[Senate Hearing 105-871]
[From the U.S. Government Publishing Office]
S. Hrg. 105-871
GAO REPORT ON HIGH PERFORMANCE COMPUTERS
=======================================================================
HEARING
before the
SUBCOMMITTEE ON INTERNATIONAL SECURITY, PROLIFERATION, AND FEDERAL
SERVICES
of the
COMMITTEE ON
GOVERNMENTAL AFFAIRS
UNITED STATES SENATE
ONE HUNDRED FIFTH CONGRESS
SECOND SESSION
__________
SEPTEMBER 16, 1998
__________
Printed for the use of the Committee on Governmental Affairs
----------
U.S. GOVERNMENT PRINTING OFFICE
51-563 CC WASHINGTON : 1999
_______________________________________________________________________
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
WILLIAM V. ROTH, Jr., Delaware JOHN GLENN, Ohio
TED STEVENS, Alaska CARL LEVIN, Michigan
SUSAN M. COLLINS, Maine JOSEPH I. LIEBERMAN, Connecticut
SAM BROWNBACK, Kansas DANIEL K. AKAKA, Hawaii
PETE V. DOMENICI, New Mexico RICHARD J. DURBIN, Illinois
THAD COCHRAN, Mississippi ROBERT G. TORRICELLI, New Jersey
DON NICKLES, Oklahoma MAX CLELAND, Georgia
ARLEN SPECTER, Pennsylvania
Hannah S. Sistare, Staff Director and Counsel
Leonard Weiss, Minority Staff Director
Lynn L. Baker, Chief Clerk
------
SUBCOMMITTEE ON INTERNATIONAL SECURITY, PROLIFERATION AND FEDERAL
SERVICES
THAD COCHRAN, Mississippi, Chairman
TED STEVENS, Alaska CARL LEVIN, Michigan
SUSAN M. COLLINS, Maine DANIEL K. AKAKA, Hawaii
PETE V. DOMENICI, New Mexico RICHARD J. DURBIN, Illinois
DON NICKLES, Oklahoma ROBERT G. TORRICELLI, New Jersey
ARLEN SPECTER, Pennsylvania MAX CLELAND, Georgia
Mitchel B. Kugler, Staff Director
Linda J. Gustitus, Minority Staff Director
Julie A. Sander, Chief Clerk
C O N T E N T S
------
Page
Opening statement:
Senator Cochran.............................................. 1
Senator Thompson [ex officio]................................ 7
Senator Levin................................................ 11
Prepared statement of Senator Levin.............................. 39
WITNESSES
Wednesday, September 16, 1998
Harold J. Johnson, Associate Director, International Relations
and Trade Issues, U.S. General Accounting Office, accompanied
by James Shafer, National Security and International Affairs
Division and Jeffrey D. Phillips, National Security and
International Affairs Division................................. 2
Hon. William Reinsch, Under Secretary for Export Administration,
U.S. Department of Commerce.................................... 24
Alphabetical List of Witnesses
Johnson, Harold J.:
Testimony.................................................... 2
Prepared statement........................................... 41
Reinsch, Hon. William:
Testimony.................................................... 24
Prepared statement........................................... 123
APPENDIX
GAO report entitled ``Export Controls, Information on the
Decision to Revise High Performance Computer Controls,''
September 1998, GAO/NSIAD-98-250, submitted by Harold J.
Johnson........................................................ 54
GAO response to the comments of the Stanford Study authors to GAO
report entitled ``Export Controls: Information on the Decision
to Revise High Performance Computer Controls with response on
the Stanford Study from Dr. Goodman............................ 104
GAO comments on the response of the Department of Defense to GAO
report entitled ``Export Controls: Information on the Decision
to Revise High Performance Computer Controls with response on
the Stanford Study from Mr. Reinsch............................ 121
Letter dated December 8, 1998 from the Secretary of Defense,
Department of Defense.......................................... 122
Article from Foreign Affairs magazine entitled ``The Great
Technology Giveaway? Trading With Potential Foes,'' by Michael
Hirsch, September/October 1998, submitted by Mr. Reinsch....... 126
GAO REPORT ON HIGH PERFORMANCE COMPUTERS
----------
WEDNESDAY, SEPTEMBER 16, 1998
U.S. Senate,
Subcommittee on International Security,
Proliferation, and Federal Services,
of the Committee on Governmental Affairs,
Washington, DC.
The Subcommittee met, pursuant to notice, at 2 p.m. in room
342, Senate Dirksen Building, Hon. Thad Cochran, Chairman of
the Subcommittee, presiding.
Present: Senators Cochran, Levin, and Thompson [ex
officio].
OPENING STATEMENT OF SENATOR COCHRAN
Senator Cochran. The Subcommittee will come to order. I
apologize for the delay in convening the hearing. We appreciate
very much the attendance of our witnesses and welcome everyone
to this hearing of our Subcommittee on International Security,
Proliferation, and Federal Services.
Today we are reviewing a report by the General Accounting
Office on the administration's 1995 decision to liberalize
export controls on high performance computers, known as
``HPCs,'' or ``supercomputers.''
In October 1995, President Clinton announced a broad
decontrol of HPCs, which subsequently reduced the number of
individual licenses granted for supercomputer exports by almost
90 percent.
After learning that the Russian and Chinese nuclear weapons
complexes had obtained American supercomputers following the
administration's change in policy, Congress acted last year to
ensure that the Federal Government would have a greater role in
determining who would be using exported high performance
computers and how they would be used in certain foreign
countries that pose proliferation threats.
At the same time, I asked the General Accounting Office to
study the analysis on which the administration based its 1995
decision to liberalize high performance computer export
controls, and to assess the soundness of the administration's
current policy. This study can also serve as a basis for
analyzing any future administration proposals to decontrol
supercomputer exports.
Our witnesses today include Harold Johnson, Associate
Director, General Accounting Office for International Relations
and Trade Issues, who will present the GAO's report.\1\ He is
joined at the witness table by two of his associates from GAO
who directed this study, James Shafer and Jeffrey Phillips.
---------------------------------------------------------------------------
\1\ GAO report entitled ``Export Controls: Information on the
Decision to Revise High Performance Computer Controls,'' September
1998, GAO/NSIAD-98-196, submitted by Mr. Johnson, appears in the
Appendix on page 54.
---------------------------------------------------------------------------
We also have with us today Commerce Under Secretary William
Reinsch, who will respond with the administration's view of the
GAO report.
I am happy to yield to the distinguished Chairman of the
Governmental Affairs Committee, Senator Thompson, for any
comments or remarks he would like to make.
Senator Thompson. Thank you, Mr. Chairman. I think I'll
wait and follow regular order.
I do have to leave in about 25 mintues for a downtown
meeting, but if we can follow the regular order, then I think I
can get my questions in. Thank you very much.
Senator Cochran. Thank you, Senator.
Let me ask the Associate Director of the General Accounting
Office, Harold Johnson, to proceed.
STATEMENT OF HAROLD J. JOHNSON,\2\ ASSOCIATE DIRECTOR,
INTERNATIONAL RELATIONS AND TRADE ISSUES, U.S. GENERAL
ACCOUNTING OFFICE, ACCOMPANIED BY JAMES SHAFER, NATIONAL
SECURITY AND INTERNATIONAL AFFAIRS DIVISION AND JEFFREY D.
PHILLIPS, NATIONAL SECURITY AND INTERNATIONAL AFFAIRS DIVISION
Mr. Johnson. Thank you very much, Mr. Chairman. We are
pleased to be here to discuss the export controls on high
performance computers.
---------------------------------------------------------------------------
\2\ The prepared statement of Mr. Johnson appears in the Appendix
on page 41.
---------------------------------------------------------------------------
As you mentioned, you had expressed concern about sales to
Russia and China and asked us to do an assessment of the basis
for the Executive Branch's revision to HPC export controls, and
to identify changes in licensing activities and export
enforcement requirements resulting from these revisions.
You also asked us to determine the current foreign
availability of HPCs, particularly for countries of national
security concern.
Because the unlicensed exports to Russia and China that you
mentioned were under investigation by Commerce, Justice, and
Customs Service, we did not specifically address this matter in
our assessment. Also, I would like to point out that our
examination did not determine the appropriate threshold levels
for controlling HPCs. Instead, as you requested, we evaluated
the process by which the Executive Branch made its decision and
the adequacy of the information available for that purpose.
Our report on the decision to revise export controls is
being released today, as is our companion report responding to
a requirement of the fiscal year 1998 National Defense
Authorization Act. Therefore, my prepared statement is fairly
short and I will summarize our principal findings. But first, I
think a little background might be useful.
As we have stated in previous testimony, the U.S. export
control system is basically about managing risk. Exports to
some countries involve less risk than to other countries, and
exports of some items involve less risk than others. The
President has the responsibility and the authority to control
and require licenses for the export of licenses that may pose a
national security or foreign policy concern, and he may remove
or revise export controls as U.S. concerns and interests
change. It should be noted that the law does not require that a
foreign availability analysis be performed when deciding to
remove or relax controls.
In 1995, the Executive Branch conducted a review of export
controls on high performance computers to determine how changes
in computer technology and its military applications should
affect U.S. export control regulations. This review was a
continuation of a process that had begun in the 1980's to take
into account the technological advances in the computer
industry. And I would point out that as recently as 1993, the
export of computers with a composite theoretic performance of
195 MTOPS were controlled. This was raised to 1,500 MTOPS in
February, so there have been changes as we have progressed
along that timeline.
A key element of the Executive Branch's review was a
Stanford University study, jointly commissioned by the
Departments of Commerce and Defense. Among other things, the
Stanford report stated that, first, U.S.-manufactured computers
with a composite theoretic performance of up to 4,000 to 5,000
MTOPS were currently widely available and uncontrollable
worldwide; second, that computers with a performance level of
up to 7,000 MTOPS would become widely available and
uncontrollable worldwide by 1997; and, third, that many high
performance computer applications used in national security
programs occur at about 7,000 MTOPS, or above 10,000 MTOPS.
The report also stated that it would be too expensive for
the government and industry to effectively control exports of
computing systems with performance below 7,000 MTOPS, and that
attempts to control HPC exports below this level would become
increasingly ineffectual and an unreasonable burden on the
computer industry.
In announcing its 1996 change to export controls, the
Executive Branch stated that one goal of the revised export
control was to permit the government to tailor the controls to
levels and licensing conditions to the national security or
proliferation risks posed at specific destinations. The revised
export control policy removed license requirements for most HPC
exports with performance levels up to 2,000 MTOPS, an increase
from what I mentioned previously, of 1,500 MTOPS.
The policy also organized the countries into four computer
tiers, with each tier--after tier 1--representing a
successively higher level of concern to U.S. security
interests.
A dual control system was established for tier 3, such
countries as Russia and China. For these countries, HPCs of up
to 7,000 MTOPS could be exported to civilian end-users without
a license, while exports at or above 2,000 MTOPS to end-users
of concern for military or proliferation of weapons of mass
destruction reasons did require a license. Exports of HPCs
above 7,000 MTOPS in this category also required a license.
These are shown graphically in the report on page 25 for
reference.
The January, 1996 regulation also made other changes. It
specified that the exporters would be responsible for
determining whether an export license is required based on the
MTOPS levels of the computer, screening end users and end uses
for military or proliferation concerns, and keeping records and
reporting on exports of computers with performance levels of
2,000 MTOPS.
The 1998 Defense Authorization Act modified the regulations
by requiring exporters to notify the Commerce Department of any
planned sales of computers with performance levels greater than
2,000 MTOPS to tier 3 countries, and the government now has 10
days to assess and notify the vendor whether a proposed HPC
sale requires a license.
The law also now requires Commerce to perform post-shipment
verification on all HPC exports with performance levels above
2,000 MTOPS to tier 3 countries.
As I indicated, one focus of our work was to assess whether
the empirical evidence presented in the Stanford study, which
was a key element in the decision to revise HPC export
controls, supported its conclusions. Our analysis showed that
it had two significant limitations.
First, the study lacked empirical evidence or analysis to
support its conclusions that HPCs were uncontrollable based on,
one, worldwide availability, and two, insufficient resources to
control them.
Second, the study did not assess the capabilities of
countries of concern to use HPCs for military or other national
security applications, as required by its tasking.
The study's principal author said that the U.S. Government
data were insufficient to make such an assessment, and the
study recommended that better data be gathered so that such an
analysis could be done in the future.
Except for nuclear weapons, the Executive Branch has not
completed an assessment of the national security risks of
exporting HPCs to tier 3 countries, and the nuclear assessment
was completed by the Department of Energy in June of this year,
more than 2 years after the export control policy for HPCs was
revised.
The Executive Branch has identified high performance
computing as having applications in such national defense areas
as nuclear weapons programs, cryptology, conventional weapons,
and military operations, and with the exception of nuclear
weapons, the Executive Branch has not identified how and at
what performance levels specific countries of concern may use
HPCs for national defense applications, an important factor in
assessing the risk of the sales of HPCs to those countries.
In December of last year, the House Committee on National
Security directed the Departments of Energy and Defense to make
that assessment. DOE's study on nuclear weapons has shown that
nuclear weapons programs in tier 3 countries, especially in
China, India, and Pakistan, could benefit from the acquisition
of HPC capabilities. The Executive Branch has not finished
identifying how specific countries of concern could use HPCs
for non-nuclear national defense applications.
