pureheart@earthlink.net

Chapter 5

From Science Policy to Research Policy

Lewis M. Branscomb

Technology policy serves to stimulate both public and private
innovation. During the Cold War, when government interests in
technical superiority were paramount, public investments in research
formed the cornerstone of technology policy. Few questioned the
appropriateness of government support for applied or technological
research for national security. Today, it is commercial innovation in
service to the economy that increasingly drives technology policy.
Does government still need to support the development of new
technologies? If so, when, if ever, is it appropriate to fund research
in private commercial firms? What kinds of technical research are
deserving of a subsidy? In what institutional settings should such
research be performed? How should government decide?

These questions lie at the heart of the debate in Congress over the
Clinton administration's science and technology initiatives. In the
political debate, few quarrel with government support for basic
science, but technology is often seen as the province of commercial
firms. Thus arguments about appropriate government roles seek to draw
distinctions between science and technology. The discussion becomes
more tractable if, instead of debating the boundaries of "science and
technology policy," we address the requirements for "research policy,"
to create technical knowledge, and "innovation policy," to cover the
important subject of incentives for innovation.

This chapter advances the idea that one should distinguish between
research—understood as an activity aimed at creating and informing
scientific and technical choices for the use of many potential (and
often unknown) beneficiaries—and narrow problem-solving, understood as
the accomplishment of specific, obtainable objectives in the service
of an identified beneficiary.

This distinction leads to a simple policy proposition: The decision on
the appropriateness of federal funding of research should rest on the
identification of the expected beneficiaries of the work, not the
level of abstractness or the practicality of the work, or the motives
of the investigator in undertaking it. When the public is the primary
intended beneficiary, public investment is appropriate, providing the
work is done under highly creative, intellectually competitive
conditions and the results are widely diffused and appreciated.

In other cases, the intended beneficiaries should normally pay for the
work, and the results may be kept proprietary. The idea that those who
will primarily benefit should pay also applies to the government, when
it intends to buy the products made possible by the research, and when
national security may require that the results be kept secret. The
criteria for public investment in research are not directly related to
how "basic" or abstract the work is, nor to its likely utility, but
rather to its net public value. However, this does not necessarily
imply that publicly funded work must be placed in the public domain;
sometimes assigning proprietary rights to the performer may be the
most effective way of assuring both the production of value and its
eventual diffusion to the public at large.

Research, in this definition, may create both new understanding of
nature and new technical opportunities. Some of those opportunities
may be immediately accessible. Others may not mature for decades. But
in general both scientific research and technological research may be
contribute to public value. Thus basic technology research is a
natural companion of basic scientific research. Indeed the two are
often interdependent and even indistinguishable. Like science,
engineering and medical research also contribute to the public stock
of knowledge and skill.

Technological as well as scientific research aimed at building
national capability, creating new opportunities, and guiding
technological decisions should motivate the government's investment in
research. When science itself is the driver to create new
opportunities and new understanding, the research community should
have the primary voice in setting goals and priorities. When, on the
other hand, it is public needs that drive the research, the political
process, informed by research, provides the funding and the overall
goals. But in both cases, researchers need an environment that favors
risk-taking and allows them considerable latitude in setting research
strategies.

This public investment in intellectual infrastructure creates
knowledge, skills, and institutions. It responds to the intellectual
challenges both of unraveling the mysteries of nature and of imagining
the uses for the knowledge gained. Scientists, engineers, physicians,
and others participate in it. Universities, government-funded
laboratories, and firms all contribute to this kind of research, which
we call "basic technological research," a companion to the more widely
understood and accepted "basic scientific research" to which the U.S.
government has contributed so effectively over the years. (I discuss
just what we mean by "basic technological research" after addressing
some reasons for caution in such public investments.)

This is a better way of thinking about the world of research in which
the U.S. government plays such an important role. Americans believe in
the importance of research to keep the nation smart, strong, and
capable, and to provide the right kinds of incentives for a
competitive, private economy. A research policy that creates new
understanding of technology as well as new science is a key to
realizing that goal and resolving some of the political conflicts over
how public dollars should be spent.

This chapter begins by seeking terminology to discuss research policy
that better reflects the criteria that should be used in
distinguishing public from private responsibilities for investment. It
then addresses the market failures that call for government
incentives, and the government failures than may frustrate otherwise
justified public expenditures. Next I discuss the selection of
publicly-funded research performers and the conditions under which
basic scientific and technological research should be performed. I
conclude that the nature of research opportunities and needs will
determine the relative influence the political process and the
research community should have in setting the levels and priorities
for public investments in research.

(1) What is Research?

Many people lump together all kinds of technical activity into
"research and development." The shorthand "R&D" is often taken as if
it were one word, despite the fact that its two components are very
different kinds of activities. Scientific and technological research
are intensely intellectual and creative activities with uncertain
outcomes and risks, performed in laboratories where the researchers
have a lot of freedom to explore and learn. Development, by contrast,
is a highly focused activity aimed at producing a design or a process
that can be realized in a specified time using specifically allocated
resources. It is typically tightly managed to minimize risks and to
achieve neither more nor less than the specified objective.

