Dr. Rita R. Colwell
Director
National Science Foundation
CRA-sponsored Computing Leadership Summit
The leaders of ACM, IEEE Computer Society, SIAM, AAAI,
USENIX, and CRA
February 11, 2002
Good morning. I'm genuinely delighted to be here today.
Not because I pretend to a sophisticated knowledge
of advanced computational systems.
In fact, until a few years ago, I wouldn't have known
"vector" and "scalar" from Laurel & Hardy. (Barnes
& Noble.) And my experience with "massively parallel"
architectures was in trying to deal with the House
and the Senate at the same time.
No, the reason I'm happy to be here is that I believe
we share a common vision for the future of 21st
century science.
We are now in an era where advanced computational systems
assist researchers and make complex problems more
tractable -- but more than that, they create unprecedented
opportunities and make new kinds of science possible.
One might even say, they create modern science, as
we know it today and see in the future.
This is increasingly true in visualization, in sharing
and mining of enormous data sets, in novel collaborations,
and in distributed research among research teams that
knew little or nothing about each other were it not
for the common ground afforded by revolutionary technology.astronomers
and sociologists, for example.
These interactions will grow with the advent of "terascale"
and GRID computing -- with data moving at around 40
gigabytes per second, hundreds of thousands of times
faster than it did even 15 years ago -- and its inevitable
expansion throughout the United States and the world.
I marvel at where we are when I think back to that
little old IBM 650 in the renovated attic computer
laboratory in the chemistry building at the University
of Washington in the 1960's.
Jim Foley has informed me that the objective of this
Summit is "to explore how the CRA member associations
can work together on issues that are of common concern
to the field of computing and to learn more about
policy issues that affect them."
He also asked me to speak about "the current and future
status of information technology research and development
at NSF, and NSF's participation in the overall federal
IT R&D effort."
There cannot be any doubt in anybody's mind that IT
holds enormous promise for S&E, as well as for society
at large. It will have profound effects on the sociology
of science.not to mention the social, behavioral and
economic sciences themselves.
Remote access to vast data sets, the ability to form
truly global collaborations of researchers, and the
elimination of barriers between disciplines will empower
the smaller, less-well-endowed institutions, boost
innovation and allow extraordinary partnerships in
which separate teams can combine resources and parcel
out parts of a problem too large for any single entity
to solve alone.
But this promise will not be reached quickly without
an expanded program of fundamental, long-term IT research.
NSF has taken on an expanded role in this important
arena.
NSF's total budget has doubled since 1990 (a rate of
growth that, in my view, is too slow). In contrast,
the CISE budget has just about doubled since 1999,
barely 2 ½ years. As I'm sure you know, I'm working
hard to double NSF's current budget (Newt Gingrich
says it should be tripled!) and CISE must participate
fully in future budget increases.
IT underpins all of S&E, in what might be called computational
S&E, taking its place right beside traditional experimentation
and theory. Whether or not that turns out to be the
most appropriate description of its importance, I
do not doubt that IT underlies the most important
new class of tools and services for carrying out S&E
that we have ever witnessed before.
All of science and engineering is in the process of
incorporating IT tools and concepts. But IT is far
from a finished structure: that is why progress in
S&E is, to a very real extent, gated by progress in
IT R&D.
That's why I'm committed to supporting continued increases
for NSF's IT R&D activities. And I must say that I'm
delighted that "CRA's own" Peter Freeman has agreed
to come to NSF to lead these activities. Peter will
be joining us on May 6 and I'm confident he will bring
vision and energy to take us on the next leap into
the future.
Let me speak briefly about some of NSF's current IT
R&D activities.
As I'm sure everyone of you is aware that NSF embarked,
in fiscal year 2000, on an NSF-wide research priority
that we called "Information Technology Research" (ITR).
Under the leadership of the CISE directorate, this
important priority area has stimulated single investigator
research and started a major effort in multi-investigator
research involving computer scientists and other scientists
and engineers.
Moreover, a major supercomputing acquisition called
"terascale computing facilities" was launched. An
award to the Pittsburgh Supercomputing Center in 2000.
The Center has successfully installed a 6-teraflop
peak performance system.
More recently, an award was made to the University
of Illinois and the University of California at San
Diego for a "distributed terascale facility" called
TeraGrid. The TeraGrid is currently under construction.
