LEXIS-NEXIS® Congressional Universe-Document
LEXIS-NEXIS® Congressional
Copyright 1999 Federal Document Clearing House, Inc.
FDCH Political Transcripts
October 28, 1999, Thursday
TYPE:
COMMITTEE HEARING
LENGTH: 26108 words
COMMITTEE:
HOUSE ARMED SERVICES COMMITTEE
HEADLINE: U.S. REPRESENTATIVE FLOYD SPENCE (R-SC) HOLDS HEARING ON TERRORISM AND
HIGH-PERFORMANCE COMPUTERS
LOCATION: WASHINGTON, D.C.
BODY:
HOUSE ARMED SERVICES COMMITTEE HOLDS HEARING ON U.S. POLICY
REGARDING THE EXPORT OF HIGH PERFORMANCE COMPUTERS
OCTOBER 28, 1999
SPEAKERS: U.S. REPRESENTATIVE FLOYD SPENCE (R-SC), CHAIRMAN
U.S. REPRESENTATIVE BOB STUMP (R-AZ)
U.S. REPRESENTATIVE DUNCAN HUNTER (R-CA)
U.S. REPRESENTATIVE JOHN R. KASICH (R-OH)
U.S. REPRESENTATIVE HERBERT H. BATEMAN (R-VA)
U.S. REPRESENTATIVE JAMES V. HANSEN (R-UT)
U.S. REPRESENTATIVE CURT WELDON (R-PA)
U.S. REPRESENTATIVE JOEL HEFLEY (R-CO)
U.S. REPRESENTATIVE JIM SAXTON (R-NJ)
U.S. REPRESENTATIVE STEPHEN E. BUYER (R-IN)
U.S. REPRESENTATIVE TILLIE K. FOWLER (R-FL)
U.S. REPRESENTATIVE JOHN M. MCHUGH (R-NY)
U.S. REPRESENTATIVE JAMES
M. TALENT (R-MO)
U.S. REPRESENTATIVE TERRY EVERETT (R-AL)
U.S. REPRESENTATIVE ROSCOE G. BARTLETT (R-MD)
U.S. REPRESENTATIVE HOWARD P.
"BUCK" MCKEON (R-CA)
U.S. REPRESENTATIVE J.C. WATTS, JR. (R-OK)
U.S. REPRESENTATIVE WILLIAM M.
"MAC" THORNBERRY (R-TX)
U.S. REPRESENTATIVE JOHN N. HOSTETTLER (R-IN)
U.S. REPRESENTATIVE SAXBY CHAMBLISS (R-GA)
U.S. REPRESENTATIVE VAN HILLEARY (R-TN)
U.S. REPRESENTATIVE JOE SCARBOROUGH (R-FL)
U.S. REPRESENTATIVE WALTER B.JONES, JR. (R-NC)
U.S. REPRESENTATIVE LINDSEY O. GRAHAM (R-SC)
U.S. REPRESENTATIVE JIM RYUN (R-KS)
U.S. REPRESENTATIVE BOB RILEY (R-AL)
U.S. REPRESENTATIVE JIM GIBBONS (R-NV)
U.S. REPRESENTATIVE
MARY BONO (R-CA)
U.S. REPRESENTATIVE JOSEPH R. PITTS (R-PA)
U.S. REPRESENTATIVE ROBIN HAYES (R-NC)
U.S. REPRESENTATIVE STEVE KUYKENDALL (R-CA)
U.S. REPRESENTATIVE DONALD L. SHERWOOD (R-PA)
U.S. REPRESENTATIVE IKE SKELTON (D-MO),
RANKING MEMBER
U.S. REPRESENTATIVE NORMAN SISISKY (D-VA)
U.S. REPRESENTATIVE JOHN M. SPRATT, JR. (D-SC)
U.S. REPRESENTATIVE SOLOMON P. ORTIZ (D-TX)
U.S. REPRESENTATIVE OWEN B. PICKETT (D-VA)
U.S. REPRESENTATIVE LANE EVANS (D-IL)
U.S. REPRESENTATIVE GENE TAYLOR (D-MS)
U.S. REPRESENTATIVE NEIL ABERCROMBIE (D-HI)
U.S. REPRESENTATIVE MARTIN T. MEEHAN (D-MA)
U.S. DELEGATE ROBERT A. UNDERWOOD (D-GU)
U.S. REPRESENTATIVE PATRICK J. KENNEDY (D-RI)
U.S. REPRESENTATIVE ROD R. BLAGOJEVICH (D-IL)
U.S. REPRESENTATIVE SILVESTRE REYES (D-TX)
U.S. REPRESENTATIVE THOMAS H. ALLEN (D-ME)
U.S. REPRESENTATIVE VIC SNYDER (D-AK)
U.S. REPRESENTATIVE JIM TURNER (D-TX)
U.S. REPRESENTATIVE ADAM SMITH (D-WA)
U.S. REPRESENTATIVE LORETTA SANCHEZ (D-CA)
U.S. REPRESENTATIVE JAMES H.
MALONEY (D-CT)
U.S. REPRESENTATIVE MIKE MCINTYRE (D-NC)
U.S. REPRESENTATIVE CIRO D. RODRIGUEZ (D-TX)
U.S. REPRESENTATIVE CYNTHIA A. MCKINNEY (D-GA)
U.S. REPRESENTATIVE ELLEN O. TAUSCHER (D-CA)
U.S. REPRESENTATIVE ROBERT A. BRADY (D-PA)
U.S. REPRESENTATIVE ROBERT E. ANDREWS (D-NJ)
U.S. REPRESENTATIVE BARON HILL (D-IN)
U.S. REPRESENTATIVE MIKE THOMPSON (D-CA)
U.S. REPRESENTATIVE JOHN B. LARSON (D-CT)
HAROLD J. JOHNSON JR., ASSOCIATE DIRECTOR, NATIONAL
SECURITY
& INTERNATIONAL AFFAIRS DIVISION,
GENERAL ACCOUNTING OFFICE
GARY MILHOLLIN, DIRECTOR, WISCONSIN PROJECT ON
NUCLEAR ARMS CONTROL
DAN HOYDYSH, CO-CHAIRMAN, COMPUTER COALITION FOR
RESPONSIBLE EXPORTS
DR. STEPHEN BRYEN, FORMER DIRECTOR,
DEFENSE
TECHNOLOGY SECURITY ADMINISTRATION, DEPARTMENT
OF DEFENSE
WILLIAM REINSCH, UNDERSECRETARY FOR EXPORT
ADMINISTRATION, DEPARTMENT OF COMMERCE
*
CHAIRMAN: Committee will please be in order.
The committee meets today to receive testimony on our policy regarding the
export of high performance or so-called supercomputers. This morning, we will
hear from two panels of witnesses. We will start with a General Accounting
Office witness, after which we'll hear from a panel of outside experts and
United State computer industry representatives. Later this afternoon we will
receive testimony from Undersecretary of Commerce William Reinsch.
Two years ago, we were all shocked by a report that United States
supercomputers, which can be used to improve nuclear weapons capabilities or to
develop advanced conventional weapons, had been shipped without any government
review or approval to military-related facilities in both
Russia and China.
For this reason, I joined with the committee's then ranking member, Mr.
Dellums, in co-authoring a provision in the fiscal year 1998 defense
authorization bill to ensure that the federal government was notified prior to
the export of supercomputers with certain capabilities to a country of
proliferation concern, a so-called Tier Three country.
Contrary to allegation one hears around this town, the provision was not
intended to and in fact has not shut down the export of supercomputers.
Instead, the provision simply requires that the government have an opportunity
to review certain proposed exports for national security reasons.
This past July, President Clinton announced his intention to exercise the
discretion afforded him under this law to increase the performance threshold
that triggers the government notification process. The president's proposal is
intended to ease the export of more capable computers to these Tier
Three countries without any government review.
Under the law, the Congress has until January the 23rd, 2000 to review this
proposal before it takes effect. Our hearing today is part of this review
process.
To assist the committee in determining the effectiveness of the 1998
legislation, earlier this year Mr. Skelton and I asked the GAO to take a look
at how the notification requirement has worked. In its recent report, GAO has
concluded that the export notification process contained in the 1998 law has
helped to prevent the shipment of United States supercomputers to potentially
dangerous end users.
Specifically, the GAO found that in more than 10 percent of industry's
notifications to the Commerce Department of an intent to export a
supercomputer, the government denied -- decided to require the exporter to
submit a formal export license application because of security concerns.
While we can debate whether
a larger or smaller number of proposed exports should have been flagged, in my
view, the GAO findings validate the fundamental importance of maintaining
government visibility into where United States high performance computers
exports are going.
The president's recently proposed revision to the notification threshold is the
third major revision to United States computer export policy proposed by the
administration. Since the president's July announcement, the committee has
been engaged with the administration, the computer industry and outside experts
in a continuing dialogue to understand the rationale behind and implications of
further decontrols of supercomputer exports.
On October the 19th, 1999, Mr. Skelton and I wrote the president asking him to
respond to a series of questions regarding the national security implications
of this proposed -- of his proposal, as well as the impact it will have on the
United States computer industry. A copy of that letter is before each member.
Given the recent history of unauthorized computer exports to potentially
dangerous end users and the administration's inclusion to weigh commercial
considerations more heavily than national security concerns on matter of export
control policy, I though it important to have this hearing prior to our
adjournment. Whether or not the proposed relaxation of supercomputer export
controls will pose increased national security risk is an issue we're looking
into and looking to our panel to help us determine.
To help us better understand the current government review process and the
president's proposal, our first witness this morning will be Mr. Jim Johnson,
associate director of the National Security and International Affairs Division
of GAO. Following Mr. Johnson's testimony and member's questions, we will move
to our second panel about outside experts and industry representatives. They
will be: Mr. Gary Milhollin, executive director, the Wisconsin
Project on Nuclear Arms Control; Dan Hoydysh, director of Trade Policy and
Public Policy in Government Affairs Resource and co-chair, Computer Coalition
for Responsible Exports; Dr. Stephen Bryen, former director of the Defense
Technology Security Administration.