When government officials gather statistics on different kinds of
technical activities in order to make policy on budgeting and managing
government R&D, distinctions as to the purpose and nature of the work
are needed and programs are divided into "basic research," "applied
research," and "development." At one end of the spectrum, basic
research is most often assumed to refer to scientific investigations,
most likely in a university or independent laboratory; at the other
end, development is assumed to be an activity of engineers, most often
in an industrial setting.

Any of these three categories of technical work may be aimed at
building society's knowledge and skill base, or creating products for
the government's use, or exploring new product or process
opportunities for a commercial market. But basic research depends
primarily on public funding, while development is almost always driven
by markets, private or public. Applied research is an ambiguous
category, usually defined as research of identifiable utility, placed
somewhere in the middle of the intellectual spectrum between
speculative theories of science at one end and predictable application
of well-verified knowledge on the other. Since most of the confusion
about the government's role in research funding concerns this gray
area of applied research, this category is not very useful for policy
makers. To call a research program "applied" is not enough to tell us
whether it is appropriate for government funding. Thus we must find a
set of categories for technical activities that better lend themselves
to the discussion of the government's role. This is the task of the
next section.

(2) Defining research

Scientific research embraces inquiry into the workings of nature
without regard to the motivation of the scientist or the investor in
the scientist's work. Within this conception of research lies all of
what is commonly called "basic" or "fundamental" research, plus much
of what some people choose to call "applied" (because it is likely to
be useful). "Research" does not normally encompass development,
testing, design, or product simulation. Research is an activity for
which the doctorate is often the appropriate training. It is carried
out primarily in laboratories managed for the purpose of conducting
scientific research and is funded by agencies or bureaus experienced
at research investment and management.

We all know that some basic research is highly abstract and
speculative, far from any kind of practical application or economic
value. No one is going to commercialize theories about

black holes any time soon. But must basic research be useless to
qualify as "basic"? Surely that would be an absurdity. Basic research
is best thought of as research to create knowledge that

expands human opportunities and understanding and informs human
choices. It may lead to a new scientific observation that raises new
questions. If black holes are found at the centers of galaxies,
including our own, what does that tell us about the ultimate fate of
our own solar system? Surely that is an important question, but only
experts will be able to see how the work might, some day, inform more
"practical" science.

Research may also lead to understanding that suggests a technological
possibility or informs a choice among alternative technologies. What
does science tell us about the chemical reactions in the earth's
stratosphere from which we might predict whether a fleet of supersonic
transports might deplete the protective ozone layer? President Nixon
asked his science advisor, Dr. E.E David, this question when
considering whether to ask the Senate to reconsider its negative vote
(by a margin of one) on the Supersonic Transport (SST). The
president's decision not to try to gain that extra vote turned, in
part, on quite basic questions of free-radical chemistry. It does not
take decades for basic research to have value when it informs
technological choices, as in this case.

Research might lead to the discovery of a new material, the
understanding of a new process, or the creation of an idea leading to
a new kind of instrument. If materials can be made that are offer no
electrical resistance at room temperature (i.e., are high-temperature
superconductors), could world demands for energy be greatly reduced in
the future? What other applications for electric current that flows
without resistance might we imagine? Such basic research may lead to
scientific or technological progress, or both.

Most often, scientific and technological research go hand in hand. A
scientist might invent a new kind of scientific instrument to explore
a poorly understood area of natural phenomena. Her colleagues build
similar instruments in their laboratories. One of them, perhaps a bit
more entrepreneurial than her fellows, decides to make a more reliable
version of the instrument, manufacture it, and sell it to other
scientists in the field. Soon this instrument is in widespread use for
****ysis, and someone, perhaps an engineer, realizes the instrument
can be used in reverse to control a process rather than measure it.
Thus an instrument designed for ****ysis becomes a tool for synthesis.

Consider, for example, the electron microscope. It was invented to
enable scientists to see very small things. It is now used in reverse
to make very small things, not only in the laboratory but in
electronics factories. In this example, science created the need for
the instrument. The resulting instrument business enabled more rapid
scientific progress. Soon the electron microscope instrument was used
in reverse as an electron-beam lithography tool for making tiny
structures on computer chips. The computers using these tiny chips are
faster and provide a more powerful tool for the advance of other
fields of science. In this example, it is very difficult to sort out
whether science was driving the technology or technology was driving
science: both were happening concurrently.

(2) Basic technological research

The justification for federal support of research that is
investigator-initiated (that is, not driven by sponsor-defined needs
or applications) is not restricted to science. Harvey Brooks points
out that "pure technology" may be as appropriate for public
investments as "pure science." There are many examples of successful
public investments in technology that preceded identification of
market-supported applications. Brooks cites as examples the
development by the Atomic Energy Commission of radio-isotopes and
stable-isotope tracers—now used for both diagnosis and treatment of
disease and for fundamental biological research—well before the
medical and biology communities had learned how to use them or defined
a need and a market. This investment enabled much of the molecular
biology revolution that followed.

