feasible. Examples include remote sensors and control systems;
integrated supply chain management systems; application service providers; full-
time, real-time stock quotes; and instant messaging.
DEFINITION OF THE INDUSTRY

The network systems and communications industry must be defined very
broadly. It clearly includes the manufacturing of telecommunications equipment
and the services that use such equipment, such as telephony, wireless telephony,
broadcast television, cable and satellite television, radio, and Internet service.
Both the equipment and services sectors increasingly require computing equip-
ment and software, and, in fact, the computer and communications industries are
no longer separate industries. For example, cellular telephony depends on a broad
range of technologies: the cell phone contains a liquid crystal display, an embed-
ded computer with a lot of software, and advanced chips that integrate most of the
components of a high-frequency radio; the transmission formats depend on ad-
vances in digital speech compression, signal modulation, and coding; the base
stations depend heavily on digital integrated circuits and computers for switching
and control and fiber-optic links between them; tracking a moving telephone
requires that computers at adjacent base stations exchange protocol messages for
the handoff; and the billing, provisioning, and maintenance of the service require
large-scale computing and software systems of the service provider. Separating
this integrated system into “communications” and “computing” components is
simply not possible. In short, computing and communications equipment and
services have converged, creating new business and technical opportunities.

NETWORK SYSTEMS AND COMMUNICATIONS INDUSTRY 31
The explosive growth of the Internet is the most visible manifestation of

this trend toward convergence. The technologies underlying the Internet—
just like those that underlie the cellular telephone—include computing and
communications. Special computers serve as routers, and network services
knit together the transmission links and implement the collection of Internet
protocols that carry Internet traffic. The explosive growth of the Internet,
however, is attributable not to these basic provisions—which existed before
1993—but to new services that created consumer demand: electronic mail,
the World Wide Web of information and its associated browser software; chat
groups; real-time delivery of audio and video media; online merchandising;
banking and financial transactions; supply-chain integration of suppliers and
customers; and numerous other applications. Some applications merely ex-
tend existing internal information technology systems to provide Internet
access. But others, such as eBay’s success with online auctions, are entirely
new business concepts. As the Internet becomes more pervasive, old ways of
computing, in which data was created, stored, and manipulated at a single
site, are giving way to networked systems in which data can be accessed
remotely and shared extensively.
The computers embedded in everyday objects—telephones, cars, televisions,
furnaces, hi-fi equipment—are becoming increasingly capable and increasingly
networked. Some cars already can connect with a diagnostic and help center by
cellular telephone or satellite communications. Home networks in which multiple
personal computers in a household are linked over existing telephone wires and
short-range wireless devices will soon make networking of appliances routine. A
world in which all devices have an Internet address is not out of reach. Thus
computers increasingly require communications to fulfill their functions, and
communications increasingly require computers to fulfill theirs.
The technologies of computing and communications are becoming indis-
tinguishable. All of them depend on software to express functions at all levels
in the network. A few years ago, a modem was a complex, integrated circuit.
Today, with more complex algorithms and faster computers, modems are writ-
ten in software embedded within digital signal processors. Many algorithms
can be used equally well in computing and communications settings. For ex-
ample, schemes to digitize and compress video signals are useful both for
manipulating and storing video information on a computer disk and for trans-
mitting it over digital communications channels. Similarly, encryption technol-
ogy can be used to protect sensitive information in a computer system or in
transit over a network.
These three trends—convergence, embedding, and network applications—
characterize the network systems and communications sector. The panel’s assess-
ment of the contributions of academic research to this industry is based on this
broad definition.

TABLE 2-1 Sales and Employment in the Information Technology Industry, 2000
Sales Number
NAICS a Revenues of Jobs
Code ($ billions) (1,000)
IT Manufacturing
Computer and peripheral equipment 3341 $110.0 190
Communications equipment 3342 119.3 291
Software 5112 88.6 331
Semiconductors and
other electronic components 3344 168.5 621
IT Services
Data processing services 5142 42.9 296
Telecommunications services 5133 354.2 1,165
aNorth American Industrial Classification System. Source: U.S. Bureau of the Census, 2002.
Size
Because our definition has vague boundaries and because the industrial
classifications used to gather statistics have not been adapted to the rapid changes
in the industry, it is difficult to determine the size of the network systems and
communications sector. Table 2-1 summarizes sales and employment in the
information technology industry based on Bureau of the Census data (U.S. Bu-
reau of the Census, 2002). Taken together, sales of computer and communica-
tions equipment and services (all information technology minus semiconduc-
tors) were about $715 billion in 2000, and the industry employed more than
2.2 million people (U.S. Bureau of the Census, 2002). Expenditures for
information-processing equipment increased almost 10 percent per year on aver-
age from 1970 to 1994; the corresponding figure for computers and peripherals
was 27.5 percent (NRC, 1999). A 1999 survey found that telecommunications
manufacturing was growing by 16.3 percent annually, computer software by
16.6 percent, and computer hardware by 9.5 percent (CTIS, 1999), however
these rates have dropped significantly since early 2000.
Structure
The role of research and innovation in the network systems and communications
sector can best be understood in the context of the structure of the industry, which
influences the mechanisms of innovation and thus how new technologies and prod-
ucts are introduced. The very general description that follows is intended only to
reveal similarities and differences with the other industries studied in this report.
Manufacturing
The structure of the computer industry is horizontal; the communications
industry was vertically integrated but has been rapidly changing to a horizontal

structure as well. In a horizontal structure, numerous suppliers manufacture parts

and components that many integrators assemble into subassemblies that are then
assembled into final products by numerous competing original equipment manu-
facturers. The multiplicity of companies at each manufacturing step ensures in-
tense competition throughout the production process, not only in terms of price
but also on a wide rage of performance characteristics. For example, manufac-
turers of personal computers buy disk drives from any one of about a dozen
suppliers. A company that needs a customized integrated circuit may design the
circuit but use one of several competing semiconductor foundries to manufacture
it. Specialized circuit board assembly firms can assemble and test complete cir-
cuit boards, giving an electronic design firm the ability to design and sell a unique
computer interface board with custom chips without having to invest in either
chip or board manufacturing facilities.
The divestiture of AT&T and the subsequent deregulation of communica-
tions services forced the communications industry to change from a vertical to a
horizontal structure. Today there are many vendors of telecommunications equip-
ment and components. Custom integrated circuits can implement very complex
communications functions; coupled with custom-built and proprietary software
designs, equipment vendors compete intensely in terms of technology, reliability,
and cost of ownership.
Another important feature of the network systems and communications sec-
tor is its reliance on components with well defined interfaces. Integrated circuits
are a good example: the physical, electrical, and logical behaviors of chip inter-
faces are specified by the manufacturer and used by the customer to determine
how to incorporate a chip into a subsystem with other components. Subsystems
then become components of still larger systems. Software, another component,
plugs into the operating system that supports it by linking the software interfaces
(sometimes called application programming interfaces, or APIs). A piece of soft-
ware that is compatible with a certain operating system adheres to the interfaces
provided by the operating system. Computer systems are built from complex
hierarchical assemblies of subsystems and components, sometimes hundreds or
even thousands of them. Some of the components are custom built, and some are
standard. Thus, interfaces give rise to components, which in turn give rise to
businesses structured around buying and selling components.
Key component interfaces become industry standards, which are usually
adopted by industry groups to hasten the spread of a new technology, increase
sales volume, and, therefore, decrease cost. Standards maintained by a group with
broad representation from competitive suppliers are said to be open standards.
For example, the Personal Computer Memory Card International Association is
an industry group that establishes standards for interchangeable interface cards
for laptop computers. The standards group includes several producers of cards
and several producers of laptops to ensure that the standard cannot be manipu-
lated to benefit one competitor over another. By contrast, standards promulgated

by a single vendor are said to be proprietary. For example, the programming

interface for Microsoft’s Windows operating system is proprietary; Microsoft
specifies it and can change it at will.
Standards play a special role in communications. Broadly speaking, they are nec-
essary to ensure that components and subsystems connected via a communications
channel can operate together (e.g., they obey the same conventions for encoding voice
signals, multiplexing many simultaneous phone calls on a single channel, performing
operation and maintenance functions). Standards of this kind are necessary for guaran-
teed, sustained interoperability, and changes must be carefully designed to avoid even
slight interruptions of network service. New versions of standards must be designed so
they can be introduced incrementally, connect new equipment to old, test new proto-
cols, and so on. The same considerations apply to Internet protocols.
Services
Communication services (e.g., voice and data transmission, switching, and
distribution) are a major portion of the network systems and communications in-
dustry. The number and structure of telecommunications service providers have
been in constant flux since the divestiture of AT&T and the deregulation of local
telephone services. First, new companies emerged offering wireless telephone ser-
vices. Then another group of new companies emerged as Internet service providers.
To increase their revenue, carriers have been developing value-added services,
such as voice mail, call forwarding, call waiting, 800 service, electronic mail, and
virtual private networking, along with conventional transmission and switching
services. Internet service providers provide national and regional portals that offer
news, chat rooms, advertising, and direct access to the World Wide Web.
Computing services are also a major element of the industry. System integration,
the design and deployment of communications and information systems for large
clients, has become a major source of revenue for many equipment vendors. In recent
years, an important service has been to implement network capabilities across compa-
nies’ existing computer systems. In some cases, networking has focused on providing
Internet access for employees and customers; in others, the focus has been on the
development of internal networks linking production and distribution facilities across
the company. So far, neither academic nor industrial research has addressed the
problems of service delivery in a structured and sustained manner.
INNOVATION SYSTEM
Most innovations are incremental improvements, such as design refinements,
improvements in technology and manufacturing processes, a better understanding of
customer needs, and integration of previously separate products. For example, impor-
tant performance metrics for communications equipment include low power and high
density (so that many circuits can be accommodated in the confined spaces of wiring

closets, boxes mounted on telephone poles, and even central switching offices). Both
power and density can be improved by advances in integrated-circuit technologies,
which in turn, derive primarily from incremental improvements in fabrication equip-
ment, processing steps, and materials. Research results may be the basis for some of
these improvements, and research has achieved major breakthroughs in these areas;
this research is performed or funded by materials, equipment, and microchip fabrica-
tors, not by the telecommunications equipment manufacturers (see Box 2-1).
BOX 2-1
The Cellular Telephone
The rapid spread of cellular telephones probably epitomizes the popular con-
ception of innovation. The use of cellular telephony started out slowly but then
exploded. There were 11 million subscribers in 1992 and 141 million in 2001. The
look, feel, and weight of cell phones have also evolved rapidly.
Although the original cellular concept was developed at AT&T Bell Laborato-
ries in 1946, cellular technology was not the outgrowth of fundamental research on
radio-frequency propagation and control. Rather, it was the result of demand-
driven technological improvements developed by corporate researchers, primarily
at Bell Laboratories and Motorola. The long delay from the initial idea to deploy-
ment was principally the result of regulatory and business decisions that de-
emphasized the development of cellular technologies. Only a few systems, such
as improved mobile telephone service, were deployed at all (in 1964). Because of
the delay, developers benefited from the microelectronics revolution and were able
to use inexpensive microprocessors and integrated circuits to make equipment
cheaper, lighter, and less power hungry.
Until recently, cellular telephony in the United States was dominated by the
advanced mobile phone service (AMPS) analog standard developed in the early
1970s. The switch to digital transmission occurred more rapidly in Europe, with
the GSM (global system for mobile communication) standard, which uses time-
division multiple access. A competing digital scheme that relies on spread-
spectrum technology (CDMA—code division multiple access) was developed as a
proprietary standard by Qualcomm. In March 1999, the firms competing over
CDMA intellectual property and products agreed to support a single worldwide
standard to promote widespread adoption. Personal communications systems
services in the United States use these techniques in a new, larger frequency
band. Many cellular telephones contain electronics that will work with more than
one of these standards and thus can operate in many areas around the world.
Academic research has played only a small role in the development of these
innovations. Until the telephone industry was deregulated, using public funds to
compete with Bell Laboratories research was considered pointless. Moreover,
Bell Laboratories did not fund academic research in this area; instead they re-
cruited graduates of broad science and engineering programs and trained them in
specialized research areas on the job. After the breakup of the Bell system, new
firms entered the telecommunications arena, demand for engineers trained in cel-
lular technologies grew, and academia began to respond. Today, considerable
academic research is being done in these technologies.
Source: CTIA, 2003; Qualcomm Corporation, 1999; Roessner et al., 1998.

