[Senate Hearing 107-28]
[From the U.S. Government Publishing Office]
S. Hrg. 107-28
HIGH PERFORMANCE COMPUTER EXPORT CONTROLS
=======================================================================
HEARING
before the
COMMITTEE ON
GOVERNMENTAL AFFAIRS
UNITED STATES SENATE
ONE HUNDRED SEVENTH CONGRESS
FIRST SESSION
__________
MARCH 15, 2001
__________
Printed for the use of the Committee on Governmental Affairs
U.S. GOVERNMENT PRINTING OFFICE
72-196 cc WASHINGTON : 2001
_______________________________________________________________________
For sale by the Superintendent of Documents, Congressional Sales Office
U.S. Government Printing Office, Washington, DC 20402
COMMITTEE ON GOVERNMENTAL AFFAIRS
FRED THOMPSON, Tennessee, Chairman
TED STEVENS, Alaska JOSEPH I. LIEBERMAN, Connecticut
SUSAN M. COLLINS, Maine CARL LEVIN, Michigan
GEORGE V. VOINOVICH, Ohio DANIEL K. AKAKA, Hawaii
PETE V. DOMENICI, New Mexico RICHARD J. DURBIN, Illinois
THAD COCHRAN, Mississippi ROBERT G. TORRICELLI, New Jersey
JUDD GREGG, New Hampshire MAX CLELAND, Georgia
ROBERT F. BENNETT, Utah THOMAS R. CARPER, Delaware
JEAN CARNAHAN, Missouri
Hannah S. Sistare, Staff Director and Counsel
Mark T. Esper, Professional Staff Member
Joyce A. Rechtschaffen, Democratic Staff Director and Counsel
Laurie Rubenstein, Democratic Chief Counsel
Darla D. Cassell, Chief Clerk
C O N T E N T S
------
Page
Opening statements:
Senator Thompson............................................. 1
Senator Bennett.............................................. 5
Senator Collins.............................................. 6
WITNESSES
Thursday, March 15, 2001
Susan S. Westin, Ph.D., Managing Director, accompanied by Stephen
M. Lord, Assistant Director, and Jeffrey D. Phillips,
International Security Analyst, International Affairs and Trade
Division, U.S. General Accounting Office:
Testimony.................................................... 9
Prepared statement........................................... 23
Questions for the Record submitted by Senator Thompson with
responses from Ms. Westin.................................. 37
HIGH PERFORMANCE COMPUTER EXPORT CONTROLS
----------
THURSDAY, MARCH 15, 2001
U.S. Senate,
Committee on Governmental Affairs,
Washington, DC.
The Committee met, pursuant to notice, at 9:35 a.m., in
room SD-342, Dirksen Senate Office Building, Hon. Fred
Thompson, Chairman of the Committee, presiding.
Present: Senators Thompson, Collins and Bennett.
OPENING STATEMENT OF SENATOR THOMPSON
Chairman Thompson. Good morning. The Committee will come to
order, please. I would welcome everyone to the hearing this
morning, of the Committee on Governmental Affairs, to assess
recent changes to our export control policy with regard to
high-performance computers. I will have a statement that I will
make a part of the record, and Senator Bennett, you or any
others, if you have any statements, we will make those a part
of the record.
I think, rather than read my statement, I will just briefly
state the reason why we are here this morning and a little bit
of the background. It seems to me like there are two or three
very important things that are converging here. One is our
increased concern about proliferation. We can hardly pick up
the paper nowadays without reading about a new agreement
between Russia and Iran, something Iraq is doing, or something
North Korea has done, and wondering whether or not they will
stop.
Most serious commentators on the subject say that the
proliferation of weapons of mass destruction is the biggest
threat that this country faces. More and more, it has become
obvious that a couple of the main sources for these items, some
of which are dual-use items, some of which are missile parts,
some of which have to do with nuclear technology, some of which
have to do with nuclear power plants, all of that, bits and
pieces and parts, some of which we know, a lot of which we
probably do not know, are coming from China and Russia. We get
our biannual intelligence reports that remind us time and time
again, year after year, that proliferation is a very big
problem; that the world's greatest suppliers of the rogue
nations are China and Russia.
Another phenomenon that has taken place is our appreciation
for the fact that high-performance computers are dual-use
items, and can play a part in this problem. The Cox Report
pointed out that it was possible, first of all, that high-
performance computers have many military applications, that it
is possible that these computers have been diverted for
unauthorized uses by other countries, which could include the
following, and they list several things here: Upgrading and
maintaining nuclear and chemical weapons; equipping mobile
forces with high-technology weapons; building a modern fleet of
combat aircraft and submarines; conducting anti-submarine
warfare; developing a reliable, accurate ballistic cruise
missile force; improving command and control communications;
and that there have been instances where the People's Republic
of China has used high-performance computers for nuclear
weapons applications, nuclear weapons design, using them to
improve and maintain their nuclear weapons, and on and on and
on.
Recognizing the proliferation problem, and recognizing the
sensitivity of high-performance computers as one of the things
that could be used in the proliferation activity, and
recognizing that one of our major new trading partners, China,
was doing things with them for the inherent activity of
improving their nuclear stockpile and so forth. This was
troublesome to us, and that their side activity of handing off
technology and other things to rogue nations. As a result,
Congress, in 1998, as a part of the Defense Authorization Act,
said that when we as a Nation--the Department of Commerce
particularly--that when you look at these high-performance
computer export situations, that Congress needs to be involved
in the process.
We use MTOPS levels in this country, and without getting
into too much detail, it has to do with the computing power of
these machines. And we have had a regime in this country for
some time now that related our export policies with regard to
these high-performance computers to that MTOPS capability. That
has come under question now and we will be talking about that
today, but that has been our policy; the idea being that the
greater the MTOPS level, the greater the power of the machine
and the more likely it would be for China, for example, to
utilize this capability--for example, their nuclear weapons
production enhancement.
Without testing, it is very important to nuclear weapons
countries, for example, to be able to simulate tests, improve
their stockpile, maintain their stockpile. We do that. They do
that. High-performance computers assist you in doing that; many
applications that the Cox Report and others have pointed this
out. So, an additional phenomenon to add to this story is that
we have greatly liberalized our export policy with regard to
high-performance computers over the last few years.
Again, we will get into a discussion of whether or not that
is wise or not, but there is no question that that has
happened. Within 1 year, the MTOPS level went from 2,000 to
85,000 under President Clinton's directives. So, backing up a
bit, in 1998, Congress, looking at all this, says that when you
raise those MTOPS levels, you need to make an assessment as to
what this is doing to national security, if anything. You need
to make an assessment as to the foreign availability of these
things. You need to make an assessment and report to Congress
as to the military applications of high performance computers
at the increased level.
In other words, I might have an opinion on whether MTOPS is
a smart way to go. You might have a different opinion. But
let's really do an assessment as to the significance of what it
is that we are about to do. Well, we now have kicked in at the
85,000 level. Under President Clinton's directives, we no
longer have the notification requirement that we had before. We
have abrogated a notice requirement that used to kick in at
below the MTOPS level.
Even though you might not need a license to export a high-
performance computer, at least we want to know; we want to be
able to have an audit trail at a lower level, and that was the
way we operated up until just recently. The MTOPS levels kicked
in. The notice requirement will kick in shortly. So now we are
at the 85,000 level with no notice requirement for any computer
exports less than the 85,000 MTOPS level. So we are saying at
that level, Mr. Exporter, we not only do not need you to have a
license, we do not need to know what you are doing for our
audit purposes or whatever.