The public investment in computer networking that led to the Internet
is a contemporary example. (See Chapter 13 by Brian Kahin.) When
Robert Kahn first developed the Internet Protocol for computer
networking at the Defense Research Projects Agency, no one imagined
that within a few years, billions of dollars would be invested in
information services offered through the World Wide Web.

Good examples of basic technological research can be found in academic
engineering research, such as might be funded in the National Science
Foundation (NSF) Engineering Research Centers, or in much of the
Department of Defense (DoD) budget category called "exploratory
development," such as the quest for efficient operating systems for
massively parallel computers. Other examples are the fusion energy
program in the Department of Energy (DOE) or the search for practical
materials that exhibit room temperature superconductivity. Within the
civilian technology programs, such as the Advanced Technology Program
(ATP) of the National Institute of Standards and Technology (NIST),
one also finds examples of high-risk science-based industrial
research, such the search for solid state lasers that radiate in the
ultra-violet, which would permit greater information storage capacity
on compact-disk-based media.

The phrase "basic technological research" is meant to direct our
attention to work that creates new capabilities as well as new
understanding, and is not simply focused on narrow problem-solving or
product development. Thus "basic technological research" takes its
place beside "basic scientific research," forming two arms of
intellectual investment into a vital capability of human society. The
appropriate public policies for investment in them are not based on
attempts to distinguish these activities by their intellectual
content, which is fortunate since they are highly interdependent and
overlapping. Rather, policies for resource allocation should derive
from a weighing of opportunities and needs, and from provisions for
diffusion and use of the information produced. The right public policy
for supporting research should not require rigid distinctions to be
made between basic scientific and basic technological research.

(2) Distinguishing research from narrow problem-solving

The criteria for public funding of basic technological research are
similar to those for basic science. Technological and scientific
research should be understood to be complementary and should receive
bipartisan support for the same reasons. Yet politicians are typically
more comfortable with advocating funding for scientific rather than
technological research, perhaps because the implicit utility of
technological research may suggest that there is—or might be—an
identifiable beneficiary who should be footing the bill. If there were
such a beneficiary, then the work, they might say, should be called
"applied research" and the government should keep hands off.

The culprit, as noted above, is the ambiguous phrase "applied
research." Much of the basic and exploratory research funded by the
Departments of Defense and Energy and by the National Aeronautics and
Space Administration (NASA) is neither specific to government
procurement nor commercially proprietary; rather it is devoted to
enhancing the U.S. capability to innovate broadly. But in government
statistics such work may be labeled "applied research" if the research
is said to be working toward a well-defined, utilitarian objective.
What matters for public policy, however, is the objective of the
investor, and its expectations of return from the investment. The fact
that work may have a useful application does not tell us whether the
government should fund it. The opposite syllogism, surely, we must
reject; government should not decline to fund research simply because
it might have a practical application.

The usual distinctions between "basic research," "applied research,"
and "development," used for many years in the formal government
statistics kept by the National Science Foundation are, unfortunately,
insufficient for discussions of policy for government investment in
technical activities. Indeed, definitions are the source of much of
the confusion over the appropriate role for government in the national
scientific and technical enterprise.

One cannot distinguish in any meaningful way "basic" from "applied
research" by observing what a scientist is doing. A scientist engaged
in testing a steel pipe for leaks—a rather routine "applied" task—may
insist his work is "basic," because his leak-free pipe might allow a
measurement of the second-order Doppler shift predicted by relativity
theory. Another scientist working on extensions to the quantum theory
of collisions of electrons with atoms—a highly sophisticated and
apparently abstract activity—may say she is engaged in "applied
research," because the use of her theory to predict collision cross
sections might be of practical assistance to fusion energy
engineering.

"Applied research" should not be used to mean "purposeful and
demonstrably useful basic research," and one should be wary of the use
of the term in government statistics. In corporate research
laboratories, such as the T.J. Watson Research Laboratories of IBM,
all of the work is referred to simply as "research." There is no need
to attempt a distinction between "basic" and "applied" research. All
of the company's research investments are motivated by corporate
interests. All of the research has a purpose. All of it is conducted
under highly creative conditions. None of it is so "pure" that there
are no expectations of value from the research investment.

We should reserve the words "applied research" for those narrowly
defined tasks in which limited time and resources are devoted to a
specific problem for an identified user who gets all the benefit and
should pay all the costs. To make this view of applied research clear
in this discussion, I use the words "problem-solving research"
instead.

Narrow problem-solving and development are activities initiated by
someone who wishes to apply research methods purposefully to exploit
an identified opportunity or solve a problem. They involve the
application of technical resources to achieve an identified goal for a
specified beneficiary, usually the investor in the work. It is a
reasonable assumption that those who engage in such activities expect
to benefit from them, and to benefit by a sufficient margin over the
cost to accommodate the technical risk that is ever-present in
research. The investor in problem-solving may be a government agency,
but is more likely to be a private firm. In most cases that firm would
be expected to be able to appropriate sufficient benefits to need no
government subsidy to take those risks.