For many businesses, vendors of materials, products, and services throughout

the supply chain are major sources of innovation. Buying an integrated-circuit chip,
for example, implicitly buys a share of the dramatic improvements in price and
performance of integrated circuits (Moore’s Law).1 Over time, innovations will
make the chip faster or cheaper or more capable. A telecommunications carrier that
wants to deploy Synchronous Optical Network (SONET), a transmission protocol
that defines optical carrier levels and their electrically equivalent synchronous
transport signals, can purchase switches, multiplexers, and test equipment from the
vendors who developed SONET technology. This pattern is a direct consequence
of the “horizontal” structure of the industry. Dell Computer, for example, does not
have in-house R&D; in effect, Dell is a broker that negotiates attractive deals to buy
components and computer-assembly services for its build-a-computer-to-order busi-
ness. Dell depends on R&D investments by its vendors, especially Intel and
Microsoft, that make the microprocessors and operating system software on which
the personal computer business depends. Dell’s innovations have been in its busi-
ness model and supply-chain management, not in its technology.
Innovation can also be purchased by acquiring other companies, especially
venture-capital-backed start-up companies that have introduced new products
with new technologies. A start-up company is a new business, often with an
innovative technology but with considerable risk. Often the innovative technol-
ogy has its origins in academic research. If the company makes good progress,
both in technology and in the market (e.g., beta testing, or success in getting its
approach adopted by standards consortia), it becomes an attractive target for a
larger company seeking to strengthen its technology or product line. For ex-
ample, Texas Instruments bought Amati; Fore Systems bought Berkeley Net-
works and Marconi bought Fore Systems; Cisco bought Granite Systems; and
Broadcom bought Epigram. Each of the acquired companies had ventured into a
new technical area. Epigram, for example, had devised a way to use home tele-
phone wiring to transmit 10Mb Ethernet traffic and had made progress in stan-
dardizing the scheme through the Home Phoneline Networking Alliance.
Broadcom, itself an innovative fabless chip company specializing in integrating
analog and digital functions of cable and twisted-wire modems, saw buying
Epigram as a natural way to enhance its core business.
Although high-tech start-ups seldom do research in the classic sense, many
behave much like “applied research” projects in an industrial laboratory. They
formulate technically aggressive plans based on established principles to pursue
and evaluate; the results of experiments often inform several products. For ex-
ample, Transwitch attempted to increase the telecommunications protocol process-
ing integrated on a single chip, as well as to partition the chip functions into an
“architecture” so that a small number of chip designs could be used to build a wide
variety of telecommunications products. Both Amati and Epigram conceived ways
of using advanced signal-processing techniques to adapt digital transmission to the
characteristics of real-world, twisted-pair copper wires (Amati) or in-house

telephone wiring (Epigram). The technology-development activities of these com-

panies are much like those in industrial research, but they are done in a commercial
setting and with strong incentives to bring innovations to market rapidly.
Industrial research is concentrated in the laboratories of a few of the largest
companies, such as Intel, Microsoft, IBM, Compaq, Lucent, AT&T, Hewlett-
Packard, Sun Microsystems, and Xerox. Although many of these firms invest 10 to
15 percent or more of revenues in R&D each year, the vast majority of this is for
“development,” that is, for the engineering of the next generation of products.
Research focused on objectives more than 18 months or one or two product cycles
out is estimated to be, at most, 5 percent of that 10 to 15 percent, or far less than
1 percent of revenue.2 A few large companies eschew research, preferring instead
to buy innovative companies (e.g., Cisco). Companies in the services sector, how-
ever, generally do not engage in or support research. For example, at MCI, which is
generally considered a technology leader, the advanced technology group is primar-
ily concerned with testing new equipment and working with vendors to solve
interoperability and operation, administration, and management problems.
Industry research is usually driven by market needs but often includes some
fundamental or long-range projects as well. For example, IBM’s research on the
Internet and electronic commerce includes some long-term work on cryptographic
systems for security and authentication. Industrial research often links advanced
technologies to emerging product needs. For example, as the Java programming
language became popular, industry laboratories at Sun, IBM, and elsewhere
launched projects to devise advanced techniques for the compilation, synchroni-
zation, and code simplification required for its implementation. Previous research
results in these areas had not adequately addressed the needs of the Java lan-
guage, of today’s large memories, or of multiprocessor servers. Some of this
research is fundamental in the sense that it can be applied to problems other than
Java language implementations. In fact, even though research in engineering
fields is usually targeted toward meeting specific engineering needs, the results
are often useful for many other applications.
One of the companies’ aims in operating research laboratories is to expand
their capability for bringing in new ideas and new people (Cohen and Levinthal,
1990). The laboratory is expected to recruit people who cannot be recruited by an
engineering organization; it is also expected to interact with the intellectual
community by attending conferences, publishing papers, collaborating with uni-
versities, or entering partnerships with other companies; and it is intended to
counteract the risk inherent in the narrow focus of engineering projects on prod-
uct development.
A Culture of Innovation
Innovation in the network systems and communications industry can take
many paths. Even when research plays an essential role, there is no linear path

1965 1970 1975 1980 1985 1990 1995 2005

Timesharing
CTSS, Multics / BSD
Unix
SDS 940, 360/67, VMS
Client/server computing
Berkeley, CMU, CERN
PARC, DEC, IBM
Novell, EMC, Sun, Oracle
Graphics
Sketchpad, Utah
GM/IBM, Xerox, Microsoft
E&S, SGI, ATI, Adobe
Entertainment
Spacewar (MIT), Trek (Rochester)

Atari, Nintendo, SGI, Pixar
Internet
ARPANET, Aloha, Internet
Pup
DECnet, TCP/IP
LANs
Rings, Hubnet
Ethernet, Datakit, Autonet
LANs, switched Ethernet
Workstations
Lisp machine, Stanford
Xerox Alto
Xerox Star, Apollo, Sun
Graphical user interfaces
Engelbart / Rochester
Alto, Smalltalk
Star, Mac, Microsoft
VLSI design
Berkeley, Caltech, MOSIS
many
RISC processors
Berkeley, Stanford
IBM 801
SUN, SGI, IBM, HP
to World Wide Web
1965 1970 1975 1980 1985 1990 1995 2005

University Industry R&D Products $1 bil. market
FIGURE 2-1 Examples of academic government-sponsored (and some industry-

sponsored) IT research and development in the creation of commercial products and
industries. Source: NRC, 2003.

1965 1970 1975 1980 1985 1990 1995 2005

from Internet
Relational databases
Berkeley, Wisconsin
IBM
Oracle, IBM, Sybase
Parallel databases
Tokyo, Wisconsin, UCLA
IBM, ICL
ICL, Teradata, Tandem
Data mining
Wisconsin, Stanford
IBM, Arbor
IRI, Arbor, Plato
Parallel computing
Illiac 4, CMU, Caltech, HPC
IBM, Intel
CM-5, Teradata, Cray T3D
RAID /disk servers
Berkeley
Striping/Datamesh, Petal
many
Portable communication
Berkeley, Purdue (CDMA)
Linkabit, Hughes
Qualcomm
World Wide Web
CERN, Illinois (Mosaic)
Alta Vista
Netscape, Yahoo, Google
Speech recognition
CMU, SRI, MIT
Bell, IBM, Dragon
Dragon, IBM
Broadbandl in last mile
Stanford, UCLA
Bellcore (Telcordia)
Amati, Alcatel, Broadcom
1965 1970 1975 1980 1985 1990 1995 2005

University Industry R&D Products $1 bil. market

from a research result to advanced development to product development to eco-

nomic return. Ideas and people tend to bounce around; new ideas can be stymied by
political or business impediments and forced to take alternative routes. The indus-
try does not have a few mechanisms for creating and exploiting innovations. In-
stead, it enjoys what has been called a “national research culture” that fosters
innovation (Lazowska, 1998). Some of the features of that culture are de-
scribed below.
A 1995 report of the National Research Council Computer Science and
Telecommunications Board documents the effect of the research culture in the
computer and high-performance computing arenas (NRC, 1995). A more recent
report (NRC, 2003), documents how technologies are born (often in academia),
are taken up and extended by other academic or industrial groups, become the
seeds of start-up companies or new products in larger companies, as well as how
the market for the technology grows and matures (Figure 2-1). All paths to
market are erratic, and often take 15 years. The diversity in the academic and
industrial sectors lends robustness to the process: a good idea is very hard to
completely eradicate. Among the findings of that report are:
• Research has kept paying off over a long period.

• The payoff from research takes time. At least 10 years, more often 15,
elapse between initial research on a major new idea and commercial
success. This is still true in spite of today’s shorter product cycles.
• Unexpected results are often the most important. Electronic mail and the
“windows” interface are only two examples.
• Research stimulates communication and interaction. Ideas flow back and
forth between research programs and development efforts and between
academia and industry.
• Research trains people, who start companies or form a pool of trained
personnel that existing companies can draw on to enter new mar-
kets quickly.
• Doing research involves taking risks. Not all public research programs
have succeeded or led to clear outcomes even after many years. But the
record of accomplishments suggests that government investment in com-
puting and communications research has been very productive.
Mobility of Ideas
As Figure 2-1 suggests, the number and types of research structures among
universities and industry provide a kind of redundancy; an idea that cannot ad-
vance in one environment may flourish in another. As an example, the path of
reduced instruction set computing (RISC) started out with John Cocke’s IBM
801, developed at IBM Research. Although the ideas were countercultural and

did not have a great impact at IBM, they spawned two research projects, one at
Berkeley and one at Stanford, to explore them further. Both university projects
resulted in prototype processor designs. The Stanford project formed the nucleus
of a start-up company—MIPS Computer—to build RISC microprocessors. The
Berkeley project led to an advanced development project at Sun Microsystems to
develop its SPARC instruction set. Both led to commercially successful products.
Moreover, publication of the work in professional journals rapidly spread aware-
ness of the technology. A related example occurred in the evolution of relational
databases. A researcher at IBM, Ted Codd, developed an idea that found little
encouragement at IBM, whose products at the time used a competing database
technology. Nevertheless, Codd was able to seed academic work at Berkeley that
enlarged the interested community and eventually led to two start-up companies,
Ingres and Oracle, and a huge industry (NRC, 1999). Both processor and data-
base technologies were later embraced by IBM.
Table 2-2 shows one way that ideas can move between academia and
industry. A key idea of this sketch is the “democratization” phase, in which a
tiny research community is deliberately enlarged into a community with a
critical mass of researchers exchanging ideas, building prototypes, and teach-
ing others. This step was clearly discernible in the histories of both RISC and
relational databases, and was, perhaps, the key step in the spread of very large-
scale integration design techniques that Carver Mead and Lynn Conway devel-
oped in the late 1970s. Democratization in that case involved writing a text-
book, teaching teachers, and starting courses in a half-dozen graduate
departments to spread the ideas. These efforts resulted in a self-sustaining
research community that built computer-aided design software, built a short-
run chip fabrication system (MOSIS), designed a number of novel chip archi-
tectures, and trained hundreds of students in the art of integrated-circuit design
and computer architecture.
TABLE 2-2 How Ideas Can Move between Academia and Industry
University Industry/Government
1. Theoretical result 2. Student graduates to industry laboratory

that encourages individual researchers
and builds a basic prototype.
3. Democratization phase, in which many

people work on the idea (e.g., RISC) 4. Advanced development leads to a
commercially successful product in a
small but significant market.
5. Academics study the details and fill in 6. Market explodes. Industrial research
the gaps [lots of difficult research here] advances technologies.
Source: Tennenhouse, 1998.

Mobility of People
People often move on to new challenges, sometimes taking with them inno-
vative ideas. The industry asserts, “technology transfer is a contact sport,” that
ideas transfer best when people carry them. Universities are, of course, a primary
source of people, graduate and undergraduate students, many of whom have had
research experience. University graduates with research experience are very valu-
able to industry, not only as staff for its research functions, but also as technical
leaders in product development organizations. Curious people with broad techni-
cal knowledge who are trained in solving technical problems through research are
extremely valuable in today’s product engineering groups.
People with ideas often feel impelled to find a receptive environment, and in
the 1990s start-up companies were a powerful attraction. University graduates
sometimes embarked immediately on a start-up company, based on ideas formu-
lated or prototyped while they were students. Faculty members often took leave
to start companies or to consult with companies that embraced their new ideas. In
fact, departing faculty members and students left some academic computer sci-
ence departments with large gaps in their curricula and research programs, espe-
cially systems and networks programs (Morris, 1998).3 The lures are not only
financial. Many students who would otherwise go to graduate school complain
that academic research is sterile and irrelevant; they prefer to actually do engi-
neering, to build a product that will change the world.
The mobility of people includes flow from industry to academia. For in-
stance, in a 2001 survey, many computer science programs reported record num-
bers of applicants for doctoral programs, attributable to the demise of many
Internet start-ups (Bryant and Vardi, 2002). Anecdotal evidence also suggests
that many industry researchers have found places in universities as industrial
research spending has declined at telecommunications and computer hardware
companies. Their knowledge and industrial experience can provide valuable in-
sight for academic research.
Open Structures
The structure of the network systems and communications industry promotes
certain kinds of innovation. When components have well defined interfaces,
innovators can offer improved components with the same interfaces, so they
remain compatible with their predecessors. Thus, there is a ready-made market
for the new product. Moreover, some interfaces are specifically designed to ac-
commodate innovation. For example, the “operating system interface” invites
application programmers to write innovative applications by eliminating the need
to deal with a myriad of details of hardware control. Standards for communica-
tions and network protocols (interfaces) allow innovative products to interoperate
with or supplant predecessors.