And this is at a time, incidentally, and the Cox Report
points this out, that our post-shipment verification situation
is virtually nil. I think for a long period of time we had one
post-shipment verification of a Chinese supercomputer. Then,
there have been some more since then, but they would not even
let us in there to check for a long period of time. That is the
background of what we are doing here. With that, the President
makes his report. The 2001 law required that the GAO take a
look at the President's report and to give Congress an
opportunity to look--a very brief opportunity, I might add--
quickly look and see what the GAO says about the President's
report in order for Congress to have a chance to act if it
wanted to.
That is what we are about here today. The President has
made his report. It is obvious that the President's report does
not make an assessment as to all of the military applications
that we know are out there, and it is clear that the
President's report has not made an adequate determination or
assessment in terms of the national security implications of
what has just been done. That is what this is about, within the
context of a much broader debate that we have had going on for
some time now and will continue, I'm sure. It is the assessment
of the President's report in compliance or lack of compliance
with the 1998 Defense Authorization Act, which required the
President--you have got the right, Mr. President--to raise
these MTOPS levels. You have got the right to cut off the
notice below the 85,000 level, if you want to, but tell
Congress what the significance of that is, so we are not flying
blind.
My conclusion is that we are still flying blind, and that
we do not have anyone in this Nation who has made an
independent, unbiased assessment of the significance of our
greatly liberalized export policy with regard to dual-use items
and, in particular, high-performance computers. My colleague,
Senator Bennett, here, will eloquently point out that the genie
is out of the bottle and everybody has got everything, etc. He
may be right, but we have 100 folks around here with 100
different opinions about that. What I am interested in is
having professionals, having folks who know this stuff to take
a look at it and make a determination and recommendation to us
as to whether or not MTOPS are outmoded. If so, what does that
mean, that we abandon everything? I think clearly not.
There is a lot of potential other things that we can do.
Nobody has taken a look at them. So, we continue, because
nothing is exploding at the moment, we continue to go down the
same path blindly, each of us, with our various opinions as the
extent of which we ought to control or decontrol, or regulate
or deregulate, with industry pounding on our doors every day,
trying to liberalize export controls further, arguing that
there is no need to try to control anything, ever, anymore,
anywhere. We are talking about Tier 3 countries here. If really
you cannot control anything anymore, why control Tier 4
countries? Why not go ahead and send computers to Syria and
Libya, if, in fact, there is nothing we can do at any stage? So
those are the issues that we have here today.
[The prepared statement of Chairman Thompson follows:]
PREPARED STATEMENT OF SENATOR THOMPSON
Good Morning. Today, the Governmental Affairs Committee is holding
a hearing on the recent changes to our export control policy with
regard to high performance computers. These changes were made by former
President Clinton in January 2001, before he left office.
High-Performance Computers (or HPCs) represent a special challenge
for our export control regime, because in many ways they are the King
of ``dual-use'' technologies--that is, technologies that are subject to
national security export controls because they are easily usable for
important civilian purposes as well as dangerous military ones. High
performance computers are particularly sensitive because they are
enabling technologies that have important applications to national
security. HPCs can be used to design more powerful nuclear weapons,
more capable ballistic missiles, quieter submarines, and high
performance jet aircraft. Computers are critical to our intelligence
agencies. They are used in cryptology, reconnaissance satellites, and
electronic eavesdropping. High performance computers are also key to
modern warfare. They are used in precision-guided munitions, theater
missile defense, battle management, and information superiority. In
short, computers are the technical keystone of our national security
apparatus; they are also important to our potential adversaries.
Given the military importance of high performance computers, and
reports at the time that foreign entities had illegally acquired U.S.-
made HPCs for military purposes, the Congress inserted language into
the FY 1998 National Defense Authorization Act that was designed to
strengthen HPC export controls. Among other things, this legislation
required the administration to provide a report to the Congress on
three factors when proposing a modification of the export control
notification thresholds for high performance computers. First, the
report is to address the availability of high performance computers
from other countries. Second, it is to address all potential uses of
military significance at the new control thresholds. And third, the
report is to assess the impact of such uses on U.S. national security
interests. It is important to note that the purpose of the law was not
to fetter international trade or hurt the computer industry, but to
ensure that the government conducted a thorough assessment of the risks
and reasons underlying any change to our HPC export control policies.
Two years later, the Cox Committee stated in its July 1999 report
that the Peoples' Republic of China was diverting U.S.-built HPCs for
unlawful military applications. Specifically, it was using American-
made computers to design, model, test, and maintain advanced nuclear
weapons. The committee clearly stated that ``The illegal diversion of
HPCs for the benefit of the PRC military is facilitated by the lack of
effective post-sale verifications of the locations and purposes for
which the computers are being used. HPC diversion for PRC military use
is also facilitated by the steady relaxation of U.S. export controls
over sales of HPCs.'' The committee added that U.S.-origin HPCs had
been obtained by PRC organizations involved in the research and
development of missiles, satellites, spacecraft, submarines, and
military aircraft, to name a few.
Despite this report and others, the Clinton Administration further
relaxed export controls on high performance computers. In January 2000,
the licensing threshold for HPCs was set at 2,000 MTOPS. Over the
course of a year additional changes were made--including removing
distinctions between civilian and military end-users--culminating in
the January 2001 decision to raise the licensing level to 85,000
MTOPS--over a 40-fold increase in a 12-month period! These changes, and
others, were made to our HPC export control policies without conducting
a thorough national security risk assessment. I might add that a risk
assessment not only would have been the responsible and prudent thing
to do, it was required by law.
It also appears that the Clinton Administration couldn't justify
these changes on economic and commercial grounds. HPCs of similar
capability are not ``foreign available'' despite industry claims that
they are ``widely available.'' Additionally, while the administration
and industry cited computer ``clustering'' and other techniques as
making computer controls ineffective, we will hear otherwise today from
the General Accounting Office. Even the Cox Committee reported 2 years
ago that ``while the PRC might attempt to perform some HPC functions by
other means, these computer work-arounds remain difficult and
imperfect.''
Surprisingly, however, little money appears to be at stake in the
decisions over HPC export controls to China and other Tier 3 countries.
Let me be clear. HPCs can be sold license-free to most countries in the
world. With regard to Tier 3 countries--China, India, the former Soviet
Union, all of the Middle East, Vietnam, and most of Eastern Europe--
computers under 85,000 MTOPS can be sold license-free. Computers over
85,000 MTOPS require a license, but even then, over 90% are approved
for sale. HPCs cannot be exported to any of the rogue states. In short,
given that sales to tier 3 countries represent less than 10% of all HPC
sales abroad, and HPC sales don't appear critical to the economic
health of most U.S. computer companies, it is hard to justify the
national security risks simply to sell a few hundred more of them to
China and other ``high risk'' destinations. It is even more difficult
to believe the computer industry's argument that the U.S. military will
be harmed if U.S. firms can't reinvest these marginal profits gained
from Tier 3 sales back into R&D programs that might produce military-
ready ``off the shelf'' technology. According to this line of
reasoning, the United States must jeopardize its national security in
order to improve its national security. This makes little sense.
The bottom line is that the computer industry is not concerned
about current sales because these sales are not as significant as we
were led to believe; however, the industry is concerned about future
sales. Any government efforts to regulate the sale of HPCs on national
security grounds is summarily and automatically rejected. The problem
with this approach is that our country faces serious threats today.
Despite the fact that the Cox Committee identified several alternative
methods to safeguard U.S.-made HPCs, the bipartisan Deutch Commission
made several recommendations to strengthen our export control regime,
and the GAO recently listed nearly a dozen alternatives to the current
MTOPS metric, we can't seem to find any consensus within the government
or private sector that export controls on computer hardware are either
needed or effective. I strongly disagree.
Given sufficient political will, ingenuity, and leadership, I am
confident that we can find a system that accommodates commerce while
protecting our national security. The first step in that process,
however, is finding an honest broker that can conduct the research,
present the facts, outline the costs and benefits, and analyze the
risks to our national security and economic prosperity of each policy
option. I am confident we will hear from such an honest broker today
when we hear from the GAO.