Public investment in the creation of new technology (technological
development, whether by research or as a product of problem-solving)
is a critical link between societal goals and the scientific research
that is pursued by virtue of society's commitment to those goals. Thus
the desire for technology is an important—perhaps the most
important—source of demand for science. The way scientific research is
used to further technological goals may profoundly affect policies for
allocating funds to science and determining the institutional settings
in which scientific research is performed. In fact, the way
innovations are brought about in industry, and the role of science in
support of innovation and productivity growth, have both substantially
changed. Thus, any discussion of technology policy must address
research policy as well.

(2) The Search for Useful Language in the Public Debate

These questions of definition may seem highly academic. But they lie
at the heart of public policy debates about technology policy, not
only because science is both a source and a product of technology, but
because the boundaries between research that leads to new technical
knowledge and research that leads to scientific understanding are
obscure and often misunderstood. Before one can create a policy for
public investment in research, one must know more about the goals of
the work, who its intended beneficiaries might be, and how these
results might reach those who can use them beneficially. These are the
attributes that should determine the role of government in funding
technical work, not the narrow distinctions between science and
technology.

As Neal Lane, director of the National Science Foundation (NSF) said
in testimony to Congress:

To my mind, the question is not, where the dividing lines are between
science and technology, or between basic and applied research, but
rather how do we take better advantage of the interrelationships in
order for the nation to reap the full benefits of its integrated
investment in science and technology?

Lane quoted Donald E. Stokes:

The annals of research so often record scientific advances
simultaneously driven by the quest for both understanding and use,
that we are increasingly led to ask how it came to be so widely
believed that these goals are inevitably in tension and that the
categories of basic and applied science are radically separated.

Conservatives in Congress are searching for the right language through
which to express their support for what they understand to be basic
research, while making clear their objections to public funding of
private goods. Congressman F. James Sensenbrenner, chairman of the
House Committee on Science, made this distinction in explaining the
Committee's report:

Federal R&D should focus on essential programs that are long term,
high risk, non-commercial, cutting edge, well-managed and have great
potential for scientific discovery. Funding for programs that do not
meet this standard should be eliminated or decreased to enable new
initiatives.

To make clear what the Committee majority does not like, Sensenbrenner
added:

Beyond the demonstration of technical feasibility, activities
associated with evolutionary advances or incremental improvements to a
product or a process, or the marketing or commercialization of a
product or process, should be left to the private sector.

Representative Sensenbrenner's views as expressed here probably do not
conflict with the consensus in both the technical and the political
communities. But it should not require six qualifying adjectives to
describe the research the government should support, and it
complicates the issue to try to restrict approval to scientific
discovery. Sensenbrenner's apparent restriction to scientific
discovery implies that he would not be equally enthusiastic about
technological discovery, even if it were "long term, high risk,
non-commercial, cutting edge, [and] well-managed." A simpler way to
distinguish appropriate opportunities for public funding from those
better left to the private sector is to focus on the intended
beneficiaries of the research.

(2) Identifying the Intended Beneficiaries of Publicly Funded Research

The rule is simple: let the primary intended beneficiary pay for the
research. The careful cir***scriptions of the precise kinds of
research that government should and should not support, described
above, are a way of implementing this principle by describing what
kind of research the speaker believes best serves the public interest.

Research to serve a firm's commercial interest will be recouped in
profits from that commercialization; no government funds should be
employed. The company pays. When the government makes the market (as
in defense procurement), the government pays. When the government
invests in the nation's skills and knowledge, going far beyond the
private investments justified by market rewards, the people benefit,
and the people's government pays. And, where firms under-invest in
relation to a defined public interest, such as reducing environmental
risk or accelerating medical progress, government and the private
sector may share the costs. As discussed in the final chapter of this
book (Chapter 18 by Lewis Branscomb and James Keller), the
cost-sharing ratio should reflect the best understanding of the likely
distribution of public and private benefits.

We have argued that public funds should be invested when the public
interest outweighs private gain, and that basic technological research
can contribute as much as basic science to national capability and
need. But that leaves a number of questions still to be answered about
the nation's research policy: What provisions should be made for
insuring that research outcomes reach intended beneficiaries? Who
should do the research? How much autonomy should be accorded
investigators in universities, national laboratories, or independent
laboratories in order to ensure a creative environment? Who sets the
priorities for different research programs? What motivations should
drive allocation of resources to different research objectives? These
will be subject of the remainder of the chapter.

(1) Conditions for gaining value from research

If research, whether in the more abstract science disciplines or in
the more "practical" fields of science and engineering, is to provide
public value, it must be conducted under conditions that ensure a high
level of creativity, accountability to sponsors of the work, and
effective access by potential users.