An “open” interface by definition is an interface controlled by a consensus of

the interested parties. Formal standards, for example, are open, and the industry
has many industry consortia that develop and maintain open standards (e.g.,
IETF, X/Open, OMG). Open interfaces promote innovation by providing innova-
tors with a dependable, stable interface for new products. The success of the
deregulation of telecommunications services depends on standard interfaces. For
example, a compatible local exchange carrier must be able to connect into
the networks of other local exchanges and long-distance carriers with predict-
able interfaces.
Software
Software is the universal building material of network systems and communi-
cations products and services, and the importance of software as an enabler of
innovation cannot be overemphasized. The vital functions of many products and
services are controlled by software. Many innovations are merely software im-
provements, but dramatic innovations sometimes come from relatively simple soft-
ware. For example, the World Wide Web is essentially a set of common standards
applied to a preexisting Internet communications infrastructure enabled by browser
software. Another example is an easy change in software to use a digital compres-
sion algorithm to increase the effective speed of a modem; this change was based
on digital signal processors becoming fast enough to implement a high-speed mo-
dem in software alone. An encryption feature could also be added easily.
Software coupled with telecommunications has another virtue—updated soft-
ware can be distributed rapidly to customers over a network. Whether the new
version fixes a bug or introduces a new feature, customers can easily upgrade
their equipment, and new software systems are frequently introduced by allowing
free downloads over the network (e.g., the Netscape browser, media players). By
distributing software widely at zero or low cost, firms count on network effects to
generate even broader use and to build a strong market base for their products.4
Software can be customized to meet the special needs of individual cus-
tomers. Although many vendors do not offer variants because of the expense of
testing and maintaining separate versions for separate customers, “open source
software,” makes the source code available to customers allowing them to inno-
vate independently. Although some large pieces of software (e.g., the Linux
operating system and Gnu software tools) are available in open source, it is too
soon to tell whether open source software will become a significant pattern in
the industry.
Multiplier Effect of Infrastructure

Infrastructure is critical to the advancement of network systems and com-
munications technologies. At any given time, the installed networking and

computing facilities are the substrate upon which further innovations are devel-
oped and introduced. The innovations may, in turn, lead to pressure to enhance
the infrastructure, thus initiating a new cycle. The early ARPAnet is an example;
the need to connect ARPAnet to other networks led to internetworking proto-
cols, most notably the transmission control protocol and Internet protocol
(TCP/IP) (Cerf and Kahn, 1974). The demand for connections led to higher
transmission speeds, faster routers, and routing protocols that could be scaled to
a larger network. The larger network and its protocols and naming conventions,
in turn, provided the near-universal connectivity that led to the creation of the
World Wide Web. The growth of the Web increased the demand on the capacity
and scale of the network, and today, the infrastructure is being challenged
to carry traffic with real-time requirements, such as VoIP (voice over IP)
and video.
Advances in network infrastructure have been a key to fostering innovation.
The federal government made the initial investments in ARPAnet and NSFnet.
As the network expanded to nonacademic customers, regional network consortia
built up the network. Today, service revenues support the network, but the federal
government continues to support experiments that may lead to significant im-
provements in performance (Internet2). A similar pattern of infrastructure invest-
ment occurred in academic computing facilities. When workstations were first
introduced, the National Science Foundation (NSF) helped equip academic re-
search centers with the new technology, which served as a substrate for academic
research in networking and interactive computing. Subsequently, parallel com-
puters were provided to encourage research on software tools for writing high-
performance parallel computing applications.
Intellectual Property
In addition to patents, the industry also issues many licenses and cross-
licenses. No company has a dominant position in the industry based on intellec-
tual property (in contrast to the way Xerox dominated the copier industry when
its basic xerography patents were still in force). Patents covering interfaces—
whether computer buses or communications protocols—must be licensed widely
because interfaces must be open to be widely used. Therefore, to receive their
endorsement, most standards bodies require that patents covering standards be
licensed liberally.
Size of the Research Investment
Government Funding
The federal government has provided major support for university comput-
ing and networking facilities. The Defense Advanced Research Projects Agency

(DARPA) and NSF were particularly active in the development and growth of
the Internet and continue to provide the bulk of support for new initiatives in
networking research and infrastructure. Federal funding for research in com-
puter science increased from $129 million in 1980 to $1.5 billion in 1999 (NSF,
2001). In 1999, roughly 33 percent of this funding was provided to universi-
ties—the rest went to industry and government agencies; more than 75 percent
of funding for basic research went to universities. In electrical engineering, of
which communications is a subset, federal funding for research remained basi-
cally flat throughout the 1980s and 1990s, peaking at $881 million in 1993 and
retreating to $699 million in 1999 (NSF, 2001). The share of total funding for
electrical engineering research that went to universities rose, however, from
10 percent to 27 percent during this period. The two federal agencies that
support research in computing and communications are the U.S. Department of
Defense (DOD) and NSF, in that order.
Federal funds support roughly two-thirds of university research in computer
science and electrical engineering. Some of this funding is used to support the
acquisition of research equipment and to support graduate students. The number
of graduate degrees in electrical engineering and computer science increased
rapidly through the 1980s and the 1990s; the number of master’s degrees awarded
more than doubled; 925 doctoral degrees in computer science and 1,596 in elec-
trical engineering were awarded in 1998 (Hill, 2001). The proportion of nonresi-
dent aliens in total doctoral degree enrollments in computer science and computer
engineering has risen steadily since 1945, up to 55 percent in 2001. Interestingly,
data from the most recent Taulbee Survey indicate that only 17 percent of new
faculty are nonresident aliens; proportionately fewer foreign students take posi-
tions at U.S. universities (Bryant and Vardi, 2002).
Industry Funding
Computer-related industries tend to be R&D intensive. Firms in this sector
spend a greater percentage of sales revenues on R&D than any other industry
except medical devices and pharmaceuticals. In computer-related industries,
roughly 10 to 20 percent of corporate R&D funds are spent on research (rather
than development). According to a 1999 report by the National Research Coun-
cil, “Such expenditures tend to derive from, and result in, the fast pace of innova-
tion characteristic of the field” (NRC, 1999). Although the volume of R&D
investment in computer-related industries has kept pace with the growth of busi-
ness over the past decade, the R&D spending of the telecommunications compo-
nent of the network systems and communications sector has contracted in the
wake of AT&T’s divestiture, deregulation, and most recently, deep recession in
the telecommunications industry.
Although the amount of industry support for university research in network
systems and communications is not known, overall industry support for research

in science and engineering in universities represents about 7 percent of the total

universities receive. The percentage is higher, perhaps as high as 15 to 20 per-
cent, in engineering; support varies by the rank and reputation of the university
program (Morgan and Strickland, 2000). For the most part, computer-related
industries have tended to draw on academic research more extensively than the
telecommunications industry.
Some of the largest firms in information technology provide significant sup-
port for university research. IBM currently spends several hundred million dol-
lars per year for research at universities. Support for university research provides
IBM with access to activities at universities and contact with potential future
employees. Microsoft has established a large research organization that empha-
sizes fundamental research. It too has established research collaboration with a
number of universities, including Southern California, Utah, Yale, and West
Virginia University. Other firms have established relationships with a small num-
ber of universities. One example is the AT&T Center for Internet Research at
UC-Berkeley that was funded by AT&T in 1998 for three years. Another ex-
ample is Intel Corporation, which has established research sites at the University
of California at Berkeley, Carnegie Mellon University, and the University of
Washington among others (http://www.intel-research.net/).
Some research organizations require or encourage both industry and govern-
ment funding. NSF supports university-based engineering research centers and
science and technology centers that must have industry contributions to supple-
ment government funding. Initiatives like Internet2 and Next Generation Internet,
which are funded principally by government, solicit industry support.
CONTRIBUTIONS OF ACADEMIC RESEARCH

Academic research has made essential contributions to the network systems
and communications industry. The special strengths of universities are reflected
in the ways they have contributed to the industry:
• Human capital. Undergraduates and graduate students educated in univer-

sities have become key players in industry as individual researchers, devel-
opment engineers, technical leaders, and entrepreneurs. Research experi-
ence in universities is highly valued by industry even for nonresearch
employees. As students and faculty flow to industry and start-up compa-
nies, they provide an effective form of “technology transfer.”
• Long-term fundamental research. With proper funding, academic re-
search is able to work on long-term problems that may be ignored by
industry or may even be anathema to dominant industry businesses, tech-
nologies, regulations, or prejudices.
• Intellectual diversity. Academia provides an open setting that can en-
gage colleagues in various disciplines and industries; the results are

reported in the open literature. Concurrent research projects and differ-

ent approaches provide a kind of redundancy and expand the commu-
nity of researchers on promising topics. Shared artifacts of experimental
research, especially software, are an important way to disseminate re-
search results.
• Collaboration with industry. Direct collaboration between industry and
academia, both on specific projects and in longer term relationships, has
produced significant contributions to network systems and communica-
tions. There are many collaborative structures but no dominant or “best”
collaboration scheme.
• Test beds. University laboratories can serve as test beds for new tech-
nologies. Most of the early participants in the ARPAnet, for example,
were universities, which played an important role in testing and refining
the technology. The pattern has continued with the Gigabit Testbed,
vBNS, and other networks, such as a campus-wide wireless network at
Carnegie Mellon University.
• Nuclei for start-up companies. University research can lead to tech-
nologies and people that become the seeds of new businesses.
Examples are Google and Yahoo, both spin-offs of research at Stan-
ford University.
At an October 1998 NAE workshop to collect information and exchange

ideas for this study, the participants came to the following conclusions: (1) the
network systems and communications sector has benefited greatly from a na-
tional research culture in which individuals move frequently between academia
and industry, thereby increasing their knowledge of both and their contributions
to both; (2) personal relationships are crucial; and (3) universities not only in-
vigorate the research culture with fresh students each year, but they also house
open research projects that anchor technical disciplines.
Human Capital
Industry looks to universities to educate and train students who will staff
industry R&D projects. Industry considers human capital to be the most impor-
tant product of universities—even more important than new knowledge captured
in research results. The question of whether industry wants students with a broad
technical education or with training in specific skills, such as programming in a
given computer language or the operation of a certain kind of computer or com-
munication device, is answered differently by different businesses. Larger com-
panies tend to prefer broadly educated candidates who can learn skills quickly on
the job. Smaller companies that do not have people to serve as mentors and
trainers prefer trained candidates.

Training in research is extremely important to innovation, even if an indi-

vidual does not continue to do research. Industry considers research experience
valuable because it demonstrates the abilities necessary for any technical en-
deavor: self-motivation, problem solving, teamwork, knowledge of related re-
search, contact with other researchers and colleagues, the ability to organize an
amorphous problem, and the perseverance to overcome setbacks. Graduates with
advanced degrees have already shown greater than average ability; and research
training is considered evidence that an individual can tackle difficult technical
problems, such as designing and building complex systems.
Students of electrical engineering and computer science have typically
been in great demand, not only by companies in the network systems and
communications sector, but also by other companies trying to modernize their
information technology. The Bureau of Labor Statistics estimates that almost
4 million information-technology workers will be needed by 2010, compared
with 1.9 million in 2000 (Bureau of Labor Statistics, 2001). If this projection
remains accurate, the current rate of graduation in U.S. universities—approxi-
mately 27,600 undergraduates and 900 Ph.D.s in computer science per year—
will not bridge the gap (Hill, 2001). Students and faculty who participate in
start-up companies are important to the culture of innovation. A significant
number of network systems and communications businesses have been founded
by students straight from universities or by faculty, who either take a leave of
absence or leave permanently. These companies are often formed to exploit a
technology developed in the university. One of the best-known examples is Sun
Microsystems, which began as a start-up company to commercialize a com-
puter workstation designed at Stanford and Unix software originally conceived
and developed at Bell Laboratories by Thompson and Ritchie then further
developed by Bill Joy at UC-Berkeley. In the summer of 1998, six (out of 60)
members of the electrical engineering faculty were on leave from Stanford
University to work with start-up companies. Faculty members who return to the
university report that their research has been stimulated greatly by their experi-
ence. A founder of Granite Systems, for example, said that he now has a far
better sense of what it takes to produce a product, as well as the state of the
industry (D. Cheriton, Stanford University, personal communication, Septem-
ber 8, 1998).
A report by BankBoston on the impact of a research university concluded
with the following statement: “If the companies founded by MIT graduates and
faculty formed an independent nation, the revenues produced by the companies
would make that nation the 24th largest economy in the world. The 4,000 MIT-
related companies employ 1.1 million people and have annual world sales of
$232 billion” (BankBoston, 1997). MIT-related firms are especially prominent in
electronics and software. More than half of the companies founded by MIT
graduates were founded by graduates in electrical engineering (which at MIT
includes computer science).