After the completion of the second panel, the committee will recess and
reconvene at 2:00 to hear from William Reinsch, undersecretary of Commerce for
Export Administration.
Before turning the floor over to our first witness, I'm going to recognize our
committee's ranking member, Mr. Skelton, for any comments he'd like to make.
SKELTON: Mr. Chairman, thank you very much. It's a pleasure for me to join you
in welcoming our witnesses today. This is the first opportunity for this
committee to review the results of the process and the proposed changes
associated with supercomputer export provisions enacted in fiscal
year 1998 National Defense Authorization Act.
Since we're all familiar with the threats to our national security that's
possible through uncontrolled exports of supercomputers, I'll not take much
time with my opening remarks, but I do want to share a few comments, Mr.
Chairman.
While today's hearing addresses the policy changes that have been proposed by
the administration, there is a more perplexing process issue before us right
now. That issue is trying to determine what is an appropriate review period
for the Congress to provide for itself before any change proposed by the
administration is implemented. One hundred and eighty days may not have been
needed at this time for us to assess national security implications of the
proposed regulations. But I'm not certain that 180 days will not be needed for
subsequent
changes.
We need to develop a rational and practical procedure, the Congress has at
least three simple objectives. One, that provides the administration
flexibility to allow for the rapidly changing technological environment. Two,
that provides the adequate review and action period for Congress to express
it's views. And finally, that let's the U.S. computer industry to remain
competitive. I'm here today to ensure the process we rely on to protect our
national security interests in the export of supercomputers achieve these three
objectives.
As the committee continues to assess the proposed changes, I look forward to
carefully examining the proposal that moves us closer to that end. And I look
forward to the testimony, Mr. Chairman, thank you so much.
CHAIRMAN: Thank you, Mr. Skelton. Without objection, the prepared remarks of
all of our witnesses will be inserted in the record. And Mr. Johnson, you can
proceed as you'd
like.
JOHNSON: Thank you very much, Mr. Chairman. I'm happy to be here this morning
to talk to you about our recently released report on high performance computers
that are exported to countries that might be of military or nuclear
proliferation concern.
The 1996 Executive -- in 1996 the executive branch removed license requirements
for most exports of high performance computers to civilian end users but
retained the licensing requirements for countries of concern. The 1996 change
also made exporters responsible for determining whether there needed to be
applied -- needed to apply for an export license based on their own knowledge
of the export's users activities.
In 1997, several U.S. exporters shipped high performance computers to Russia
nuclear -- to Russian nuclear weapons laboratories and to military end users in
China without licenses. Because the Congress believed that U.S. exporters may
be unaware of the
end users activities, it included a provision in the 1998 Defense Authorization
Act requiring exporters to notify the Department of Commerce of any proposed
exports of high performance computers to countries of concern so that a
determination could be made whether the export needed to be licensed.
Countries that pose such a concern include such countries as China, Russia,
India, Pakistan, Israel and with a total of 50 countries. The 1998 Act also
required Commerce to verify the high performance computer exported to countries
of concern, regardless of whether a license was required were being -- had been
installed where they were intended.
To help you evaluate how this legislation is working, you asked us to determine
two things. First, were the exporters notifications to Commerce of proposed
sales of high performance computers to countries of concern have resulted in
license applications. And what actions were taken on these applications. And
secondly, whether Congress is verifying -- whether Commerce is verifying the
use of
high performance computers after the export to these countries.
I'll briefly summarize the findings of our report. But first I think a little
background might be helpful. The legislation authorized 10 days upon receiving
notification for Commerce to circulate among the Departments of Defense, State
and Energy, the exporters notification of the export of a high performance
computer, or an HPC. The Act requires the license to export -- requires a
license to export if any of these agencies raise a written objection to the
export without a license. If no objection is raised during the 10 day period,
the exporter may ship the computer without a license.
Exporters that plan to ship HPC's to users that are already known to be of
military or proliferation concern must apply directly for a export license.
They do not need to go through the notification process. To indicate the
level of concern the United States has with regard to the export of HPC's, the
executive branch has organized countries into four tiers.
Each tier after Tier One represents a successively higher level of concern to
the United States interests. We have in our -- an appendix to the statement, a
list of countries in the four tiers. Tier Three contains 50 countries that are
of concern for military or proliferation reasons.
The executive branch also established separate control levels for different
types of end users in Tier Three. For end users of military or proliferation
concern, the controls required a license for export high performance computers
that are over 2,000 million theoretic operations per second or MTOPS. For
civilian end users in Tier Three countries, a license was required to export
computers that performed over 7,000
MTOPS.
For exports of HPC's performing between two and 7,000 MTOPS, an exporter could
ship the computer without a license provided the exporter determined that the
recipient was a civilian end user.
In summary, we found that most of the 938 proposed exports of high performance
computers to civilian end users in countries of concern from February 3 of
1998, when procedures for implementing the 1998 Authorization Act became
effective, to March 19, 1999, which was the cutoff date for our audit. Most of
those did not require a license. The agencies that reviewed the export
proposals, which are the Department of Commerce, Energy, Defense, State, and
until March, the Arms Control and Disarmament Agency, allowed 828 proposed
performance -- high performance computers to be exported without a license.
But the required license applications for 101 proposed exports. Nine of
exports were categorized --
proposals were categorized as incomplete and returned to the exporter.
The majority of the agencies objections to the 101 proposed exports were based
on concerns that proposed end users of the computers might be involved in
military or proliferation related activities. Of the 101 license applications
required, 16 were approved and six were denied. The remaining 79 were returned
to the exporter without action, which essentially blocks the proposed export.
Licenses that were approved had additional conditions placed on the re-export
or the use of the computer. The majority of these applications involved China,
India and Israel. Licenses were required in nine case where the end user had
previously received a computer without a license before the Authorization Act
was implemented.
The Act contains no time limit for the completion of post- shipment
verifications. As of November 17, 1998, Commerce had performed post-shipment
verifications of 104 expert -- export -- exported high
performance computers, or about 27 percent of the verifications required on the
390 high performance computers exported during fiscal year 1998.
In their report to Congress, Commerce stated that all 104 shipment
verifications were favorable, that is the computer had been seen during an
on-site visit and nothing was inconsistent with the license or license
exceptions. However, a verification conducted by Commerce but not yet
completed, detected the possible diversion of two computers to a military end
users in apparent violation of U.S. export control regulations. At the time we
completed our work, the Commerce Department was investigating these diversions.
Of the 286 high performance computers, computer exports where post-shipment
verifications had not been completed, almost two-thirds involved exports to
China. According to Commerce, the verifications had not been done because
China's policy prior to June of 1998 did not
permit post-shipment verifications. Or because the exports did not meet the
requirements agreed upon in the June 1998 Memorandum of Understanding between
the Department of Commerce and China's Ministry of Foreign Trade and Economic
Cooperation.
The Departments of Commerce, Energy, Defense and State were provided an
opportunity to comment on our report. Energy did not comment and State
provided helpful technical comments. The Department of Defense reviewed the
report but had no comments. Commerce said that our report had not acknowledged
that first it had to divert enforcement resources from investigations of other
-- or from other preventative enforcement activities to conduct the
legislatively mandated post-shipment verifications. And second that it would
soon be impossible to perform the investigations mandated by law. Commerce
also stated that the most -- that most of the uncompleted verifications were in
China. And that 103 of the 200 cases outside of China were complete.
Although the 1998 Act requires post-shipment verifications on all high performance computers exported since
November 18, 1997, whether licensed or not, Commerce told us that it believes
that it is futile to seek to verify the use of the high performance computers
exported to China before the end use visit arrangement, or where there were no
end use certificates. This is particularly true in view of the proposed change
to the control levels for exports to military end users in countries of
concern.
The July 1999 announcement to change export control levels will move future
licensing requirements from any high performance computers that have already
been exported to China. Notwithstanding the new control levels established by
the executive branch, the Act requires Commerce to conduct post-shipment
verifications on all licensed and unlicensed high performance computers at
certain performance levels that are exported to
countries including China.
Mr. Chairman, that concludes my summary of our report. I would be pleased to
try and respond to your questions.
CHAIRMAN: Thank you, Mr. Johnson. I think we'll go directly into the second
panel and then have all of you come back for questions at that time.
JOHNSON: Fine.
CHAIRMAN: Mr. Milhollin, Mr. Hoydysh, Mr. Bryen, please come up front.
Mr. Milhollin, we'll start with you, whenever your ready.
MILHOLLIN: Thank you very much. I'm pleased to be able to appear again before
this distinguished committee. As you probably recall, I appeared before you
back in 1997 to recommend that the NDAA process, which the committee
spearheaded, be enacted into law. And I'm very proud to have participated in
that -- in that process.
I'd like to make three points today. First, that the NDAA process has
performed brilliantly -- I think that's what the GAO report
establishes -- and that it should be preserved.
Second, I'd like to say that if the control level is raised for high
performance computers next year to 6,500, which the president has announced he
plans to do, it's virtually certain that at least 100 American high performance
computers will go to military or mass destruction weapons sites in countries
like India, China, Russia and Pakistan.
Third, the third point I want to make today is that the industry has shown no
reason -- no reason -- why the control level should be changed. The present
system is not a burden to industry, and there's no reason to think it will be
in the future.
I'd like to invite the committee to consider what would happened had the NDAA
process not come into effect. What would have happened had we had no NDAA
process beginning in 1998? And I'd like to start by considering the case of
the Digital Equipment Corporation.
Digital has since been acquired by COMPAQ Computer Corporation.
Digital applied for permission in 1998 to sell a high performance computer to
the Nuclear Power Corporation of India. The Nuclear Power Corporation of India
runs a string of plutonium producing reactors. And that plutonium in India is
available for atomic bombs because India is not a member of the
Non-Proliferation Treaty. As we all know, India tested a number of nuclear
weapons last year. This application was widely -- was wisely denied after
being objected to by most federal agencies, if not all.