(2) A creative environment for research

Research is most fruitful when pursued by highly trained people who
are accorded freedom to decide what are the most important scientific
questions and how to pursue them. When engaged in research, scientists
need latitude to set research strategies and need the feedback that
comes from exposure of their work to the praise and criticism of their
peers. Development and narrow, task-oriented problem-solving also
benefit from a creative environment, but the ******** nature of the
goals and the time pressure to reach them substantially reduces the
freedom to shift directions in response to unexpected opportunities
for discovery. Thus the cir***stances under which the work is
performed help to distinguish "basic scientific and technological
research" from "problem-solving," "testing," and "development."

Researchers insist that the need for academic autonomy, balanced with
accountability, is indeed legitimate when the work is research and the
goal is to maximize learning. A government agency may lay out its view
of the areas of research it believes most fruitful and in which it is
prepared to invest. A policy of responding to unsolicited proposals
from scientists with good ideas in response to the agency's challenge
creates a wealth of valuable research opportunities. Selecting from
among these ideas by submitting them to the critical review of other
scientists knowledgeable in the same field and by technically expert
users who can assess the likely value of the work sustains the quality
of the work, its likely value to society, and the fairness of the
allocation of government funds. The peer review process also expands
awareness of the work and promotes its subsequent diffusion. The
primary diffusion mechanisms are horizontal, to others in the same or
nearby fields. The employment of recent graduates trained in research,
the secondary literature, and various conferences and study groups are
relied upon to diffuse the work to practitioners such as engineers,
clinicians, and scientists in "downstream" disciplines.

This system of selection of projects and of performers is strongly
defended by the scientific community, and for good reasons. One of
these arguments seeks to tie the need for autonomy to the
unpredictable nature of science in pursuit of understanding and new
possibilities. Scientists fear that Congressional pressure for
immediately useful results from publicly supported research may lead
to a loss of the conditions necessary for creative work. If research
is treated like problem-solving or like development, they fear,
micromanagement and unrealistic expectations for quick results cannot
be far behind. Three examples illustrate the basis for this fear.

In the early 1970s the government tried to respond to political
concerns that government research contributed too little to social
well-being by creating a program at NSF called Research Applied to
National Needs (RANN). This program addressed concerns such as fire
research, and both social and natural scientists were expected to
participate. As the name implies, this was NSF's attempt to induce
scientists accustomed to opportunity-based research to redirect their
attention to what the government viewed as need-driven. In point of
fact, some the research conducted under RANN received generally high
marks from independent evaluators. However, scientists remained
concerned that NSF should be undertaking need-driven research at all.
This was a legitimate concern. It is far from obvious that NSF is best
equipped to set such priorities, but no other agency seemed to be
available to do so.

A second example is the "war on cancer," which threatened to divert
biomedical research from a fundamental attack on molecular biology and
immunology in the quest for a "quick fix" solution to cancer (a fear
that only an extremely generous Congress and tenacious National
Institutes of Health (NIH) management averted). Similar concerns
emerged when AIDS research was given a special priority by the
Congress.

The third example arose in 1992 when Walter Massey, then director of
the NSF, requested that the National Science Board (NSB) establish a
"Blue Ribbon Commission on the Future of NSF" to examine whether NSF
should attempt to contribute to the government's effort to enhance the
competitiveness of American industry. Pessimists jumped to the
conclusion that such a mission priority would lead to the displacement
of individual-investigator, opportunity-driven research by political
perceptions of industrial need. A second fear was that this preference
for need-driven research (in this case responding to industrial needs)
at NSF might result in government officials selecting investigators
without expert peer review, and evaluating both outputs and outcomes
of the research. The Commission concluded, however, that scientific
autonomy—based on peer review evaluation of competing, unsolicited
proposals—can and should be preserved, even as NSF seeks to balance
both intrinsic and extrinsic values by appropriate priority setting
for its fields and areas of research.

Neal Lane, director of the NSF, is careful to avoid this trap. In his
1995 testimony, he said:

NSF support of research focuses almost exclusively on answers to
fundamental questions that defy our ability to predict the outcomes.
Still, it is important to recognize that taxpayer-funded fundamental
research can and should have a conscious relationship to the nation's
priorities and societal needs. This does not mean a narrowly directed
agenda of targeted research, but rather, a program of fundamental
science and engineering [research] that clearly is in and for the
national interest, in its most comprehensive interpretation.

In all of these cases it was research autonomy that was at stake. It
was not an argument over the importance of the social goals to which
research contributes so much. The concern of the researchers had been
elevated by the tendency of politicians to imply that research of high
public value should be managed differently from more conceptual or
theoretical work. When the work might be of economic value, the
concern escalates to the fear that politicians will conclude that such
useful work could and should also be paid for from non-governmental
sources. Where public goals are driving research investments, the
agencies do feel accountable for the ultimate delivery of public
benefits, but they must not let this lead them to micromanage the
creative research on which they are depending.