Research in Engineering and Computer Science

University research in electrical engineering and computer science has made
significant contributions to the network systems and communications industry. In
some cases, academic research projects have been essential to the creation of
billion-dollar businesses (see Figure 2-1 for some examples).
Academic research has also built a foundation of techniques and analysis
tools that are widely used as enabling tools by the industry. These include tech-
niques for the optimization of computer programs, for the automatic design of
integrated-circuit chips, and for the verification of bus specifications. These tech-
niques are not dramatic developments that spawned businesses, but they have
been important to the industry as a whole. The digital cryptographic techniques
widely used today to ensure privacy and authentication in electronic commerce
and related applications were invented in academia (NRC, 1996). The develop-
ment of object-oriented programming, from the first step (Simula 67) to the most
recent form (Java), took 30 years. Most of the research was necessarily conducted
in academia, because industry typically does not invest in risky research that
offers only long-term prospects for payoff.
The two case studies below illustrate how academic research has contributed
to the network systems and communications sector: (1) the Internet, which shows
a 30-year trajectory of academic and industrial R&D to build a revolutionary
communications technology; and (2) research in signal processing that led to
a start-up company, Amati Communications, that successfully exploited
the technology.
Case Study: The Internet

Academic research played a key role in the development of internetworking,
the connection of disparate networks into a worldwide, scalable, packet-switched
network. The Internet, which now connects more than 100 million people and
computers, was the direct result of government-funded research begun in the late
1960s to link different kinds of academic research computers. Although industry
was essential to the scaling of the Internet and the development of services, the
early technical development depended almost entirely on university research (for
a more detailed case study, see SRI International, 1997; for a brief chronology,
see Box 2-2). The ARPA funded research, development, and deployment of this
revolutionary packet-switching technology, because the telecommunications in-
dustry showed no interest in participating.
The story begins with the ARPAnet, which was initiated by DARPA in the
late 1960s as a way to share access to expensive or special-purpose research
computers around the country. Precursor ideas for packet-switching networks
had been developed at MIT and UCLA, but Paul Baran is credited with discover-
ing packet switching while at the RAND Corporation in the early 1960s. Donald
Davies, a researcher at the U.K.’s National Physical Laboratory independently

BOX 2-2
Chronology of the Internet Development
1969 DARPA commissions ARPAnet to promote networking research.
1974 Vinton Cerf and Robert Kahn publish a paper specifying the TCP/IP
protocol for data networks.
1981 NSF provides seed money for CSnet (Computer Science NETwork) to
connect U.S. computer science departments.
1982 DARPA establishes the TCP/IP protocol as standard.
1984 The number of hosts (computers) connected to the Internet exceeds 1,000.
1986 NSFnet and five NSF-funded supercomputer centers are created. NSFnet
backbone operates at 56 kb/s.
1989 Number of hosts exceeds 100,000.
1991 NSF lifts restrictions on commercial use of the Internet. World Wide Web
software is released by CERN, the European Laboratory for Particle Physics.
1993 Mosaic browser developed at NSF-funded National Center for Super-
computer Applications at the University of Illinois is released.
1995 U.S. Internet traffic is carried by commercial Internet service providers.
1996 Number of Internet hosts reaches 12.8 million.
Source: SRI International, 1997.
discovered the idea of packet switching in 1965, deciding upon some of the same
parameters for his network design as Baran, such as a packet size of 1024 bits.
ARPA built an early network by contracting with Bolt, Beranek, and Newman to
build packet switches (IMPs) at research computers at about a dozen universities.
In addition, academic research projects were initiated to develop protocols by
which different types of computers could communicate, to outfit the computers
with suitable hardware and software interfaces to the network, and to measure the
performance of the operating network. Similar networks were also developed,
such as a network using satellite or radio-transmission links to connect the
packet switches.
Early protocol experiments, together with the clear need for interconnecting
the various kinds of networks being developed, pointed to a need for
“internetworking.” The key idea is the Internet datagram, a universal way of
formatting network packets, together with associated protocols (TCP/IP), intro-
duced in a paper by Cerf and Kahn (1974) while Cerf was a member of the
Stanford faculty. This paper provided the first definition of Internet architecture
and led to implementations and experiments at several universities. With the
TCP/IP implementation developed in “Berkeley Unix” software (at the Univer-
sity of California at Berkeley) and released in the late 1970s, it was easy to
connect academic research computers to the network. Ad hoc committees of
academic researchers refined the TCP/IP protocol standards, including applica-
tion protocols.
In the late 1970s, academic computer science research centers not served by
the ARPAnet banded together to form CSnet, also using the TCP/IP protocol. In

1983, the new network was linked to the ARPAnet, an event that could be called
the birth of the Internet. Subsequently, the network continued to grow, and the
demand for connections increased. In 1986, with the creation of NSFnet, the
responsibility for the principal “backbone” of the nationwide network shifted
from DARPA to NSF. The network could be used only for research and educa-
tion, and academics continued to play a major role in network governance
and engineering.
With the emergence of the World Wide Web, the Internet was no longer only
for research and education but became a worldwide network connecting busi-
nesses and consumers, as well as researchers. The idea of browsing text docu-
ments obtained in a uniform way from any machine connected to the Internet was
developed in 1991 by Tim Berners-Lee, then at the CERN nuclear research
facility in Geneva, Switzerland. Documents on the Web may contain “hyperlinks”
to other documents, thus linking documents into a complex “web.” Marc
Andreesen and other researchers at the University of Illinois later extended the
Web to include pictures and other types of data. They also built a graphically
oriented browser, called Mosaic, that allows users to “click” to follow a hyperlink,
thus opening browsing to a wide range of people. Jim Clark recruited Andreesen
to cofound Netscape Communications, which developed Netscape’s Navigator
browser product based on Mosaic. Microsoft soon developed a competing prod-
uct, Internet Explorer. The combination of pictures, ease of use, and supported
products enabled the Web to grow with astonishing speed. It quickly developed
into a mechanism for publishing, for finding information, and for transacting
business electronically.
As more computers were connected to the network, the bandwidth and
switching capacity had to be expanded. DARPA and NSF, with university and
industry support, organized a series of test beds to explore high-speed networking
technologies and test emerging products and protocols. Between 1990 and 1994,
NSF and DARPA funded the Gigabit Testbed Initiative, a university-industry-
government effort to explore networking technologies at speeds of 155 Mb/s and
higher; one of these test beds achieved long-haul transmission at 800 Mb/s. NSF
operated the vBNS network (very high-speed backbone service) in conjunction
with MCI to link more than 75 universities in a network with backbone speeds of
622 Mb/s to 2.4 Gb/s and access links of 43 Mb/s to 155 Mb/s until the vBNS
network was terminated in April 2000. The participants explored new applica-
tions of advanced communication bandwidth and protocols, as well as opera-
tional and governance issues. Universities are presently engaged in a new round
of infrastructure enhancement, Internet2, designed to meet a full range of aca-
demic research needs.
As the Internet expanded, commercial businesses and services grew up along-
side government and academic operations. New companies were started to sell
packet switches (routers), application software, authoring tools, and network
services. The leaders were not the telecommunication companies, but start-up

companies, such as Cisco and Netscape, using technologies developed in univer-

sities. Network service providers emerged as regional networks (e.g., NEARnet,
BARRnet) and were encouraged to connect to the national backbone operated as
part of NSFnet. International connections were also developed. As the Internet
grew, major telecommunications companies began to offer Internet service as
well. In the United States today, businesses and residents in most densely popu-
lated areas have a choice of several Internet connection methods and ser-
vice providers.
The governance and engineering of the Internet are unique: a governing body
(the Internet Activities Board) and engineering standards organization (Internet
Engineering Task Force) consist of volunteers from Internet participants. At first,
committees of academics and the few government contractors that were building
and operating the network set protocol and other engineering standards. Today,
these committees have broad participation from academia, industry, and non-
profit organizations.
Academic Contributions
An SRI study commissioned to analyze the nature of the research that con-
tributed to the Internet described the contribution of academic research in some
detail (SRI International, 1997):
The Internet appears, overall, to be primarily a problem-driven, technology-based
innovation that required little direct input from fundamental research for its real-
ization. The driving forces, interestingly, were not profit incentives in the private
market, but public goods, first in the realm of national defense and subsequently
in the university and government research infrastructure, as a means of fostering
communication among computer scientists. What we are calling the Internet’s
intrinsic technologies—network design, packet switching, routers, protocols,
browsers—were the products of problem-driven research conducted in universi-
ties and government contractor laboratories with government support. One pos-
sible exception is the research conducted at the University of Illinois’ NCSA
[National Center for Supercomputer Applications], which took place in an envi-
ronment (according to Andreesen) that enabled researchers to head off in direc-
tions that looked “interesting” without seeking justification. Nonetheless, the con-
text was one of application, as suggested by the Center’s name. Although the
evolution of the Internet did not encounter technical roadblocks that required
fundamental research for their resolution before further progress could be made,
there is obvious, fundamental research content in both the Internet’s intrinsic and
supporting technologies. The electronic and physical infrastructures that comprise
the Internet clearly depend on information theory, solid-state physics, electro-
optics, and other fields on which modern communications technology is based
and for which NSF has provided substantial support.
The SRI study stresses the importance of government, industry, and univer-
sities in the development of the Internet and points out that, as the focus of

technical and organizational innovation shifted from government to industry over

a 30-year period, universities played a constant supportive role. By following the
career trajectories of some key individuals who moved among the three sectors,
the SRI study highlights the importance of networks of individuals and the im-
portance of human capital. The study also cites the frequent opportunities for
interactions and linkages among the three sectors and the ease with which indi-
viduals can move across sectors in the United States as important factors to
success (SRI International, 1997).
The Internet is an example of the major impact academic research can have
in the creation of an entirely new technology. Although the enormous impact of
the Internet is unusual, this case illustrates how government, academia, and in-
dustry can contribute to technological—and therefore economic—success.
Case Study: Amati Communications

Amati Communications was founded in 1991 by Stanford University profes-
sor John M. Cioffi to commercialize technology for transmitting high-speed digi-
tal signals over copper telephone wires. The technology, named discrete multitone
(DMT) modulation, was one of several technologies competing to be adopted as
an industry standard. DMT is now an accepted standard for providing DSL (digi-
tal subscriber line) service and a commercial success.
Research
The original work on DMT was conducted in 1987 by a research group at
Stanford directed by Cioffi, who was then an assistant professor. Cioffi used
funds from an NSF Presidential Investigator Award (1987–1992), with matching
funds from several companies, including Bell Communications Research, to in-
vestigate asymmetric digital subscriber lines. The initial objective was to develop
reliable transmission of high-quality digital movies over phone lines, which re-
quired speeds about 10 times faster than integrated service digital network (ISDN)
lines, the existing technology. Later, the objective evolved to encompass high-
speed Internet access and other data applications. The researchers investigated
many methods and focused on an old encoding technique called multitone trans-
mission, in which separate frequency channels (tones) carry separate digital sig-
nals. A crude analogy would be sending several channels of Morse code over a
telephone line, with each channel using a different audible frequency (like the
seven separate tones used in touch-tone dialing). A receiver can split out the
separate tone frequencies and decode each Morse sequence.
The goal of Cioffi’s research was to transmit data as fast as possible, which
would require using many separate frequency bands and sophisticated signaling
techniques (not Morse code!) in each band. This objective led Cioffi and his
team to seek fundamental improvements in digital signal processing algorithms

that could be applied to various channels. But the most important innovation was
the adaptation of each band to the band’s transmission characteristics. In effect,
DMT measures the properties of signals transmitted over each band and then
allocates data accordingly. A band that attenuates signals less than another band
carries more data. A band that introduces less noise than another also carries
more data. DMT also measures and compensates for the transmission character-
istics of each pair of copper wires. This complex scheme is practical because of
inexpensive digital signal-processing hardware.
Some of the key innovations for DMT were patented while Cioffi’s group
worked to refine, promote, and publish the method in IEEE journals and at
American National Standards Institute standards meetings. Three of the patents,
which are assigned to Stanford, are still valuable and are licensed throughout the
telecommunications and data communications industry. At least one patent, on
pioneering artificial intelligence techniques used in the adaptation, is considered
necessary to comply with any of the existing or emerging DMT standards. In
addition to patents, Cioffi and his graduate students acquired valuable know-how
that would benefit any company that attempted to use DMT.
Amati Start-up Company