Based on Commerce's view of the worldwide availability of
computing power and the technological advancements in this
area, the Executive Branch raised the MTOPS threshold for HPCs.
The 1996 revision to export control policies had three key
consequences.
First, by increasing the performance threshold of computers
that require license, the 1996 revision decreased the number of
license applications from 459 in 1995 to 125 in 1997, and of
approved export license applications for HPCs from 395 in 1995
to 42 in 1997.
Second, as I have indicated, the revision shifted some of
the government's end-use screening responsibilities from the
government to the computer industry. In essence, the exporter
had to decide whether a license was required, since the
decision is made on the basis of end use, the end user, and a
computer's performance capability. This decision could be
particularly difficult for some tier 3 countries like China,
where identifying the distinction between a civilian and a
military end user can be very difficult.
In response to several allegations of improper sales to
Russia and China, Congress partly reversed this situation by
passing the fiscal year 1998 National Defense Authorization
Act, which requires exporters to notify Commerce of all HPC
sales over 2,000 MTOPS to tier 3 countries prior to their
export.
Third, the regulation required HPC manufacturers to keep
records of end users of all HPC exports over 2,000 MTOPS. Based
on our review of records provided by the manufacturers to the
Commerce Department from January 1996 through September 1997,
we noted that China ranked first in the number of HPCs acquired
by tier 3 countries, having purchased a total of 77 HPCs during
this period. These exports were all made without an individual
license being required; in other words, they were supposedly to
civilian end users, and below 7,000 MTOPS. Examining how these
machines are being used, however, was beyond the scope of our
review.
Responsibility for post-shipment verification checks on
exports remained with the government, but information on these
exports reported to the government has been incomplete. Post-
shipment verifications for computers generally have been of
somewhat reduced value because of how the process is
implemented. First, post-shipment verification really only
verifies the physical location of the HPC, not how it is used.
There are some exceptions to that. Also, some governments, such
as China, have not allowed the United States to conduct post-
shipment verifications.
With regard to foreign availability, we found that
subsidiaries of U.S. computer manufacturers dominate the
overseas HPC market, and they must comply with U.S. controls.
Russia, China, and India have developed HPCs, but their
capabilities are believed to be limited.
Thus, our analysis suggests that HPCs over 2,000 MTOPS are
not readily available to tier 3 countries from foreign sources
without restriction.
Our report contained two recommendations, one that requires
action by the Secretary of Defense and one that requires action
by the Secretary of Commerce, with support from Defense,
Energy, State, and the Arms Control and Disarmament Agency.
First, we recommended that to compliment the studies
undertaken by DOD and DOE for the House Committee on National
Security, that the Secretary of Defense should assess and
report on the national security threat and proliferation impact
of U.S. exports of HPCs to countries of national security and
proliferation concern. It seems to us that this assessment
should, at a minimum: one, address how and at what performance
levels countries of concern use HPCs for military modernization
and proliferation activities; two, whether such uses are a
threat to U.S. national security interests; and three, the
extent to which such HPCs are controllable.
Second, and these would have to come in sequence, upon
completion of that analysis we would recommend that the
Secretary of Commerce, in conjunction with the other agencies
that I mentioned, jointly evaluate and report on options to
safeguard U.S. national security interests regarding HPCs. Such
options should include, but not be limited to, one, requiring
government review and control of the export of computers at
their highest scalable MTOPS performance levels, and two,
requiring that HPCs destined for tier 3 countries be physically
modified to prevent their upgrade beyond allowable levels.
These are just some suggestions as to areas that ought to
be considered. There may be others.
I would like to comment just briefly on the agency's
response to our report. In addition to Commerce and Defense,
the Departments of Energy and State, as well as the Arms
Control and Disarmament Agency, offered some views.
Commerce said that the President's decision was intended to
change the computer export policy from what it referred to as
``a relic of the Cold War'' to one more in tune with today's
technology and international security environment, and was
based on, one, rapid technological change in the computer
industry; two, wide availability; three, limited
controllability; and four, limited national security
applications for HPCs. Commerce further stated that our report
focused too much on how countries might use HPCs for
proliferation and military uses rather than on what it called
``an outdated Cold War concept of foreign availability.'' The
Department said that our analysis of foreign availability was
too narrow and that foreign availability is not an adequate
measure of the problem.
Now, we agree that rapid technological advances in the
computer industry have made the controllability of HPCs a more
difficult problem. However, we disagree that foreign
availability is necessarily an ``outdated Cold War concept''
that has no relevance in today's environment. While the threats
to U.S. security may have changed, they have not been
eliminated. Commerce itself recognized this in its March, 1998
annual report to the Congress in which it stated, ``The key to
effective export controls is setting control levels above
foreign availability.''
Commerce also commented that the need to control the export
of HPCs because of their importance for national security
applications is limited. It stated that many national security
applications can be performed satisfactorily on uncontrollable
low-level technology, and that computers ``are not a choke
point for military production.'' Commerce said that having
access to HPCs alone will not improve a country's military and
industrial capabilities.
While it may be true that many of our modern weapons
systems were designed with less powerful computer systems,
Commerce's view seems to be inconsistent with the requirement
for DOD to identify militarily-critical technologies. In
assessing these militarily-critical technologies, DOD has
determined that high performance computing is an enabling
technology for modern tactical and strategic warfare, and also
important in the development, deployment, and use of weapons of
mass destruction.
High performance computing has also played a major role in
the ability of the United States to maintain and increase the
technological superiority of its war-fighting support systems.
DOD has noted in its High Performance Computing Modernization
Program Annual Plan that ``the use of HPC technology has led to
lower costs for systems deployment and improved the
effectiveness of complex weapons systems.''
DOD further stated that as it transitions its weapons
systems design and testing process to rely more heavily on
modelling and simulation, the Nation can expect more examples
of the profound effect that HPC capability has on both military
and civilian applications.
In DOD's comments on our report, it said that it had
considered the threats associated with HPC exports to countries
of national security and proliferation concern. DOD referred to
its identification of how HPCs in the United States are used
for national security applications.
While our report recognized that such an assessment of
domestic uses had been done, this did not answer our concern.
We reported that the Stanford study did not assess the
capabilities of countries of concern, such as China, Russia,
Indian, or Pakistan, to use HPCs for military or other national
security applications, as required by its tasking, and the
Executive Branch did not undertake a specific threat analysis
of providing HPCs to such countries.
As we reported, the principal author of the Stanford study
noted that no assessment had been done of the national security
impact of allowing HPCs to go to particular countries of
concern, and of what military advantage such countries could
achieve. In fact, in its most recent report--its April, 1998
report--on HPC export controls, the same principal author also
noted that identifying which countries could use HPCs to pursue
which military applications remained a critical issue on which
the Executive Branch provided little information. State,
Energy, and ACDA all generally agreed with our report.
That concludes my prepared statement.
Senator Cochran. Thank you, Mr. Johnson, for your statement
and for your work on this report.
With the concurrence of the distinguished Ranking Member of
the Subcommittee, we're going to yield first to the
distinguished Chairman of the full Committee, Senator Thompson,
for any questions or remarks he would like to make.
OPENING STATEMENT OF SENATOR THOMPSON
Senator Thompson. Thank you very much, Mr. Chairman.
First of all, let's establish exactly what it is we're
dealing with regarding these sophisticated computers and their
significance with regard to nuclear proliferation and military
uses.
Referring to the final paragraph on page 7 of your report,
you say, ``Prior to the Executive Branch's decision to change
computer thresholds, scientists at the Department of Energy
National Laboratories and other U.S. Government officials had
accumulated information to show how countries of concern could
use HPCs to facilitate the design of nuclear weapons and to
improve advanced nuclear weapons in the absence of tests of
nuclear explosives. However, this information was not used as
part of the decisionmaking process.'' Is that correct?
Mr. Johnson. That's what was reported to us, yes, sir.
Senator Thompson. So the Department of Energy's report came
down after the new rules went into effect in January 1996?
Mr. Johnson. Well, the report came down after the new rules
came into effect, but the information that was available that
you referred to was available prior to that.
Senator Thompson. All right.
Mr. Johnson. As we understand it, it wasn't necessarily put
into the form of a report.
Senator Thompson. So that information had been available,
certainly----
Mr. Johnson. It was available, had it been sought.
Senator Thompson [continuing]. At the time that the
Stanford report was being written. Is that correct?
Mr. Johnson. That is correct, yes.
Senator Thompson. That these computers could facilitate the
design of nuclear weapons.
I refer to page 17 of your report, where the Department of
Defense comments on this issue. It says, ``Moreover, Commerce's
position on this matter is not consistent with that of''--
they're talking about another position of Commerce, different
from the one I just articulated--``DOD, in its Militarily
Critical Technologies List, has determined that high
performance computing is an enabling technology for modern
tactical and strategic warfare, and is also important in the
development, deployment, and use of weapons of mass
destruction. High performance computing has also played a major
role in the ability of the United States to maintain and
increase the technological superiority of its war-fighting
support systems.'' And then you go on to point out that more
and more we have to rely upon simulation and modelling and so
forth, that it is more significant.
So you have got both the Department of Energy and the
Department--when was this Militarily Critical Technologies List
published?
Mr. Johnson. Well, they do one periodically, I think on an
annual basis.
Mr. Phillips. It's essentially on an ongoing basis, and
information from the one during the 1995 period was carried
over into 1996.
Senator Thompson. So this is the one that was extant at the
time that this report was produced?
Mr. Johnson. Right.
Senator Thompson. So you have both the Department of Energy
and the Department of Defense saying that these are computers
that can assist people in their design of nuclear weapons and
is an enabling technology for modern tactical and strategic
warfare and the use of weapons of mass destruction.
So this is the background that we have when the
administration decided to change the rules of the game in
January 1996 and to lower the bar and make it easier for us to
export these computers?
Mr. Johnson. Right.
Senator Thompson. So the answer, as I understand it, is
multi-fold. I am amazed that the Department of Defense is under
the impression that the issue of whether or not these can be
converted for military uses was addressed by the Stanford
report, when in fact it obviously was not.
Mr. Johnson. Well, one small part of it, I think, to give
them their due. They did consider how HPCs were used
domestically in our weapons system development and other
aspects. But that doesn't convert, right.
Senator Thompson. Well, our major concern is not in that
area, is it? I mean, it's what China is doing with them, it's
what India is doing with them.
Mr. Johnson. The countries of concern----
Senator Thompson. That was not addressed?
Mr. Johnson. That was not addressed.
Senator Thompson. So as I understand it, they rely upon the
Stanford study to justify what they did, a study which to me is
pretty clearly based on commercial interests more than anything
else, as evidenced once again by the lead of the Department of
Commerce. But they rely on this study, and you found when you
looked at this study, it was faulty in more than one respect.
First of all, there is the argument that ``everybody does it,
everybody has it.'' You looked at the lay of the land, and
first of all, there is nothing in the report--and, I take it,
nothing in your conversations with the people who did the
report--to indicate any backup or any basis for the conclusions
that they came to. Is that correct?
Mr. Johnson. The concern that we had was that there was a
lack of empirical evidence to support the conclusions that they
had come to.
Senator Thompson. And when you looked at the situation, you
found that--referring to page 12--the only global competitors
for general computer technology are three Japanese companies,
two of which primarily compete for sales of high-end computers,
systems sold in small volumes and performing at advanced
levels. Two of those companies reported no HPC exports to tier
3 countries; and China and Pakistan, of course, are tier 3
countries----
Mr. Johnson. Correct.
Senator Thompson [continuing]. While the third company
reported some exports on a regional, rather than country,
basis. That's Japan.
One German company sells these computers, primarily in
Europe, and has reported several sales over 2,000 MTOPS to tier
3 countries.
One was a British company, but this company said that it
never sold a system outside the European Union.
You say that a 1995 Commerce Department study of this
global market showed that American dominance had prevailed at
that time, as well. So basically what you're saying is that
America has the clear dominance in this area?
Mr. Johnson. That's right.
Senator Thompson. And has had for some time?
Mr. Johnson. Yes.
Senator Thompson. And you also state, ``Available
information indicates the capabilities of China, India, and
Russia to build their own HPCs still lag well behind that of
the United States, Japan, and European countries.'' Is that
correct?
Mr. Johnson. That is what was reported to us, right.
Senator Thompson. Another problem you had with the report,
besides the fact that there was not supporting documentation
for the conclusion that this was uncontrollable and everybody
has it or is going to soon have it, is that this report was
supposed to assess the capability of countries of concern to
use these computers for military purposes. That was supposed to
be a part of this Stanford report, wasn't it?
Mr. Johnson. Yes. That was one of the taskings, but we also
recognize and acknowledge that that tasking could not have been
fulfilled because of a lack of information that only the U.S.
Government can provide.
Senator Thompson. Well, I understand. I'm not asking you
why it wasn't fulfilled, I'm asking, was it supposed to be in
there? And the answer to that is yes?
Mr. Johnson. Yes.
Senator Thompson. And the next question is, was it in
there?
Mr. Johnson. No, it was not.
Senator Thompson. And the answer to that is no. In fact, in
the report they refer to the fact that it's not in there. They
acknowledge that they do not address that, and they have
certain recommendations as to what they feel needs to be done
in order for this government to determine whether or not the
countries to which these are being shipped could convert them
for military purposes. But as far as we know, that was not
done.
Mr. Johnson. That has not been done, and that
recommendation is contained in Dr. Goodman's April, 1998 study,
as well. So that is still an area of concern.