(2) Ensuring user access to research results

The linkages between outputs from research and outcomes for society
are often ill-defined (see Chapter 3 by Adam Jaffe on metrics) and
operate outside the province of government control. This does not free
government from the obligation to understand the processes that create
public value from its research investments. When government is
investing in research to build national capability, the extent of
returns to public value is strongly influenced by the ease with which
users can access the results and put them to effective use. Research
environments that foster creativity also offer effective mechanisms of
information and skill diffusion.

When basic technological research is performed in universities,
students, project referees, academic visitors, and collaborating
companies all contribute to the diffusion of new knowledge, adding
greatly to the effectiveness of formal publication. The government
provides special institutional mechanisms to foster information
diffusion; among these are Cooperative Research and Development
Agreements (CRADAs; see Chapter 9 by David Guston); government support
for University-Industry Research Centers; and funding of post-doctoral
research fellowships. When basic technology research is funded in
industry, the use of consortia of firms, perhaps in collaboration with
universities and state technology programs, provides an effective
diffusion mechanism (see Chapter 18). Thus even where the most
immediate public benefit from research might be the creation of jobs
resulting from commercialization of the ideas, the employment of
consortia is a way to increase the ratio of public to private
benefits.

(2) When should private firms be funded to perform research?

The selection of performers of research should be based on competence,
taking into account the productivity of the work, the skills and
experience that will be developed, and the effectiveness of the
diffusion of research outputs. If one is persuaded that government
should fund technological as well as scientific research, the next
question is, When, if ever, should private firms be funded to perform
the work? Here policy-makers confront a dilemma. If one is to avoid
politically sensitive choices among competing firms, the safe way out
is to fund research only in government laboratories, universities, and
other not-for-profit institutions. But avoiding this Scylla of choice
delivers one to the Charybdis of needlessly isolating the research
from its ultimate users in private industry. If new research is to be
quickly and widely accessible, it should be conducted in industry
laboratories or in institutions with effective links to industry. The
most effective mechanism to foster commercialization of new ideas is
to have the ideas arise inside industry itself. For these reasons we
conclude it would be a serious mistake to categorically exclude
industrial organizations from eligibility to perform basic scientific
and technological research for the government. However, we support the
NSF policy of giving its priority to universities, while the research
programs of the "mission" agencies should cover the broader spectrum
of research institutions, both public and private.

(1) Allocating resources to publicly funded research

The growing budget pressure on public funding of scientific research
exacerbates tensions that have accompanied the public funding of
scientific research for a long time. Going all the way back to the
writings of Francis Bacon, policy makers have struggled with the
balance between resource allocation strategies supportive of
scientific autonomy and those derived from identified public goals and
values. This was the subject of major academic debates in the 1920s,
with the protagonists represented by J.D. Bernal and Michael Polyani.
As Harvey Brooks has observed, research motivations fall generally
into two categories: opportunity-driven research—pursuing the visions
of scientists, and need-driven research—responding to the needs of
society. How should these two sources of motivation for the government
research investor be balanced? To what extent can—or should—scientific
research investments be based on government constructed plans, and to
what extent should public investors rely on the intrinsic values of
scientific research to ensure outcomes of maximum benefit to society?

(2) Research Motivation: The Clinton-Gore Science Policy

In August of 1994 the Clinton-Gore administration issued its
long-awaited "science policy," a companion to the technology policy
declaration that appeared almost instantly after the 1993
inauguration. The science policy was issued in a well-illustrated
paper entitled Science in the National Interest (referred to as SNI).
SNI makes the case that there are two valid criteria that should be
invoked in allocating resources for science, one involving
centralized, goal-driven decisions, the other aimed at creating a
strong scientific infrastructure on which all goal-oriented research
can draw. Investments in this infrastructure, SNI says, should be
based on the intrinsic values of science, reflecting the opportunities
for conceptual progress identified by scientists.

SNI drives home the importance of opportunism in basic science. It
says:

It has seldom proved possible to anticipate which areas of science
will bring forward surprising and important breakthroughs at any given
time. Therefore U.S. scientists must be among those working at the
leading edge in all major fields in order for us to retain and improve
our competitive position in the long term.… [N]ature yields her most
precious secrets in surprising ways, to those who are well prepared
and persistent, and with a schedule not often amenable to detailed
planning. Thus although we can and must do more to identify and
coordinate research thrusts aimed at strategic goals, we must not
limit our future by restricting the range of our inquiry. Vibrant
scientific disciplines are best guaranteed by the initiatives of
talented investigators and in turn provide the strongest and most
enduring foundation for science in the national interest.

This very sensible vision correctly associates creation of a strong
intellectual base for society's future use (policy for science) with
the need to ensure that science serves the goals to which it can make
decisive contributions (science for policy). It does not spell out,
however, how the two criteria for choice—intrinsic scientific merit
and extrinsic social utility—will be placed in balance. How much of
each do we need, and who decides?