And so, I think the question for Digital and now for COMPAQ, is whether they
would be happier today if that sale had gone through. Would COMPAQ or Digital
be happier if its product was helping make plutonium in India? And that's what
would have happened without the NDAA process.
Digital also applied for permission to
sell a supercomputer to the Harbin (ph) Institute of Technology in China. The
Harbin (ph) Institute is part of China's missile program. It makes rocket
cases and other components for long-range missiles.
The application was denied after objections by the Arms Control and Disarmament
Agency and the State Department. The question is, would Digital or COMPAQ be
happier today if their equipment were helping China make long-range missiles.
That's what would have happened had we had no NDAA process.
I've also mentioned a third case in my testimony. Digital also asked for
permission to supply a high speed computer to the Weitzman (ph) Institute in
Israel. The Weitzman (ph) Institute has always been part of nuclear -- India's
-- I'm sorry -- of Israel's nuclear weapon effort. It does the research
necessary to perfect the design of nuclear weapons. And so, we have the same
question here. Would Digital be happier if that
sale had gone through?
To sum these cases up, I think we can say that if we had not had the NDAA
process, the COMPAQ Computer Corporation would now look out across the world
and see machines made by a company that it just acquired operating at nuclear
weapon and missile sites in Asia, South Asia and the Middle East. That's
surely not something that COMPAQ wants to be known for.
And how much money are we talking about. The sale to India was valued at
$250,000; the sale to China at $348,000; and the sale to Israel, quite a bit
less than that. COMPAQ's annual revenue is $31 billion.
So if you're looking at things from the point of view of a CEO, you have to ask
yourself, does it make sense for my $31 billion company to supply sites that
are making nuclear weapons and long-range
missiles in three areas of the world for a few hundred thousand dollars. And I
think the answer is no. It doesn't make any business sense to do that.
Therefore, the CEO -- in my view -- the CEO of COMPAQ should be supporting the
NDAA process. It should be sending representatives here the urge that the
process be preserved in order to spare COMPAQ embarrassment over sales that
really aren't worth very much.
The second case I'd like to talk about is Silicon Graphics. Silicon Graphics
applied for permission to sell a high performance computer to India's Tata (ph)
Institute of Fundamental Research. It's part of India's nuclear program. It's
also on the British list of entities linked to mass destruction weapon
programs. And it's also listed by our State Department as involved in nuclear
or missile activities. This application, too, was blocked under the NDAA.
SGI also
applied for permission to sell a high performance computer to India's Space
Application Center. The center is part of India's rocket program. It was also
listed by both the British and U.S. government as a site that is not safe to
export to. And this application too was blocked after an objection by the Arms
Control and Disarmament Agency.
For these two sales combined, SGI received about $500,000, or would have
received had they gone through. Its annual revenue is $2.7 billion -- billion
dollars. So again the question is, does it make sense for SGI to supply
equipment to nuclear and missile sites in India for $500,000. And I think the
answer is no. SGI ought to be pleased that the NDAA process prevented the
sales from going out.
I'd like to also mention Sun, Sun Microsystems. Sun asked for permission to
sell to the Indian Institute of Technology, which is
helping develop India's biggest rockets. Those rockets will someday, if India
persists in its present policy, those rockets will someday carry nuclear
weapons to intercontinental range. That application was also blocked.
The second application from Sun was to Rafael (ph). Rafael (ph), helps make
India's largest nuclear capable missile and also the re- entry vehicle that
carries Israeli nuclear warheads. For those two sales, Sun would have received
$52,000. Sun's annual revenue is $7 billion. Again, does it make business
sense to supply products to these programs in exchange for what amounts to the
price of one luxury automobile.
Why have I brought out the sales information? To make a simple point that the
NDAA process helps exporters. It saves them from embarrassing sales which are
not good business. These companies should be the first ones to be urging that
the process be preserved.
In my testimony
I've recommended to the committee that he invite the CEO's of these companies,
of all companies whose sales were blocked by the NDAA process, to testify
before this committee and to ask the CEO's whether they think it's better that
these sales did not go out. If we agree that it was better that these sales
not go out, these CEO's should go on record as supporting the present process.
They should also endorse the request by this committee to the White House to
rescind its plan to weaken the NDAA by changing the control level.
I think this is a very public service that this committee could perform, would
be to call these companies in, call the CEO's in, and ask them whether they
think the NDAA has protected them from mistaken sales in the past and whether
it'll protect them in the future. And if they agree, then they should be
willing to endorse the process and ask that it be preserved as it is.
Last point I'd like to make is that the process is working and it doesn't need
to be fixed. There is no evidence that there will be
an increased burden on industry next year if the process is not changed. Right
now there's no burden on industry whatsoever, very little burden, an
insignificant burden. We're asking industry to fill out a form and wait 10
days before shipping a computer.
Right now the process is producing about three to four applications or
notifications per day. Even if the number of notifications doubled because of
chip speed increases, as industry has contended, we would be looking at seven
notifications per work day. The Commerce Department is now handling about 40
per work day. This is not going to make a significant difference in Commerce's
workload.
Commerce could very easily handle any notifications produced by the present
process if it remained at the same level for an additional year. If it did
remain at the same level for additional -- for an additional year, it's certain
that at least a hundred and probably more, U.S. supercomputers would not go to
military and mass destruction weapons programs across the world. If the White
House goes through with it's change in the process, that is exactly what will
happen.
The last time industry predicted that the system was going to be overwhelmed
was in 1995. The industry predicted that computers running at 7,000 MTOPS
would be available by 1997, widely available in the world. That's still not
true today. The industry was wrong in 1995 with its predictions, so why should
we believe industry predictions now.
There is simply no evidence to support the idea that the system is going to be
overwhelmed unless it is changed. The real question is whether there is
foreign availability for high speed computers. I don't think the industry has
proved its case that there is or that there will be. And the reason industry
can't prove that is that it's all based on suppositions.
The most recent studies, that is actual studies that have been done on foreign
availability, both show that there is no foreign availability for high
performance computers. These are the real questions that have to be answered.
That is, is there foreign availability and can you control computer exports
with the present system. And industry isn't answering these questions because
the answer to those questions is clear, and that is there is not foreign
availability now and there's not likely to be in the future. And if there --
and we, the United States, can still control very easily the level of
applications we can foresee during the next year if the present system is
retained as it is.
Thank you very much.
CHAIRMAN: Thank you, Mr. Milhollin. Mr. Hoydysh, now if you'll pull the
microphone a little closer to you before you start please.
HOYDYSH: And thank you, Mr. Chairman. I represent the Coalition -- Computer
Coalition
for Responsible Exports, and we would like to thank the committee for inviting
us here to give our views on the proposed change in Tier Three
computer export control threshold.
My testimony will focus on four key points. First, we believe that the
proposed control threshold is entirely consistent with technology and
competitive realities. And contrary to what Mr. Milhollin said, I think we
will be able to demonstrate that there is substantial foreign availability of
these products.
We also -- the second point we would like to make is that the proposed control
threshold should be implemented immediately. Third point is, we think the
six-month review period is too long. And fourth, we would like to extend --
rather we are prepared to work with the committee to develop a more responsive
and effective export control system.
We're not here to abolish the NDAA process, we're here to try to make sure that
it works and that it works in a streamlined
fashion so that we can both protect national security and remain competitive in
the real world.
Before I get to the specifics of my testimony, I want to make absolutely clear
a couple of points. The companies represented by the coalition take their
national security obligations very seriously. We fully support the need to
protect national security and prevent the proliferation of weapons of mass
destruction. And we fully support rational, responsive and effective export
controls.
Now to the first of my points. The 6,500 MTOPS control level is fully
consistent with technology and market realities. In the first instance,
technology continues to advance at a dramatic pace. Over the past 12 months,
the speed of microprocessors that are used in single processor and
multiprocessor systems has increased by over 300 percent from somewhere on the
order of 500 MTOPS to about 1,700
MTOPS, as just announced this week by the INTEL Corporation. These chips are
designed to work in multiprocessor systems containing up to eight chips.
Another example is that in September of this year, Apple Computer introduced
the first single processor PC that operates at over 2,000 MTOPS. In fact, it
operates at about 2,700 MTOPS. Early next year, this same PC will operate at
over 3,000 MTOPS. So we're halfway into the 6,500 level with a single
processor PC. And just this week, IBM also announced a line of their Aptiva PC
which operates with one processor over 2,000 MTOPS.
The reality is, we are rapidly approaching a world in which there will be no
products below 2,000 MTOPS, no leading-edge PC or multiprocessor products.
Just to clarify one thing, and I think that we really
need to make sure what we're talking about. These are not supercomputers. The
committee knows that supercomputers are those kinds of products that you looked
at -- I believe it was last week you had a hearing about a product sold at a
computer show by the Energy Department. That computer operated, according to
the Energy Department, 150,000 to 250,000 plus MTOPS. That's were
supercomputers begin, not at 3,000 to 6,000, which is the PC and the ordinary
server range.
These computers are, in fact, commodities. They are available in large
quantities worldwide. A study by the Gartner Group projects that about five
million of these boxes will be sold during the year 2000. This number does not
include the hundreds of thousands, or possibly millions of single processor
PC's which have just come on the market. When we commissioned the Gartner
Study, we did not anticipate that these single processor machines would be on
the market as
quickly as they were.
And finally, possibly to the most important point, these computers are readily
available from foreign manufacturers. Of the five million machines that'll be
sold in the year 2000, over one million will be sold by foreign manufacturers,
including companies from Japan, Germany, Taiwan and France. These are major,
world-class companies that can fit any market void left by U.S. companies.
All you need to do is hook up your computer, get on the Internet and take a
look at the web sites of Acer, which offers two and four processor machines
that we're talking about; Comparex from Germany; Toshiba; Bull from France;
Hitachi; Olivetti and NEC. And these are only the major companies. These
companies can provide any number of these type of systems on 24-hours notice.