The Committee has been closely involved with nonproliferation
policy and export control issues for many years. In legislation passed
by the Congress last year at my urging, the General Accounting Office
was tasked to assess the President's compliance with the reporting
requirements outlined in the FY 98 NDAA, and to evaluate the adequacy
of the stated justification for any proposed changes to HPC export
controls. This hearing, therefore, will ask the GAO to report on its
findings pursuant to these taskings.
I look forward to hearing today, therefore, from our witnesses, who
can shed some much-needed light on the complex, yet critical, issue of
high performance computer export controls.
Senator Thompson. Senator Bennett.
OPENING STATEMENT OF SENATOR BENNETT
Senator Bennett. Thank you, Mr. Chairman.
Chairman Thompson. Would you like to state your own case?
Senator Bennett. Well, my case is not that much different
from yours. The only other item that I would put on the record
before we hear from the witnesses is that other countries have
been involved in the manufacture of these same devices.
Particularly, when you get into the issue of MTOPS, which are a
function of the speed and excellence of individual chips--there
are individual chips being manufactured in France, Japan,
Germany, and, interestingly, China itself, that have the same
computing power of some of the chips that we previously sought
to control.
Part of the problem facing American manufacturers is that
if we restrict them from exporting into the world, on the basis
that some of their products can be used by these rogue
countries, we also cut them out of competing in the world in
benign areas, with other countries who have the ability to
manufacture the same level of capability that we have. The
Chinese will use their own chips, instead of ours, to do some
of the things that we might not want them to do. Then, with
that same manufacturing capability, come into our markets, take
our markets away from us, and we have, therefore, accomplished
little or nothing in terms of our concern.
Aside from putting that statement into the record, Mr.
Chairman, I agree with you completely that the proliferation of
weapons of mass destruction is the most serious challenge we
face in the world today, in terms of our national security, and
I welcome your initiative in getting the GAO involved in doing
an analysis and study of this. As I said, we may not be quite
as far apart as you might think or as we may have appeared to
be in the past.
Chairman Thompson. Thank you very much.
Senator Collins, do you have an opening comment.
OPENING STATEMENT OF SENATOR COLLINS
Senator Collins. Thank you very much, Mr. Chairman. I first
want to thank you for holding this hearing on this very
important issue on which you have been a real leader in the
Senate.
It is often observed that we live today in the information
economy, one in which our individual and collective economic
success depends increasingly upon the ability of computer
systems to store and retrieve data, to control sophisticated
manufacturing processes, to perform complex mathematical
calculations, and to connect us together, seamlessly, without
regard to the bounds of geography. Ever faster computer
processing power is, in many respects, the backbone of this
information economy. As a result, the health of our computer
industry, which currently leads the world, should be of very
great concern to all of us. At the same time, however, we
cannot forget that the very processing speed that makes
computers so useful in the private sector also makes them
crucial enabling technologies for advanced military
developments.
The atomic bomb was designed by scientists using slide
rules, and it only took a comparatively primitive computer to
design the first Stealth bomber. The sophisticated weapons
systems of tomorrow and the complex command and control schemes
into which they will all be fitted, however, require state-of-
the-art technology. For countries seriously attempting to take
advantage of what many military theorists have described as the
information-driven revolution in military affairs now underway,
acquiring more and better high-performance computers is an
absolute necessity. We encourage such developments in potential
adversary countries only at our great peril. High-performance
computers are, thus, the classic dual-use technology. They are,
indeed, useful for numerous legitimate purposes, but they can
be very dangerous in the wrong hands.
The same computing power that can model complex atmospheric
weather patterns on a continental scale, for example, might be
just as useful in modeling critical aspects of a nuclear
explosion, perhaps thus helping a country improve its nuclear
stockpile or develop sophisticated nuclear weapons without
telltale test explosions.
This is the source of the real tension in national security
export control law. To be sure, export control restrictions can
do real harm to high-tech, highly competitive sectors of our
economy, by making overseas sales slower, more difficult and
less certain, ceding ground to our foreign competitors. At the
same time, however, it is imperative that our understandable
eagerness to promote commercial sales not lead us to permit
American companies to help arm actual or potential foreign
adversaries.
That is why today's topic is so interesting. We saw the
actions taken by the previous administration, which I am
describing in some detail in my statement, and which I would
ask be put within the record in the interest of time. This is
clearly a significant issue and it is one that Chairman
Thompson has been closely involved with for some time. As a
Member of this Committee myself and as a new Member of the
Committee on Armed Services, I feel strongly that we can strike
an appropriate balance.
So, with that, Mr. Chairman, I would ask that the remainder
of my statement be included in the record and I look forward to
the hearing today.
[The prepared statement of Senator Collins follows:]
PREPARED STATEMENT OF SENATOR COLLINS
It is often observed that we live today in an ``Information
Economy,'' one in which our individual and collective economic success
depends increasingly upon the ability of computer systems to store and
retrieve data, to control sophisticated manufacturing processes, to
perform complex mathematical calculations, and to connect us seamlessly
together without regard to the bounds of geography. Ever-faster
computer processing power is in many respects the backbone of this
information economy. As a result, the health of our computer industry--
which currently leads the world--should thus be of very great concern
to all of us.
At the same time, however, we cannot forget that the very
processing speed that makes computers so useful in the private sector
also makes them crucial ``enabling technologies'' for advanced military
developments. The atom bomb was designed by scientists using slide
rules, and it took only comparatively primitive computers to design the
first Stealth bomber. The sophisticated weapons systems of tomorrow and
the complex command-and-control schemes into which they will be fitted,
however, require the state of the art. For countries seriously
attempting to take advantage of what many military theorists describe
as the information-driven ``Revolution in Military Affairs'' now
getting underway, acquiring more and better high-performance computers
is an absolute necessity. We encourage such developments in potential
adversary countries only at our great peril.
High-performance computers are thus the classic ``dual-use''
technology: They are useful for innumerable legitimate purposes, but
they can be very dangerous in the wrong hands. The same computing power
that can model complex atmospheric weather patterns on a continental
scale, for example, might be just as useful in modeling crucial aspects
of a nuclear explosion--perhaps thus helping a country improve its
nuclear stockpile or develop sophisticated thermonuclear weapons
without tell-tale test explosions.
This is the source of the real tension behind national security
export control law. To be sure, export control restrictions can do real
harm to high-tech, highly-competitive sectors of our economy by making
overseas sales slower, more difficult, and less certain--ceding ground
to our foreign competitors. At the same time, however, it is imperative
that our eagerness to promote commercial sales not lead us to permit
American companies to help arm actual or potential foreign adversaries.
This is why today's topic is so interesting. On several occasions,
the previous administration saw fit to dramatically loosen the export
control restrictions that govern sales of high-performance computers to
so-called ``Tier 3'' countries such as Russia and the People's Republic
of China. Last year, in fact, the Clinton Administration eliminated the
distinction between ``civilian'' and ``military'' end-users for Tier 3
computer sales--with the effect that military research laboratories and
intelligence agencies in these countries can now freely buy American
computers at the same high performance level at which we permit
license-free sales for less dangerous end-user such as banks or
hospitals. Moreover, the performance level of computers available
without any export license at all has steadily been raised in lockstep
not with the actual availability of such equipment from foreign
competitors but rather with the American computer industry's
anticipated marketing strategy for ever more powerful processors.
Less than 2 years ago an export license was required in order to
sell computers to military end-users in a country such as Russia or
China at performance levels of more than 2,000 Million Theoretical
Operations Per Second (MTOPS). This, I should emphasize, did not mean
that sales above that level were barred--merely that in order to sell
such a computer to such an end-user, one had to obtain permission from
the U.S. Government, which reserved the right to object on national
security grounds. As of January this year, however, the export license
threshold level stood at 28,000 MTOPS. With President Clinton's last
minute decision on January 18, 2001 to implement yet another loosening
of export controls, this level is rising to 85,000 MTOPS. That amounts
to more than a 40-fold increase in less than 2 years! Some of these
changes, moreover, have been made over the objection of our allies and
in violation of our government's commitment to observe export licensing
levels agreed upon by the multilateral Wassenaar Arrangement
headquartered in Vienna.