We anticipate it will come into service next year.
Future NSF opportunities for IT R&D
The ITR initiative is a five-year program and we are
currently in its fourth year. We are gathering input
research priorities for follow-on programs. Clearly,
cybersecurity will play a big role in that future,
as will other advanced uses of IT for homeland defense.
Congress is aware of those issues, and is already responding.
Last week, the House of Representatives passed H.R.
3394, the Cyber Security Research and Development
Act, by an overwhelming margin of 400 to 12.
This bill, introduced by Science Committee chairman
Sherwood Boehlert, authorizes $880 million over five
years for research into computer security, funded
through the National Science Foundation and the National
Institute of Standards and Technology. Action is expected
soon in the Senate, and related bills are moving in
both houses.
In addition, sensor systems for bioterrorism threats,
smart materials in buildings and elsewhere, and data
mining of huge data sets all require additional research
before they can be put into service for defense of
our country.
In a discussion with me recently, a noted computer
scientist suggested that "we need a major initiative
to redesign the computer, with security as a design
component rather than a later add-on".
And at a conference last month on post-Moore's law
computing, another noted computer scientist referred
to needed advances in high-end computing by saying
that "we don't need to reinvent the transistor [yet];
we need to reinvent the computer." Perhaps these research
dimensions will be added to PITAC's call for research
in programming, scalable information systems, high-end
computing, and the socio-economic consequences of
the IT revolution.
In addition to interdisciplinary research involving
computer scientists, the development of advanced IT
tools and services for the conduct of science and
engineering is of increasing importance.
We have chartered a special advisory committee on "cyberinfrastructure"
chaired by Dan Atkins of the University of Michigan
to advise us on coming opportunities in this area.
This activity must be led by CISE researchers working
in conjunction with researchers from all S&E disciplines.
We anticipate that it will generalize on the advanced
computing and networking infrastructure programs that
have been part of CISE since its beginnings in the
mid-80s.
NSF's role in the federal IT R&D effort
NSF continues to lead the interagency IT R&D effort
by chairing the interagency committee, co-chairing
virtually all of its subcommittees, and hosting the
National Coordinating Office, which provides support
for both interagency activities and the PITAC. Because
of its broad charter, NSF is the only federal agency
that is active in all areas of IT research and education.
We expect that our leadership will continue in this
area.
Although this conference is focused on R&D for the
advance of information research, systems, and technologies,
we must also view your work in the larger context
of all science and engineering.
In my remaining time, I want to explore what I believe
to be issues that we must address within science,
and issues external to science that need to be tackled
to make science more effective.
First, what we need to do within science and engineering
We know that science brings fresh knowledge of our
planet, and ourselves, thus what is newly possible.
But, what do we need to do within science and engineering
to be most effective in that journey?
Our community should be first-line responsive to the
changing context of society. To do this, we will need
to strengthen the links between the natural sciences
and the social and behavioral sciences. Sept. 11 made
that clear.
We have already seen the convergence of knowledge among
the natural sciences in the expansion of interdisciplinary
research. So too we must recognize that only the social
and behavioral sciences can help us understand and
anticipate the responses of the human universe. The
terrorist's attacks make that all too apparent.
Our accrued knowledge from decades of research support
is already serving new objectives brought about by
the events that began on September 11th.
And the nation's science policy will continue to move
in the direction of national necessity.
However, in the long sweep of civilization, we've utilized
most of our science and engineering knowledge to remediate
an existing problem or to address a current need.
We now recognize that we must draw on one of science's
most potent capacities -- prediction. If we can predict,
we frequently can prevent. The centuries of our accrued
knowledge can and should increasingly be directed
toward prevention.
NSF had a team of earthquake disaster specialists at
"ground zero" within a few days of the attacks. They
were there to assess the reasons for the Twin Tower's
utter collapse to the ground.
As it turns out, it was not the impact of the crashes
into the structures. Rather, it was the heat coming
from the jet fuel melting the steel superstructure
of the towers and their design that brought them down.
This is new and important knowledge for future building
materials, and to prevent or minimize loss in the
future.
The national directive for "homeland security" will
involve every sector of society, but especially the
federal government.