In 1991, after an unsuccessful search for corporate partners, Cioffi founded
Amati Communications Corporation with Stanford University and a Stanford
alumnus, Mr. Kim Maxwell, who was the company’s first CEO. Amati’s vision
was to “get DMT on every phone line in the world.” To be successful, Amati had
to make DMT an industry standard, which involved working with national and
international standards organizations and competing with other technologies. On
March 10, 1993, after competitive testing against several other technologies,
including an alternative promoted by AT&T, DMT technology was selected
unanimously as a U.S. standard.
As part of Stanford’s contribution to the company, Cioffi was given three years
of leave (with 50 percent leave spread over two years to make up the third year). In
exchange for an exclusive licensing and sublicensing privilege on the DMT patents,
Stanford received stock and a promise of royalties. In 1997–1998, Stanford re-
ceived $7.9 million when it liquidated its holdings in Amati; royalties totaled more
than $8 million in 2001 (Stanford University Corporate Guide, 2001).
Between 1991 and 1998, Amati employed several of Cioffi’s graduate stu-
dents, at least four of whom were directly involved in the Stanford research and
had considerable knowledge of DMT. Stanford’s then dean of engineering, Jim
Gibbons, was chairman of Amati’s Board of Directors until 1998.
Amati went public late in 1995 and was a growth leader on the NASDAQ for
approximately one year. In March 1998, Texas Instruments acquired Amati and
the DMT license for approximately $450 million in cash. At the end of 2001,
3.6 million residential DSL (digital subscriber lines) using DMT technology were

installed in the United States, which was then projected to grow to 13.5 million
by 2005 (InternetWeek, 2001).
Academic Research in Economics, Social Sciences,

Management, Design, and Policy
Academic research in a variety of nonengineering disciplines has also contrib-
uted to the network systems and communications sector. This research has focused
on how computers and communications systems fit into larger socioeconomic sys-
tems. Examples include how information technology systems increase business
productivity; the effects of e-mail on people and organizations; how people use the
Internet; the effect of the Internet on family structures; the effect of prices on
communications services; and the value of communication services to consumers.
Some of these studies have formulated and tested models to explain the
behavior of people and systems in operation. For example, the GOMS model of
performance grounded in cognitive psychology (Card et al., 1983) has been used
to design interactive interfaces. John Anderson has built several successful com-
puterized tutoring systems based on a detailed understanding of how people
model specific subjects (e.g., geometry and algebra) and the errors in these
models. Results of social psychology research on electronic communication
have been used to improve training for new users, in effect teaching them about
the social norms and effects that spring from the technology.
In some cases, the models only described what had been observed, but in a few
cases they were used to predict the behavior of future systems. For example,
models can predict how well people will perform simple interactive computer
tasks. Although models cannot predict whether one chat room or e-mail system will
be more popular than another or the details of e-mail usage, general principles can
be learned. For example, people are generally less inhibited when using an elec-
tronic communication medium than in face-to-face interactions. Observations such
as this can lead to a better understanding of network systems and communications,
but they cannot be used as a guide to design. At a workshop held in connection with
this study, the consensus was that these research topics should be given more
attention, especially as the services provided by network systems and communica-
tions become more important and affect a greater portion of society.
Economics, Policy, and Regulation

Because communication systems have historically been operated as regu-
lated monopolies, researchers in economics and policy were able to study them
extensively. The work of one academic, Alfred Kahn of Cornell University, was
used as a basis for the deregulation of several industries, including trucking,
airlines, and communications (Kahn, 1970, 1971). Nevertheless, a great many
questions about economics and policy remain to be answered (see Box 2-3).

BOX 2-3
Contributions of Economics and Other Social Science
Research to the Development of Information Technology
Role of regulation. Economics research by Alfred Kahn (1970, 1971), Paul
Joskow, Roger Noll, and Kip Viscusi redefines the role of regulation from protect-
ing the public interest to stepping in when markets fail to drive prices to marginal
costs. This redefinition has helped spur deregulation in a number of industries,
including communications.
Network externalities. Work by Hal Varian, Paul David, Brian Arthur, Garth
Saloner, David Shapiro, and others shows that the network industries and infor-
mation industries are characterized by “network externalities” that make the value
to a consumer of a particular product or service increase as more people use it.
An example is an Internet connection that becomes more valuable with the amount
of information available and the number of people connected. This insight re-
inforces the importance of getting products and services to the marketplace quick-
ly, pricing them low at first to establish markets, and then raising prices as more
units are sold and their value grows.
Internet Economics. Research by McKnight and Bailey (1997) and others
address the implications of the pricing of Internet-based resources and ser-
vices, such as the allocation of resources based on the willingness of users
to pay.
Group dynamics and decision making. Research by Sara Kiesler, Suzanne
Iaconno, Wanda Orlikowski, and others (e.g., Siegel et al., 1986) on group dy-
namics and decision making in small electronic groups informs the design of group
decision-support systems.
Diffusion of applications. Research by M. Lynne Markus (1987) and others
examine how critical mass predicts the diffusion of networked applications within
organizations and informs the deployment decisions for information technol-
ogy applications.
Distribution of the benefits of information technology. Research by Tora
Bikson, Lee Sproull, and others demonstrates that peripheral members of social
systems benefit more from using electronic communication than central members
(e.g., Sproull and Kiesler, 1991) influencing policy decisions about subsidies for
access to the Internet.
Information sharing. Research by Paul Attewell, Tora Bikson, Sara Kiesler,
Robert Kraut, Lee Sproull, and others demonstrates how personal attributes and
organizational characteristics such as incentive systems influence peoples’ use of
information technology for information sharing. Research by Julian Orr (1990) and
others demonstrates that service technicians often have more useful technical
expertise than system designers and share their knowledge in a community of
practice. This work influenced the design of a community-based troubleshooting
database at Xerox Corporation that has significantly improved service perfor-
mance (Bell et al., 1997).
Sources: Sirbu, 1998; NRC, 2000.

The Telecommunications Policy Research Conference is an annual forum for

scholars engaged in research on policy-relevant telecommunications and information
issues and public-sector and private-sector decision makers engaged in telecommuni-
cations and information policy. The wide range of topics at the 2002 conference
reflected the intense academic interest in telecommunications policy. Topics included:
comparative telecommunications policies in the United States and abroad; broadband
deployment and uptake; spectrum management; computer and Internet security; wire-
less communications standards; mergers and acquisition; intellectual property; basic
research in telecommunications; mass media; and numerous other topics. University
researchers presented the majority of research results at these sessions, reflecting the
active involvement of academic research.
Business and Management

Business schools have long been concerned with how information technol-
ogy can be exploited for the benefit of businesses (see Box 2-4). For example,
research on decision-support systems not only developed techniques for collect-
ing and presenting relevant business data to management, but also compared the
quality of decisions made with different kinds of supporting technology. The
rapid development of the Internet has opened up new avenues for study, such as
supply-chain integration, which uses the network to connect a manufacturer’s
process-planning system to the corresponding systems of its suppliers to ensure a
smooth flow of the component parts required to fulfill orders. Success will de-
pend on solving problems related to information technology, network protocols,
and control theory. Electronic commerce will certainly face new problems that
must be addressed.
Optimizing network design in network systems and communications busi-
nesses is a similar problem to the transportation problems studied by operations
research. In the early development of the ARPAnet, attention was focused on
optimization of network design. Today, the emphasis is on techniques to expand
networks to meet burgeoning demand.
Psychology and Social Sciences

Research by psychologists and social scientists has focused on how people use
computer and communication systems, the effects of these systems on people, how
people interact with each other, and how they work in organizations. These studies
are retrospective, conducted after the technology has been deployed long enough
for transient behaviors to abate. De Sola Pool’s classic book, The Social Impact of
the Telephone (1997), is a fine example. Other examples are Computers in Class-
room Culture (Schofield, 1995), Connections (Sproull and Kiesler, 1991), The
HomeNet Field Trial of Residential Internet Services (Kraut et al., 1996), and The
Second Self (Turkle, 1984), a study of personal interactions with computers.

BOX 2-4
Contributions of Business Research to the Development of
Information Technologies
Critical success factors. Rockart (1981) identifies factors critical to the success
of information systems in business settings. The author addresses the question of
which information is critical to the success of a business; the questions a database
should answer; and how information systems can be designed to support busi-
ness objectives.
Decision-support systems. Work by Keen and Scott Morton (1978) promul-
gates the idea of using information systems to support corporate decision making
at a variety of levels.
Information technology and strategic management. Research by Earl (1988)
stresses that IT is not just for back-office operations but contributes to a firm’s
competitive advantage. Companies that deploy and employ information technol-
ogy systems wisely can benefit in the marketplace.
Computer-supported cooperative work. This research introduces the idea of
using information technology to allow people to work cooperatively within and across
organizations, thereby overcoming differences in geography or time.
Productivity and information technology. Productivity gains from investments
in information technology have been hard to measure, but Brynjolfson’s (1991)
analysis of firm-level data (as opposed to industry-level data) indicates that invest-
ments may have large payoffs, but not immediately. The author identifies factors
that contribute to positive returns from investments and the characteristics of firms
that do and do not experience increased productivity.
Software development methodologies. Research by Cusumano (1991) pro-
vides guidance on software development methodologies from the point of view
of management.
Process handbook. This repository of business-process knowledge developed by
the MIT Process Handbook Project (Malone et al., 1999) can facilitate further re-
search and help determine best practices for deploying information technology. The
classification and structure of the database itself is a powerful tool.
Source: Based on Malone, 1998.
Research has also helped guide the design of computer and communication
systems. The Psychology of Human-Computer Interaction, a classic work by
Card, Moran, and Newell (1983), showed how studies in cognitive psychology
could be used to estimate human performance when interacting with a computer.
These and other performance studies have influenced the design of graphical user
interfaces. Ethnographic studies of the behavior of boys and girls at play have
been used to inform the designs of many products.