These have been remarkable changes. But have the previous
administration's dramatic relaxation of computer export controls
managed to strike an appropriate balance between commerce and security?
Disturbingly, it is remarkably hard to tell. The unanimously-adopted
Cox Report of 2 years ago discussed some of these issues in one of its
chapters, warning of the many military uses to which the PRC is putting
high-performance computers purchased from the United States. Press
accounts have also reported the appearance of U.S. machines in Russian
nuclear weapons design laboratories--an occurrence about which the head
of Russia's weapons program actually boasted publicly. Remarkably,
however, a thorough assessment of the national security implications of
specific decisions to loosen computer export controls has never been
done.
This is very troubling. Despite repeated GAO studies and
Congressional hearings highlighting this fact and urging corrective
action, not one of the previous administration's repeated relaxations
of computer export control thresholds was accompanied by a serious
analysis of the national security implications of the change.
Fortunately, Congress saw fit last year to require such a national
security assessment by GAO. Our witnesses today will present us with
the GAO's findings pursuant to this Congressional requirement.
According to conference report language adopted late last year,
Congress expects that GAO will provide an assessment of the adequacy of
the stated justifications for a change in computer export license
threshold levels, as well as an evaluation of this change's likely
impact upon our national security and foreign policy interests, the
security of our friends and allies, upon multilateral regimes such as
Wassenaar, and upon our nonproliferation policy. That is quite a tall
order, but I have great faith in our experts who join us from GAO
today. This is apparently the first time such factors have ever
formally been assessed in connection with a change to computer
licensing rules, so I am pleased that this Committee is providing this
opportunity for them to testify.
This is clearly a significant issue, and one with which Chairman
Thompson has been closely involved for some time. As a Member of this
Committee for some time, and as a new Member of the Armed Services
Committee, I feel that it is possible to strike an appropriate balance
between commerce and security. Today's hearing will contribute to these
debates in important ways, and should help ensure that these national
security issues receive the attention they deserve. I look forward to
hearing the testimony of our witnesses.
Thank you, Mr. Chairman.
Chairman Thompson. Without objection, it will be made a
part of the record. Thank you very much, Senator.
At this time I would like to recognize our panel. We are
pleased to have three witnesses with us today from the General
Accounting Office, Dr. Susan Westin, Managing Director of
International Affairs and Trade Division. She will present the
GAO's findings with regard to former President Clinton's
January 2001 changes to the United States high-performance
computer export control policy.
Dr. Westin will be joined by Steve Lord, Assistant Director
of the International Affairs and Trade Division, and Jeff
Phillips, a senior analyst also in the International Affairs
and Trade Division. I would like to read the resumes of all you
folks. It would be extremely impressive, I assure you, if you
have not already read them, but unless you insist, we will just
skip those and just say that you are, obviously, highly
qualified in these very technical, complex areas. We thank you
for being with us today.
Dr. Westin, would you please proceed with your testimony?
TESTIMONY OF SUSAN S. WESTIN,\1\ Ph.D., MANAGING DIRECTOR,
ACCOMPANIED BY STEPHEN M. LORD, ASSISTANT DIRECTOR, AND JEFFREY
D. PHILLIPS, INTERNATIONAL SECURITY ANALYST, INTERNATIONAL
AFFAIRS AND TRADE DIVISION, U.S. GENERAL ACCOUNTING OFFICE
Ms. Westin. Yes, thank you, Senator. Mr. Chairman, I ask
that my entire written statement be put into the hearing
record, but I will just summarize my remarks.
---------------------------------------------------------------------------
\1\ The prepared statement of Ms. Westin appears in the Appendix on
page 23.
---------------------------------------------------------------------------
Chairman Thompson. Without objection.
Ms. Westin. I am pleased to be here today to discuss export
control issues for high-performance computers. As you know,
export controls continue to be a contentious part of our
national security debate. Over the past several years, there
has been continuing congressional concern over the rationale
for revising these controls, and their effect on our national
security.
U.S. policy with respect to the export of high-performance
computers seeks to balance the national interest in promoting
economic growth in trade against the essential national
security requirement to prevent the proliferation of
technologies related to weapons of mass destruction, missiles,
and advanced conventional weapons. The United States has long
controlled the export of high-performance computers to
sensitive destinations, such as Russia and China. Recent
technological advancements in computing power have been rapid.
The regulations for high-performance computer control
thresholds are currently based on a computer's composite
theoretical performance, as measured in millions of theoretical
operations per second, or MTOPS.
We recently reported that the growing ability of other
countries to cluster lower performance computers has made the
current computer export control system ineffective in limiting
countries of concern from obtaining high-performance computing
capabilities for military applications. In that report, we also
reported that MTOPS is an outdated and invalid means for
determining whether individual high-performance computers
should be licensed for export.
On January 10, 2001, President Clinton announced a
significant relaxation in export control licensing and
notification thresholds for high-performance computers, and a
proposal to eliminate almost all controls on exports of
computer hardware. Today, I will first discuss our evaluation
of the President's justification for changing the computer
control thresholds. I will then turn to our evaluation of the
support of the President's policy proposal to shift the
emphasis of controls from hardware to software controls.
Finally, I will discuss some of the implications of these
changes.
The main message of my testimony this morning is that the
inadequate justification for the relaxation of export controls
for high-performance computers demonstrates the need for a
comprehensive study of export controls. As we have previously
recommended, a panel of experts is needed to conduct a
comprehensive assessment of available options to address
weaknesses of computer export controls, as well as to determine
what U.S. countermeasures might be necessary to respond to
computer-related enhancements of military or proliferation
capabilities in sensitive countries.
Turning to the first point, the President's January 2001
changes in the export control thresholds for high-performance
computer exports are not adequately justified. While the
President's report recognizes that high-performance computing
capabilities will become increasingly available to other
countries through computer clustering, the report fails to
address all militarily significant uses for computers at the
new thresholds, and assess the national security impact of such
uses as required by law. The inadequacies of the President's
report are further compounded by continued use of a flawed
measure, MTOPS, for assessing computer performance.
Second, the President's policy proposal for relaxed U.S.
computer controls also is not adequately supported. Although
the new policy was based on a conclusion that computer hardware
exports can no longer be controlled, the Executive Branch has
failed to adequately assess alternative control options. In
addition, the new policy would focus more attention on using
new software controls to protect U.S. national security
interests, even though such controls have yet to be identified
and developed.
Finally, we identified several implications of the changes
to the control thresholds and the proposed change in U.S.
computer export control policy related to increased risk for
U.S. national security. I would like to emphasize three of
these implications. The first implication is that the
inadequacies of the President's justifications again
demonstrate the need for a comprehensive study of the issues
involved. In our December 2000 report on high-performance
computers, we recommended that Executive Branch agencies
comprehensively assess ways of addressing the shortcomings of
computer export controls, including the development of new
performance measures and of countering the negative effects on
U.S. national security of increased availability of high-
performance computing. Although the Departments of Defense and
Commerce did not agree with this recommendation, we believe
this recommended course of action is still valid. Without such
a comprehensive study, the United States will not be prepared
to fully assess and mitigate the risks that it may face by
revising U.S. high-performance export controls.
The second implication is that the new rules effectively
eliminate prior U.S. Government review of any computer exports
below the licensing threshold to sensitive countries. The
National Defense Authorization Act of 1998 requires computer
exporters to notify the Commerce Department in advance of any
proposed high-performance computer exports to sensitive
countries, such as China, Russia and India that pose a concern
for military or proliferation reasons.