Senator Thompson. I notice on page 8 of your report, the
DOE study concluded that ``The acquisition and application of
these computers to nuclear weapons development would have the
greatest potential impact on the Chinese nuclear program,''----
Mr. Johnson. Yes.
Senator Thompson [continuing]. ``Particularly in the event
of a ban on all nuclear weapons testing.'' Also, the study
indicates that India and Pakistan may now be able to make
better use of the HPCs in the 1,000 to 4,000 MTOPS range for
their nuclear weapons programs because of the testing data that
they acquired in May 1998 from underground detonations of
nuclear devices. Is that right?
Mr. Johnson. That is correct.
Senator Thompson. So now it appears, as of the time of this
report, the opinion is that India and Pakistan can use these
computers in the 1,000 to 4,000 MTOPS range for their nuclear
programs.
On page 10, where you're dealing with the fact that license
applications have declined, you say, ``Also during this period,
77 HPCs were exported to China and 19 were exported to India,
all without individual licenses. Most U.S. HPCs exported in
this period, about 85 percent, had performance levels between
2,000 and 5,000 MTOPS.''
Mr. Johnson. Yes.
Senator Thompson. So am I to understand that some computers
with 2,000 and 5,000 MTOPS were exported to both China and
India?
Mr. Johnson. Yes. In fact, all of the ones that were
exported during this period of time to China and India were
within that range, that are included in that 77.
Senator Thompson. And that is the range that has the
greatest potential impact on the Chinese program, and India and
Pakistan may better use computers in that range for their
nuclear weapons programs. I find that amazing.
I have no further questions.
Senator Cochran. Thank you very much, Mr. Chairman.
Senator Levin, I'm going to call on you now, if you would
like to proceed with your questions, then I will follow you.
OPENING STATEMENT OF SENATOR LEVIN
Senator Levin. Thank you, Mr. Chairman. Thank you, and I
also have a prepared statement which I would appreciate being
made a part of the record.\1\
---------------------------------------------------------------------------
\1\ The prepared statement of Senator Levin appears in the Appendix
on page 39.
---------------------------------------------------------------------------
Senator Cochran. Without objection it will be printed in
the record.
Senator Levin. Thank you, Mr. Chairman.
First of all, in terms of the national security questions
which are the most significant questions, as part of the
studies which you have done for this Subcommittee and for the
Senate Armed Services Committee, did the GAO assess and reach a
conclusion as to whether the 1996 restructuring of computer
export controls has had any negative effect on U.S. national
security?
Mr. Johnson. We don't have that kind of capability. But
what we looked at was whether or not the Executive Branch had
performed that, and they had not.
Senator Levin. Let me get to the DOD questions, because you
said that they have not done it, and yet the DOD says that it
did. In their letter to Mr. Nelson, the DOD says that ``The GAO
draft report inaccurately states that DOD did not consider the
threats associated with high performance computer exports. DOD
did take into account the security risks associated with the
export of HPCs to countries of national security and
proliferation concern. DOD identified numerous national
security applications that require various levels of computing
power, which helped to determine licensing policies for the
various country groups and to establish specific safeguards on
computer exports. Countries of greatest national security and
proliferation concern are subject to the most stringent
licensing and safeguard requirements.''
Do you disagree with that statement of the DOD?
Mr. Johnson. Yes. Substantially, what DOD is referring to
are the analyses and the information that they provided on how
HPCs are used for domestic--for our own weapons systems
development. Now, when the tier system was created, they did
take into account--they had six different criteria, and took
into account the potential of how a country might use an HPC,
whether they were a member of the nonproliferation treaty, and
a number of other criteria. But they did not do a specific
threat analysis of how individual countries might use HPCs or
how that might threaten our own national security. That's the
part that was not done.
Senator Levin. Are you disagreeing with the statement of
the DOD which says that ``the GAO draft report inaccurately
states that DOD did not consider the threats associated with
high performance computer exports.'' You disagree with that?
Mr. Johnson. Yes.
Senator Levin. So there's just a difference here between
the DOD and the GAO on that specific statement?
Mr. Johnson. Right.
Senator Levin. And then it says in the next sentence, ``The
DOD did take into account the security risks associated with
the exports of HPCs to countries of national security and
proliferation concern.'' On that specific sentence, do you
disagree with that?
Mr. Johnson. Not entirely. Like I said, they did take some
aspects of that into account as part of their development of
the tier system.
Senator Levin. On that sentence, then, you agree in part
and disagree in part?
Mr. Johnson. Right.
Senator Levin. And on the first sentence, you say you just
disagree with it.
On the third sentence, ``DOD identified numerous national
security applications that require various levels of computing
power, which helped to determine licensing policies for the
various country groups and to establish specific safeguards on
computer exports.'' You disagree in part, or totally----
Mr. Johnson. I totally disagree with the specific words
that they use there because, as I mentioned, they are referring
to the determination of how we use those HPCs in weapons
development.
Senator Levin. And the fourth statement in that paragraph,
``Countries of greatest national security and proliferation
concern are subject to the most stringent licensing and
safeguard requirements.'' Do you disagree with that?
Mr. Johnson. Well, I really can't respond to that because,
like I said, we didn't ourselves develop special levels. The
export controls that have been established are being enforced
to the extent that they can, but I can't comment specifically
on that assertion.
Senator Levin. OK. The next paragraph says the following,
``The GAO recommends that the Secretary of Defense assess how
and at what performance levels countries of concern use HPCs
for military modernization and proliferation activities,'' and
then they go on to say, ``These factors were taken into account
by DOD and the interagency process in the 1995 review of
computer export controls.'' Do you disagree with that?
Mr. Johnson. Well, I don't know what they mean, ``taken
into account,'' but I do know that they didn't do an assessment
of countries of concern and how they might use high performance
computers for the development of their systems.
Senator Levin. Well, let me re-read it to you. It seems to
me there's just a conflict on this one----
Mr. Johnson. Well, there may be. What I don't know is what
they mean by they ``took it into account.'' Did they think
about it? We were unable to see any study, any documentation
where they took that into account.
Senator Levin. So you don't deny that they took that into
account, you just haven't seen the documents in which they did
take them into account?
Mr. Johnson. Yes, it's like suggesting that somebody
consider something, and they may consider it for 30 seconds,
and go on. But we did not see a study or an analysis that would
satisfy what we were intending in our recommendation.
Senator Levin. So again, I just want to be really clear on
this, because we ought to ask the DOD these questions, Mr.
Chairman. I would hope we would ask the DOD for the information
here, because there seems to be a conflict between the
Department of Defense and the GAO here, and I just want to
repeat this sentence.
Senator Cochran. We received a request to invite a witness
from the Department of Defense. Our staff called and talked to
Sandi Stewart over in the Secretary's office to convey that
request. The Defense Department was unable to provide a
witness, but we would be happy to hear from them at some future
point.
Senator Levin. Well, I don't know if anyone's here. I have
no idea. But in any event, I think we ought to ask the
Department of Defense that question because there is a
difference here between the GAO and the Department of Defense.
The Department of Defense----
Senator Cochran. We can submit that to them in letter form,
and I will join you in the request.
Senator Levin. All right. Because when the DOD represents
here that it represents to you folks--that these factors--I'd
better repeat the whole sentence.
``The GAO recommends that the Secretary of Defense assess
how and at what performance levels countries of concern use
HPCs for military modernization and proliferation activities.
These factors were taken into account by DOD and the
interagency process in the 1995 review of computer export
controls.''
It seems to me there's a conflict here between the GAO and
the DOD, and we ought to ask the DOD what they mean because
they obviously just differ with the GAO on this point.
It's clear there's a difference here, isn't that right?
Mr. Johnson. Yes.
Senator Levin. Now, in terms of the DOD, they also say the
following----
Your comment is that the Stanford University study did not
incorporate--did not consider certain material.
Here's what they say about the Stanford study. They say,
``The Stanford University study referred to in the GAO report
was just one of many inputs considered by the Executive Branch
in its 1995 assessment of computer export controls. Information
and analysis was also provided by various Defense components,
as well as other U.S. Government agencies, including the
intelligence community.''
Do you agree with that statement?
Mr. Johnson. Yes, but the key component, the key study, was
the Stanford study.
Senator Levin. Do you know what other inputs there were
from other U.S. Government agencies, including the intelligence
community, on the part that the Stanford study said it did not
have adequate information on?
Mr. Johnson. We know that there was a study by the
Institute for Defense Analysis, IDA, and that there were other
documents that we reviewed that were provided to us, because we
asked for everything, the whole range of considerations from
all the agencies--not just DOD, but from Commerce and Energy as
well, and we evaluated those.
Senator Levin. Did that include the intelligence community?
Mr. Johnson. Yes.
Senator Levin. So you would agree, then, with the statement
of the DOD that the Stanford University study was just one of
many inputs, and that information and analysis was also
provided by various Defense components, as well as other
government agencies, including the intelligence community?
Mr. Johnson. Right.
Senator Levin. You would agree with that.
Mr. Phillips. May I add an elaboration to that, a little
more detail?
Senator Levin. Sure.
Mr. Phillips. There were a few items that were presented to
us as being part of the study. One was information from the
intelligence community which was largely a ranking of countries
by general proliferation activities and concerns. There was
nothing from the intelligence community or DOD that looked at
how HPCs, particularly, were of a threat nature or were going
to be used. So it was more of general proliferation activity.
They looked at the effectiveness of the export controls of
different countries, that type of thing.
There was another study, one from Institute for Defense
Analysis, which really talked about a technical feasibility
study, on a very technical level, that did not seem to be a
major part of the information in the decision.
Finally, they alluded to some DOD analyses related to
Defense applications. Again, there was no specific document
that anyone provided to us to show us what that was, and the
information seemed to overlap considerably with what was
presented in the Stanford study. Some people told us that
whatever DOD did, they did collaboratively with the Stanford
study, and it was, again, incorporated in there together. They
also mentioned some Internet search, but they had not kept that
material.
That was about the extent of what was presented to us.
Senator Levin. Is there certain information that the
National Security Council had that was not made available to
you?
Mr. Johnson. Well, we don't know, because we asked to meet
with a National Security Council representative on this issue
and were told that they would not meet with us.
Senator Levin. All right.
You have a statement here that ``We requested, but were
denied access to, information from the National Security
Council and data and analyses that were used in the interagency
forum to reach the final decision to revise controls.'' So,
according to your statement here, anyway, there was certain
information from them that they would not release----
Mr. Johnson. That may exist----
Senator Levin [continuing]. That may exist that was not
provided to you?
Mr. Johnson. Right.
Mr. Phillips. The one point I would want to make on that,
when we talked to the NSC representatives, they wouldn't talk
to us but they referred us back to the agencies involved and
said that they would have whatever analysis they were using.
Senator Levin. Well, did they say there was no additional
analysis? Because your statement here says they would not give
you the data and analysis that were used, it says here, in the
interagency forum. So it implies that there may have been data
that was used there other than the data that you received from
the agencies. That's the implication here.
Mr. Phillips. No, we have no knowledge if there was
anything else that was used. Again, they wouldn't discuss the
decisionmaking with us, so we can't say what they actually
used. But as far as what was provided to us, or what we were
referred to, again, they said go back to the agencies and ask
them.
Senator Levin. Did you, or could you, list for the National
Security Council the data and analysis which you received from
the agencies and ask them whether there was any other material
which they used? Could you do that?
Mr. Johnson. Yes, we could.
Senator Levin. Because I think that would be useful to us,
so that we can find out whether or not there is additional data
which they don't wish to disclose, other than what you've
already seen from the agencies.
As I understand it, the DOD played a critical role in
designing this four-tier structure, is that correct?
Mr. Johnson. That's my understanding, yes.
Senator Levin. So this was essentially a DOD design?
Mr. Johnson. I'm not sure I would consider it essentially a
DOD design. I would say it's an Executive Branch design. It was
through an interagency process that that was done, but some of
the groundwork was by DOD.
Senator Levin. They support this design?
Mr. Johnson. Yes.
Senator Levin. And finally--I think I'm probably over my
time, Mr. Chairman, I'm sorry, I haven't tracked it--in the
tier 3 licenses, as I understand it, if a license is granted
and any of the agencies that are involved here, including DOD
or DOE, have an objection to the license, there is an appeal
process. Is that correct?
Mr. Johnson. That's correct.
Senator Levin. So any one of these agencies that are
involved here can effectively stop a sale if it doesn't approve
of it, and then kick it up two or three levels, is that
correct?
Mr. Johnson. That's correct.
Senator Levin. All right. I have additional questions but I
am way over my time limit, Mr. Chairman.
Senator Cochran. Let me go back if I may, Mr. Johnson, and
review the conclusions you reached about this report on which
the administration based much of its reason for making the
change in export control policy, the so-called Goodman Report
or the Stanford University report that was led by Dr. Goodman.
Your conclusion, as I understand it, is that the
administration placed great weight on that report----
Mr. Johnson. Yes.
Senator Cochran [continuing]. And that they----
Mr. Johnson. That was a key component of the decision
process.
Senator Cochran. That was what?
Mr. Johnson. A key component, I'm sorry.
Senator Cochran. One could conclude that it was a very
important part of the decisionmaking process, is that correct?
Mr. Johnson. That's correct.
Senator Cochran. And in your analysis of the Stanford
study, you tell us in your report that there was a lack of, in
your words, ``empirical evidence or analysis'' to support the
conclusions reached by that report. Is that correct?