Design Research
Several universities have developed broad multidisciplinary programs aimed
at harnessing developments in information technology to human needs. A leader
in this area, the MIT Media Laboratory, brings together individuals from a broad
spectrum of disciplines, including the humanities and fine arts, to conduct re-
search and application development. For example, the News in the Future Project
explored innovative ways to present the news to people using electronic media by
tailoring content, presentation, and structure to the needs of the viewer. In addi-
tion to developing prototype applications, the laboratory often works on funda-
mental technologies, such as video compression or image understanding.
MECHANISMS FOR UNIVERSITY-INDUSTRY COOPERATION

Collaborations between industry and academia raise some obvious questions
about the kinds of organizations and mechanisms that work best. A questionnaire
on the subject sent to 60 researchers for this study elicited a wide range of
responses. For example, one respondent felt that “centers which promote close
interactions between academic researchers and knowledgeable industrial spon-
sors are probably a prerequisite for making progress.” Another mentioned several
collaborative arrangements: joint research programs, like MIT’s Project Athena;
experimental test beds and university centers, like the NSF-supported
supercomputing centers; and consortia. Another respondent felt that “Centers
have an indifferent record in communications . . . I doubt that such forms of
collaboration will ever be a success.” Still others felt that the structure of the
organization didn’t make much difference as long as the participants understood
each other’s value systems.
NSF has several programs to create university-based, industry-university
research centers and engineering research centers (ERCs), both of which require
industrial participation. The ERCs, which are designed to integrate research and
education, have generally received favorable reviews (Parker, 1997). However,
the Telecommunications Research Center at Columbia University, funded by
NSF from 1985 through 1995, was the only ERC established in the network
systems and communications area.
The network systems and communications sector does not have an institu-
tion comparable to the Semiconductor Research Corporation, an organization
that provides industrial support for university research relevant to semiconduc-
tors, based on a 10-year technology “road map” to help guide research funding
decisions (Bailey et al., 1998). Although the establishment of a consortium of
network systems and communications businesses has been discussed, nothing has
come of it so far. A consortium of computer storage peripheral companies, Na-
tional Storage Industry Consortium, has been established to support academic
research through focused programs like optical storage. In addition, some firms

have targeted their research support for a small number of universities; for ex-
ample, in late 1998, AT&T and the International Computer Science Institute at
UC-Berkeley announced formation of the AT&T Center for Internet Research
(ACIR), a multimillion-dollar research center that AT&T agreed to fund for at
least three years (AT&T and International Computer Science Institute, 1998).
Recently, Intel has sited research operations at UC-Berkeley, Carnegie Mellon
University, and University of Washington, all centers where researchers are Intel
employees and university professors are engaged as laboratory directors and
technical leaders (Intel, 2003).
The Microelectronics Innovation and Computer Research Opportunities
(MICRO) Program in the University of California system is an example of a state
government effort to encourage university-industry cooperation. The MICRO
Program was established in 1981 by the state of California to support innovative
research in microelectronics technology, its applications in computer and infor-
mation sciences, and its necessary antecedents in other physical science disci-
plines. The program is a partnership between industry and the state in which the
state supplements industry funds and waives overhead on university research
funding. In 2001–2002, 96 companies contributed approximately $6 million in
cash and equipment to fund 98 different projects (MICRO, 2002). In some cases,
MICRO support has led to increased federal funding, as well as long-term part-
nerships between universities and industry. For instance, after an initial concept
phase, the RAID (redundant arrays of inexpensive disks) project at UC-Berkeley
received MICRO support, which led to the creation of an industrial consortium in
1988–1990. The federal government became a research sponsor in 1990. By
1996, RAID was a $10 billion industry.
In general, university-industry collaborative arrangements in network sys-
tems have received mixed reviews. No structure has emerged as the “best,” nor
has any scheme emerged that works robustly in different circumstances. It ap-
pears that strong personal leadership and a collaborative spirit between an aca-
demic researcher and his or her industrial counterpart are the elements essential
to success. The problem, of course, is that a good collaborative project can
founder if one key individual (e.g., the “champion” in the firm) is transferred or
moves elsewhere.
FINDINGS AND RECOMMENDATIONS

Academic research has made essential contributions to the network systems
and communications sector. Although these contributions—trained researchers,
new technologies, algorithms, and prototype systems; early operating experience;
studies of social and economic effects—cannot be quantified, they have undoubt-
edly had a substantial impact. In this industry, the academic ivory tower has been
heavily populated by entrepreneurial engineer-researchers.

Findings
Finding 2-1. Academic research has played and will continue to play an important
role in the research culture of the network systems and communications industry.
University-industry collaborations are fostered by a vigorous research cul-
ture, and academic research has been crucial to the technical evolution of the
industry, especially in the development and deployment of the Internet. To be
sure, the recent deep recession in the telecommunications sector, which has forced
significant reductions in corporate R&D budgets and manpower, has further
diminished the industrial research contributions of this important subsector of the
network systems and communications industry—a trend begun with the breakup
of the Bell system and further deregulation. Given the historical reliance of the
telecommunications sector on internal industrial research, changes may be
needed. If trends in industrial research persist, academic research in telecommu-
nications will have to be increased. However, for the most part, the research
culture that supports the network systems and communications industry is func-
tioning well and needs no major repairs.
People are the key components of this research culture. Collaboration be-
tween universities and industry often depend on sometimes fragile personal rela-
tionships that can be threatened if an industry researcher is reassigned or an
academic researcher goes off in a new direction. Students, faculty, postdoctoral
students, researchers with long-term visions, and researchers who focus on ap-
plied problems play different roles. Contrary to popular opinion, university-
industry projects are not devoted exclusively to long-term basic research; teams
of faculty and students often address pragmatic, applied problems in close coop-
eration with industry.
The flow of people from academia to industry and vice versa is essential to
the well-being of the industry and to academic research. The university’s role of
fueling the research culture with trained students is unique, and training in re-
search is extremely important for innovation, even if the researcher does not
continue to perform research but becomes part of an academic-like cadre that
pursues innovations in industry (such as the groups awarded the Association for
Computing Machinery’s Software System Award).
Universities also have a very broad research culture, and network systems
and communications systems have increasingly drawn on the wide range of tech-
nologies and expertise available at research universities. Electrical engineering
and computer science are, of course, central to the industry, but other areas, such
as cognitive science, social science, economics, and business modeling, are be-
coming increasingly important, especially as the importance of information
technology-delivered services increases. Some research universities (e.g., UC-
Berkeley, University of Michigan, Indiana University) have created information-
centered schools that focus on the social, political, and organizational context
of information.

To participate in the research culture, a company must have a capacity to

absorb innovation, an industrial research laboratory, for example, that can absorb
people and ideas from outside the firm and exploit them within the firm. Despite
the trend in industry research toward applied problems, industry laboratories
have so far retained this absorptive capacity. In the last 15 years, innovations in
the industry have focused on completing the digital revolution in telecommunica-
tions (e.g., new switching gear); new transmission protocols (e.g., SONET and
Asynchronous Transfer Mode), and transmission equipment; faster modems; the
refinement and deployment of fiber optics; the refinement of IP protocols; and
the widespread deployment of the Internet.
The focus is now shifting toward innovative services, which requires an
understanding of psychology, consumer behavior, social phenomena, and other
disciplines that can inform the design and operation of new services. The explo-
sive growth of the World Wide Web can be attributed to its social properties more
than to its technical capabilities. Industry is most likely to devise and launch new
services, but formal research on the uses and effects of these services is most
likely to be undertaken in academia.
Finding 2-2. Innovation cycles in the network systems and communications
industry have worked well.
An astonishing number of incremental changes have cumulatively taken on
the character of breakthroughs. The Internet, for example, had its roots in the
ARPAnet in 1969, was developed and deployed incrementally, and was
launched into the public arena by the World Wide Web and browsers in 1993.
The effect on the industry was of a breakthrough; the telecommunications
industry today is utterly different from the industry of 10 years ago. Similarly,
the incremental evolution of technologies (e.g., batteries, low-power circuits,
integrated radio-frequency elements) to support small portable devices spawned
breakthrough products, such as pagers, cellular telephones, and packet-
radio modems.
Innovations in the network systems and communications industry have been
characterized by the integration of a wide range of technologies: chip designs
with increasing levels of integration; digital-analog integration in wireless and
wireline communications (e.g., cellular and satellite telephony, wireless devices,
such as pagers and security devices, modems and cable modems, local-area net-
work and intranetwork systems and communications receivers, optical network
interfaces); internetworking technologies; and, above all, the increasing use of
software technologies of all kinds. The convergence of computing and telecom-
munications has brought together a wide array of technologies for new products.
Academic contributions to these innovations have varied widely. Some inno-
vations originated in university research and were spun off into venture-capital-
backed start-up companies. This route is supported by established university
policies, a workforce eager to engage in risky start-ups, and a mature venture-

capital industry that has been willing to back telecommunications and network-
ing businesses. Many academics have been consultants for network systems and
communications companies. Academic design departments have worked on in-
dustry projects, and some companies have supported academic research, often
directed toward solving specific problems. Sometimes, new companies or prod-
ucts have emerged from business-school entrepreneurship programs.
Finding 2-3. The success of industry-academic collaboration (as defined by
participants) depends more on leadership and people than on the type of collabo-
rative structure.
Organizational structures, such as research centers that foster university-
industry collaboration, receive mixed reviews in network systems and communi-
cations. Success appears to depend less on the choice of structure, the funding
arrangements, or the legal agreements than on the leadership and passion of the
people involved. A committed leader is essential for establishing personal rela-
tionships, and, in general, researchers consider inducements to individuals (as
opposed to institutions) more effective than collaborative organizational struc-
tures. Dependence on personal relationships can sometimes lead to problems,
however. A collaborative effort between industry and a university can founder if
a key individual (the “champion” within the firm) is transferred or moved else-
where within the company. Overall, therefore, the diversity of approaches to
industry-academic collaboration is healthy for both partners.
Finding 2-4. Creating standards is an important aspect of innovation in commu-
nication systems.
Standards are necessary for interoperability, which is essential to the indus-
try. The more interoperability, the faster the growth of the user base and the faster
the increase in value of the system. The success of many businesses depends on
the number of other entities that can communicate in a network—the value of
network externalities.
Committees of industry members, sometimes with academic participation,
usually determine standards. Many Internet standards groups, such as the Internet
Engineering Task Force (IETF); the ATM Forum, which was organized to pro-
mote data-networking uses of Asynchronous Transfer Mode (ATM); and the
discrete multitone modulation standard for asymmetric digital subscriber lines,
have academic participants. In the United States, university researchers often
have difficulty participating in setting standards because of time and travel de-
mands. In Europe, academic participation has been stronger. U.S. researchers
could be helpful in gathering data and analyzing standards proposals; good data
and independent analysis could reduce squabbling over business biases and focus
more attention on design issues.

Finding 2-5. Academic institutions have been at the forefront of network infra-
structure deployment.
The United States led the way in deploying advanced infrastructure critical
to enabling research (e.g., ARPAnet, NSFnet, vBNS, NGI, Internet2). The de-
ployed infrastructure has led to further developments. For example, the World
Wide Web was successful partly because networking infrastructure had already
been installed.
Academic institutions have played a crucial role in the deployment of net-
work infrastructure. Ever since NSFnet was formed, universities have recognized
the importance of the Internet to academic endeavors of all kinds, not just com-
puter science and engineering research. With funding from NSF and other sources,
universities have been willing to deploy leading-edge technologies. Deployments
such as Internet2, which was spurred in large part by universities, are likely to
increase the speed and throughput of network services available to universities
and thus to support research that requires high-performance networking infra-
structure. However, this may not necessarily stimulate research on networking.
Conflicting demands on these systems has created some tension between provid-
ing robust services for other research fields and experimentation for networking
research. A state-of-the-art network that can be used for experiments in network-
ing has not been developed.
Finding 2-6. The network systems and communications industry is evolving in
directions that may require new kinds of university-industry partnerships to ex-
ploit research.
As communicating appliances proliferate, the need for harmonizing the tech-
nological and human elements increases. Examples include: designing communi-
cation services that users can understand and exploit; integrating multiple devices
and services to create personalized configurations; designing new user interfaces;
and streamlining or automating customer service functions. Even within a net-
work, areas that are not purely technical will also require research: the provision
of services; the quality of service; incompatibilities between heterogeneous prod-
ucts and services; security; and network management and operation. Optimal
interactions among product design, network organization and management, ser-
vice provision, and technology will require close collaboration between univer-
sity researchers in the social, behavioral, and management sciences on one hand
and engineers and scientists on the other. As the industry moves toward offering
more “information services” rather than “communications devices,” it must turn
to the market rather than to research for guidance. In the future, university re-
search might focus on how individuals and society as a whole value and use
network systems and communications services.
Finding 2-7. Many Internet service providers are not willing to make their data
available to researchers.