Since the threshold for notifications was set lower than
the threshold for licensing, it served as a tripwire for
triggering advanced U.S. Government review of exports to
potentially sensitive end-users. Under this procedure,
exporters must apply for a license if any Executive Branch
agency in the licensing process objects to a proposed computer
export. Next week, the threshold for notification takes effect
at 85,000 MTOPS. This will be the first time that the
notification threshold is the same as the licensing threshold
for computer exports to sensitive countries.
In the past, the notification threshold lower than the
licensing threshold allowed the U.S. Government to deny or
defer some exports, or make other exports conditional on
additional safeguards. There were over 3,700 notifications of
intent to export computers between February 1998 and September
2000. Of the 204 notifications that required subsequent license
applications, 165 were not approved, while the remaining 39
were approved with conditions. Reasons for non-approval
included concerns over end-users.
Admittedly, only a small percentage of the notifications
were not approved for licenses to export. Commerce and Defense
officials point to this small percentage as evidence that the
notification process has not been useful. However, we are not
able to assess the significance of the non-approved licenses,
in part because of lack of the national security analysis of
the impact of these exports to sensitive countries.
The third implication is the U.S. Government will have to
rely more heavily on computer vendors to know your customers,
and assess their intentions to use computers for proliferation
purposes. Past evidence has shown this reliance may be
misplaced. In summary, we recognize the need for U.S. policy on
the export of sensitive technology to balance economic and
trade interests against national security interest. However,
the justification for relaxing the export controls did not
address all computer uses of military significance, to which
high-performance computers could be applied at the new
thresholds, and did not assess the national security impact of
such uses. The lack of justification makes it difficult to
determine if the appropriate balance between economic interest
and national security interests has been achieved.
Mr. Chairman, Members of this Committee, this concludes my
prepared remarks. My colleagues and I will be pleased to
respond to any questions you may have.
Chairman Thompson. Thank you, Dr. Westin. I appreciate that
very much and I especially appreciate GAO's attention to this
in a short period of time. The legislative framework that we
set up for your getting involved and doing this was short, and
our review period is short. In fact, I think we have to act by
tomorrow as a Congress, and I think the whole process is very
deficient in my mind, but it is what it is, and you are doing
your part and we are trying to do ours.
Let me see if I can basically summarize your testimony and
see if I am accurate. First, is that with regard to three basic
requirements under the 1998 Defense Authorization Act, that two
of those requirements were not fulfilled, or were deficient in
the President's report; that is, the requirement to assess the
military applications and the other requirement to assess the
national security implications. Is that basically correct?
Ms. Westin. That is right, the justifications that were to
be included, according to the NDAA, were not there.
Chairman Thompson. All right. Also, next week will be the
first time that the notification level is the same as the
license level at 85,000 MTOPS. Is that correct?
Ms. Westin. That is right.
Chairman Thompson. And you point out the significance of
that, in that in times past notification has been lower, and we
have been able, as a country, to catch some, presumably, that
should not be approved that, otherwise, would have been
approved had it not been for a lower notification level. You
mentioned 165, I believe, licenses were denied that otherwise,
presumably, would have been approved.
Ms. Westin. Yes, Senator, 165 were not approved. I think it
is 12 that were outright denied and the others were returned
without action.
Chairman Thompson. Sometimes I think that is done for
national security purposes and people do not want to talk
about, in too much detail, why they are doing what they are
doing. They just do not get their license. You also say that
you believe that the MTOPS metric is no longer very useful in
making these determinations, primarily, as I take it, because
of the clustering capability that countries have now. Is that
basically correct? I know I am paraphrasing.
Ms. Westin. That is basically right, that it is an outdated
measure.
Chairman Thompson. It is an outdated measure. I am also
correct, though, in concluding that you are not suggesting
that, therefore, we should have no measure, or that we should
have no control; are you?
Ms. Westin. No, definitely, that is not part of our
conclusion, nor, Senator, are we concluding that 85,000 MTOPS
is an appropriate level or not an appropriate level. Our
analysis was to look at the justifications that were behind
changing these export control levels.
Chairman Thompson. Justification in these two important
categories simply are not there. We are continuing to
liberalize our export policy without the knowledge of the
national security impact it might have, or without an
assessment as to all of the military applications that such a
raising of MTOPS-level licensing might have.
Ms. Westin. That is right.
Chairman Thompson. I think it is important to remind
ourselves, too, that when we are talking about 85,000 MTOPS,
that does not mean anybody is cut off at 85,000. That just
means that when you get below 85,000, you do not even have to
have a license, but when you get to the 85,000 and above, 90-
something percent of the license applications--I think that is
correct--are approved. The Cox Report suggested several
alternative ways of measuring these computers, in terms of
their significance. The Deutch report had several possible
suggestions. Your report in December of last year had almost a
dozen possible suggestions as to different ways that we could
possibly control or have a control regime, which would really
keep the stuff we wanted out of the bad guys' hands, or
potential bad guys, in terms of end-users--I am not talking
about countries necessarily--that there were options out there,
but that nobody has really sat down and gone through and made a
determination as to which one of these might be viable and
which ones might not be; is that correct?
Ms. Westin. That is right, and we include ourselves in
that, Senator. GAO did not do that.
Chairman Thompson. Nobody has done that. We are flying
blind, essentially, with regard to that. Some of those
suggested possibilities are hardware-type solutions, counting
of processors instead of MTOPS, things of that nature; but
suffice it to say, there are a variety of suggestions out there
from Cox to Deutch to GAO, things that are on the table, that
knowledgeable people have said, ``We might ought to take a look
at this. This might work.''
All kinds of ideas are out there. None of us are wed
forever to the MTOPS criteria or anything else. We just know
that whatever the criteria is, it is becoming more and more
liberal. It went from 2,000 MTOPS to 85,000 MTOPS in 1 year.
The notice requirement has been liberalized, shall we say, or
there is less and less notice, less time for Congress--Congress
used to have 6 months to review this. Now we have 60 days.
Ms. Westin. 60 days.
Chairman Thompson. Everything is being pushed toward more
high-performance computers out the door to Russia and China. As
I said, we might conclude, someday, that we might as well do
that and we might as well go ahead to give them to Iran and
Iraq, while we are at it, but we were not there yet, because
nobody, including the President, even though the law required
it, has made a national security assessment or an assessment of
all the military applications that might have significance in
terms of the MTOPS level being raised. Is that essentially
correct, a summary?
Ms. Westin. Yes, that is a good summary, Senator.
Chairman Thompson. I noticed here in your September 1998
report on export controls, national security issues and foreign
availability for high-performance computer exports, GAO has
done a lot of good work in this area. It is unfortunately under
the radar screen. Nobody pays a whole lot of attention to it
now, except the people in the business, but it is very
important and one of the things that you have looked at is how
we stand in comparison to other countries.
I noticed here on page four, it said, ``Based on EAA's
description of foreign availability, we found that subsidiaries
of U.S. companies dominate overseas sales of high-performance
computers.'' According to U.S. high-performance computer
exporters, there were no instances where U.S. companies had
lost sales to foreign high-performance computer vendors in Tier
3 countries. We also obtained information on the capability of
certain Tier 3 countries to build their own high-performance
computers, and found it to be limited. Tier 3 countries are not
as capable of producing machines of comparable quantity and of
comparable quality and power as major high-performance computer
supply countries.