Mr. Johnson. That's right, yes.
Senator Cochran. Well, do you conclude, therefore, after
you've reviewed the study and you've looked to find the
evidence on which the policy changes were based--do you
conclude that the Stanford study lacks credibility because its
assumptions and conclusions are not supported by the empirical
evidence or any analysis?
Mr. Johnson. I'm not sure I would want to characterize it
quite like that, but what I would say is that our conclusion is
that more information should have been made available to the
Executive Branch in making that decision. There should have
been further analysis done on the cost of implementing the
export controls, and on the aspects of controllability. The
conclusion that Dr. Goodman came to, that certain levels were
uncontrollable, simply didn't have the empirical data to
support that. I don't want to be so harsh as to say it lacks
credibility, but it did have those limitations.
Senator Cochran. One example that your report points out is
that officials could not explain nor provide documentation as
to how they arrived at the decision to set the license
requirements for exports of HPCs to tier 3 countries for
military or proliferation end users at 2,000 MTOPS.
Mr. Johnson. Right.
Senator Cochran. The study concluded that the computing
power 4,000 or 5,000 MTOPS was uncontrollable.
Mr. Johnson. That's right. Essentially, we were told that
that was an interagency process that was undocumented, and that
it was done that way to be conservative.
Senator Cochran. Could you tell us which administrative
officials or Executive Branch officials you are referring to?
Mr. Johnson. Well, we asked a number of Executive Branch
officials, but primarily in the Department of Commerce, and at
several levels, including the higher levels in the Department
of Commerce, Bureau for Export Administration.
Senator Cochran. Did these officials give you any
explanation for the recommendations that resulted in the
licensing parameters for tier 3 countries?
Mr. Johnson. Well, they did, in terms of the 2,000. They
explained that it was an interagency process and it was
established at 2,000 rather than at the higher levels, to be
more conservative. We asked for documentation to help us track
through that decisionmaking process, and we were not provided
documentation.
Senator Cochran. Were you told that there was any? What was
the response? Did you actually know of some documentation, or
suspect that there was some?
Mr. Johnson. No, we did not know of any.
Senator Cochran. The suspicion was the other way, then,
that there wasn't any?
Mr. Johnson. Yes.
Senator Cochran. I'm speculating, now, but----
Mr. Johnson. We did ask for the actual minutes of the
interagency meetings and were told that there were none.
Senator Cochran. Would it be fair to conclude that this was
a result of the influence of the computer industry and its
pressure on the administration to set that parameter as it was
set?
Mr. Johnson. I'm not sure that I would necessarily agree
with that conclusion. I would think that if there was strong
influence by the computer industry, that the level would have
been set higher than at the 2,000 level. But I'm purely
speculating. I can't speak for what was in their minds in
establishing the control levels.
Senator Cochran. You say in the report that the Stanford
study described uncontrollability as ``the relationship between
the difficulty of controlling computers, and the willingness of
government and industry to meet the costs of tracking and
controlling them.''
Mr. Johnson. Right.
Senator Cochran. Does this mean that computers could be
uncontrollable, simply based on a limited willingness on the
part of government and industry to control them?
Mr. Johnson. One could certainly interpret it that way. If
they decided that they--either government or industry--decided
that they didn't want to spend anything on controlling HPC's,
that would indicate an unwillingness to----
Senator Cochran. That means that based on the Stanford
study, then--the description is that ``U.S. high performance
computers are uncontrollable.''
Mr. Johnson. Well, if the conclusion were reached that they
didn't want to spend any money, then they clearly be
uncontrollable.
Senator Cochran. Does GAO agree with the Stanford study's
description of ``uncontrollability''?
Mr. Johnson. Well, in a sense we would, but we think that
in reaching that conclusion there should be some data on how
much is spent, what the cost is, and a comparison of cost to
risk. We didn't see any of that in the study.
Senator Cochran. Your report also says, Mr. Johnson, that
the Stanford study was ``tasked with assessing the capabilities
of countries of concern to use HPCs for military and other
national security applications, and it did not do so.''
Mr. Johnson. Correct.
Senator Cochran. Is that based on a written directive to
Dr. Goodman from the Department of Defense and the Department
of Energy?
Mr. Johnson. It's in the tasking, yes.
Senator Cochran. This was in the formal letter? So this was
a specific request that was made of the university?
Mr. Johnson. That's correct. It was in its tasking, right.
Senator Cochran. Were any officials in the Executive Branch
able to identify for you other studies that substantially
informed the October, 1995 decision to decontrol high
performance computers?
Mr. Johnson. No.
Senator Cochran. Even though I know, as Senator Levin has
pointed out, there is the Department of Defense letter
responding to the report,\1\ saying that there were other
considerations. But even in your efforts to find out what these
sources were or what the information was, no one ever produced
anything?
---------------------------------------------------------------------------
\1\ The letter referred to, dated December 8, 1998, appears in the
Appendix on page 122.
---------------------------------------------------------------------------
Mr. Johnson. That's right. When we talk about the
limitations of that study, I think we would come to the
conclusion that the most important limitation has to do with
the lack of determination as to national security risks
involved for providing HPCs to countries of concern.
It seems to us that if that analysis were made, and a
decision is made that there are no risks, then there is no
point in controlling computers, regardless of the capability.
There needs to be, first of all, the establishment of some
national security reason for doing that, for controlling high
performance computers. We see that as a significant limitation
in the analysis.
Senator Cochran. You use the terms in your report of
``scalability,'' ``upgrading,'' and ``clustering.''
Mr. Johnson. Right.
Senator Cochran. Could you tell us if there are problems
associated with these efforts and difficulties in adapting
software or the like?
Mr. Johnson. Yes, we can respond to that, but I'm going to
ask my expert to.
Mr. Phillips. Yes. As far as clustering goes, there is a
lot of disagreement within the computer community that we came
across about how mature that technology actually is. Clustering
refers to linking computers together, usually by some kind of
hard-wire interface. And the problems that come in--in theory,
the more processors you put together, the more power you should
get out of a computer system. But in reality, as you add more,
you also have limitations based on the memory and on the
communications speed between each of the processors
communicating with each other. So in reality, when you are
dealing with these systems in parallel, the benefit should be
that they can deal with different problems at different speeds,
but in reality the slowest part of the system is going to slow
down the whole system.
These are limiting factors in the capabilities of
clustering machines. Most people in the community think that
they are not insurmountable problems, and that technological
advances have been getting better. But still, at this point, a
number of people--including the authors of the Stanford study--
did not think that clustering technology in and of itself
should be driving the thresholds for the export controls.
Senator Cochran. In regard to scalability, could this be
detected in a post-shipment verification to try to determine
whether or not these computers have been upgraded by the end
user?
Mr. Johnson. If there is adequate expertise by those people
doing the post-shipment verification, it could be. The problem
with post-shipment verifications for the most part is that
people don't have the expertise to do that kind of testing or
analysis.
Senator Cochran. On that subject, who under the current
regime is doing the post-shipment verification? The industry?
Or government officials?
Mr. Johnson. Ordinarily it is government officials from the
local embassy. There are teams that are sent out from
Washington, called Special Verification Teams, and they
probably would most likely have the capability to detect a
machine that had been scaled up. But most of the post-shipment
verifications are done by officials of embassies that may not
have that kind of capability. So that is a limitation, but at
the very least they know whether or not the machine is
physically located there, so there is value to the post-
shipment verification.
Senator Cochran. Senator Levin, that's a good stopping
place for me in this round, if you would like to ask additional
questions.
Senator Levin. Thank you, Mr. Chairman. Just a few
additional questions.
We've gone through some of the disagreements that you have
with the Department of Defense. In their letter--it's not
dated, but it's about August 9----
Mr. Johnson. Yes, thereabouts. I think we received their
letter a few days after we received the response from Commerce.
Senator Levin. OK. But it's addressed to Benjamin Nelson of
the GAO.
Mr. Johnson. Yes.
Senator Levin. We've gone through some of the disagreements
between you and the Department of Defense. There is an
additional statement in that letter, and I want to see whether
you agree or disagree with that one as well.
That letter says that ``We in the Department of Defense
determined that computers with performance below the current
license threshold for tier 3 countries are widely available
globally.''
That's the first one. From what you've testified to
already, I take it that you disagree with that statement of the
DOD, that computers with performance below the tier 3 license
threshold are widely available globally. Would you disagree
with that?
Mr. Johnson. We don't have data on that, nor was data
presented in the Goodman study----
Senator Levin. No, I mean--it says here that the DOD
determined that. Do you disagree that they determined that?
Mr. Johnson. We did not see where they had made that
determination.
Senator Levin. All right. And you yourselves have not
determined whether or not computers below the threshold level
for tier 3 are widely available globally? You have not made
your own determination of that?
Mr. Johnson. No, but below the license level I would
believe that they would be widely available. But we don't have
that data, either.
Senator Levin. We're talking about tier 3 licensing?
Mr. Johnson. Yes.
Senator Levin. So you would believe that computers below
the 2,000 level are widely available?
Mr. Johnson. Probably.
Senator Levin. OK. That would sound like you agree with
them on that one.
Now, on Stanford, I gather that you had some discussions
with the Stanford study folks----
Mr. Johnson. Yes, indeed.
Senator Levin [continuing]. But their response to your
comments is not attached because it's not the practice of the
GAO----
Mr. Johnson. We----
Senator Levin. If I could finish.
Mr. Johnson. I'm sorry.
Senator Levin [continuing]. As I understand it, it's not
the practice of the GAO to incorporate the comments of
contractors in your own reports, whereas you would incorporate
comments of agencies that you're looking at? Is that accurate?
Mr. Johnson. That's generally our policy and our practice.
There are times when we do incorporate comments of contractors.
In this case, we didn't specifically seek written comments on
our reports. If the Commerce Department wished to do that, that
was their prerogative.
Senator Levin. The Commerce Department could have
incorporated the Stanford response in their response?
Mr. Johnson. That's correct. They could have done that. I
don't know that they did.
Senator Levin. All right.
Mr. Johnson. But we had several discussions with the
authors of the Stanford study and tried as best we could to
reflect their comments in the text of our report.
Senator Levin. Your report criticizes the Stanford study
for concluding that computers in the 4,000 to 5,000 MTOPS range
were widely available, without citing any empirical evidence
for their conclusions. Is that correct?
Mr. Johnson. Yes.
Senator Levin. And do they--when you talk to them, do they
agree with you?
Mr. Johnson. They felt that the information that they had
available, working from the theoretic basis that they were
working from, was sufficient.
Senator Levin. So they felt they did have adequate evidence
to reach that conclusion?
Mr. Johnson. Yes, but not specifically empirical data on
what the installed base is at any particular level, they did
not have.
Senator Levin. Just so that I am clear on this, they felt
that they did have an adequate basis for their conclusion, but
you felt they did not?
Mr. Johnson. Yes.
Senator Levin. So you disagree with the Stanford folks on
that specific issue.
Mr. Phillips. I'd like to add a point on that, Senator.
What they did feel less certain about, though, was the
precision of their own definition of what ``uncontrollable''
means, and that was the term that they used predominantly
throughout the report. Availability was just one aspect of
uncontrollability.
In reaching a conclusion that computers within the 4,000 to
5,000 MTOPS range were uncontrollable, they themselves admitted
that they did not fulfill the terms of the definition that they
had set up to do that.
Senator Levin. OK.
I think it would be useful if the Commerce--I don't think
the Commerce Department incorporated the Stanford response. I
may be wrong on this, but if they haven't, I think it would be
useful for the Subcommittee, Mr. Chairman, to request either
the Commerce Department or Stanford to comment on the findings
of the GAO relative to the Stanford study. \1\
---------------------------------------------------------------------------
\1\ GAO responses to the comments of the Stanford Study and DOD
appear in the Appendix on pages 104 and 121 respectively.
---------------------------------------------------------------------------
Senator Cochran. We have the Commerce Department
represented here today, so we can ask that question now.
Senator Levin. All right.
Senator Cochran. Do you want to write a letter to Stanford
and ask them for a response to the report?
Senator Levin. I think so.
Mr. Johnson. I think that would be fair.
Senator Cochran. Is Dr. Goodman still around?
Mr. Johnson. Yes.
Senator Cochran. He's been a witness in here, hasn't he?
Not yet?
Senator Levin. We have disagreements between GAO and DOD.
We have disagreements between GAO and Stanford, and it seems to
me we ought to give both the DOD and Stanford--ask them for
their explanation of their position, given the GAO report.
Thank you very much, Mr. Chairman, for doing that, and for
holding this hearing.
Senator Cochran. Mr. Johnson, one of the assumptions
underlying the decontrol of HPCs in 1995 was that there would
be widespread availability of HPCs, at least to the level of
7,000 MTOPS, by 1997. But according to your report, ``Our
analysis shows that subsidiaries of U.S. companies dominate the
overseas sales of HPCs.''
If U.S. companies dominate the overseas sale of HPCs, can
there be widespread availability of HPCs overseas only if the
United States elects not to control the computers?
Mr. Johnson. Not on an unrestricted basis.
Senator Cochran. The Department of Commerce suggested to
GAO in the course of conducting this study that foreign
availability is an outmoded Cold War concept--or that was your
report?
Mr. Johnson. No, that was our characterization of it.
Senator Cochran. Right. You included that in your report?
Mr. Johnson. Yes.