The second point is, we urge the committee to support making the new control
threshold effective as soon as possible. By law this new Tier Three threshold
cannot become effective until January 2000. This additional three-month delay
will seriously affect the ability of our members to market commodity systems
during the last quarter of this year, a time when sales volumes are
traditionally heavy.
Our members who sell commodity products often sell them through distributors
who then resell them to other distributors and ultimately to the final end
user.
According to the regulations that are in place now, you cannot sell to a Tier
Three country without identifying the ultimate end-user at that time of export.
The marketing model used by many of our members simply do not allow this to
happen. As a result, we have been forced to lose sales, curtail our activity
in these areas; or, in some cases, dumb down our products by
removing capability in order so that they fall below the threshold.
To underscore this competitive dilemma, let me just read the press release that
was issued in Tokyo on August 24th by the Fujitsu Corporation. It says,
Fujitsu today announced that it will be shipping 8-way upgradeable Intel
Pentium III Xeon processors -- processor-based servers before Christmas. These
are computers that have an MTOPS rating of 9,000, not the 6,500 that we're
talking about. And you can find the same kind of statement on the press
releases of all of these foreign companies.
The third point that I really would like to make is that we ask the committee
to support a permanent reduction in the 6-month waiting period. We think that
waiting a half a year before (OFF-MIKE) are implemented is simply too long.
Six months is longer than the life cycle of
many of our products. We note that Congress has a limit of 30 days for
reviewing hardware military sales. And so we urge the committee to consider
making this review process consistent with the 30-day military hardware review.
Finally, I would like to note that we are prepared to work with the
administration and the Congress to develop a responsive and effective export
control system that balances national security technology and competitive
interests. In that regard, I would note that our industry's CEOs and members
of the administration and members of Congress have had preliminary discussions
on this topic under the auspices of Secretary Hamre. I understand that
Congressman Weldon, who I don't believe is here right now, has played an
important role in these discussions.
We are prepared to continue this dialogue or, perhaps I should say, trialogue,
to make sure we get the export control system right.
Thank you very much. I'd be happy to answer
any questions that you may have.
CHAIRMAN: Thank you, Mr. Hoydysh. (Inaudible) for us to recess the committee
and to sound the gavel, we have a vote on the floor. The meeting will stand
recess at the sound of the gavel.
(RECESS)
CHAIRMAN: The meeting will please come to order.
Mr. Hoydysh, you had completed your statement, is that correct?
HOYDYSH: Yes, I just wanted to point out to the committee one thing. We have a
chart here that shows that the relative positioning of the foreign companies
and the U.S. companies in terms of sales of these systems.
And I also wanted to make one final point. The systems that we're talking
about are of this size and magnitude. This is a two- processor system; a
four-processor system is not much bigger than this. And this is the Apple G-4
chip, which will have an MTOPS rating of over 3,000 by January.
So we're not talking
about the computer that you folks talked about last week, which had to be
brought out on a flatbed truck, as I understand it from reading the testimony.
You don't need a flatbed truck to ship these around. You put them in an
Airborne Express box and you ship them overseas or wherever in 24 hours. So,
just to refocus on what is really a supercomputer. This is not a
supercomputer.
CHAIRMAN: Thank you.
HOYDYSH: Thank you.
CHAIRMAN: Mr. Bryen.
BRYEN: Well, thank you, Mr. Chairman. I have a prepared statement. I think
you copies of it. I would like to ask it be put in the record today
CHAIRMAN: The entire statement will be put in the record, sir.
BRYEN: Mr. Chairman, I'm batting clean-up today, as so I will try to summarize
some of the aspects of my statement and also respond to
my colleagues, if I can.
High performance computers, Mr. Chairman, are a vital element in United States
industrial, economic, political and military leadership. The United States
developed the first high performance computers -- and I, by the way, equate
high performance and supercomputers as the same thing. And we continue to lead
the world in the development of high performance computers, both hardware and
software.
By the way, Mr. Chairman, if we are unrestrained in how we handle the export of
such computers to what I would call dangerous countries, then we can only
expect that our allies and friends who may also have these capabilities will do
the -- will not be restrained, as well. And I think this is a terribly
important issue and I hope it's one that the committee will keep in mind.
Now, my testimony focuses on the export of high performance computers where
there is strategic risk
involved. And there are a number of countries that are embarked on nuclear
programs, have advanced missile developments underway. And, as you know, high
performance computers play a vital role in assisting in the development of
these modern weapon systems.
Now, I want to distinguish for the committee the difference between a high
performance computer and a desktop computer. I want to make my position very
clear. I don't think there's any purpose today in trying to control a desktop
computer, even though all over the world use American processors. The fact is
that the manufacture of these devices occurs, really, on a global basis. Quite
a lot in Asia. In fact, I would guess that this one and many others are made
are made in Asia. Many of the peripherals are made in Asia. So, trying to use
expert controls for that kind of grid or item is simply not going to work.
Can't be done.
I don't think there's any point in having an export control system that tilts
at windmills. I think you have to have controls that make sense, that can be
enforced and that protect our strategic interest.
So the next question obviously is, is there a difference between a high
performance or a supercomputer and one of these things. I believe that there
is significant difference. And it's not related just to this measurement which
is called MTOPS -- millions of theoretical operations per second.
MTOPS, by the way, is an invention of the government. It's not a measurement
that is used in industry. It replaced another invention of the government,
which was called process data rater, PDR. Both of these measurements are
extraordinarily confusing and neither of them, it seems to me, really measures
what matters.
And one of the things that this committee
can be very helpful on in working with industry, as Dan Hoydysh has offered,
and I think it's a good offer, is coming up with formula to differentiate
between the desktop and the super high performance computer, one that will then
differentiate the export control mechanisms in place.
Let me just point out a few of the differences so the committee understands and
has on record, because I don't think anyone has -- in all the debates that I've
heard on the subject, this aspect has not been brought out, I think, clearly.
First of all, high performance computers have tens or hundreds of processors.
Desktop machines mostly have one or two processors. Now, in the next few
years, we'll see more four and eight processor desktop computers or server
computers. But these still will fall short of the category that we normally
think of as
high performance computers.
Most high performance computers today, it has not always been true, but today
are parallel processor machines. Most desktop machines, in fact, all of them
as far as I know, are not parallel processor machines. Most of the advance
work, for example, at the Department of Energy and at various universities that
are working with the Department of Energy to develop nuclear codes and nuclear
weapons, to develop new types of missiles, are parallel machines. Most of the
software that's been developed for this purpose runs only on parallel machines.
I'm not worried about this, this little old machine. I'm worried about the big
machine. Now, high performance computers, supercomputers feature very tight
coupling between the processor -- that's the microprocessor, this device here
-- and the memory in the machine. This tight coupling is something that you
don't find in a desktop
computer.
What that means in practical terms is that when you have a high performance
computer, the memory can in real-time be addressed, the whole memory can be
addressed, and very complex calculations can be run. You can't do it on one of
these things.
Now the best high performance computers have random, non- interruptible
communication between the machine's processor and memory; desktop machines have
interrupts in their operating system. They don't have that kind of
communication.
Now, there are other features, as well. But the main point that I want to make
is that we can't just rely on this MTOP measurement to distinguish between the
high performance machine and the not so high performance machine.
Second point I want to make is that what has to done here and what's really
vital is to make sure that a potential adversary can't get hold of the machine
and the kind of software that's needed on those machines, but will enable them
to do
advance nuclear research or build advance missile systems. It's a logical
place to be, unfortunately our expert control regime right now addresses
neither issue very well. In fact, I'd say it doesn't address them at all.
First of all, most of the software is readily available and is not export
licensed, so that this software is flying around all over the world. There are
conferences going on, supercomputer conferences, where all of this is being
discussed. In fact, in this May, there will be one in Beijing, with heavy
attendance from the United States, I might add, from some in the military. And
these are just tremendous events in which scientist, including from the Chinese
Academy of Sciences, which is well-known to be involved as Gary knows in
nuclear research and development, military nuclear work.
I've been there. I visited that center in China. In
fact, I was shown their nuclear accelerator. So I'm quite well aware of what
they do.
And these kinds of conferences and meetings and exchange of information and the
software that's being legally transferred is an area I have great concern
about, and it's not being addressed by the expert control mechanism.
Second -- the second point I think that goes with that is that the current --
the current controls that have been proposed, no one has analyzed the impact
that will have on the ability of a country like China to run that kind of
software on the machines they have already acquired or that they will acquire,
and to achieve the rapid prototyping of new -- new missiles and new weapons
programs.
There's a very interesting article -- it's in my testimony today -- that
appeared in the Hong Kong Standard just a week ago, where China announced that
it had conducted laboratory simulations of a test launch of a multi-warhead intercontinental ballistic missile called the DF-41.
I don't know if you've seen that article, but as far as I know, no one has
refuted the authenticity of the information, so I'm accepting it as valid. But
the committee should ask the intelligence community to assess this and to
assess it carefully, because what they're saying, basically, and let me read
from the reporter's account.
She says: Many sources told her that the computer simulations of the launches
of the solid-fuel Dong Fang-41, the DF-41 ICBM had been completed and proved
successful.
This is the first time China has publicly acknowledged that it has the ability
to simulate -- and I don't know how well they've done it, but they claim to
have done it well -- to simulate the design of a new ballistic missile and
nuclear weapon.
So if it's accurate and if it's true, we're seeing the first results of the
transfer of
supercomputers to China.
You know there's more 600 of them there that have gotten there since 1996.
Before 1996 there were none.
And this is where the threat I think the future threat's coming from and this
is where the problems coming from. There are many ways that supercomputers can
help a country like China not only advance its nuclear programs and its missile
programs, but to take it very hard for us to contend with it. We're talking in
this country about building a missile defense. But if China can reduce
fracture side of incoming multiple warhead, incoming nuclear warheads that may
be aimed at the United States, if China can build better penetration aids using
the supercomputers to design them, then we're in a lot of trouble because
missile defense won't work.