I think that one of the issues we are going to have to deal
with as a Congress is that even acknowledging the world is a
smaller place, and many of the technological genies are
certainly out of the bottle, is there anything to be derived
from still trying to keep certain things out of the hands of
certain people for certain lengths of time, with a full
concession that maybe eventually everybody will have
everything? If not, why not send it to Saddam Hussein right now
and make the profit off of it and put it back into our system,
so that we can strengthen our own system over here? Obviously,
nobody would advocate that. Where I think we come to is that
there is something to be said for certain reasonable
restrictions that comport with reality, but that, if nothing
else, slows down this process that, among other things, has
caused us to conclude we need a national missile defense
system. That is the debate we are having here and I think you
have made an important contribution to it. Thank you.
Senator Bennett.
Senator Bennett. Thank you, Mr. Chairman.
Thank you, Dr. Westin. I agree, absolutely, there must be
careful study of this, further study of it. The primary
challenge in this world is how rapidly it is changing. A report
that comes out is obsolete almost as fast as the product that
it examines. Let me share with you and the Committee a press
release that came out on the 1st of February, so it is 45 days
old now, a company I had never heard of until this was called
to my attention, called Juno Online Services. Are you familiar
with that? Are any of you familiar with that?
Ms. Westin. Yes, I have my chief scientist here and he is
familiar with that.
Senator Bennett. Here is what it says, and is pretty
staggering in its implications: ``Preliminary studies recently
completed by the company suggest that if the computers of all
of Juno's active free subscriber base were simultaneously
working on a single computational problem, they would together
represent the world's fastest supercomputer and might approach
or break the petahertz barrier.'' I have never heard of the
petahertz barrier, but things keep coming at us in this world
all the time--that is a hypothetical effective processor speed
in the order of a billion megahertz.
Then you go on in the press release, and this is
fascinating to contemplate, it says, ``Applications will run as
screen savers on the computers of participating subscribers
when their machines would otherwise be idle, performing
calculations when the computer is on, but not in use.'' Now, I
have a screen saver on my computer that shows the beautiful
scenery of the State of Utah and changes every few seconds from
one magnificent vista to the next, but that is obviously not
taxing the computer very much to put those pictures up there.
So, apparently, if I am connected with Juno, while my screen
saver is on, they will get in and use the computing power to
work on this problem. It says Juno's free, basic service may
ultimately be required to make their unused computing power
available to the project as a condition for using that service.
In other words, if you subscribe to Juno, you have to say while
we have our screen saver up, you have access to the computing
power in our computer, and Juno has enough subscribers that
altogether, they would represent computing power greater than
the world's existing supercomputer, the biggest supercomputer.
This is an idea that is 45 days old as far as the public is
concerned.
I bring this up to illustrate how rapidly everything is
changing. This is a circumstance that was not in the debate
when the Congress met in the year 2000. We did not have any
concept--at least I did not have any concept--I do not know
about my colleagues, but I did not have any kind of concept
that this kind of connection could even be conceived of, let
alone discussed in early 2001 as a legitimate situation.
Whether or not Saddam Hussein would somehow plug into this and
take advantage of it, without anything being exported at all,
no physical box ever shows up in Iraq, he just makes an
electronic connection with a subscriber somewhere, and then
makes use of it, is a further demonstration of how badly we
need the study that you are talking about, and it needs to be
an ongoing study; that the conclusions arrived at by July of
this year will be obsolete by October of this year.
Maybe that is an issue, Mr. Chairman, we should raise with
Condoleeza Rice as an ongoing thing in the National Security
Council. I have talked to her about the whole issue of critical
infrastructure protection, because I think that the next war
will not be fought with kinetic weapons. It will be fought with
computers. If I were somebody who wished this country ill, I
would not try to take on the U.S. military, because I would
lose. Instead, I would hire the best hackers I could find, and
say let's find a way to get into the telephone system and shut
down the Fed wire, so that the American economy comes to its
knees overnight, because no checks can clear, no money can pass
and so on.
I think that is where the next war is being fought, and I
was delighted that Condoleeza Rice not only understood that,
but she had thought very deeply about it and was ahead of me.
That is all a little humbling, for a Senator, to have somebody
come to your office, and you are going to raise an issue, and
have that individual know more about the issue than you do.
I think your recommendation here is a very sound one and we
clearly need to pay attention to it, and the only amendment I
would add is the one I have suggested, that it not just be a
single date and the report issued and then you go onto other
things; that we find some way that this is an ongoing
assessment, because the Juno announcement demonstrates how
rapidly moving and changing everything is all of the time. Now,
if you have any comment or reaction to that, I would be happy
to hear it.
Ms. Westin. Senator, I was going to say that I think you
really have given an excellent example that does back our
recommendation, because our recommendation for this expert
panel is really twofold; to address any weaknesses in export
control measures--but I think what your example was pointing
to--to make a determination of what U.S. countermeasures are
necessary to deal with these weaknesses.
It may very well be that advances in technology are
occurring so quickly that there is not going to be a good
measure for controlling the export of computers, which is why I
think the second part of the recommendation, as I have said, is
equally important, to determine what countermeasures are we
going to take. It all does hinge on having an analysis of the
military uses of high-performance computers, and what impact
these uses have on our national security interest.
Senator Bennett. The only other comment I would make, Mr.
Chairman, that I would like the panel's reaction to, when Dr.
Hamre testified in a hearing on the Export Controls Act, he was
no longer the Deputy Secretary of Defense, but testified in his
position as the President of CSIS, but he talked about his own
transition on this issue, how, as an official of the Department
of Defense, he was firmly opposed to export of any of these
items. And then he said he woke up one night and realized that
what is happening around the world, in terms of other countries
developing these capabilities, could mean a situation where the
United States might be dependent on a foreign manufacturer,
because the foreign manufacturer would have access to markets
that the American manufacturer would not.
He said I began to change my position pretty clearly
because I did not want to run the risk of American
manufacturers falling behind foreign manufacturers, so that we
might eventually, at DOD, have to turn to a foreign source.
Would you comment on that or give us your reactions to that?
Mr. Phillips. That is a legitimate concern. There is
another trend, a market trend, that in some ways contradicts
that. In the computer area the trend has been toward DOD
acquiring more mass-market commercially available components
and systems, and relying less and less on specially designed,
specially built computers. As a result, it may be that at some
point DOD will be relying on foreign suppliers, at any rate,
because those commercially available mass-market systems that
they are getting will become more and more available from
everybody.
The idea that by keeping the U.S. computer industry
competitive overseas is going to allow money to come back and
be reinvested in R&D, may well be true. But the kind of high-
end, state-of-the-art systems that DOD used to be getting from
the more traditional computer companies, like Cray and Silicon
Graphics, it is moving away from, into the more commercially
available lower-end systems and linking those together.
Senator Bennett. Yes, because where the competition is
coming is in the lower-end, where you are seeing French
machines and Japanese machines competing with the Americans,
and it is not a question of the money coming back to fund R&D.
It is a question of the company drying up because they cannot
get market share.
Mr. Phillips. Well, at this point, based on the last work
that we did, the market share situation still fairly well
favors the United States.
Senator Bennett. That is true, and Dr. Hamre wants to keep
it that way.
Chairman Thompson. I might add, do you have any figures on
the high-performance computer sales to Tier 3 countries?
Mr. Phillips. We do not have any updated numbers on those,
but----
Senator Bennett. From any source, if I might add, Mr.
Chairman, not just American sales, but any sales.
Mr. Phillips. I will start with U.S. sales. You can think
of them as a sort of inverted pyramid. The vast majority of
U.S. sales has been to the European countries--and of course,
in the United States--to the Tier 1 countries, then to the Tier
2 countries, Tier 3 sales as a relative proportion of overall
sales, has been very, very small. I think we talked about five
or six percent as of a few years ago. Tier 3 has been a growing
market, but I think those ratios are still fairly constant.
As far as overseas suppliers, the Japanese are usually
considered our next biggest competitors, and we have an
arrangement with them where they are supposed to report to us
any sales of their high-performance computers to Tier 3
countries. We have been told by the State Department that they
have received no reports in the past several years, perhaps as
many as 5 years, of sales to Tier 3 countries.