Senator Cochran. Has the Commerce Department, to your
knowledge, reported otherwise in the recent past to Congress?
Or does GAO agree with this suggestion, that it's an outmoded
Cold War concept?
Mr. Johnson. No, we don't agree that it's an outmoded
concept. And yes, Commerce has used that same terminology and
concept, which is described in the Export Administration Act,
in its most recent report to the Congress in March. It makes
reference to foreign availability being established at a level
above what is available in the market elsewhere.
So that concept still exists. I would mention, though, that
foreign availability--and I think I mentioned that in my
statement--the notion of doing a foreign availability analysis
is not specifically required for relaxation of controls. That
concept is used more often in determining that controls ought
to remain on a commodity rather than being taken off. The
concept still exists.
Senator Cochran. Your report also notes that there is a
German company that has had some sales of HPCs over 2,000 MTOPS
to tier 3 countries. Does that, in your opinion, constitute
widespread foreign availability?
Mr. Johnson. No.
Senator Cochran. Let me ask you how similar to the United
States are the export control regimes of Germany and Japan for
HPCs.
Mr. Johnson. They are quite similar.
Senator Cochran. If a German company, therefore, has made
limited sales of HPCs capable above 2,000 MTOPS to tier 3
countries, why don't more tier 3 countries buy computers from
Germany instead of from the United States to avoid dealing with
the U.S. export controls?
Mr. Johnson. What we're told is that we have better
technology, and we'll accept that. But they also would have to
deal with the export control systems of Germany, as well.
Senator Cochran. The report says also that ``available
information indicates that''--this is on page 13 of your
report--``available information indicates that the capabilities
of China, India, and Russia to build their own HPCs still lag
well behind that of the United States, Japan, and European
countries.''
For how long does GAO expect that these countries will lag
behind U.S. HPC capabilities?
Mr. Johnson. Well, we don't have information on China and
India. Russia, we were told, is 3 to 7 years behind.
Mr. Phillips. Yes, from various sources, as little as 3
years and as many as 10 years, perhaps.
Senator Cochran. You note on page 42 in your report that
you asked the Commerce Department to provide data to support
its assertion on foreign manufacturers, but ``we received no
documentary support.'' What did Commerce say about this?
Mr. Phillips. When we talked with them, they said they were
relying largely on some HPC world market studies that had been
done in the 1995 time period. That was the essential
explanation.
Senator Cochran. And you had requested data from the
Commerce Department on this subject to support its assertion?
Mr. Phillips. That's correct.
Senator Cochran. And that's what you got in response?
Mr. Phillips. Yes.
Senator Cochran. Did you get copies of the data?
Mr. Phillips. We had those reports earlier. They had
provided them to us at an earlier time, and we in fact even
referred to them in our report.
Senator Cochran. Mr. Johnson, during the preparation of
your report, did any U.S. exporters provide GAO with evidence
of a lost sale due to implementation of the requirement in the
recent Defense Authorization Act to notify the Commerce
Department of planned sales above 2,000 MTOPS to a tier 3
country?
Mr. Johnson. No, they did not, and we queried companies to
see if there had been an effect and we were told that nothing
had changed.
Senator Cochran. This is off the subject, but did you get
at all into the recent decision that led to a fine, and an
agreement to pay a fine, by IBM for violating export
regulations?
Mr. Johnson. No, we did not. That was under investigation
during most of the course of our work, so if a case is under
investigation, we stay away from it. We are aware of the
settlement that was reached, but----
Senator Cochran. What was the settlement, can you tell us?
Mr. Johnson. As I understand it, IBM--their East European
subsidiary--entered a plea and paid a fine of $8.5 million,
which I understand is the steepest fine that has been levied.
Senator Cochran. And what was that fine imposed for? What
regulation was violated? Was it the regulations as decontrolled
by the administration, as relaxed, or was it earlier
regulations?
Mr. Johnson. No, as relaxed. They had violated the controls
in existence.
Senator Cochran. The new controls?
Mr. Johnson. Right.
Senator Cochran. The new limited controls that exist for
exports?
Mr. Johnson. Yes.
Senator Cochran. OK.
Thank you very much. If we have some other questions that
we need to ask you to clarify provisions in the report, we will
submit them to you. But we appreciate your doing this study and
we appreciate your delivering this, and you have instructions
here on how you can get more copies and everything, right? It's
available?
Mr. Johnson. Yes, on the back page.
Senator Cochran. On the back page here. So if anybody wants
copies for friends, that's how you can order them.
Thank you very much.
Mr. Johnson. Thank you, Mr. Chairman.
Senator Cochran. Our next panel is the Hon. William
Reinsch, Under Secretary for Export Administration, U.S.
Department of Commerce.
Mr. Reinsch, we thank you for being here and your patience
and your cooperation with our Subcommittee.
You are free to proceed with any comments or statement you
would care to make in response to the GAO report or the subject
at hand, and then I will have a few questions for you.
STATEMENT OF HON. WILLIAM REINSCH,\1\ UNDER SECRETARY FOR
EXPORT ADMINISTRATION, U.S. DEPARTMENT OF COMMERCE
Mr. Reinsch. Thank you, Mr. Chairman.
---------------------------------------------------------------------------
\1\ The prepared statement of Mr. Reinsch appears in the Appendix
on page 123.
---------------------------------------------------------------------------
Let me say first that this is the third time I've been here
this year, not counting times previous, and I want to begin by
expressing my appreciation for your courtesy on every occasion.
We haven't always agreed on these things, and I have a
lingering suspicion that we won't agree today, but you have
been on every occasion fair and generous in giving me time to
present the administration's point of view, and I want you to
know that I appreciate that.
Mr. Johnson. I also harbor the lingering--probably naive--
hope that one of these days I'm going to convince you that
we're right and you're wrong---- [Laughter.]
But I suspect that will be another day.
Mr. Reinsch. I am tempted to dump my statement and respond
to some of the questions that were asked of the panel, but what
I would like to do is deliver an abbreviated version of it, if
I may, and then hope that you ask me some of the same questions
that you asked them, because I would be pleased to respond to
some of this.
The debate over computer export policy has been marked by a
good deal of smoke and heat, but perhaps not as much light as
one could wish. I want to try today to explain our policy and
its derivations, and in doing so will comment--albeit
indirectly--on the substance of most of GAO's comments.
The fundamental premises of our policy are that, like it or
not, rapid technological progress has rendered control of high
performance computers increasingly difficult, and that it is
more important to our national security to have a healthy
computer industry supplying state-of-the-art products to our
military and intelligence services than it is to attempt to
``control the uncontrollable'' and jeopardize our companies'
futures in the process.
Four factors have shaped our thinking.
The first is technological change. Improvements in
microprocessor design have allowed high performance computers
to become ever smaller, cheaper, and faster. At the same time,
improvements in microprocessors have made routine desktop PCs
capable of performing at what were considered supercomputer
levels a few years ago.
The second factor is global diffusion. We must assess
realistically our ability to control the distribution of
computers when they are produced in the thousands, even tens of
thousands, and sold from a variety of sources around the world.
Third is the growth of parallel processing, which greatly
increases computer performance and the concomitant ability of
users to easily upgrade performance.
Finally, there is our conclusion, based on our research and
the 1995 and 1998 studies, that computer power is a secondary
consideration for many applications of national security
concern.
We have also kept in mind the nature of the computer
market, which is a vital element of U.S. economic strength. We
are world leaders in this very competitive market, with $2
billion a year in revenue, and this leadership helps us across
the board in the information technology sector. The policy
adopted by the United States in 1995 affected more than $10
billion in exports, which supported 140,000 jobs annually. If
misapplied, export controls can profoundly damage this
important sector, put these jobs at risk, and--relevant to your
consideration today--seriously damage our national security by
crippling our companies, just as our national security
establishment's reliance on them is growing.
The competitive and increasingly global market has strong
implications for controls. Roughly half the computers made in
the United States are exported, and the sales fall in the
ranges below--my written statement provides that. Computers up
to 400 MTOPS, which are your standard PCs, have been sold in
the millions; 400 to 1,000 MTOPS, in the tens of thousands;
1,000 to 5,000 MTOPS, in the thousands range; and a few
thousand computers capable of 5,000 to 10,000 MTOPS have been
sold, and some hundreds at levels beyond that.
Some of these computers can be reconfigured by their users
to have much higher performance, and in the future, in response
to market demands, more and more computers will be scalable.
Our fundamental reality is that computers which are available
in the thousands in markets around the world cannot be
effectively controlled, even if they are built in the United
States or based on U.S. technology. The 1995 study predicted
many of these developments, and everything we have learned
since then confirms them. And I want to emphasize that point,
Mr. Chairman.
It seems to me that the previous discussion had a certain
note of unreality because it concerned itself primarily with
whether or not the Stanford study presented empirical data.
What GAO has not really commented on is a look at the
marketplace in 1996, 1997, and 1998 to see reality, because the
1995 study was predictive. And I would suggest that if you look
at reality today, you will find that what the 1995 study
predicted came true in their assessment of what the marketplace
would look like this year. It is confirmed by the facts of what
we have right now. To me, that's the best empirical evidence
that you have. I can address the study later in more detail.
Now, on technological change particularly, the term means
that computer performance is constantly improving, creating
unavoidable pressure on export controls. There are few sectors
that we deal with where technological change has been so rapid
and so dramatic. I want to make the point here that our
reference to GAO about foreign availability being an outdated
Cold War concept was a reference that was specifically confined
to the question of computers. That is not a statement that we
subscribe to as a general matter with respect to what we do. It
is a statement that we think is accurate with respect to this
sector.
Five years ago, the U.S. controlled as a supercomputer
machines with a performance of 195 MTOPS. Today's average
desktop PC is more powerful; if you have a Pentium II in your
machine, it is probably twice as powerful, and the software
which can run on it is more sophisticated.
The engine of change is the microprocessor. Computer chips
are produced in the millions in plants in the United States and
overseas, and I want to emphasize this point. This discussion
should not be about the box; it should be about the chips,
because the chips are the critical ingredient. You are all
familiar with Moore's Law, which states that the performance of
chips doubles every 18 months. These performance increases are
the result of both improved design and improved manufacturing
techniques. As of August, 1998, chips capable of roughly 500
MTOPS alone are being produced in the millions, and chips
capable of 1,800 MTOPS are being produced in the tens of
thousands. Although the United States is the most advanced
producer, plants around the world can make these chips. Within
12 months, if industry projections are correct, we can expect
to see chips capable of 2,000 MTOPS entered into mass
production. It happens, Mr. Chairman, that I have one of them
with me. For the record, this little goodie is smaller than the
size of my fingernail. This is IBM's latest announced chip, a
copper-based chip, which operates at 1 gigahertz, or 1,000
megahertz. It is scheduled to start going into IBM's product
line toward the end of next year or in early 2000. It will
probably start in the RS-6000 workstation line and mainframe
server lines, which are the kinds of technologies that we are
talking about in this debate. Thereafter, it and chips like it
will start to appear in our PC server line. The actual MTOPS
rating of this chip is proprietary, but I can tell you that it
will operate at well over 2,000 MTOPS, and that's a single
chip.
In addition, I can tell you that advances in processor
design, and also advances in semiconductor process technology,
will allow Intel to create processors with double their current
MTOPS rating of approximately 500 MTOPS today, to over 1,000
MTOPS by early 1999. Intel will be able to provide specific
information about these developments next year. Much of the
details of these things, including the exact ratings, are
proprietary.
Other technological changes have made it easier to upgrade
performance. These include the increased sophistication of
software, which is very important here, including for
clustering, and the increased availability of interconnect
technologies which offer substantial improvements in
performance and which may allow numbers of low level
workstations to be clustered together to give high performance.
The spread of parallel processing, which allows many
microprocessors to work simultaneously on the same problem, has
also reduced the controllability of high performance computers.
You may have read, Mr. Chairman, of several recent
occasions in which the ``encryption community,'' if you will,
announced success in breaking 56-bit encryption. One of the
devices that they used in the first contest was, essentially,
parallel processing, to get thousands of computers working
together in a coordinated fashion to attack a particularly
complicated problem.
Another element of technological change could be called
attainable performance, or scalability, which you alluded to
earlier. Manufacturers have sought to build platforms which can
be easily upgraded through the addition of new boards. This
allows users to buy computers at one performance threshold, and
then increase the performance later through upgrades. Some
computers are designed to allow these upgrades to be performed
without even turning off the machine, and with system software
that automatically adjusts to the higher performance levels.
The result is that it is possible to buy a number of systems
that perform well below 2,000 MTOPS, or even below 1,000 MTOPS,
and thus do not require a license for export, and then upgrade
these machines to 5,000 or 6,000 MTOPS or more.
Now, on foreign availability, the availability of high
performance computers built by foreign manufacturers with
foreign parts and technology was a key determinant of our
export policy during the Cold War. We believe for computers it
now makes little sense and is of secondary importance in
determining policy. We cannot realistically control the many
thousands of U.S.-made computers sold freely in Europe, Asia
and elsewhere. Many countries we sell to do not have re-export
controls; in fact, the New York Times recently quoted an
official from a close European ally as saying that they advise
their exporters to ignore U.S. re-export controls. We know
there is a flourishing market in second-hand high performance
computers overseas; some can be ordered directly over the
Internet. In fact, Mr. Chairman, I asked one of my staff to do
a little surfing. We've been able to come up with a substantial
number of web sites here, including sites in the Netherlands,
Canada, Austria, and Israel, all of which offer for sale
substantial numbers of computers, many of them in excess of
2,000 MTOPS.