So Mr. Chairman, the theme of what I have to say is that I would urge the
committee -- I appreciate the work, the good
work, that this committee has done, it's very important work in trying to get a
hold of this problem -- but the current export control parameters simply don't
do the job. And raising them higher without really studying this issue in a
strategic sense, having a proper strategic analysis, is irresponsible. It's
irresponsible by the administration, and it's really something that needs badly
to be fixed.
So my conclusion is this: That before there's any change in these parameters,
the committee should demand that -- I would propose a Joint Chiefs of Staff
study, a Defense Department study, of the strategic implications of potential
adversaries acquiring even more powerful supercomputers and how that will
impact us, in particular in our ability to be able to field the missile defense
system in the future.
And thank you for -- Mr. Chairman, for inviting me here today, and I'll be glad
to answer your questions with
my colleagues.
CHAIRMAN: Thank you, Mr. Bryen.
The chair recognizes the ranking member, Mr. Skelton.
SKELTON: I pass at this moment.
CHAIRMAN: The gentleman from Virginia, Mr. Bateman.
BATEMAN: Thank you, Mr. Chairman, and I thank the witnesses.
I find myself in the very, very unusual position for me when I have the benefit
of hearing people with great expertise expound on the subject on which I had
none. I'm somewhat perplexed with Mr. Johnson, I believe, and Dr. Milhollin
telling us that we have no concerns, that no one in the foreign marketplace is
going to be able to sell these high speed computers and therefore we don't need
to regulate or that -- it's useless for us to regulate and control them.
On the other hand, I hear the opposite. Who am I supposed to believe on this?
Is there a marketplace out there and in a number of other
areas and companies in the world where a high-speed, high- performance
computers will routinely be available, even if we were to continue to restrict
the sale of them by American companies?
Is there a problem? Or is there another problem?
HOYDYSH: I guess I could take a crack at answering this.
There is clearly a market for the kinds of systems that we're talking about.
And for the purposes of the threshold we're talking about, two- and
four-processor systems, these are sold in the millions worldwide, and anywhere
from 25 to 30 percent of the market is provided by foreign manufacturers.
And again, I have -- we have commissioned a study that looks at this.
There are industry -- there are firms that are in the business of cataloguing
sales by U.S. and foreign companies that clearly substantiate this.
The companies that we are in competition with are large world- class companies,
whether it's Fujitsu or NEC or Hitachi or Siemens or Olivetti -- all of them
are capable of manufacturing this because the
industry has moved to allowing -- or has actually the technology to where these
can be assembled from commodity parts. There's no technological impediment for
anybody with a modest amount of technical expertise to put one of these
together by buying the boards on the open market and the chips on the open
market.
What keeps people from getting into the business is a question of
entrepreneurialships and economics. In order to sell these and make money, you
have to have personnel, you have to have advertising, you have to have a
service department.
But for someone -- whether it's the Chinese National Academy of Sciences or
anyone else -- to buy a bunch of these boards and plug this in is a trivial
exercise. So there is absolutely -- in our judgment -- a large market for
these that can be supplied by foreign companies.
MILHOLLIN: Mr. Bateman, can I respond to that?
BATEMAN: I'd like for you to.
MILHOLLIN: I think that what we've just heard is -- I must
say this with all due respect to my colleague -- quite misleading. It's not
true that there are millions of computers available from foreign sources that
operate above 2,000 MTOPS.
The most recent studies were done by GAO and by -- actually a group that the
government commissioned to do several studies. They both found, I think within
the last year or so, that there was no foreign availability in the ranges we're
talking about, between 2,000 and 6,500, from companies that could supply in
sufficient quantity and sufficient quality with sufficient backup so that a
person -- so that a buyer would feel confident acquiring such a machine.
The industry is doing -- Mr. Hoydysh is making the same argument now the
industry made in '95, which is to say that the sky hasn't fallen yet but it's
just about to fall. We're just about to see a
hoard of foreign assemblers who are going to take their little solder guns and
they're going to buy all these parts, they're going to put them together, and
they're going to bury our industry with competition.
It didn't happen the last time they predicted it, and I don't believe it's
going to happen this time either, and I don't think there's any evidence that
it will happen.
BATEMAN: Let me turn to another aspect in the testimony that puzzles me.
Dr. Bryen, you made reference to MTOPS being a manufactured term by the
government, having no industrial or scientific basis.
BRYEN: Well, I didn't say it didn't have any basis. What I mean was that it is
not a term that is commonly used as a measurement of performance by industry.
BATEMAN: And you criticized that as being the measure of what should be
licensed as exportable.
BRYEN: Yes.
BATEMAN: Do you have
another measure that you would advance as being the superior or the appropriate
measure?
BRYEN: I think we need to include the other parameters. For example, one
measurement -- and there's probably a good scientific way to put it, I'm not
sure how to write the rule, but the rule has to do with the coupling of memory
to the processor. This seems to be a very vital issue.
If the coupling was very tight and there's the ability of the processor to
address all the memory at one time, then that's one kind of machine we have to
be concerned with.
If the coupling is loose and can address the memory at once, then I think the
machine is far less likely to prove problematic.
And as I said in my testimony, most of the -- most of the -- these kinds of
machines that Dan has here -- he promised to give me this to take home, by the
way -- most of the machines like this have very loose coupling of memory. And
they don't really pose a strategic threat.
The Commerce Department could fix this whole
problem in a minute. All they really have to do is to put a line in the
regulations that says anything less than eight processors is not covered by
these regulations, and the whole problem is solved.
From the industry point of view, they get to exploit these things which they
think are getting out there in large numbers. I'm not going to say it's in the
millions, but it's in large numbers in any case.
And that will satisfy I think a lot of their problem. And at the same time we
can then focus on dealing with controlling the parallel processor
supercomputers, which is the strategic problem. And it doesn't -- I don't
understand why we haven't done that. I'm very frustrated about that, and I
don't see why the Commerce Department and Defense can't get together on that
and come up with a regulation that makes sense for everybody.
HOYDYSH: Congressman, I'd
like to add one more thing. I just wanted to make clear for the record that my
offer to give this to Dr. Bryen have nothing to do with his statement that he
just made.
BATEMAN: Mr. Chairman, could I be indulged for one brief moment?
CHAIRMAN: You may.
BATEMAN: Mr. Hoydysh, you indicated that you are not advocating that we
completely dispense with any of our regulations or control over supercomputers,
and that we do need a rational program, but you apparently take issue with the
rationale of the one that's in place.
Have you and industry sat down and undertaken to write what you thought would
be the ideal regulatory regimen?
HOYDYSH: We have certainly been thinking a lot about it. And we have been
discussing it among ourselves, and we've been discussing it with the executive
branch and even with some members of Congress.
I think Steve is on the right track here. I think some of the elements that we
have clearly identified as being part of the
system is: You should not control what is not controllable. And so you have to
look from the bottom up, you have to look at what are mass-market products, and
I think Steve is on to something in terms of looking at the number of
processors as one alternative.
One of the problems with the MTOPS rating is that the numbers have become so
inflated that no one really understands what it is, but, by God, 6,000 is sure
a lot of them. So it scares people.
So this is probably not the best method of looking at it. And maybe we even
need to look at architecture, some of the other things that Steve looked at.
We clearly agree that from the top down we ought to be controlling the highest,
most powerful computers available. And the things that you saw, that you
talked about last week, which required the flatbed truck, was up in the
150,000, 200,000-plus, and that's an old computer.
So there are things -- the
Deep Blue, Pacific Blue, from IBM and Silicon Graphics that have thousands of
processors. We're not talking about that in any sense. We're talking about
the products that are essential to make the Information Age work. These are
the products that make the Internet work, that make all kind of commerce work,
and that are in fact -- and I have to vehemently disagree with my friend, Mr.
Milhollin here -- that these products are available. They are available from
world-class companies, companies that are Fortune 100 companies, that have
thousands of employees worldwide and produce thousands of products and make
billions of dollars in revenue. There is no impediment to getting a
two-processor or four-processor or even an eight-processor system from any of a
dozen foreign manufacturers.
And we can document this, and we can document this with independent sources
from the IDC Corporation and from Gartner and whatever. There are industry
measures of this. These products are available and they're being sold now and
they
will be sold. And the GAO report that Mr. Milhollin refers to is something
that was in 1997, and since that time, microprocessor speed has increased by
300 percent.
CHAIRMAN: The gentleman from Missouri, Mr. Skelton is recognized.
SKELTON: Let me ask: When did the industry first ask the White House to boost
the threshold -- anybody?
HOYDYSH: This current...
SKELTON: Yes, if you know.
HOYDYSH: Well, we kind of asked for it on a continuous basis. I'm not sure that
I could actually pinpoint a date for you. I don't know if anybody else maybe
-- maybe -- I can't give you a date.
SKELTON: Now, one of the major problems is the time limit, am I correct, 180
days?
HOYDYSH: Right.
SKELTON: Now, if I want to go backwards and recite numbers from 180 down to
zero, at what point would you raise your
hand and say,
"That's about right," the time limit.
Let me start with you, Mr. Milhollin.
MILHOLLIN: I think the time limit of -- you're talking about the time limit
for...
(CROSSTALK)
SKELTON: Yes, that's correct.
MILHOLLIN: I'm fairly happy with the present system.
SKELTON: All right, Mr. Hoydysh.
HOYDYSH: I would say we prefer 30 days, but let me just expand on that. I
think that we have to start thinking about doing what we're doing not in serial
fashion, like the old computers, but in parallel.
There's no reason why information cannot be presented to the committee by the
administration, by the GAO, by us, on a continuing basis so that we don't have
to wait for a decision, let me know for a report, let me write something else.
Of the 180 days that we are -- or six months,
half a year, that we are waiting, most of the time is dead time when nothing
much is happening.
So I think that we can significantly reduce the time period and still get the
information to you that you need to review this.
SKELTON: Mr. Bryen.