(2) Need-driven research

The source of priority evaluation for need-driven exploratory research
may be quite different from that for opportunity-oriented science,
even when the motivations of the researchers are the same. Whereas
opportunity-oriented research is proposed and evaluated in a
competitive horizontal environment (peer review being the dominant
mechanism), need-driven research derives its priority from vertical
relationships. The need is typically expressed in terms of some
capability to be realized through a technology, which in turn may
derive its conceptual structure and future evolution from new science.
The level of priority derives from the initial goal, but, as a
practical matter, is derivative of the congressionally-authorized
investment in the technology to which the science contributes. Thus,
for example, the importance of new sources of energy and the
likelihood that fusion can provide a welcome solution should determine
the scale of investment in fusion. The technical agenda for fusion
research will determine what fields of science are relevant, and will
suggest where research might offer new options that can increase the
likelihood of success.

DOE's criteria for deciding what kinds of plasma physics research to
pursue are not the same as might be assigned to an opportunity-driven
NSF project in, let us say, the plasma phenomena in formation of
stars. Pertinence to fusion technology is clearly relevant for the
DOE; not all forms of serendipity are equally welcome. Thus, equally
competent investigators working on the same problem will find their
work evaluated differently by DOE and by NSF—and, indeed, quite
differently again if funded by a private commercial firm.

The fact that science is almost always embedded in a fabric of other
technical activities—both in space and in time—means that allocation
decisions also follow those chains of consequence and pertinence. A
given research project in plasma physics might find itself justified
by a commercial and public interest in new energy sources, by an
academic curiosity about how stars form, or by the government's
concern about the effects of atmospheric re-entry on space craft or
military missiles. Which of these justifications dominates the public
investment decision does say a lot about the amount of money to be
invested and how the overall program of research is directed. But for
this research to reveal the richness of potential options, it must be
carried out in much the same style as one would pursue
opportunity-oriented science. Thus the scientists performing it will
describe their work as basic research, if their work environment is
similar to that expected for opportunity-oriented science.

If all exploratory research, whether need-driven or opportunity
driven, requires similar staffing skills and research environments,
and if the sources of judgment about resource allocation are distinct
but understood for each, why is it necessary to ask whether the
research activities look more "scientific" or more "technological"? It
is the motive of the investor that matters, not that of the
investigator. The answer is that there remains one other dimension of
policy to be addressed. How does government decide how to allocate its
resources among the different kinds of scientific research goals?

One answer to that question we have already given: government should
sponsor a level of need-driven exploratory research appropriate to
support the missions that created the need. It is time-honored U.S.
policy that every federal agency should invest in basic scientific
research in proportion to the agency's dependence on the skills and
knowledge of the relevant scientific field. This policy has provided a
level of diversity of sources and perspectives in science that have
greatly enriched the U.S. scientific enterprise.

(2) Opportunity-driven research

But what about opportunity-driven research? How far does the scope of
our conception of research driven by intrinsic scientific interest
take us toward utility? The justification for government support of
"basic research" may be at a maximum when the likelihood that basic
concepts may be altered or extended is a maximum: when the "laws" of
science are created or repealed. But must its level of applicability
to more practical matters also be at a minimum to satisfy the
requirements for public support?

Is high energy physics "better" science than materials science and
engineering? What about a project to characterize the properties of a
new material, or research on a new idea for a scientific instrument,
or measurements of the thermodynamic properties of a new polymer? This
kind of research may be both opportunity-driven and need-driven. It
may be very interesting science while also utilitarian. It usually
requires the kinds of people and environments that are characteristic
of "basic" research or exploratory science.

This is the kind of research the National Institute for Standards and
Technology and its predecessor agency, the National Bureau of
Standards, have done for decades to support the productivity growth of
U.S. industry. Such research is deserving of public support for the
same reason that other areas of need-driven research are supported,
except that in this case the customers for the results are widely
dispersed, like the customers for other "basic" research.

Such "basic and useful" work falls into the category that has been
called "infrastructural," and has a counterpart in "infrastructural
technology." Unhappily for clean policy distinctions, but happily for
the health of science, there is no discontinuity between intrinsic and
extrinsic values in science. One cannot sort out the fields of science
on a line, with the most prestigious at one end and the most
utilitarian at the other. Quite often the best scientists working in a
utilitarian field make a remarkable scientific discovery. For example,
work on the electrical properties of materials at low temperature,
driven by the search for new electronics technologies for the computer
industry, led to the discovery of high temperature superconductivity,
one of the most startling events in physics in this century. It
revitalized a field of physics that had become largely moribund
because it had been prematurely assigned the "useful but not so
interesting" label. Furthermore the most sophisticated science
occasionally creates byproducts that are quite utilitarian; witness
the evolution of the storage ring of high-energy physics into the
x-ray lithography tool of the integrated circuit manufacturer.

How should decisions about investment in such work be made? Figure 5-1
suggests that research draws on both understanding and technology, and
contributes both to understanding and to improved technology. Both
outcomes are appropriate motivations for public investment in
research. There must be two elements of motivation in research: the
investor must be motivated to take financial risks, and the
investigator must be motivate to take professional risks. These risks
are generally higher in opportunity-driven research, but so too may be
the rewards, both to the research and to society. A reasonable public
policy involves a balance of risk and reward.