We will need to develop a broader, more anticipatory
perspective in our research. We will need to increase
our emphasis on envisioning future possibilities,
good or ill, as a mechanism to predict. Work in IT
will contribute heavily to our success.
Undoubtedly, the emphasis on prediction and prevention
will open new pathways in exploration and discovery
-- at the same time that the research community maintains
its freedom and passion to explore new frontiers,
within the rigor of merit review. Our ability to use
foresight gives us a kind of early warning system
-- a guard against unintended consequences.
Over the past few years, NSF has been developing a
program called NEON (National Ecological Network).
It is distributed instrumentation of sensors that
collect data from the entire ecological spectrum.
The sensors will constantly monitor the environment,
serving both short-term and long-term objectives.
From moment to moment, they will be an early warning
system for biological or chemical threats, such as
invasive disease and poisonous toxins. For the long
run, they will develop the base-line data that determines
the parameters of what is a healthy environment for
an area. NEON is clearly a foresight project.
As scientists, we also know that current knowledge
is never the final word on a subject or a security
blanket for the future. It will help us in the present
but in the words of Alfred North Whitehead "Knowledge
doesn't keep any better than fish." Tomorrow, new
more complete knowledge will always replace today's
-- a process of constant renewal, at an ever-accelerating
pace.
This makes an unshakable case for consistent
research in all eras, at all times. It also means
that we, as a community, face the challenge of aiding
policymakers and the public to better understand the
continuously evolving nature of scientific knowledge.
Second, what will science need to produce most effectively
for the nation and for humankind?
There are three primary components that will help determine
the effectiveness of science in the future. They are
stable funding, a balanced portfolio, and an expanding
talented workforce.
If you examine the history of federal funding for research
and development, you know that it looks like an erratic
electrocardiograph. And yet, we know that a steadily
advancing momentum of discovery depends on stable
funding.
The throttle forward, throttle back approach to research
funding is wasteful in terms of dollars spent, damaging
to the thrust of scientific activity, and disillusioning
to the pool of scientific and technical personnel.
Stable funding obviously does not exclude funding increases
because the more we expect from science the more we
have to provide for the expansions of its breadth
and depth. This is self-evident.
The necessity for a balanced portfolio is less well
understood. Today, the convergence of knowledge across
disciplines requires that all disciplines are able
to move forward at a healthy pace. If they don't,
then it is very possible that a neglect of chemistry,
for example, could in the long run inhibit future
advances in biology.
We are also witnessing the proliferation of fields
of research, an important indication of expansion
in scientific understanding.
And advances in physics, biology, chemistry -- the
core natural sciences -- undergird all of the biomedical
sciences on which we depend to understand disease,
find cures, develop vaccines, and initiate preventive
strategies.
Thus, the case for a balanced portfolio is yet another
self-evident premise for a viable science enterprise.
The scientific workforce issue is perhaps the most
complicated of the three components and will require
a lot of hands-on initiative on our part.
As scientists and engineers, your own background can
likely attest to your excitement for science not beginning
at age 18 or 20, but most typically at a very early
age. This means that if we want more talent like all
of you, we have to reach children to enhance that
excitement when they're young... and develop the background
for them to do science.
The future of our country depends on attracting more
women and our diverse minority populations to science
and engineering ... a profoundly significant challenge
in our primary schools.
We must build our broader base of science talent from
the very young, and scientists and engineers have
to roll up their sleeves and get to work on this.
We need to make a commitment to a home-grown science
and engineering workforce that uses the diversity
of our nation as the talent pool.
If the science community can be hands-on to inspire
young people to a future in science, we would be performing
one of the most enduring acts of patriotism for the
nation. The future of the United States promises to
be spectacular, but there is a growing community of
nations with equally capable workers. Globalization
has proven this repeatedly in the last decade. There
is a reservoir of talent in other cultures of which
we know little. They too will join the ranks of our
economic competitors.
The workforce issue will be the most formidable for
us. Our engagement will determine its success.
I hope and trust that you will be among the leaders
in bringing about some of these changes. As IT leaders
and specialists, you are the scholars and practitioners
in a new, very powerful discipline. All of science
is now dependent on your science. We will need your
collaboration for our future success.
Thank you for the opportunity to speak with you. I
would be glad to try to answer any questions and would
be pleased to hear your comments regarding future
research directions.
|