Senator Bennett. Thank you, Mr. Chairman. My time has
expired.
Chairman Thompson. Senator Collins.
Senator Collins. Thank you, Mr. Chairman.
Dr. Westin, like the Chairman, I am very troubled by the
fact that the GAO found that the Clinton Administration failed
to conduct a serious national security analysis before making
the decision to loosen national security export controls upon
high-performance computers. Does the GAO know why the
administration failed to conduct the kind of in-depth review
that many in Congress urged be conducted?
Ms. Westin. Senator, if the officials from the Defense
Department and Commerce Department were here, I am sure they
would tell you they disagree with us. They would say that the
justification was adequate in the President's report. We look
at what was written in the law, and the law says for the second
and third requirements, the President's report should include a
list of all the militarily significant uses of high-performance
computers, and conduct an analysis of the impact on national
security of all of these uses. That was clearly not there.
In the written statement, we have given a few examples of
things we would have expected to be included in having a
national security impact of those examples.
Senator Collins. Just for the record, GAO's finding is that
the Clinton Administration failed to comply with the legal
requirement to conduct a thorough review of the national
security implications; is that correct?
Ms. Westin. Yes.
Senator Collins. Could you give us some idea of what kinds
of military uses computers at the 85,000 MTOPS level could be
put to, for those of us who are not as expert in this area as
the Senator from Utah? For example, could they be used in
nuclear weapons design? Are they powerful enough to process
radar or other sensor information from multiple sources, in
order to help a country build up its air defense system against
stealth aircraft? Could you give us some idea of what the
potential military uses of computers at that level could be,
Mr. Phillips?
Mr. Phillips. There are several--again, we can base this in
part on the studies that DOD and Commerce commissioned in 1995
and 1998, which did have a list of about 194 known militarily
significant applications at that time. At about 85,000 MTOPS or
so, they talk about global ocean modeling and weather
forecasting. They talk about on-board data processing for UAVs
(Unmanned Aerial Vehicles).
There are physics simulations, three-dimensional modeling,
three-dimensional simulations of submarines, submarine
modeling. There are a number of things on that list which we
understand, has grown since then in militarily useful
applications. So there are a lot more than what are on that
list at different levels.
Senator Collins. With regard to the examples that I gave of
probably helping to develop nuclear weapons design, probably
the answer to that is yes; is that fair?
Mr. Phillips. Nuclear weapons design, typically, can be
done at fairly low levels. We reported in 1998, based on
Department of Energy information, that a country like China
could use computers as low as, I believe, 6,000 or 7,000 MTOPS,
to help design new nuclear warhead designs.
Senator Collins. So, at 85,000 MTOPS, we are talking about
very powerful computers, then.
Mr. Phillips. Right, and the kinds of uses that would lend
itself to are stockpile stewardship uses, that the United
States is trying to be able to simulate without nuclear weapons
testing.
Senator Collins. Does the GAO know how the Clinton
Administration arrived at the 85,000 MTOPS level, Mr. Lord?
Mr. Lord. Sure. The Department of Defense officials we met
with indicated they extrapolated--or they projected through the
end of this year what they thought the foreign availability was
going to be. So, basically by December of this year, according
to their calculations, Tier 3 countries would be able to
achieve that capability. They had some modeling. We have not
looked at their assumptions in great detail, but that was their
analysis at the Department of Defense--with clustering, they
could achieve that capability. So, they have a unit over there
who performed those calculations.
Senator Collins. As I recall, there was previous GAO
testimony before this Committee that suggested that the levels
were set, not by ascertaining what was actually available from
foreign computer makers, but rather according to what new
processors U.S. manufacturers might hope to be able to market,
but did not currently market at the time the decision was
announced. Mr. Phillips, is that accurate?
Mr. Phillips. That is correct. The process, until this last
go-around, was that the administration would contact the
vendors and identify, within about a 6-month period, what their
production schedules would be for the next state-of-the-art
processors. Then they would base their projections on the MTOPS
levels of those processors and different configurations.
Senator Collins. So, in some ways, this is a prospective--
--
Mr. Phillips. Right, they have intended it to be that way.
Their reasoning has been they wanted the export control system
to not lag behind the technical capabilities that the companies
could achieve.
Senator Collins. Dr. Westin, the United States helped to
build, and concurrently participates in, a multilateral export
control regime that is called Wassenaar. As I understand it,
under that multilateral system, there is a requirement for
export license controls on high-performance computers; is that
correct?
Ms. Westin. To the Tier 3 countries, right.
Senator Collins. To the Tier 3 countries.
Ms. Westin. I will let Jeff clarify that.
Mr. Phillips. They do not normally have tiers of countries
in Wassenaar. It is not a system where there is a requirement,
per se. Wassenaar attempts to have common standards, common
guidelines, common control lists among its members, but it is
left to the national discretion of the countries to put into
their own laws and regulations the standards that Wassenaar
agrees to.
Senator Collins. Given those common controls and that
cooperative regime, I am curious whether the United States
consulted with the members of Wassenaar, for example, with our
NATO allies or with the Japanese, before announcing that
decontrol measure in January 2001. Do you know whether there
was consultation?
Mr. Phillips. We have talked to the State Department about
that. I am not quite sure of the sequence, whether they were
consulted ahead of time or were told that this is the decision
we are about to make. They did notify in December 2000 at a
regular session of Wassenaar, at which the United States was
convincing Wassenaar to go along with the more recent U.S.
changes. Levels were changed in August to 28,000 MTOPS. Either
late December or January, the State Department made it known to
Wassenaar that the U.S. was going to raise the levels up to
85,000 MTOPS.
Senator Collins. So it went from 28,000 MTOPS to 85,000
MTOPS in a period of 6 months or so.
Mr. Phillips. Less than 6 months.
Senator Collins. Less than 6 months. Are you aware of
reports that there was friction with our allies, who adhered to
the regime because of the United States moving unilaterally in
this area?
Mr. Phillips. Yes, we have heard things--and this is sort
of secondhand information--but the State Department officials
have told us that there was some concern among the members of
Wassenaar. I cannot say it was because they did not want,
necessarily, the levels to change, but at least they were
concerned because the levels had recently changed and, now, in
such a short time, were being changed again to a rather high
level.
Senator Collins. Thank you very much. I very much
appreciate GAO's excellent work in this area. Thank you, Mr.
Chairman.
Chairman Thompson. Thank you very much, Senator Collins.
Let me add to that report, Senator Collins, if I may. Some of
us went over and talked to our Wassenaar allies last year,
about this very thing, and I can add to your assessment. You
are getting your information from the State Department. It
sounds like they have been pretty candid with you, and I just
want to back that up. Folks that we talked to over there were
pretty consistent in being concerned that the United States,
after having so many of us complain about our efforts to get us
out of COCOM, and how that Wassenaar was not tight enough, and
there were no under-cutting rules or anything like that. It was
much too loose.
Here we are, in their eyes, always pushing the envelope in
terms of high-performance computers, and I think we are doing
it strictly to get a competitive advantage. I can verify that
the ones we talked to were not pleased with it at all and they
think we are quite hypocritical in complaining to want to have
control of sensitive items and have a multilateral approach to
it on the one hand and continue to push the envelope past where
they are willing or capable, I guess, of going on the other
hand.
Let me get back to another point, and maybe we can take
advantage of your expertise while we have you here, in more of
a technical area. It seems to me from what I have read, from
the work that the GAO has done and others, that the computing
power of these high-performance computers obviously is very
important. If you string that power together, you get more
power. But there are two things that seems to me like we need
to say about that.
First, is that clustering at the 70,000 MTOPS level does
not give you the same situation or the capability as having a
high-performance computer at the 70,000 MTOPS level, because
clustering is more difficult to operate. It is more difficult
to maintain. The power is not exactly the same. The application
is different. Is that a correct assessment?