[Figure 5-1 about here]

(1) Conclusions and Recommendations

This leads us to our conclusions: First, the correct criterion for the
appropriateness of federal investments in research is the expectation
that the primary beneficiary will be the public, that is, the national
interest. Where government makes the market (either through
procurement, or in some cases through regulation), it may be
appropriate for government to fund not only research but development
as well, and to manage the projects to specific objectives, costs, and
schedules. Where government is investing in the knowledge base,
whether in response to learning and discovery opportunities or in
response to identified public needs, government should invest in
research performed under conditions appropriate for high productivity
and creativity.

Second, research is not "pure" or "basic" because no uses for the
results are expected. Instead, it is research because the knowledge
gained is to a significant degree unpredictable and serendipitous and
is expected to be widely diffused and therefore broadly beneficial.
The primary distinction between research and narrow problem-solving is
not found in the level of intellectual sophistication or in the level
of utility of the work, but rather in the prior identification of the
beneficiaries of the work.

Third, basic technology research is intimately related to basic
scientific research and should receive resources and be assigned to
performers using similar criteria to those used for basic science.
Creative conditions of work are just as necessary for creating new
technologies as for new science.

Fourth, resource allocation decisions for need-driven research must be
made by the funding agencies based on their legislative mandates.
Agencies authorized to address specific problem areas (such as energy,
health, space, or defense) should further the nation's capabilities to
address those problems by funding basic scientific and technological
research in the relevant technical areas. The level of research
investment should reflect the priority accorded to the mission
objective by the political process and the opportunity that research
offers for enabling mission success.

Fifth, resource allocation for opportunity-driven research should be
based on professional assessment of the likelihood that success will
create new and important intellectual as well as practical
opportunities. The magnitude of investment should be proportionate to
the need for research training in the universities and to the demand
for research progress reflected in technological development
commitments, both public and private.

Finally, the criteria for investing in basic technology research, like
that for science, must be originality, intellectual rigor, and
practical value. Like science, if technology research is to be
creative, it must not be micromanaged by government.

If the consensus behind the federal support of basic scientific
research is extended to basic technological research, and it is
understood that the federal government subsidizes development of
products and services only when these outputs are required to fulfill
federal missions such as defense, health, and environment, then it
should be relatively easy to come to a general understanding about
federal support for basic research that is relevant to commercial as
well as public purposes.

What is the practical effect of these conclusions? Before 1980 there
were frequent complaints from the engineering professions that NSF
treated engineering research (or technological research) as a
lower-priority activity than the "hard" sciences. This debate even
found expression in Congressman George Brown's threat (as chairman of
the House Science and Technology Committee) to create a National
Technology Foundation, which would have competed with NSF for funds
and attention. This proposal was dropped and instead the Congress
revised the NSF enabling statute to add the words "and engineering"
everywhere the word "science" appeared. The National Science Board
removed the words "applied research" from the Engineering and Applied
Research Division, thus making it clear that NSF does not engage in
narrow problem-solving research but does regard engineering research
as parallel to chemistry and oceanography, and not simply a branch of
applied research. Today, it is fair to say that the Science Board's
policy is to view research in its broad sweep, embracing fields as
different as mathematics, chemical engineering, and econometrics,
without intended intellectual prejudice.

What is required now is a more robust process for justifying budgets
and allocating resources over the full range of opportunity-driven and
need-driven criteria for investment. With this in mind, the NSB should
consider reinstituting, perhaps in improved form, the "COSEPUP"
studies chartered by the Council of the National Academy of Sciences
and the Council of the Academy of Engineering and funded by the NSF.
Performed by teams of research scientists within each discipline, they
were intended to map out the most fruitful lines of research in the
next five to ten years and to inform the research investments of
federal agencies. These studies typically covered the full range of
criteria for investment, and the experts themselves set priorities.
Two new features might be valuable. First, a stronger effort to engage
the field sciences, engineering, and clinical communities in
identifying need-driven priorities that might pay off in better
balance in the overall NSF program. Second, selected interdisciplinary
subjects should be systematically studied. Such studies might also
explore the usefulness of technology roadmaps, as discussed in Chapter
18.

It is our hope that this discussion will prove most useful to the
committees of the Congress that have struggled so long and hard to
communicate their policy objectives to the public and to the agencies
and—most important—to a nervous and sometimes defensive science and
engineering community. If Congress can get comfortable with the
support of research, without trying to deconstruct its intellectual
content, while fulfilling the full Congressional responsibility to
address the motives of its investments in research for the nation's
future, most of the rancor can be eliminated and a bi-partisan
research policy can be realized.

What the nation needs is not a science policy and a companion
technology policy, but rather a research policy to support a
research-based innovation policy.

http://www.ksg.harvard.edu/iip/techproj/chapter5.htm