Many pressing research questions in network systems and communications

require studies of the characteristics of networks under normal operating condi-
tions as a basis of comparison. Research on limiting congestion, improving the
quality of service, and improving routing requires traces or other logs of actual
network activity. Although data are available for some experimental networks,
many Internet service providers have not been willing to make their data avail-
able to researchers. This has hampered university research that might lead to
improved network design and operation. The problem could get worse if service
providers become more vertically organized and less open about their operations,
problems, and needs.
Finding 2-8. The high cost of protecting intellectual property could im-
pede research.
Industry and academia are both becoming increasingly protective of their
intellectual property rights, and the enormous economic activity in the network
systems and communications sector encourages this trend. The processes of
working out licensing and sharing agreements could impede the free flow of
ideas necessary for research to flourish. Whenever universities band together in
a research consortium or a single university-industry collaboration is started,
researchers spend much time and energy working out intellectual prop-
erty agreements.
The trend can be counteracted in several ways. “Open source” software licenses
that implicitly recognize that unused intellectual property has no value explicitly
promote sharing. Other ways of reducing the costs of intellectual property protection
could include standardized forms of collaborative research agreements.
Finding 2-9. Long-term research is important to the future of the network sys-
tems and communications industry.
Despite the apparent success of network systems and communications tech-
nologies, many difficult problems must be solved for the industry to continue to
grow and prosper. Continued rapid expansion of a sophisticated communications
infrastructure with millions or billions of network elements is bound to face some
difficult problems. The industry needs software-engineering discipline to ensure
that modules intended to fit together do so and that upgraded modules can be
introduced without disrupting network operation. Distributed systems must be
designed to be robust under failure, to remain stable under all operating condi-
tions, and to guarantee performance requirements. Some long-term problems,
such as the quality of service and security, are related to scale limitations. Current
data networks require many people with sophisticated skills; and the design
complexity and deployment scales of these systems exceed our engineering
knowledge. As the extraordinary growth of the industry slows, new problems will
have to be addressed, such as the impact of microelectronic components with
different costs and technical properties, new computational needs as the rate of

miniaturization slows, and the technical implications of these changes. Indeed, as

industrial research investments change, and as human capital stresses wax and
wane, it is important to keep long-term academic research activities alive, pre-
cisely because they are the long-lived seeds from which both ideas and people
can spring, regardless of the short-term financial health of the industry.
Recommendations
Recommendation 2-1. Universities and industry should take steps to ensure that
faculty and students are available to carry on research in computer science and
other information technology fields in the future.
Innovation, either from research or incremental engineering, depends on trained
researchers. Projected demand for computer science and other information technol-
ogy graduates indicates periodic shortages in coming years. To maintain the pipeline
of both academic and industrial researchers, the following measures could be taken:
• Universities should provide early research experiences for undergradu-

ates or even secondary school students.
• Universities should provide career-development support for young fac-
ulty members.
• Fellowships should be provided for graduate students to encourage them
to pursue research degrees; industry should provide some of this support.
• Universities and industry should provide incentives for industry engineers
to return to academia for training in research.
• Universities should develop cooperative programs in which master’s de-
grees are based not only on course work, but also on research experience.
• Training in academic research should include training in some of the
qualities students will need for jobs in industry.
• Research should involve addressing not only small technical puzzles in
isolation, but also complex systems problems in context. Students should
be encouraged to confront complexity and to address real-world data and
operational problems.
• Research should encourage teamwork.
• High-caliber industry researchers and engineers should be encouraged to
take sabbaticals to work in academia, thus bringing real-world research
problems into academic settings.
Recommendation 2-2. Universities and industry should continue to develop

diverse collaborative arrangements.

Industry and universities should resist the temptation to impose standard struc-
tural mechanisms to promote collaboration. Incentives for personal interactions
between university and industry should be encouraged in the following ways:
• Provide support for strong, committed leaders and the collaborative orga-
nizations they lead.
• Encourage sabbaticals in both directions, enabling academics to spend
time in industry, especially in start-up companies.
• Support people and projects that involve academic and industry research-
ers in essential ways.
• Explore new ways to support personal interactions across academic-
industry boundaries, including using technology to support collaboration.
Recommendation 2-3. Universities and industry should make every effort to
invigorate academic research on networking.
The extraordinary success of the Internet and the lure of Internet-related
start-up companies have tended to focus attention on short-term goals, caus-
ing long-term research to suffer. The situation could be improved in the
following ways:
• Acknowledge that the research community must take risks.

• Focus academic research on the thorny problems of large systems: model-
ing, maintenance, upgrades, quality-of-service, security, and so on. Both
funding agencies and academics must recognize that large-scale systems
can best be addressed in a university setting. Even applied systems
research can be structured in a way that accommodates a long-
term approach.
• Universities and funding agencies (and industry) should support long-
term, radical research on networks.
• Universities and industry should encourage interdisciplinary research that
combines network technologies with design and social science disciplines.
Networked devices (especially hand-held mobile devices) will have to
meet both technical and human requirements.
• Universities should recognize that valuable innovations and engineering
in the field are often not channeled through traditional peer-reviewed
publications. Therefore, effective industry interaction should be more
highly valued in decisions about academic promotion and tenure.
• To revitalize academic research on networking, the National Science
Foundation should consider sponsoring a workshop on the subject that
brings together academic and industry participants. A new agenda could
provide a strong argument for industry support, either by individual firms
or by a consortium.

NOTES
1Gordon Moore (cofounder of Intel) predicted in 1965 that the transistor density of semicon-
ductor chips would double roughly every year. See Moore, 1965.
2For example, during fiscal year 2001, Microsoft spent $4.38 billion on product research and
development activities excluding funding of joint venture activity. This represented 17.3 percent of
revenue that year. Microsoft Research, the part of the company that looks more than one or two
product cycles out, has around 600 employees and a budget of roughly $200 million, less than
5 percent of the $4.38 billion, or less than 0.8 percent of total revenue.
3The loss of faculty to commercial endeavors was limited in time and to only a few programs.
Data from the most recent Taulbee Survey of computer science and computer engineering depart-
ments indicate that faculty numbers have grown and are anticipated to grow through 2004. The
survey also indicates that faculty departures have ranged from 2.3 to 2.6 percent over the last several
years (Bryant and Vardi, 2002).
4Economists have long acknowledged “externalities,” factors that alter the value of a good viewed
in isolation. Shapiro and Varian (1998) applied the idea to networks, so-called “network effects.”
Robert Metcalfe, a popular speaker on the value of networks, has often said that the usefulness, or
utility, of a network equals the square of the number of users. This observation has been dubbed
“Metcalfe’s law” (Gilder, 1993).
REFERENCES
AT&T and the International Computer Science Institute. 1998. AT&T Labs, ICSI establish Internet
research center. Press release October 8, 1998. Available online at: http://www.icir.org/
aciri.html. [June 24, 2003]
Bailey, D.E., F.S. Settles, and D. Sanrow. 1998. Designing, Controlling, and Improving SRC Re-
search Quality. Presentation at the NAE-Committee on Science, Engineering, and Public Policy
Workshop on the Role of Human Capital in Capitalizing on Research, Beckman Center, Irvine,
California, January 21, 1998.
BankBoston. 1997. MIT: The Impact of Innovation. Special Report of the BankBoston Economics
Department. Available online at: http://web.mit.edu/newsoffice/founders. [June 24, 2003]
Bell, D.G., D.G. Bobrow, O. Raiman, and M.H. Shirley. 1997. Dynamic Documents and Situated
Processes: Building on Local Knowledge in Field Service. Pp. 261–276 in Information and
Process Integration in Enterprises: Rethinking Documents, T. Wakayama, S. Kannapan, C.M.
Khoong, S. Navathe, and J. Yates, eds. Norwell, Mass.: Kluwer Academic Publishers.
Bryant, R.E., and M.Y. Vardi. 2002. 2000–2001 Taulbee Survey: Hope for More Balance in Supply
and Demand. Computing Research News 14(2): 4–11.
Brynjolfson, E. 1991. Information Technology and the “Productivity Paradox”: What We Know and
What We Don’t Know. Cambridge, Mass.: MIT Sloan School of Management.
Bureau of Labor Statistics. 2001. Table 3b. Fastest Growing Occupations, 2000–10. Available online
at: http://www.bls.gov/news.release/ecopro.t06.htm. [June 24, 2003]
Card, S.K., T.P. Moran, and A. Newell. 1983. The Psychology of Human-Computer Interaction.
Hillsdale, N.J.: L. Erlbaum Assoc.
Cerf, V., and R. Kahn. 1974. A protocol for packet network intercommunication. IEEE Transactions
on Communications 22(5): 637–642.
Cohen, W.M., and D.A. Levinthal. 1990. Absorptive capacity: a new perspective on learning and
innovation. Administrative Science Quarterly 35(1): 128–152.
CTIA (Cellular Telecommunications and Internet Association). 2003. CTIA’s Semi-Annual Wire-
less Industry Survey. Available online at: http://www.wow-com.com/pdf/CTIA_Survey_
Yearend_2002.pdf. [June 24, 2003]

CTIS (Corporate Technology Information Services, Inc.). 1999. Spotlight. Woburn, Mass.: Corpo-
rate Technology Information Services, Inc.
Cusumano, M. 1991. Japan’s Software Factories: A Challenge to U.S. Management. Oxford, U.K.:
Oxford University Press.
de Sola Pool, I., ed. 1977. The Social Impact of the Telephone. Cambridge, Mass.: MIT Press.
Earl, M., ed. 1988. Information Management: The Strategic Dimension. Oxford, U.K.: Clarendon
Press.
Gilder, G. 1993. Metcalfe’s Law and Legacy. Forbes ASAP, September 13. Available online at:
http://www.seas.upenn.edu/~gaj1/metgg.html. [June 24, 2003]
Hill, S. 2001. Science and Engineering Degrees: 1966–1998. Arlington, Va.: Division of Science
Resources Studies, National Science Foundation.
Intel. 2003. Intel Research Network of Laboratories. Available online at: http://www.intel-research.net.
[June 24, 2003]
InternetWeek. 2001. Report: DSL Market to Rebound Next Year. October 24, 2001. Available online
at: http://www.internetweek.com/story/INW20011024S0005. [June 24, 2003]
Kahn, A. 1970. The Economics of Regulation, Vol. 1. New York: John Wiley & Sons.
Kahn, A. 1971. The Economics of Regulation, Vol. 2. New York: John Wiley & Sons.
Keen, P.G.W., and M.S. Scott Morton. 1978. Decision Support Systems: An Organizational Perspec-
tive. Reading, Mass.: Addison-Wesley.
Kraut, R., W. Scherlis, T. Mukhopadhyay, J. Manning, and S. Kiesler. 1996. The HomeNet field trial
of residential Internet services. Communications of the ACM 39(12): 55–63.
Lazowska, E. 1998. Remarks made during panel session on Changing the Interaction Between Aca-
demic Research and Industry: University, Industry, and Government Perspectives. Presented at
the workshop How Can Academic Research Best Contribute to Network Systems and Commu-
nications?, National Academy of Engineering, Washington, D.C., October 30, 1998.
Malone, T. 1998. Remarks made during panel session on Contributions and Impact of Academic
Research: Social, Management, and Policy Sciences. Presented at the workshop How Can
Academic Research Best Contribute to Network Systems and Communications?, National
Academy of Engineering, Washington, D.C., October 30, 1998.
Malone, T.W., K. Crowston, J. Lee, B. Pentland, C. Dellarocas, G. Wyner, J. Quimby, C.S. Osborn,
A. Bernstein, G. Herman, M. Klein, and E. O’Donnell. 1999. Tools for inventing organizations:
toward a handbook of organizational processes. Management Science 45(3): 425–443.
Markus, M.L. 1987. Toward a “critical mass” theory of interactive media: universal access, inter-
dependence, and diffusion. Communications Research 14: 491–511.
McKnight, L.W., and J.P. Bailey, eds. 1997. Internet Economics. Cambridge, Mass.: MIT Press.
Messerschmitt, D.G. 1996. The convergence of telecommunications and computing: what are the
implications today? IEEE Proceedings (August). Available online at: http://www.informatik.
tu-darmstadt.de/VS/Lehre/WS97-98/JavaTK/convergence.html. [June 24, 2003]
MICRO. 2002. University of California Microelectronics Innovation and Computer Research Oppor-
tunities. Available online at: http://www.ucop.edu/research/micro/. [June 24, 2003]
Moore, G.E. 1965. Cramming more components onto integrated circuits. Electronics 38(8): 114–
117. Available online at: http://www.cybraryn.com/news/documents/moore-paper.pdf. [June 24,
2003]
Morgan, R.P., and D.E. Strickland. 2000. U.S. university research contributions to industry: findings
and conjectures. Science and Public Policy 28(2): 113–121.
Morris, J. 1998. Remarks made during panel session on Changing the Interaction Between Academic
Research and Industry: University, Industry and Government Perspectives. Presented at the
workshop How Can Academic Research Best Contribute to Network Systems and Communica-
tions?, National Academy of Engineering, Washington, D.C., October 30, 1998