Ms. Westin. Yes, as I understand it, Senator, not all
applications will run the same on a high-performance computer
that is 70,000 MTOPS, as opposed to a system where the
computers have been clustered to achieve the 70,000 MTOPS
level.
Chairman Thompson. As a matter-of-fact, is it not true
that, as far as we can determine, the clustering ceiling, as
best we can determine, is at about the 70,000 MTOPS level
capability?
Mr. Phillips. Yes, that is with the caveat that is using
mass-market commercially available components. That is how we
have tried to address the question. There are higher levels,
sometimes considerably higher levels of performance you can
achieve through clusters, that have been achieved. But that
higher level sometimes is reached using items that would be
controlled by export controls or would be proprietarily
controlled.
Chairman Thompson. All right. So, if they are able to get--
a clustering country is able to get a higher MTOPS level
computer, they can cluster more?
Mr. Phillips. That is possible.
Chairman Thompson. We are at the 70,000 MTOPS cluster level
with the caveat, and we are about to decontrol altogether at
the 85,000 MTOPS level. We are decontrolling at a level higher
than what we believe they can cluster at right now.
Mr. Phillips. That is correct. The projections we have seen
from the Defense Department are based on projections of
availability in the last quarter of 2001, not now.
Chairman Thompson. The second thing it seems to me we can
say about this is that while the power--and this is used in
layman's terms, but hopefully, one of the things we can do is
help the general public understand this better, and I am a heck
of a lot closer to the general public than I am an expert on
this stuff. While the computing power is very important and
stringing it together is very important, and is evidently
unlimited, it seems to me like there is a big difference
between that and a military application, for example--that how
you use that state of facts that has just been described, in
terms of application, in terms of your ability to maintain it,
in terms of your ability to repair it--the software that you
use--it seems to me, from all I can tell, that there is an
awful lot of stuff that goes into an ultimate ability to have,
say, a battlefield application other than just the raw power of
the computer; is that not correct? Could you elaborate on that
somewhat?
Mr. Phillips. Again, depending on the application,
sometimes you want enhanced graphics capability, sometimes you
want enhanced communications capability. If you are dealing
with real-time or near real-time command and control
communications capabilities, clearly, you need to have real-
time or near real-time computing. You have to have that
processing done when you need it, as opposed to research and
development projects, which is what most high-performance
computers have traditionally been used for. There is
flexibility in R&D in waiting longer for your results with a
clustered system than you might otherwise have needed with a
vendor-supplied system.
Chairman Thompson. In terms of all of that, the
applications and so forth, can anyone in Tier 3 or Tier 4
countries touch the United States in terms of that ability
right now?
Mr. Phillips. We have not done any recent work on that, but
traditionally their computing capabilities have been lower,
sometimes considerably lower, than the U.S. capabilities.
Chairman Thompson. You are talking about not just the power
of the MTOPS power, but all of the applications and all of that
we have been talking about. You made another distinction, and
that is in terms of the usefulness of, say, using clusters and
things of that nature; that if you are in an R&D application,
that timing is not that important. You can wait longer for more
results and so forth. But if you are in a battlefield
application, you have got to have results pretty quickly. So
when--it is not just if you get the capability; it is when you
get the capability might be very important. Is that a correct
assessment?
Mr. Phillips. That is right.
Chairman Thompson. Senator Bennett, do you have anything?
Senator Bennett. Mr. Chairman, I have nothing further, but
I appreciate the panel and your expertise, and I think, Mr.
Chairman, you were on the right track again on this last series
of questions, because it is not just the box. You can have the
box and it can have all of this capability, and if you do not
have the intellectual capability to program it properly or use
it properly, you are going to be in real trouble.
The only other comment that I would make--that comes to me
as I sit here and listen to this. I remember my interview with
General Horner, who was the head of the Air Force war in the
Gulf War, and then became the Commanding General of the Space
Command, he told me something that I think applies here, that I
found fascinating. He said the Gulf War was the first war
fought from space. All of the command and control was done from
space. They used satellites to direct tanks. He said we so
filled up all of the military satellites that we had to go out
and lease space on commercial satellites, because we did not
have enough space. Then with a grin he said one of the
commercial satellites we leased was leasing the other half of
its capability to Saddam Hussein. I think that is kind of a
precursor of what we are going to be seeing here.
Again, back to Juno. They may end up--somebody comes in and
says they want to buy ``X'' amount of time and they turn out to
be buying it for and on behalf of somebody else, but the main
protection we have in America is not just the ability to
produce a better box, but all of the other things, Mr.
Chairman, that you were going through that have an impact on
how the box is used. Again, I am grateful to the panel and for
the expertise.
Chairman Thompson. Thank you, Senator, for your expertise.
I think you are right, implicitly, in what you have always
said, and that is that among the important things, probably the
most important thing has to do with our ability to run faster
as a Nation, that the world does not stand still, and we have
to continue to run faster and outstrip the other guys. The
question here is whether or not we want to assist them to run
faster, too, even though they are going to be behind us or
whether or not we can do some things that might slow them down
a bit.
I hope this has not been a session to beat up on the prior
administration. I am hopeful that we can look forward from
here, and that the current administration, can, over a period
of time, when they are able to get all the people in that they
need to get--I can only imagine what they are going through
over there in trying to get their people confirmed and so
forth--will take these issues into consideration and conduct a
thorough national security risk assessment. Get some people
together, take a little time, start from scratch. If MTOPS
needs to go, let it go, but look and see what else is
available. Instead of concentrating on the hardware, maybe
concentrate on the software. Maybe there are other things that
we can do, or if we are just to throw up our hands and
decontrol everything to everybody, let us do it on top of the
table and quit claiming that we have an expert control system,
when really we do not have any system at all, because whatever
we can manufacture, we are going to exceed that MTOPS level
under the assumption everybody is going to get it.
At least we can do it honestly and after having made an
assessment, and I believe that they will do that and that is
basically, I think, all we can ask.
Thank you very much for being with us. The Committee is
adjourned.
[Whereupon, at 10:52 a.m., the Committee was adjourned.]
A P P E N D I X
----------
[GRAPHIC] [TIFF OMITTED] T2196.001
[GRAPHIC] [TIFF OMITTED] T2196.002
[GRAPHIC] [TIFF OMITTED] T2196.003
[GRAPHIC] [TIFF OMITTED] T2196.004
[GRAPHIC] [TIFF OMITTED] T2196.005
[GRAPHIC] [TIFF OMITTED] T2196.006
[GRAPHIC] [TIFF OMITTED] T2196.007
[GRAPHIC] [TIFF OMITTED] T2196.008
[GRAPHIC] [TIFF OMITTED] T2196.009
[GRAPHIC] [TIFF OMITTED] T2196.010
[GRAPHIC] [TIFF OMITTED] T2196.011
[GRAPHIC] [TIFF OMITTED] T2196.012
[GRAPHIC] [TIFF OMITTED] T2196.013
[GRAPHIC] [TIFF OMITTED] T2196.014
[GRAPHIC] [TIFF OMITTED] T2196.015
[GRAPHIC] [TIFF OMITTED] T2196.016
[GRAPHIC] [TIFF OMITTED] T2196.017
[GRAPHIC] [TIFF OMITTED] T2196.018
[GRAPHIC] [TIFF OMITTED] T2196.019
[GRAPHIC] [TIFF OMITTED] T2196.020
[GRAPHIC] [TIFF OMITTED] T2196.021
[GRAPHIC] [TIFF OMITTED] T2196.022
[GRAPHIC] [TIFF OMITTED] T2196.023
[GRAPHIC] [TIFF OMITTED] T2196.024
[GRAPHIC] [TIFF OMITTED] T2196.025
[GRAPHIC] [TIFF OMITTED] T2196.026
-