Now, as a result, even though the United States today
dominates the market for high performance computers, there is a
performance threshold below which we cannot realistically
expect to maintain control of computers unless we restrict
sales to everybody, including our closest allies. That is why
we have focused on controllability--whether licensing can be
effective in restricting access to HPCs. The studies in 1995
and the new one in 1998 suggest that HPCs are becoming less and
less controllable because they are becoming smaller, cheaper,
more powerful, and more reliable, requiring less vendor
support. The availability of fast, well-designed
microprocessors has allowed manufacturers to build more and
better computers. Machines capable of 4,000 to 5,000 MTOPS are
small and easily transportable. Computers well above 2,000
MTOPS are freely available on the global secondhand market, as
I just alluded to. We cannot realistically expect to keep the
organizations responsible for weapons development in states of
concern, organizations that are technically sophisticated and
well funded and which enjoy strong government support, from
clandestinely obtaining computers with a performance much below
10,000 MTOPS.
Let me also make the point, Mr. Chairman, that one of the
other problems with the old standard of foreign availability is
its basis of searching for comparable quality and sufficient
quantity, as the basis for availability. The point when it
comes to proliferation isn't sufficient quantity. In order to
determine availability, we don't have to determine that the
Chinese have an industry that can compete with IBM or HP or
SGI. The Chinese need four or five, or maybe ten, computers to
meet a lot of their proliferation needs. That doesn't require a
full-scale industry that is competitive with ours in third
markets. It requires the capability to produce those machines
at a small scale, and perhaps significant expense. But that's
very different from the statutory definition.
In addition, let me say, in our judgment computers are not
a ``choke point'' for military production. High performance
computers have attained a symbolic importance in our export
control debates which their real utility may not warrant. The
fundamental premise of critics of our policy is that high
performance computers are essential for making advanced
weaponry. We believe this is wrong. The weapon systems found in
the U.S. arsenal today--the tanks, the airplanes, the missiles,
the ships--were built with computers whose performance was
below 1,000 MTOPS. These were the supercomputers of the 1980's,
but today you can find more capable machines on many office
desktops.
We have found that the amount of computing power needed to
design and manufacture modern weapons, once you get over a few
hundred MTOPS, is not significant. For example, the level of
computational power used to develop all the bombs in the
current U.S. nuclear arsenal is less than that found in many
workstations. Other factors--skill in software design, access
to sophisticated manufacturing techniques, experience in
building weapons, and good test data--are much more important
than a high performance computer.
To say that this is an enabling technology, which is what
the MCTL does, which is what GAO commented on, is very
different from saying that this is a choke point.
There are a number of national security applications--
precise weather forecasting; computational fluid dynamics, and
particle dynamics in particular, which is what happens when you
set off an explosion or what happens when you set off a
chemical weapon and where do the particles go--which are areas
where high-level computing power are significant, and our
policy attempts to identify those and, where possible, protect
them. This differs from those who have argued that high
performance computers will give countries like China the
ability to leap forward in military production. HPCs no doubt
provide some incremental benefit, as would a wide range of
items, but we do not believe that they constitute a choke point
in weapons development and, as stated previously, even if they
did, there are serious limitations on our ability to control
them at all but the highest levels. In fact, none of the
proliferation regimes--the Missile Technology Control Regime,
the Nuclear Suppliers Group, or the Australia Group, which
controls chemical and biological items--consider computers
important enough to control. The only regime which controls
computers is the Wassenaar Arrangement, which inherited the old
economic warfare controls aimed at the Soviet Union. In the
end, these controls did not work very well in preventing the
Soviet Bloc from getting its hands on widely-available
computers. They were helpful in keeping big machines that
require extensive support out of enemy hands, but they failed
to stop computers available in the thousands, sold freely at
relatively low prices around the globe without significant
vendor support.
We would do well to remember this lesson, because what we
are experiencing now, as demonstrated in both the 1995 study
and the 1998 study, and what anybody in this industry will tell
you, is essentially an acceleration in the rate that ever-
higher performing computers are becoming available. In other
words, we have controlled the high end; we can still control
the high end. But whereas that high end used to be measured in
hundreds of MTOPs, now it is measured in ten thousands of
MTOPS.
Maintaining our status as world leader in information
technology and computer manufacturing is critical to both our
economic growth and our national security. Exports account for
roughly half the revenues of U.S. computer companies. Ill-
advised export controls would put this vital sector at risk,
and at the same time compromise our security by making it
harder for the Pentagon to obtain the cutting edge technology
it needs. Events since 1995 have confirmed that we are on the
right course, and I hope the Congress will support the
administration as it moves into a new review of control
parameters.
Now, let me emphasize that final point, Mr. Chairman,
because that was mentioned in the previous dialogue. In fact,
the proposal that the President adopted in 1995 was the
recommendation of the Department of Defense. There were other
recommendations made by other agencies; it was the Department
of Defense recommendation that everyone coalesced around and
supported, and it was the one that the President adopted.
Now, the Department of Defense is in the best position to
explain to you why they came to that conclusion, and I can add
my own thoughts. But let me put in my own words a little bit of
what I think their reasoning is.
They have realized that as warfare becomes more electronic-
based--meaning not only smart bombs and electronic-based
weapons systems, but also more sophisticated command and
control and communications--their reliance on high performance
computers is ever increasing and becoming much more important.
They also know that they don't buy enough to keep any of these
companies in business. Military business for these companies is
less than 10 percent of their total business.
What the Pentagon figured out in 1995, and what is
reflected in its letter embodied in the GAO report, and what
they still believe based on my most recent discussions with
them, is that the most useful thing they can do to make sure
that they have access to cutting edge technology, to make sure
that their weapons systems and C\3\ facilities are generations
ahead of everybody else's, is to have a strong, healthy
domestic computing industry, which means one that is
successfully exporting, because that's where they make their
money. And the money they make there, they plow back into R&D,
which is of direct benefit to the Pentagon.
In closing, Mr. Chairman, I would ask your indulgence and
ask to submit into the record an article which I want to
commend to everyone's attention, which I think is quite
thoughtful. It's in the current issue of Foreign Affairs
magazine. It's by Michael Hirsh, who is a reporter for
Newsweek, and it's entitled ``The Great Technology Giveaway?''
with a question mark after it, and it is one of the best
dissections that I've seen of what has happened to the economy
as it is globalized, and what has happened to the relationship
between our military and our defense establishment and our
intelligence establishment, and the civilian sector of the
economy. Whereas in the 1960's and 1970's you had MILSPEC and
military technologies driving the train of R&D, now it is
reversed. It is the civilian sector that is driving the train
and the military that is buying commercial off-the-shelf items,
which means the health of these sectors is critical--not just
for jobs, which I care about and you care about--and not just
for the economy, but for our national security. If the
companies whose representatives are in the back of the room
here go broke, the biggest loser is the Department of Defense,
not the Chinese.
With that, I am happy to answer your questions.
Senator Cochran. Thank you very much.
The article by Michael Hirsch you referred to will be made
a part of the hearing record.\1\
---------------------------------------------------------------------------
\1\ The article referred to from Foreign Affairs magazine by
Michael Hirst, appears in the Appendix on page 126.
---------------------------------------------------------------------------
To what extent have the changes that the Congress recently
made affected the ability of our computer industry to make
sales in the foreign markets? Do you know of any downturn in
sales that could be attributed to the changes that the Congress
made?
Mr. Reinsch. We have not specifically asked them that
question, Mr. Chairman, and I would recommend that you do. I
don't know of any in particular. I would say that as far as the
10-day notification process is concerned, as a process matter,
it appears to be working; that is, companies are submitting
notifications and agencies are responding within the timeframe,
and the system appears to be going forward. The objection rate,
if you will, has been steady from the beginning at around 15 or
16 percent, and a little higher in some cases and lower in
other cases. That seems to be going forward.
What I can tell you happens is that when there is an
objection, that then converts that notification into an
application for an individual license, which means that it
takes a much longer time. We are discovering that most of them
are being returned without action because we don't have
sufficient information in order to process an individual
application. And what we are then finding is that many of them
are not coming back; that is, the company is just folding and
not going forward. Whether that is because it's too much
trouble, whether it's because the sale went away anyway,
whether it's because it was a shaky sale and they didn't think
they could get it approved, I couldn't tell you. But I can just
say, that's one of the results that we've noticed.
Senator Cochran. What about the review of a sale after it
has been consummated? There is a process now where you have
people from embassies around the world going in and checking to
see if the end use is what you expected it to be, and the end
user is who you expected it to be. How is that working? Are
there problems there that need to be discussed?
Mr. Reinsch. The problem that we will ultimately encounter,
Mr. Chairman, is volume, because we were tasked to visit every
one of them without regard to whether or not it made sense to
visit every one of them. And I can give you an illustration.
A number of these, particularly--and when I say every one
of them, every one at tier 3; I don't mean the other tiers--a
lot of these are sold, not all of them, in fact a declining
portion, but a number of them are sold under warranty or with
established vendor support or with a relationship where
something breaks, they call up the American company and they go
out and examine it and fix it.
In a number of cases where we've had some questions about
the nature of the end user, one of the things we've done is
consult with the companies, and we discovered that they very
recently--within 2 or 3 weeks, often--have made a warranty
visit or a repair visit and have been able to provide us some
reassurance as to both the location of the computer and its
use, without us visiting it. Yet, we're going to have to visit
all those anyway.
Senator Cochran. You're not taking their word for it, are
you?
Mr. Reinsch. No, we're not taking their word for it. The
law doesn't permit us to take their word for it. But I will
tell you that, from the standpoint of management, with limited
resources and limited people, I would be inclined to put those
at the bottom of the list, and put at the top of the list the
ones where we have no feedback from companies, that are more
problematical.
Now, in point of fact our obligation statutorily stems from
those sold since November 19, 1997, which is a smaller portion
of the universe. We have, I believe--and I can't speak for the
last few days--I believe we've conducted about 66 post-shipment
visits in tier 3 countries. Let me put it this way: All of
those have checked out favorably with the exception of one,
which was not a Chinese case. But I would prefer not to go into
detail about that one in public session.
Senator Cochran. Do you think that's a useful thing to do,
given the fact that we have seen some situations, particularly
in Russia, with the computers winding up in the weapons
laboratories. Where a Russian official, the Minister of Atomic
Energy, almost gloated over the fact that they had obtained
these high performance computers----
Mr. Reinsch. Let me say first, Mr. Chairman, that, of
course, was in our judgment an outright violation of our rules,
not something that fell between the cracks. We thought our
rules were clear, and they were violated, and that relates to
the settlement in the one case that you referred to earlier.
We have always felt that post-shipment visits are a useful
enforcement tool. We have used them in the past on a wide
variety of merchandise and we intend to continue to use them,
because we think that they provide a lot of information. While
on computers GAO makes a valid point--certainly, what they tell
you is whether it's still there; that's the most obvious thing
they tell you--and while it is also true that without someone
specially trained in computers, you might not learn a lot more
about exactly what the machine is being used for, I can tell
you two other things.
First, our people are being trained; in fact, we have a
group of about 55 of them in town who are being trained
tomorrow on this very subject.
Second, without going into a lot of detail in open session,
I can tell you that the encounters that we have in these visits
give us a great deal of information above and beyond the mere
physical presence of the item.
My objection to the provision is based not on the concept
but on the flat requirement that we visit every one of them,
whether we believe it makes sense from a management and control
perspective or not. My enforcement people tell me that they
would much prefer to be able to exercise some judgment and
visit more intensively the ones that they think are
problematical and deserve visits, and not have to waste time
and money going to places where a vendor that we have
confidence in was there 2 weeks ago, or where we have other
means of ascertaining what that computer is for. They don't
have that choice, and as you know from this list of tier 3
countries, these places are not all easy to get to. These are
expensive visits. And if I sent somebody out for 6 weeks on a
Safeguards mission to do 20 of these, you're talking about
$30,000 or $40,000 when you add up the air fares and hotel
bills and per diem and all the other stuff.
We're going to do the best we can because the statute
requires us to, but that's our problem with it.
Senator Cochran. Is there any limitation in the statute in
terms of the time within which the visit has to be conducted? I
don't recall there being.
Mr. Reinsch. No, sir, there is not. And our general
practice is not to arrange the visit until sometime after it
arrived, because it would be stupid to show up before it was
installed. Our general rule of thumb is, I think, to make those
requests between 4 and 8 months after we've ascertained that it
has arrived.
Now, in the case of computers, what we have put into place
is essentially a double reporting system. That is, under the
current system, notifications have to be reported to us, as you
pointed out, under the law. A notification doesn't constitute a
sale, and these things often fall through, even after they've
been notified. We get notified on specs, sometimes. So our
enforcement agents also require companies to notify us of
actual sales, actual deliveries, and we base our visiting
requirements on those reports, not on the notification reports.
Senator Cochran. Can you tell us the extent to which you
have encountered difficulty of access? Are some countries just
saying that you can't have access?
Mr. Reinsch. I think with respect to the one I alluded to,
that was part of the problem. I think there is another case
where one is pending that we have not yet received an answer,
or we may have gotten an adverse answer.
In the case of China, which you may be building up to, as
of the President's summit we have an agreement with them on
this issue--which is not computer-specific--and we have thus
far put forward one computer post-shipment request, and the
nature of the agreement gives us a certain amount of time to
put that request forward, which we did, and gives them a
certain amount of time to respond, and we're in the middle of
that time period right now.