NRC (National Research Council). 1995. Evolving the High-Performance Computing and Commu-
nications Initiative to Support the Nation’s Information Infrastructure. Washington, D.C.: Na-
tional Academy Press.
NRC. 1996. Cryptography’s Role in Securing the Information Society. Washington, D.C.: National
Academy Press.
NRC. 1999. Funding a Revolution: Government Support for Computing Research. Washington,
D.C.: National Academy Press.
NRC. 2000. Making IT Better: Expanding Information Technology Research to Meet Society’s
Needs. Washington, D.C.: National Academy Press.
NRC. 2003. Innovation and Information Technology. Washington, D.C.: National Academies Press.
NSF (National Science Foundation). 2001. Survey of Federal Funds for Research and Development:
Fiscal Years 1999, 2000, and 2001. Arlington, Va.: National Science Foundation.
Orr, J. 1990. Sharing Knowledge, Celebrating Identity: War Stories and Community Memory in a
Service Culture. Pp. 169–189 in Collective Remembering: Memory in Society, D.S. Middleton
and D. Edwards, eds. Beverly Hills, Calif.: Sage Publications.
Parker, L. 1997. The Engineering Research Centers Program: An Assessment of Benefits and Out-
comes. Arlington, Va.: National Science Foundation.
Qualcomm Corporation. 1999. ERICSSON and QUALCOMM Reach Global CDMA Resolution.
Press release. March 25, 1999. Available online at: http://www.qualcomm.com/press/pr/
releases1999/press457.html. [June 24, 2003]
Rockart, J.F. 1981. The changing role of the information systems executive: a critical success factors
perspective. Sloan Management Review 22(2): 15–25.
Roessner, D., R. Carr, I. Feller, M. McGeary, and N. Newman. 1998. The Role of NSF’s Support of
Engineering in Enabling Technological Innovation: Final Report-Phase II. Arlington, Va.: SRI
International. Available online at: http://www.sri.com/policy/stp/techin2/. [June 24, 2003]
Schofield, J.W. 1995. Computers and Classroom Culture. New York: Cambridge University Press.
Shapiro, C., and H. Varian. 1998. Information Rules: A Strategic Guide to the Network Economy.
Cambridge, Mass.: Harvard Business School Press.
Siegel, J.L., V. Dubrovsky, S. Kiesler, and T. McGuire. 1986. Group Processes in Computer-
Mediated Communication. Organizational Behavior and Human Decision Processes 37:
157–187.
Sirbu, M. 1998. Remarks made during panel session on Contributions and Impact of Academic
Research: Social, Management, and Policy Sciences. Presented at the workshop How Can
Academic Research Best Contribute to Network Systems and Communications?, National
Academy of Engineering, Washington, D.C., October 30, 1998.
Sproull, L., and S. Kiesler. 1991. Connections: New Ways of Working in the Networked Organiza-
tion. Cambridge, Mass.: MIT Press.
SRI International. 1997. The Role of NSF’s Support of Engineering in Enabling Technological
Innovation. IV. The Internet. Available online at: http://www.sri.com/policy/stp/techin/
inter1.html. [June 24, 2003]
Stanford University Corporate Guide. 2001. Top 10 Stanford Inventions. Available online at: http://
corporate.stanford.edu/innovations/invent.html. [June 24, 2003]
Tennenhouse, D. 1998. Diagram presented during panel session on Changing the Interaction Be-
tween Academic Research and Industry: University, Industry and Government Perspectives at
the workshop How Can Academic Research Best Contribute to Network Systems and Commu-
nications?, National Academy of Engineering, Washington, D.C., October 30, 1998.
Turkle, S. 1984. The Second Self: Computers and the Human Spirit. New York: Simon & Schuster.
U.S. Bureau of the Census. 2002. Statistical Abstract of the United States–2002. Washington, D.C.:
U.S. Government Printing Office. Available online at: http://www.census.gov/prod/www/
statistical-abstract-02.html. [June 24, 2003]

71
ADDENDUM
E-Mail Questionnaire
The following questionnaire was sent to individuals selected from various
parts of the network systems and communication industry, some of whom at-
tended the October 1998 workshop. Included among the questionnaire respon-
dents were senior executives at AT&T Laboratories, Bell Atlantic, Bellcore,
MCI, and Motorola, and professors with expertise in computer science and engi-
neering, network systems, and telecommunications from Stanford University,
University of Delaware, University of California-Berkeley, University of Cali-
fornia, Los Angeles, University of Virginia, and University of Washington.
THE IMPACT OF ACADEMIC RESEARCH ON INDUSTRIAL

PERFORMANCE
NETWORK SYSTEMS AND COMMUNICATIONS PANEL
We invite your responses to the following questions. Your responses will be

used by our Panel as background information for our report. Any material used
verbatim will not be attributed to you without seeking your permission.
1. Could you describe briefly significant academic research contributions to

the network systems and communications industry? (If possible, please supply
references to published information that outlines the contributions.)
2. Overall, would you describe the impact of academic research on industrial

performance in the network systems and communications industry as (Please put
an X in one box):
□ 1. very large
□ 2. large
□ 3. medium
□ 4. small
□ 5. very small/non-existent
3. What is the role of academic research in educating people who work in

your industry? (Please focus on university research activities, rather than univer-
sity education generally.)
4. What structural forms of university-industry collaboration lead to good

results in your industry? An example of such a structure might be a discipline- or
industry-oriented “center” that solicits industry sponsors for a collection of

projects that span a varied research program. What seem to be the essential
determinants of success of such structures?
5. What are significant emerging trends or problems that the network sys-
tems and communications industry will face in the future that could benefit from
academic research?
6. What changes are required, if any, in academic research if it is to be

responsive to these industrial trends and problems?
7. What single step could be taken by universities to enhance the impact of

academic research on the industry?
8. What single step could be taken by companies to enhance the impact of

academic research on industry?
9. What single step could be taken by government to enhance the impact of

academic research on industry?
10. Do you see any downside to enhanced university-industry research col-

laboration? Things to be avoided?
11. Other comments? Any comments, pointers to other studies, or sugges-

tions would be appreciated.

WORKSHOP AGENDA
HOW CAN ACADEMIC RESEARCH BEST CONTRIBUTE TO

NETWORK SYSTEMS AND COMMUNICATIONS?
October 30, 1998
National Academies Building

2101 Constitution Avenue N.W.
Washington, D.C.
9:00 am Welcoming remarks and self-introductions

Wm. A. Wulf, President, National Academy of Engineering
9:15 am Overview of the work of the Network Systems and Communications

Panel and description of the wider NAE study
Bob Sproull, Panel Chair
10:00 am Break
10:15 am Session I. Contributions and impacts of academic research on

performance in the network systems and communications indus-
try: Engineering and the Physical Sciences
David Forney, Ambuj Goyal, Robert Kahn, H.T. Kung, David Mills
11:45 am Lunch in Meeting Room
12:30 pm Session II. Contributions and impacts of academic research on

performance in the network systems and communications indus-
try: Design, Social, Management, and Policy Sciences
Dan Atkins, Walter Bender, Robert Kraut, Tom Malone,
Marvin Sirbu
1:30 pm Session III. Structures for university-industry collaboration

James Flanagan, Stewart Personick, David Roessner,
Donald Strickland, Stephen Wolff
2:30 pm Break

2:45 pm Session IV. Changing the interaction between academic research

and industry: University, Industry, and Government Perspectives
Hamid Ahmadi, Ed Lazowska, James Morris, Rick Rashid,
George Strawn, David Tennenhouse
4:30 pm Discussion, conclusions and recommendations

Bob Sproull

WORKSHOP ATTENDEES
Robert Sproull, chair * Ambuj Goyal

Vice President and Sun Fellow IBM Corporation
Sun Microsystems, Inc. T.J. Watson Research Center
Hamid Ahmadi George H. Heilmeier *

AT&T Labs Chairman Emeritus
Bellcore
Alfred V. Aho *
Associate Research Vice President Robert Kahn
Communications Sciences Corporation for National Research
Research Division Initiatives
Lucent Technologies
Bell Labs Innovations Robert Kraut
Department of Social and Decision
Daniel Atkins Sciences
School of Information Carnegie Mellon University
University of Michigan
H.T. Kung
Walter Bender Department of Electrical Engineering
MIT Media Lab and Computer Science
Harvard University
John Cioffi *
Associate Professor Ed Lazowska
Department of Electrical Engineering Department of Computer Science and
Stanford University Engineering
University of Washington
David J. Farber *
Alfred Fitler Moore Professor of Tom Malone
Telecommunications Sloan School of Management
University of Pennsylvania Massachusetts Institute of Technology
James Flanagan David Mills

Center for Computer Aids Department of Electrical and
Rutgers University Computer Engineering
University of Delaware
G. David Forney, Jr.
Motorola, Inc.
*Panel member

James Morris George Strawn

School of Computer Science National Science Foundation
Carnegie Mellon University
Stewart Personick Donald E. Strickland

Drexel University Chair, Management Department
Southern Illinois University
Richard Rashid
Advanced Technical Research David Tennenhouse
Microsoft Defense Advanced Research Projects
Agency
J. David Roessner
School of Public Policy Stephen Wolff *
Georgia Institute of Technology Executive Director
Advanced Internet Initiatives Division
Jerrard Sheehan Cisco Systems, Inc.
National Research Council
Wm. A. Wulf
Marvin Sirbu President
Information Networking Institute National Academy of Engineering
Carnegie Mellon University
NAE Program Office Staff

Tom Weimer, Director
Proctor Reid, Associate Director
Nathan Kahl, Project Assistant
Robert Morgan, NAE Fellow and Senior Analyst
*Panel member

29-76.the Impact of Academic Research and Industrial Performance

Uploaded by

Copyright:

Available Formats

29-76.the Impact of Academic Research and Industrial Performance

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

29-76.the Impact of Academic Research and Industrial Performance

Uploaded by

Copyright:

Available Formats

THE NATIONAL ACADEMIES PRESS

This PDF is available at http://nap.edu/10805 SHARE

The Impact of Academic Research on Industrial Performance

264 pages | 6 x 9 | HARDBACK

FIND RELATED TITLES

National Academy of Engineering 2003. The Impact of Academic Research on

– Access to free PDF downloads of thousands of scientiﬁc reports

Copyright © National Academy of Sciences. All rights reserved.

Report of the Panel on the Network

The Panel on Network Systems and Communications, one of five panels

Copyright National Academy of Sciences. All rights reserved.

30 THE IMPACT OF ACADEMIC RESEARCH ON INDUSTRIAL PERFORMANCE

market via start-up companies; and some were incremental improvements to

DEFINITION OF THE INDUSTRY

Copyright National Academy of Sciences. All rights reserved.

NETWORK SYSTEMS AND COMMUNICATIONS INDUSTRY 31

The explosive growth of the Internet is the most visible manifestation of

Copyright National Academy of Sciences. All rights reserved.

32 THE IMPACT OF ACADEMIC RESEARCH ON INDUSTRIAL PERFORMANCE

Copyright National Academy of Sciences. All rights reserved.

NETWORK SYSTEMS AND COMMUNICATIONS INDUSTRY 33

structure as well. In a horizontal structure, numerous suppliers manufacture parts

Copyright National Academy of Sciences. All rights reserved.

34 THE IMPACT OF ACADEMIC RESEARCH ON INDUSTRIAL PERFORMANCE

by a single vendor are said to be proprietary. For example, the programming

Copyright National Academy of Sciences. All rights reserved.

NETWORK SYSTEMS AND COMMUNICATIONS INDUSTRY 35

Copyright National Academy of Sciences. All rights reserved.

36 THE IMPACT OF ACADEMIC RESEARCH ON INDUSTRIAL PERFORMANCE

For many businesses, vendors of materials, products, and services throughout

Copyright National Academy of Sciences. All rights reserved.

NETWORK SYSTEMS AND COMMUNICATIONS INDUSTRY 37

telephone wiring (Epigram). The technology-development activities of these com-

Copyright National Academy of Sciences. All rights reserved.

38 THE IMPACT OF ACADEMIC RESEARCH ON INDUSTRIAL PERFORMANCE

1965 1970 1975 1980 1985 1990 1995 2005

Spacewar (MIT), Trek (Rochester)

to World Wide Web

1965 1970 1975 1980 1985 1990 1995 2005

FIGURE 2-1 Examples of academic government-sponsored (and some industry-

Copyright National Academy of Sciences. All rights reserved.

NETWORK SYSTEMS AND COMMUNICATIONS INDUSTRY 39

1965 1970 1975 1980 1985 1990 1995 2005

1965 1970 1975 1980 1985 1990 1995 2005

Copyright National Academy of Sciences. All rights reserved.

40 THE IMPACT OF ACADEMIC RESEARCH ON INDUSTRIAL PERFORMANCE

from a research result to advanced development to product development to eco-

• Research has kept paying off over a long period.

Copyright National Academy of Sciences. All rights reserved.

NETWORK SYSTEMS AND COMMUNICATIONS INDUSTRY 41

1. Theoretical result 2. Student graduates to industry laboratory

3. Democratization phase, in which many

Source: Tennenhouse, 1998.

Copyright National Academy of Sciences. All rights reserved.

42 THE IMPACT OF ACADEMIC RESEARCH ON INDUSTRIAL PERFORMANCE

Copyright National Academy of Sciences. All rights reserved.

NETWORK SYSTEMS AND COMMUNICATIONS INDUSTRY 43

An “open” interface by definition is an interface controlled by a consensus of

Multiplier Effect of Infrastructure