BRYEN: Well, I think there's two issues. One is that 180 days I think is a
good timeframe if we're dealing with really true high- performance computers.
It's a bad timeframe if we're dealing with these small processor machines.
So if you make a distinction between the two, then you need to change the rule,
which I think the problem goes away.
SKELTON: Thank you.
CHAIRMAN: Ms. Fowler is recognized.
FOWLER: Thank you, Mr. Chairman.
I want to thank the gentleman for being here. Mr. Skelton was getting -- my
question I had because I've heard different timeframes floating around. I met
with
some industry people last week, and they said, well, they preferred 30 days,
they could live with 60 days, but 180 is too long, you know, in trying to find
some compromise here.
Because I think what we're all looking at is to come up with something that
protects our national security but is not doing harm to our computer industry
and to the valid need to sell the ones that are readily available.
And there doesn't seem to be anyone out there that's keeping real good track of
how readily available are some of these, and, you know, all sorts of numbers
that float around.
And then -- and none of you really answered his question, which he was trying
to get at as to what the days would be unless we change it.
So I would just like to ask you: What factors -- what the response will be of
reviewing the administration's proposed change and what was put into the fiscal
year in '98 defense bill on notification? What factors should be looking at?
How did you assess the
importance of each of those factors? Should we be insisting on any safeguards
as a condition of agreeing to any proposed threshold changes? If so, what
should they be?
Because this train is moving down the station and we need to, you know, somehow
work out a system that's going to work. And I'm very concerned that we aren't
coming towards that. I would appreciate some of your expertise.
MILHOLLIN: If I might -- again, I would say that the first factor that Congress
should look is whether the government can actually control the flow of
notifications that is presently coming in and is likely to come in if the
present control level is kept as it is or it's changed to some other level.
I mean, the ability of the government to process this paper and to make
decisions is a vital consideration. Mr. Hoydysh says that controllability
should be the main question, and I agree, it should be.
And the real question of controllability is: Can the Commerce Department and
the other agencies process this information?
Now, DOE, I think, is processing around 10,000 to 12,000 applications a year.
The GAO has just established that this new process, the NDAA process, is
generating less than 1,000 a year.
So we're looking at something than a 10 percent increase in the work load.
So far I haven't heard anybody say that that's not doable.
The second criterion I think you should look at is: What difference does it
make? What impact will it have on foreign programs if we don't control or if
we move the level to some other level?
In my testimony I have cited a report by DOE which found that -- it's a recent
report by DOE -- which found that computers operating at about 4,000 MTOPS
would directly aid the nuclear weapon design efforts today of India, Pakistan
and China, and would aid
China most of all.
So if we're talking about what level of computing power is important
strategically, the DOE has given us a number. They say around 4,000 would aid
these nuclear weapons missile programs.
So those are two criteria I think that the Congress and the committee ought to
look at very carefully: controllability, by virtue of government work; and
second, the impact of letting these computers go out with control.
FOWLER: Does anyone else have any comment on this?
BRYEN: Yes, thank you.
I think it would be reckless to agree to any change in the parameters without
receiving with that proposed change a strategic analysis of its impact. And
there isn't any. There simply isn't any. No one's done it.
And by that I mean, what's the impact on -- particularly China, I'm very
concerned about China -- from a nuclear defense point of view, what's the
impact? What does that
mean as a practical matter? We don't have that kind of analysis. I don't see
how you can then raise the threshold, and I don't understand how the
administration could even dream up changing the numbers, because it's -- it has
no grounding in any reality. It's simply taking a number -- you know, take a
number, take any number and bet on this. And that's a crazy way to do
business.
I don't think that -- Dan represents a group for responsible computing. They
don't want to be in an irresponsible position.
I really believe that industry follows the flag on these things. They want to
do things that are responsible, that are right. And where we're being put is
in a situation where we don't know. So I would say the answer to the question
is that we have to demand such a study before anything is done.
HOYDYSH: I would like to respond to your
question about the time period. In the first instance, what we would like to
say is the current threshold proposal can be implemented as quickly as
possible, without dealing with the question of a permanent change to the thing.
And that can be done by a one-time waiver of some kind. So what we need to
have is this implemented immediately.
The problem that Dr. Gary Milhollin refers to all the time about the volume --
we are not seeing the volumes in some of the product sales that we have in
terms of asking for approvals, because as I stated earlier, the Apple
Corporation, for example, cannot market its products effectively the way --
through its business model -- selling to distributors and then to third parties
and ultimately to the end- user. So they are simply foregoing sales into Tier
Three until this threshold is raised. And that amounts to tens of thousands of
units over this last quarter.
So the first thing we'd like to do is
see the change immediately. Then we would like to see if there is a way of
shortening the permanent review period because during a six-month period, we --
we did not anticipate that the Apple machine would be ready in this time
period, and it came on the market during the six-month hiatus. So we would
like to see if there's a way we could streamline this process by providing the
information sooner by engaging in discussions -- whatever it is -- so that you
would have the information you needed to review it, but it wouldn't take six
months worth of wait.
FOWLER: Thank you, Mr. Chairman, thank you.
CHAIRMAN: The gentleman from Mississippi, Mr. Taylor, is recognized.
TAYLOR: Thank you, Mr. Chairman. And I -- I do appreciate you all being here.
I've got to admit a very strong sense of skepticism. I remember very vividly
sitting on the House floor when a former colleague from California was
passing around a letter saying that we needed to allow a company to launch
telecommunications satellites over in China. And I remember in that letter
were some assurances that nothing can possibly go wrong; that we're going to
watch it every step of the way. And it was signed -- I did not sign it -- but
by Republicans and Democrats alike. Of course, that's where the Hughes- Loral
snafu that gave the Chinese just an incredible amount of information came from.
I hear the testimony of a few of you that says we've got to do it or someone
else will. As I was listening to that, I turned to my colleague from Hawaii
and said: At what point does industry come to us and say fill in the blank,
we've got to sell nuclear weapons because if we don't, somebody else will and
there's 15 other countries out there that know how to do
it.
I mean, where do you draw the line? We spend $53 billion today trying to
defend ourselves from other people's missiles. Why would we want to make it
easier for other people to develop them? And I'm asking this as a question.
I'll give you an opportunity to respond. Tell me why I'm wrong and why you're
right. Mr. Hoydysh?
HOYDYSH: Well, Congressman, let me try to answer the question. I mean, I don't
think anybody on our coalition wants to sell nuclear weapons. And certainly we
don't want to help anyone design nuclear weapons. In fact, we cannot even sell
a pencil to an entity in Tier Three countries if we know that it will be used
to develop nuclear weapons or weapons of mass destruction. We cannot come in
with even the lowest level of computer, and somebody said
"I have this nuclear
weapons lab but these are (inaudible) computers; help me hook them up." We can't do that and we wouldn't do that. So first of all, we have no
interest in doing that.
The real question that has to be asked, and I grant that the answer is
complicated and it's a difficult question is: At what point does restricting
our sales of these commercially available, low-end computers hurt us more than
it protects national security? Because we depend -- our national security
depends on the technological superiority of our computer industry and our other
high-tech industries. And the fact is that technological superiority is
derived from profit, which is poured back into R&D, which creates new products. If you break that cycle, you risk damaging the
computer industry and other high-tech industries, and you risk in the long term
eroding our technological superiority.
Now where you draw that line and where the balance is, that's a difficult
thing. But you have to understand that when you're selling this and if we make
a profit,
it is in our interest. You know, at what point do the restrictions end up
hurting us more than helping us. And I think that's the key that has to be
looked at and that's where we are all struggling with. We think the line where
it is drawn now helps us more than hurts us.
BRYEN: Could I respond?
TAYLOR: Dr. Bryen.
BRYEN: Thank you. First of all, I think what you said is right on. I couldn't
agree with you more. I really -- I think there's some confusion. You don't
balance national security with competitive interests. National security is
fundamental. Competitive interests are not fundamental. They're important,
but not fundamental. If you're dead, competitive interests won't help you
much. You have to be sensible here.
And the second point is if industry is sincere about this, there's an easy
solution. All they have to do is accept the difference between a high
performance computer which can hurt you, and a
small machine which won't bother you. And the committee can be helpful here in
a very simple way. All the committee has to tell the administration is we
won't approve changing the threshold, and you can solve the problem of the
small machines by writing an exception in the regulations. It's a simple
exception.
These functions are not a strategic threat -- not a significant strategic
threat. The high-performance computers, the parallel processing machines are a
strategic threat. That's what you design nuclear weapons with. That's what
you design multiple reentry warheads with. That's what you design new
generations of missiles with. That's why you break crypto-codes with. Those
are just some examples. What you design germ warfare with.
I mean, these are the kind of machines that can really cause great difficulty,
and we're seeing evidence that the first wave of transfers of high performance
computers over the past three years is causing that kind of problem already.
It's very important
for this committee to respond, and I think -- and as I said in response to an
earlier question, we must have a strategic study. The administration has to
prove that these increases are not harmful to national security, and they
haven't done so. There is no Defense Department study -- none has been done.
It simply doesn't exist. And that's crazy.
TAYLOR: Mr. Milhollin?
MILHOLLIN: I would like to respond as well. I think that's an excellent
question and I think it puts its -- it shows a very perceptive understanding of
a process that we've been dealing with for a long time. Here's how I see it.
What we're really doing in these cases where we allow imprudent exports is that
we're competing against ourselves. Taxpayer dollars are used to develop the
high-tech equipment that we're not -- that we're now concerned with. Taxpayer
dollars are used to develop high-speed
computers -- always have been -- along with many, many other defense items.
So after receiving these taxpayer dollars, the industry develops a high-end
product then it comes in and argues to Congress that in order to stay
competitive, it ought to be able to sell those products to foreign countries.
Congress says yes. The products are then sold to foreign countries. And then
we hear that we're not as far ahead of those foreign countries militarily as we
should be. And so the industry comes in and says: Well, we need even more
money to build even higher tech products so we can maintain our advantage.