We have put other post-shipment visit requests forward to
the Chinese on non-computer items and they have agreed to
permit those visits, and those visits have been performed
successfully and favorably. So I have every expectation that
they will agree to our computer requests and that they will be
performed.
Senator Cochran. I did notice that it was advertised, after
the recent visit to China by the President, that there was a
computer-specific--I thought--agreement that would permit us to
have access or information that we otherwise didn't have.
It says, ``Press reports indicated an agreement was reached
that would allow the United States to conduct computer PSVs.''
Does it provide for anytime, anywhere PSVs in China? Or if not,
what does the agreement permit, if you know?
Mr. Reinsch. First of all, it is not exclusive to
computers. It includes computers, but we insisted that it cover
anything else we want to look at, because we have other
problems with the Chinese besides computers, as you well know.
It sets up rules. The basic one is a question of
sovereignty, and frankly, it's the same rule we would impose on
countries seeking to do the same thing here; we have to get the
government's permission. So we can't just show up at the
factory door and say, ``Surprise.'' We have to get the
government's permission. We have to ask within a certain period
of time, and then they have a certain period of time to respond
and to arrange the visit and conduct the visit. Thus far I
believe they've accompanied us on the visits. I believe that's
their current preference.
The ministry in question that is controlling all this is
undergoing a substantial reorganization, and, we're told, a
reduction in staff of approximately 47 percent, which is
significant, so it's a little difficult to say exactly what
impact that is going to have on this.
We have some things that we would like to do to this
agreement to broaden it a bit, which I would prefer to discuss
with you privately and which I'm happy to do. The Joint
Commission on Commerce and Trade, which is a joint meeting
which happens annually between the Secretary of Commerce and
his counterpart--and this is the ministry which does this in
China--is meeting in Washington in late October and this issue
is on the agenda. The Chinese agreed to come here for a week-
long seminar on export controls in April. They did so; they
appeared; they agreed to invite us for a return this fall. We
will be going later in the fall. This is a relationship that is
a work in progress, and we do have further agenda items we want
to get from them.
Senator Cochran. Well, that's encouraging. I think progress
is being made and we ought to continue to pursue that.
The report that we are talking about today, I think, can be
summarized in this way. The GAO says that the decision that was
made by the administration to decontrol high performance
computer exports was really subject to challenge on three
grounds: First, that it was based on the Stanford study, which
lacked empirical evidence or analysis to support its
conclusions that HPCs were uncontrollable; second, that there
would be widespread availability of foreign high performance
computers; and third, that the administration did not undertake
a threat analysis of providing HPCs to countries of concern.
I want to give you an opportunity, if you can, to give us
your reaction to those three bases of criticism.
Mr. Reinsch. First, I guess, Mr. Chairman, empiricism is in
the eye of the beholder. I am familiar with the 1995 study. I
think the 1998 study is much more detailed. But I think it
provided a good bit of data and information with respect to the
state of the computer industry worldwide. It did not assert--
and we don't assert, although we would differ somewhat with
GAO--we don't assert that there is indigenous production of
computers that compete successfully with American computers in
China or India or in other countries that you might name.
I think we would assess the level of production of those
computers in China as somewhat higher than GAO would, but I
would recommend you consult with the intelligence community on
that because they do have information that discusses the state
of the computer industry in those two countries.
And I would urge you to keep in mind what I said
previously. The issue isn't whether they compete with IBM; the
issue is whether they can make a dozen of them. There were
statements by some Indian officials after their nuclear tests
that they were employing Indian computers to analyze those test
results, which I think suggests that they have some
capability--or at least they say they have some capability.
I think the study has plenty of data about the state of the
art and empirical data. I think the 1998 study has more. I
would point out that in fact the authors of the study have
substantial proprietary data that they were not permitted to
put into either study because it was provided confidentially by
individual companies who gave them quite a bit of information
about actual activities, sales, and competition. They were not
permitted to provide it, but that information exists; the
authors have some of it, and we have some of it.
On the second issue of availability and controllability, as
I've made clear, I think we have a disagreement with GAO over
what the relevant standard is. My judgment of their report on
that point is that they've asked the wrong question; they
probably provided the right answer to the wrong question. If
you use the statutory standard, it is probably a fair judgment
to say that availability is not there, there is not foreign
indigenous production in comparable quality and sufficient
quantity so as to offset the effectiveness of our controls.
Our point is a better term, and this is not 180 degrees
different. This is a 40 degree difference, or maybe 50. Our
term is controllability, by which we mean the ubiquity of the
ingredients and the technology. As I said in my testimony, Mr.
Chairman, the secret here is the chips, it's not the box. It's
the ability to take a lot of these things that I have now put
away, a lot of these chips, together to operate quickly and
successfully as a high performance computer. What a high
performance computer is, is a computer that has more processors
working faster than a PC. There are a lot of ways to put those
together. Chips have been effectively decontrolled, except at
very high levels, since 1989, as I recall. There are lots of
people in the world who make chips, not just American
companies. There are a lot of American companies overseas that
make chips; there are other American companies that make clone
chips or clone PCs.
What you are seeing in this industry is the same thing that
you have seen in the textile industry, the automobile industry,
the television industry, a host of industries over the last 40
years. Developing countries come in at the low end; they make
the basic chips. Maybe they stole the design, maybe they
licensed it, who knows, and they make low-end products and they
gradually work their way up the value added chain. And that's
happening in this sector no differently than it has happened in
all the other sectors I just named, and I know that there a lot
of people in your State that know about moving up the value
added chain in a number of those sectors.
This is no different. Part of our concern here is that we
lead the world in these computers, particularly at the high
end. A viable competitor is the Japanese. And we're proud of
that, and we want to keep it that way. What I would say is, if
you tie our guys' hands--and you have not suggested doing
that--but if the Congress or our controls were to tie our
producers' hands and say ``You can't sell in such-and-such a
market'' through one means or another, what will happen, I
would say, is that in 9 to 12 months you will have the Koreans,
the Taiwanese, and a variety of European nations competing at
the high end, because you will have opened the door and made it
economically viable for them to do that. It is not economically
viable for them to do that now because they don't produce
products that are as good as ours at this end, and they don't
have any particular economic incentive to go to the high end.
If you create the economic incentive by knocking us out of
the market, then you are handing them the market, and that goes
back to controllability. These things are out there. They can
make these products.
So we would submit that the GAO has asked the wrong
question, albeit having come up with the right answer. The
studies asked the right question, and if you want empirical
data, look at the marketplace, invite some of the people in the
back row up to talk and tell you what's going on in the
marketplace, and you will see that what Goodman said in 1995
has happened. That's the best empirical information that I can
find.
Now, the last point is the most interesting one, which is
the question of whether we performed a national security
effects analysis. I guess I would say a couple things about
that. I would make the point that Senator Levin made; the
Department of Defense argues that that was taken into account,
and I think that's right. I think if you want to talk to the
Department of Defense, that would be a constructive thing to
do. They are the agency that is most responsible for making
those judgments, and they indicated that they have done so. My
recollection of the interagency discussion suggests that they
made clear that the levels of decontrol that they were
proposing--or liberalization that they were proposing--was
consistent with their sense of what the security threats were.
And they can elaborate on that.
Let me also say that I would disagree with the GAO in the
order of events. Mr. Johnson said, ``We ought to look at
national security first, and then figure out what can be
controlled.'' And I will say, I think, to the extent the
administration did something differently, I would acknowledge
that we did it in reverse order. We asked ourselves first, what
is controllable? Then we said, let's look at the national
security issues. Because I am perfectly willing to tell you--
and this is not a surprise, nor is it a revelation--if you want
to look at nuclear weapons design, you can do nuclear weapons
design on a PC. You can do nuclear weapons design on a PC in
your office. So if you ask me, is there an adverse national
security impact of selling PCs, the answer is yes.
But that debate reminds me a little bit about this guy I
was reading about the other day who got frustrated over all the
hoopla about the movie Titanic, so he printed out 500 T-shirts
that said, ``It sank. Get over it.'' [Laughter.]
This debate is a little bit like that. This technology is
out of the box. We can perform a national security analysis,
for example, on computers below 2,000 MTOPS, and we can tell
you that, yes, those exports are going to have an adverse
national security impact here and here and here and here and
here. It's kind of an arid discussion because there's not an
awful lot we can do about it.
What we can do about it is not stop it for all the reasons
I outlined in my testimony. What we can do about it is make
sure we stay ahead of the curve, and that we're prepared to
deal with the consequences of that. And GAO missed that entire
point.
Senator Cochran. Well, you've given a very spirited defense
of the liberalization of export controls----
Mr. Reinsch. This comes from the heart, Mr. Chairman.
Senator Cochran. I can tell you're sincere about that, and
I respect that. I think it also means, though, that because we
are seeing an evolution of technology in many areas, and this
is just one of them, we have to be very determined in our
effort to do what we can do to safeguard America's security
interests, and we can't just ignore the fact that we have an
obligation. I think that national security is the highest
priority of our government, that the agencies that have the
responsibility of enforcing our laws on this subject have the
obligation to do the best they possibly can with the resources
they have, to see that they faithfully carry out these laws.
For that reason, I was interested in what is being done to
carry out the change in the law that the Congress approved on
this subject. There are many who remain concerned that the
policy is flawed and that we could do a better job of
controlling the sale of these HPCs to countries that we know
are going to use them to improve the capabilities of their
nuclear weapons or their missile systems, particularly if we
think that they can proliferate these systems to countries that
are emerging as threats to our national security interests. I
am thinking right now of North Korea, for example; Iran is
another example, countries that have expressly promised to
either destroy the United States or kill Americans or the like.
This is a situation that makes this a very, very dangerous
world. I know we can't pass a statute or an amendment to a law
and end it or make it right or fix it, but we do have an
obligation--and I know you agree with this--to continue to work
as hard as we can to be sure that our policies and our export
laws are consistent with that national security interest, as
well as our economic interest. We are interested here in having
a regulatory environment that permits our businesses to
flourish, to grow, to expand, to remain the best in the world,
the most advanced in the world. We're all for that; I'm
certainly for that. It is good to have this interaction, I
think, and this discussion of these competing interests and how
we sort them out.
We will continue to work with the Department of Commerce
and with you and with GAO and with the Department of Defense
and others to try to do our part to help make sure we do our
part to help make sure we do a good job of preventing the
proliferation of weapons and the means of developing those
weapons that threaten our security interests.
We thank you for your continued cooperation and we will
stay in touch with you. If we have any other questions of you,
we will submit them for the record.
Mr. Reinsch. Thank you, Mr. Chairman. It's always a
pleasure to be here. I'll be glad to come back if you want.
Senator Cochran. Thank you very much. Our hearing is
adjourned.
[Whereupon, at 4:15 p.m., the Subcommittee was adjourned,
to reconvene at the call of the Chair.]
A P P E N D I X
----------
[GRAPHIC] [TIFF OMITTED] SH871.001
[GRAPHIC] [TIFF OMITTED] SH871.002
[GRAPHIC] [TIFF OMITTED] SH871.003
[GRAPHIC] [TIFF OMITTED] SH871.004
[GRAPHIC] [TIFF OMITTED] SH871.005
[GRAPHIC] [TIFF OMITTED] SH871.006
[GRAPHIC] [TIFF OMITTED] SH871.007
[GRAPHIC] [TIFF OMITTED] SH871.008
[GRAPHIC] [TIFF OMITTED] SH871.009
[GRAPHIC] [TIFF OMITTED] SH871.010
[GRAPHIC] [TIFF OMITTED] SH871.011
[GRAPHIC] [TIFF OMITTED] SH871.012
[GRAPHIC] [TIFF OMITTED] SH871.013
[GRAPHIC] [TIFF OMITTED] SH871.014
[GRAPHIC] [TIFF OMITTED] SH871.015
[GRAPHIC] [TIFF OMITTED] SH871.016
[GRAPHIC] [TIFF OMITTED] SH871.017
[GRAPHIC] [TIFF OMITTED] SH871.018
[GRAPHIC] [TIFF OMITTED] SH871.019
[GRAPHIC] [TIFF OMITTED] SH871.020
[GRAPHIC] [TIFF OMITTED] SH871.021
[GRAPHIC] [TIFF OMITTED] SH871.022
[GRAPHIC] [TIFF OMITTED] SH871.023
[GRAPHIC] [TIFF OMITTED] SH871.024
[GRAPHIC] [TIFF OMITTED] SH871.025
[GRAPHIC] [TIFF OMITTED] SH871.026
[GRAPHIC] [TIFF OMITTED] SH871.027
[GRAPHIC] [TIFF OMITTED] SH871.028
[GRAPHIC] [TIFF OMITTED] SH871.029
[GRAPHIC] [TIFF OMITTED] SH871.030
[GRAPHIC] [TIFF OMITTED] SH871.031
[GRAPHIC] [TIFF OMITTED] SH871.032
[GRAPHIC] [TIFF OMITTED] SH871.033
[GRAPHIC] [TIFF OMITTED] SH871.034
[GRAPHIC] [TIFF OMITTED] SH871.035
[GRAPHIC] [TIFF OMITTED] SH871.036
[GRAPHIC] [TIFF OMITTED] SH871.037
[GRAPHIC] [TIFF OMITTED] SH871.038
[GRAPHIC] [TIFF OMITTED] SH871.039
[GRAPHIC] [TIFF OMITTED] SH871.040
[GRAPHIC] [TIFF OMITTED] SH871.041
[GRAPHIC] [TIFF OMITTED] SH871.042
[GRAPHIC] [TIFF OMITTED] SH871.043
[GRAPHIC] [TIFF OMITTED] SH871.044
[GRAPHIC] [TIFF OMITTED] SH871.045
[GRAPHIC] [TIFF OMITTED] SH871.046
[GRAPHIC] [TIFF OMITTED] SH871.047
[GRAPHIC] [TIFF OMITTED] SH871.048
[GRAPHIC] [TIFF OMITTED] SH871.049
[GRAPHIC] [TIFF OMITTED] SH871.050
[GRAPHIC] [TIFF OMITTED] SH871.051
[GRAPHIC] [TIFF OMITTED] SH871.052
[GRAPHIC] [TIFF OMITTED] SH871.053
[GRAPHIC] [TIFF OMITTED] SH871.054
[GRAPHIC] [TIFF OMITTED] SH871.055
[GRAPHIC] [TIFF OMITTED] SH871.056
[GRAPHIC] [TIFF OMITTED] SH871.057
[GRAPHIC] [TIFF OMITTED] SH871.058
[GRAPHIC] [TIFF OMITTED] SH871.059
[GRAPHIC] [TIFF OMITTED] SH871.060
[GRAPHIC] [TIFF OMITTED] SH871.061
[GRAPHIC] [TIFF OMITTED] SH871.062
[GRAPHIC] [TIFF OMITTED] SH871.063
[GRAPHIC] [TIFF OMITTED] SH871.064
[GRAPHIC] [TIFF OMITTED] SH871.065
[GRAPHIC] [TIFF OMITTED] SH871.066
[GRAPHIC] [TIFF OMITTED] SH871.067
[GRAPHIC] [TIFF OMITTED] SH871.068
[GRAPHIC] [TIFF OMITTED] SH871.069
[GRAPHIC] [TIFF OMITTED] SH871.070
[GRAPHIC] [TIFF OMITTED] SH871.071
[GRAPHIC] [TIFF OMITTED] SH871.072
[GRAPHIC] [TIFF OMITTED] SH871.073
[GRAPHIC] [TIFF OMITTED] SH871.074
[GRAPHIC] [TIFF OMITTED] SH871.075
[GRAPHIC] [TIFF OMITTED] SH871.076
[GRAPHIC] [TIFF OMITTED] SH871.077
[GRAPHIC] [TIFF OMITTED] SH871.078
[GRAPHIC] [TIFF OMITTED] SH871.079
[GRAPHIC] [TIFF OMITTED] SH871.080
[GRAPHIC] [TIFF OMITTED] SH871.081
[GRAPHIC] [TIFF OMITTED] SH871.082
[GRAPHIC] [TIFF OMITTED] SH871.083
[GRAPHIC] [TIFF OMITTED] SH871.084
PREPARED STATEMENT OF MR. REINSCH
Mr. Chairman, thank you for the opportunity to testify before the
Subcommittee on export policy for high performance computers. The
debate over computer export policy has been marked by a good deal of
smoke and heat, but perhaps not as much light as one could wish. I will
try today to explain on the Administration's policy. Its fundamental
premises are that like it or not, rapid technological progress has
rendered control of high performance computers increasingly difficult,
and that it is more important to our national security to have a
healthy computer industry supplying state of the art products to our
military and intelligence services than it is to attempt to control the
uncontrollable and jeopardize our companies' futures in the process.
Four factors have shaped our thinking. The first is technological
change. Improvements in microprocessor design have allowed high
performance computers to become ever smaller, cheaper and faster. At
the same time, improvements in microprocessors have made routine desk
top computers capable of performing at were considered supercomputer
levels a few years ago. The second factor is global diffusion. We must
assess realistically our ability to control the distribution of
computers when they are produced in the thousands or even tens of
thousands and sold from a variety of sources around the world. Third is
the growth of parallel processing , which greatly increases computer
performance, and the concomitant ability of users to easily upgrade
performance. Finally, there is our conclusion, based on research and
the 1995 and 1998 studies, that computer power is a secondary
consideration for many applications of national security concern. These
factors--rapid technological change, limited controllability,
scalability and limited national security application--have shaped our
efforts to keep our policy in tune with today's technology and
international security environment.
In doing so we have kept in mind the nature of the computer market,
which is a vital element of U.S. economic strength. We are world
leaders in the very competitive computer market with $2 billion a year
in revenue, and this leadership helps us across the board in the
information technology sector. The policy adopted by the United States
in 1995 affected more than ten billion dollars in exports, which
supported 140,000 jobs annually. If misapplied, export controls can
profoundly damage this important sector, put these jobs at risk and
seriously damage our national security by crippling our companies just
as our national security establishment's reliance on them grows.
The competitive and increasingly global market has strong
implications for export controls. Roughly half of the computers made in
the United States are exported, and the sales fall in the ranges below:
LComputers capable of up to 400 MTOPS have been sold in
the millions.
LComputers capable of 400 to 1,000 MTOPS have been sold in
the tens of thousands.
LThousands of computers capable of 1,000 to 5,000 MTOPS
range have been sold.
LA few thousand computers capable of 5,000 to 10,000 MTOPS
have been sold.
LSome hundreds of machines capable of more than 10,000
MTOPS have been sold.
Some of these computers can be reconfigured by their users to have
much higher performance, and in the future, in response to market
demands, more and more computers will be scalable. Our fundamental
reality is that computers which are available in the thousands in
markets around the world cannot be effectively controlled, even if they
are built in the United States or based on U.S. technology. The 1995
study predicted many of these developments, and everything we have
learned since then confirms them.
Let me turn now to the issue of technological change. Technological
change means that computer performance is constantly improving,
creating unavoidable pressure on export controls. In few other areas
has the pace of technological change been so rapid and so dramatic as
computers. Five years ago, the U.S. controlled as a supercomputer
machines with a performance of 195 MTOPS. Today's average desktop PC is
more powerful, and the software which can be run on it more
sophisticated.
The engine of change is the microprocessor. Computer chips are
produced in the millions in plants in the United States and overseas.
You are all familiar with Moore's Law, which states that the
performance of chips doubles every eighteen months. These performance
increases are the result of both improved design and improved
manufacturing techniques. As of August 1998, chips capable of roughly
500 MTOPS are being produced in the millions and chips capable of 1,800
MTOPS are being produced in the tens of thousands. Although the United
States is the most advanced producer, plants around the world can make
these chips. Within twelve months, if industry projections are correct,
we can expect to see chips capable of 2,000 MTOPS enter into mass
production. When this occurs, we will see sales of 2,000 MTOPS
computers numbered in the thousands, making the limitations on our
ability to maintain effective controls even more obvious than they are
now.
Other technological changes have made it easier to upgrade
performance. These include the increased sophistication of software,
the increased availability of interconnect technologies which offers
substantial improvements in performance and which may allow numbers of
low level workstations to be clustered together to give high
performance. The spread of parallel processing, which allows many
microprocessors to work simultaneously on the same problem, has also
reduced the controllability of high performance computers.
Rapid advances in microprocessors, software, interconnects and
parallel processing mean that the performance levels once associated
with giant machines can now be obtained by smaller and relatively
inexpensive computers. The implications of technological progress go
beyond performance. High performance computers are smaller, cheaper,
simpler to install and maintain and more reliable. These attributes are
desirable in the marketplace, but they degrade our ability to control.
Another element of technological change could be called attainable
performance or scalability. Manufacturers have sought to build
platforms which can be easily upgraded through the addition of new
boards. This allows users to buy computers at one performance threshold
and then increase the performance later through upgrades. Some
computers are designed to allow these upgrades to be performed without
even turning off the machine and with system software that
automatically adjusts to the higher performance levels. The result is
that it is possible to buy a number of systems that perform well below
2,000 or even 1,000 MTOPS, and thus do not require a license for
export, and then upgrade these machines to 5,000 or 6,000 MTOPS or
more.
These technical developments pose real problems for
controllability. Faster chips available in the millions; smaller,
cheaper and more reliable computers with performance up to 7,000 MTOPS
and computers which can be exported without a license and then upgraded
to HPC performance, all have created serious limitations on our ability
to control computer exports.
Foreign availability--the availability of high performance
computers built by foreign manufacturers with foreign parts and
technology--was a key determinant of our export policy during the Cold
War. It now makes little sense and is of secondary importance in
determining policy. We cannot realistically control the many thousands
of U.S.-made computers sold freely in Europe, Asia and elsewhere. Many
countries we sell to do not have re-export controls--in fact the New
York Times recently quoted an official from a close European ally as
saying that they advise their exporters to ignore U.S. re-export
controls. We know there is a flourishing market in secondhand high
performance computers overseas--some can be ordered directly over the
Internet. As a result, even though the United States today dominates
the market for high performance computers, there is a performance
threshold below which we cannot realistically expect to maintain
control of computers unless we restrict sales to even our closest
allies.
That is why we have focussed on controllability--whether licensing
can be effective in restricting access to high performance computers.
The studies in 1995 and this year suggest that HPCs are becoming less
and less controllable because they are becoming smaller, cheaper, more
powerful, and more reliable, requiring less vendor support. The
availability of fast, well-designed microprocessors has allowed
manufacturers to build more and better computers. Machines capable of
4,000 to 5,000 MTOPS are small and easily transported. Computers well
above 2,000 MTOPS are freely available on the global second hand
market. We cannot realistically expect to keep the organizations
responsible for weapons development in states of concern, organizations
that are technically sophisticated, well funded and which enjoy strong
government support, from clandestinely obtaining computers with a
performance much below 10,000 MTOPS.
Computers are not a choke point for military production. High
performance computers have attained a symbolic importance in our export
control debates which their real utility may not warrant. The
fundamental premise of critics of our policy is that high performance
computers are essential for making advanced weaponry. This critique is
wrong. The weapon systems found in the U.S. arsenal today--the tanks,
airplanes, missiles and ships--were built with computers whose
performance was below 1,000 MTOPS. These were the supercomputers of the
1980's, but today you can find more capable machines on many office
desktops.
We have found that the amount of computing power needed to design
and manufacture modern weapons, once you get over a few hundred MTOPS,
is not significant. For example, the level of computational power used
to develop all the bombs in the current U.S. nuclear arsenal is less
than that found today in many workstations. Other factors--skill in
software design, access to sophisticated manufacturing techniques,
experience in building weapons and good test data--are much more
important than a high performance computer.
There are a number of applications--precise weather forecasting,
computational fluid dynamics, and particle dynamics in particular--
where high levels of computing power are significant, and our policy
attempts to identify those and, where possible, protect them. This
differs from those who have argued that high performance computers will
give countries like China the ability to leap forward in military
production. While HPCs no doubt provide some incremental benefit, as
would a wide range of items, we do not believe they constitute a choke
point in weapons development, and, as stated previously, even if they
did, there are serious limitations on our ability to control them at
all but the highest levels. There is no evidence that you need a high
performance computer to make most modern weapons, including nuclear
weapons, or that having access to high performance computers alone will
give you improved military-industrial capabilities.
In fact, none of the nonproliferation regimes, the Missile
Technology Control Regime, the Nuclear Suppliers Group, or the
Australia Group consider computers important enough to control. The
members of these regimes decided that computers are not essential for
production of these weapons systems. The only regime which controls
computers is the Wassenaar Arrangement, which inherited the old
economic warfare controls aimed at the Soviet Union. These controls did
not, in the end, work very well in preventing the Soviet bloc from
getting its hands on widely available computers. They were helpful in
keeping big machines that require extensive support out of enemy hands
but failed to stop computers available in the thousands sold freely at
relatively low prices around the globe without significant vendor
support. We would do well to remember this lesson because what we are
experiencing now is essentially an acceleration in the rate that ever-
higher performing computers are becoming widely available. In other
words, we can still control the high end, but whereas that used to be
measured in hundreds of MTOPS; now it is measured in ten thousands.
The United States currently dominates the high performance computer
market, in part because of the realistic computer export policy we
adopted in 1995. Tighter controls would penalize U.S. firms for winning
the high performance computer competition. Our firms' strength has
driven most producers from all but the low end of the market and
discouraged others from entering. Export controls, like any government
intervention, can, however, reverse that situation. Controls at too low
a level act as a subsidy for our foreign competitors, damage our
national security, and cost the American economy exports and jobs.
Maintaining our status as world leader in information technology
and computer manufacturing is critical to both our economic growth and
our national security. Exports account for roughly half the revenues of
U.S. computer companies. Ill-advised export controls would put this
vital sector at risk and at the same time compromise our security by
making it harder for the Pentagon to obtain the cutting edge technology
it needs. Events since 1995 have confirmed we are on the right course,
and I hope the Congress will support the administration as it moves
into a new review of control parameters.
[GRAPHIC] [TIFF OMITTED] SH871.085
[GRAPHIC] [TIFF OMITTED] SH871.086
[GRAPHIC] [TIFF OMITTED] SH871.087
[GRAPHIC] [TIFF OMITTED] SH871.088
[GRAPHIC] [TIFF OMITTED] SH871.089
[GRAPHIC] [TIFF OMITTED] SH871.090
[GRAPHIC] [TIFF OMITTED] SH871.091
[GRAPHIC] [TIFF OMITTED] SH871.092
-