What happens is the tax dollars are being spent to compete against ourselves.
We know that nuclear weapons have been manufactured in the tens of thousands.
There are probably more nuclear weapons than there are high-speed computers, if
we follow the industry's definition. Does that mean nuclear weapons are a mass
market
item, and uncontrollable because, as you've pointed out, other countries also
manufacture them? Under the industry's logic, we should simply decontrol
nuclear weapons because there are just so many of them, we can't keep track of
them and we know other countries are making them.
I think that the industry ought to be happy with the NDAA process because in
exchange for waiting 10 days, the government tells industry: Look, this product
is going to help make a nuclear weapon or a ballistic missile in a foreign
country. The industry gets that for free. The government provides a free
bureaucrat to keep the industry out of trouble. But the industry keeps arguing
that they don't want to know. They don't want to wait that 10 days. They
don't want to find out that their product is going to a missile site or a
nuclear site. They just want to decontrol.
I can't remember a
time when industry wasn't arguing that whatever system we had was too
restrictive, to answer a previous question. I've been working on export
controls for a long time. I asked a bunch of industry people one day to give
me a list of things that they thought were really important, and they had never
thought about it. Whatever it is -- whatever is being controlled, they want to
-- they're against. They want that control to be dropped. I think it's as
simple as that. And I don't think that's a responsible position. And I think
we do have a system that allows them to find out where the dangers are, and I
just don't understand why they don't support it.
CHAIRMAN: The gentleman from Indiana, Mr. Hill, is recognized.
HILL: Thank you, Mr. Chairman. Mr. Milhollin, you made -- went through a
series of examples of companies that orders were canceled or not authorized.
And then you went on to make the
statement that just because they have done that doesn't mean the rules as
relates to the NDAA process should be changed. Could you kind of expand upon
that and tell me what your line of thinking was there?
CHAIRMAN: Mr. Milhollin.
MILHOLLIN: The idea is that is you look at the GAO report, you see that about
85 cases were blocked -- either denied or returned without action. If you add
up the -- if you add up the value of those cases, I think you come out to
something like maybe $30 million or $50 million. I recommend that the
committee ask the GAO to total them. And my point was that that's a very small
amount of money compared to the billions of dollars that the computer industry
takes in. And that it's a good deal. That is, for giving up the insignificant
amount of sales, the companies are avoiding having their products go to the
wrong places. And that was my point.
HILL: But I guess my question to you is is -- you made the statement that just
because those sales were
not authorized, that that does not mean that we should change the rules of the
NDAA...
MILHOLLIN: Oh, my point is that if we know that with the present level of
control, we stopped 85 machines that should not have gone out, if we raise the
level to 6,500 we have to -- we can be virtually certain that at least that
many machines will be stopped next year, and won't go to missile and nuclear
and other sites that we would prefer for them not to go to.
And so I think that the fact that we stopped those machines is an argument for
keeping the level where it is.
HILL: Are you saying then that if we raise the threshold that there will not be
enough people to analyze whether or not those sales ought to be authorized?
MILHOLLIN: No, I'm saying that if we raise the level to 6,500, there's no
evidence that we couldn't -- that the people we now have could not handle the
workload. It seems to me that they probably could. The current workload
increase for them is about 10 percent with the present level. Even if they get
a
doubling of the increase in the workload, that would only be a 20 percent
increase in the workload. And I haven't -- I'm not aware that we couldn't
handle that.
I hope I've answered your question.
HILL: Well, you have, but not to your advantage, it seems to me. It seems to me
that if they can handle the workload by an increased threshold, why shouldn't
be go ahead and do it?
MILHOLLIN: Well the point is that -- the industry's arguing that unless we
increase the threshold, we're going to be swamped. Well -- I'm sorry, what I'm
arguing is that we can keep the threshold where it is and handle the increased
workload. But even if the workload -- because chips increase in power and the
workload doubles next year if we keep the level where it is now, we'll be able
to handle it. That's my point.
HILL: OK. Thank you, Mr. Chairman.
CHAIRMAN: The gentleman from Hawaii, Mr. Abercrombie.
ABERCROMBIE:
Thank you very much, Mr. Chairman. Gentlemen, I was kindly -- a kind reference
was made to me by my colleague -- about me by my colleague, Mr. Taylor, and I
have to return the compliment by -- when he said to me at one point during this
discussion -- he said: This reminds me of somebody standing up and saying we're
for responsible binge drinking.
(LAUGHTER)
And I have to agree with that. I have a great deal of difficulty in
understanding how we're supposed to put forward additional billions and
billions and billions of dollars for a national missile defense to defend
against weapons that in great measure -- at least in some appreciable measure
are going to be constructed by those to whom we have sold the technology that
enables them to build these weapons. I don't understand the logic of it. I
understand the logic of greed. And I don't mean that personally against
anybody because I think it's worse than that. If it's personal greed, you can
dismiss it. You just say it's a human failing.
But what I see happening here is institutional and programmatic, and a
deliberate policy of the United States of America. And that is something that
I think needs our attention, Mr. Chairman. I can't say it in any more severe
terms than that. I don't want to personalize this. I'm not looking for
villains. I don't have to look further than the institutional network that
we're looking at to find a villain.
I believe, Mr. Chairman, that government is an instrument of our will, and if
we fail to exercise our will in a responsible way, in this instance we're going
to be -- we'll have only ourselves to blame.
Now having said that, Mr. Hoydysh is wondering: Oh God, what's he going to ask
me?
(LAUGHTER)
I
hope you will accept my point, Mr. Hoydysh, that I see this in institutional
terms, not in any personal terms. I don't attribute any -- a base motive to
your presentation whatsoever. When -- and I'm pulling this out of context, but
not to cause you difficulty, the necessity of remaining competitive in the real
world -- are you speaking in a illustrative sense about what the real world is?
Surely you don't mean to contend that what I just cited as weapons that may be
used against us are constructed at least in part from what we might give them
technologically speaking is not real -- is theoretical or hypothetical?
HOYDYSH: Congressman, I'm not in any sense suggesting that we give any
potential enemies any technology, any information that would substantially
enhance their capability to do us damage. What I am suggesting is that the
export control
system that we have in place now does not work well, and I agree with Dr. Bryen
that -- what we are talking about are the low-end of the spectrum, which is not
of concern. Dr. Bryen is in many cases on the opposite site from where
industry is -- we agree on this.
ABERCROMBIE: Let me ask you this, then. Do you agree with Dr. Bryan -- because
that was going to be my next point and my final point on this -- do you agree
with Dr. Bryen that there were some relatively direct -- I won't say simple
because things are never simple and the simpler they are, probably the more
difficult the consequences to deal with -- were some rather direct or
(inaudible) -type language that this could be solved at this lower level?
HOYDYSH: At first blush it appears that there might be a solution along the
lines that Dr. Bryen...
ABERCROMBIE: Let me ask Dr. Bryen...
HOYDYSH: But again, like you said, it is -- you'd have to look very carefully
at
exactly the implications for the industry to make sure that there are no
competitive disadvantages.
ABERCROMBIE: All right. Mr. Hoydysh, I haven't the slightest doubt that there
will be a real close (inaudible) there is any competitive disadvantages. I
don't think that's going to be a problem.
Dr. Bryen, do you in fact have language or have you thought about language or
asked someone with some legislative experience to put together language that
would address the commerce questions that you raised? Because you did take a
position, if I understood you correctly, that you didn't feel that from a
policy point of view this was that difficult to deal with if we're all
operating on the level.
BRYEN: I haven't drafted any language. I think I could very quickly. I
certainly could work with the committee staff to do that. The regulatory
language is really very simple. I mean, all you need to write on the rule is
an exception that says any
machine of less than eight processors is not controlled under this rule.
Period. It solves the problem...
ABERCROMBIE: Mr. Milhollin, do you think that that is at least a useful
exercise to undertake?
MILHOLLIN: I think that before we amend the rules to provide that kind of an
exception, we have to -- I think we should do the other thing that Dr. Bryen
recommends, which is to make a strategic assessment of the impact of such a
change.
ABERCROMBIE: Well, I'm all for that,
MILHOLLIN: But on -- on foreign recipients of our computer technology...
ABERCROMBIE: OK.
MILHOLLIN: ... I would have to sit down and be convinced that this change would
not put in the hands of people we don't want to have them high-speed machines.
ABERCROMBIE: Mr. Chairman, I suggest that those two points are something that
we could use in the immediate future the question of strategic assessment and
the question of whether or not having done that, whether or not there is
language that could be usefully be put forward
by this committee to help resolve at least this portion of the overall
question.
And I want to conclude, Mr. Chairman, by saying I see very close parallels to
the encryption question as well and the marketing of encryption.
CHAIRMAN: I thank the gentleman for his advice, and we will follow through.
ABERCROMBIE: Thank you.
CHAIRMAN: Let me say to both the witnesses and the two members that are left,
we have a vote in about 10, 15 minutes, so if we can be brief in our questions
and direct in our answers, we can maybe finish before we have to come back.
The lady from California, Ms. Tauscher is recognized.
TAUSCHER: Thank you, Mr. Chairman. I'm glad we've gotten to a place where we
can talk some resolutions as opposed to the kind of troublesome -- well I feel
when the nature of the relationship between the competitive forces and the
companies that predominantly work in my
district in the high-technology business, and their integrity and their
patriotism are questioned, and the ability to try to be competitive.
I think it's important that we discuss what Dr. Bryen has suggested. I have
been concerned for a very long time, as someone that came here from private
industry and represents the district where Silicon Valley sleeps and the home
of two national security labs, that MTOPS is something that no one ever looked
at as a measurement; that MTOPS was something that is esoteric and specifically
a government invention, and is not something that is nationally or
internationally known as the things that you might look at to decide if this
thing's going to hurt us or not.
So I do think it is important for us to look at the opportunity to have
definitional terms that people will understand -- that is not only
industry-based, but has a relevance in national security; that we understand
we're talking donuts and donuts and holes and
holes.
UNKNOWN: And apples and apples. Right.
TAUSCHER: So I would encourage, Mr. Chairman, our committee to look at that. I
would also encourage the opportunity for us to look for some very simple
language that will alleviate the redressing on a constant basis of the moving
target. We obviously have a nano-second life cycle in many of these processor
businesses, and I think that if we can come to some simple understanding over a
short band-width of time about what's good and what's bad, and at the same time
we have got to have some understanding, once and for all, which haven't
accomplished today, as to what is available worldwide.
You know, Japan has a ban against selling high-performance computers to two or
three countries and we know it's a farce. We know that many times, Japanese
companies assemble their computers in China. And you know, some of them are
high-performance computers. So are you meant to suggest to me that what
they're assembling in China isn't left on that shore, and that there isn't
availability there? What do we do about the distribution channels that are
becoming not a specialty, but a norm for the computer and the PC business, and
their worldwide distribution capabilities? How do you prevent a company from
saying, well, you know, I'm not in that business. I make them. I don't sell
them. And how do you control distribution?
Dr. Bryen?
BRYEN: I think what you say is very constructive. I'd like to comment for a
minute on Japan and also on the U.S. Japan has -- I was involved with Japan on
many levels for many years, including the famous case called the Toshiba case,
which I uncovered and prosecuted, I guess, is the right word. But I have a
high level of confidence that the Japanese government
will do the right thing if they understand what the right thing is.
TAUSCHER: Right.
BRYEN: And right now the problem is they don't understand what the right thing
is because we don't...
TAUSCHER: We don't either.
BRYEN: ... we don't seem to be able to articulate it, and we keep changing the
rules. We tore up the supercomputer agreement we had with Japan unilaterally
-- threw it in the trash, I might add. I've never seen anything more reckless
in my life and I don't understand why we did it. So we need to get global
cooperation. You can't do this by yourself. And the way you get it is to show
leadership and to show responsibility, and I really think that's very
important.
I'd like to also comment a little bit about Silicon Valley in California. The
pace of technological change and inventiveness is -- it's terribly important to
our economy. It's vital to our national security. And I think we all have to
take our
hats off and say that this is part of a new variable in our economy. We have
to understand that.
At the same time, California is one of those places that is under the most
direct threat from hostile missile attack, let's say from China. That's -- Mr.
Hunter knows that. We know that. And so we have to be careful here. And so
-- what I'm trying to suggest is that these supercomputer controls have a lot
to do with the ability of a country like China to attack us in the future. So
we've got to take this really seriously. We need that strategic study. We've
got to have it. And it has to be done right.
TAUSCHER: And we have to have a definitional agreement on what an apple looks
like; what a donut looks like.
BRYEN: Absolutely. That's what I'm suggesting.
TAUSCHER: And we have to, I believe, also be very aggressive about making sure
that on a timely basis, we here on the legislative side and the executive
branch and our security people and certainly the industry
come together, not episodically, but regularly so that we are dealing with the
vagaries of competition and the kinds of technology advances that happen quite
outside of our control, so that we can be appropriate about what we're doing.
BRYEN: And I -- you know, I'm a strict constructionist when it comes to export
controls. I'm well known for it. But I don't believe in cutting off your nose
to spite your face. And so I think we need to change the controls, and I've
said that. But another -- we need to tighten them up some, particularly in
areas like software that's being developed for nuclear weapons. It's very
sensitive and right now we don't have any export controls on it. It's all just
disappearing -- let's put it that way. Whoa -- we're getting ourselves in a
lot of
trouble. We're getting ourselves in a lot of trouble. We can avoid that.
And it doesn't have any economic impact. It doesn't have -- Apple could are
less, you know, or IBM or whatever. It's -- the issue is getting a responsible
export control program that the members of Congress can defend, which the
administration can defend, and which they can sell to our allies.
TAUSCHER: Yes.
BRYEN: And I think if we have something that's coherent to our allies, they'll
go along with it. They've done it before. They didn't like it, but they did
it. And I think we'll get the cooperation we need to have.
TAUSCHER: Mr. Hoydysh, I would assume that the business community would be very
interested in participating in a very aggressive...
HOYDYSH: Absolutely. I mean, we've made the offer and the offer's still on the
table. We will work with the administration, with the Congress, to develop
something that is responsive and that's rational and that works.
TAUSCHER: Thank you, Mr. Chairman.
CHAIRMAN: The gentleman from California, Mr. Hunter.
HUNTER: Thank you, Mr. Chairman -- appreciate it. I apologize to all of our
witnesses for being here -- only catching a part of the hearing. But Dr.
Bryen, we've worked a lot in the past. I want to applaud you for all of your
work as a guy whose worked in government, often against the tide, and against
the political tide to do what's right for the country.
Mr. Milhollin, I've always regarded you as kind of a national treasure. You've
-- you're a good honest broker who kind of holds our feet to the fire, and
thank you. You represent the best of what we have to offer in this country in
terms of an organization that comes in offers commentary and positions that
would have no economic interest behind them, but do have an important national
security
interest. Thanks for being with us.
And a couple of things have come out that I think need to be addressed. One is
the old argument of foreign availability, which has always been thrown up as a
justification for lowering the gates and letting the rest of the horses out of
the barn. Often we've found, whether you're talking about the French pipeline
-- of how the French were going to build the gas pipeline for the then-Soviet
Union and they could do it with all the technology that we had, and we ended up
discovering that in fact the protestations of foreign availability weren't as
accurate as they had been. And that's been kind of a historic thing. It's a
simple thing to say: Everybody else does it, therefore we should be able to do
it.
With respect to supercomputers, foreign manufacturers of supercomputers -- how
many of them have a policy of selling to Tier Three countries? I think that's
an important question. If you gentlemen could answer that one -- how many of
them have a
policy of selling to Tier Three countries?
Secondly, Dr. Bryen, it appears to me that we need to have a high-tech COCOM,
if you will -- another COCOM. This idea -- and I agree we have to show
leadership because if we don't show it, the game's over. This thing's going to
have a cascading effect and everybody will rush off to sell Winchesters to the
(inaudible) as fast as they can.
But how do we put that together? Do you think there's a need for that? How do
we put it together and what are reasonable parameters for that? How would it
differ from the old Cold War COCOM that we had?
So gentlemen, if you could -- and lastly, end-use on supercomputers -- during
one 12-month stretch of the 160-some supercomputers that went to China, they
allowed us to check end-use of precisely one. Now unless that changes, does it
make sense to continue to allow the continuation of those transactions, because
obviously that's not what this
Congress had in mind.
We want to check the end-use because end-use is so fragile to begin with -- the
idea that a nation that has an interest to use a supercomputer is going to
develop a nuclear weapon with it, will tell you that it's going to the weather
service. That's a fragile -- that's a fragile position to begin with. And the
idea then that we take their word for it, when they don't even allow us to go
to the weather service and see if it's there, enters the realm of a total
fiction.
So three questions. Thank you.
BRYEN: Want me to start?
HUNTER: Dr. Bryen, can you start?
BRYEN: Let me...
HUNTER: The first one was -- go ahead.
BRYEN: ... well I'm not sure whether I'm going to do this or...
(LAUGHTER)
HUNTER: Well, the first one -- the foreign availability -- how many countries
ship to -- foreign manufacturers of supercomputers now have a policy of
shipping to two or three nations.
BRYEN: I think -- of course this all depends on what you want to call a
supercomputer. If you're talking about power processing machines, which is
what I'm primarily concerned with in terms of countries like China, and China's
one of the biggest customers for these kind of machines these days, we ship
most of them. We ship most of them. There aren't that many countries that
make them, anyhow. There's some in Germany; there's a few in England, Japan.
But most of them are made here. So I think to the extent that the evidence is
publicly available, we're -- this is -- this whole thing is principally an
American problem, from just about any angle you look at it.
HUNTER: But I mean that goes right to foreign availability. Japan has a policy
now of not shipping supercomputers to China, does it not?
BRYEN: That's right.
HUNTER: And Mr.
Johnson did a study on that a little while back -- GAO did a study, I believe,
on foreign manufacturers so you may want to give us a...
JOHNSON: What we found about a year...
HUNTER: Excuse me.
JOHNSON: ... what we found about a year and a half ago is that only Germany had
sold high-performance computers to Tier Three countries. Japan, the UK --
neither one had sold to Tier Three countries. And Germany and Japan and the UK
all have export controls similar to what we have. But you're right, we do need
a COCOM arrangement for that.
BRYEN: Coming back to the COCOM arrangement, you could start by taking the
supercomputer agreement that was trashed by the administration and revise it
and expand it to the other countries that make such machines. There's only two
others. So three countries could join in an agreement and
it wouldn't be very taxing on the other countries because as Mr. Johnson has
said, they're not shipping them anyway for the most part.
So it's really an American problem. If we want to do this, we can do it. Yes,
we do need to have -- in the future we do need to have something like COCOM.
COCOM was very effective. I worked for many years. Dan worked for many years
on that as well. We were very familiar. And he used to work in the
administration before he went to industry and...
HUNTER: Mr. Hoydysh?
BRYEN: He has two hats, as I do, and we know it can work, and we know that we
can make -- I think that everyone who's analyzed the results of that process,
particularly with respect to the then-Soviet Union, and who's looked at the
technology base and what we achieved, it was quite successful.
We have to do it in a different dimension now, because the problem's different,
but I think your
suggestion is well taken and it's important to do it.
Now you made one other point that I want to come to is end use, an end user.
In China, literally all the supercomputers that are there from what we
understand are in networks.
There's about ten major networks that we know about in China. That means that a
machine at a Chinese toy factory, if it's a supercomputer, can be used by a
military research institute to carry out calculations on a new weapons system.
So when you check end use, it's unlike a machine tool, because with a machine
tool, we put it in a place and it cuts metal for a specific purpose and you can
look at the parts and you know what comes out the other end. But with a
computer in a network,
it could be almost anywhere. And so I'm not real convinced that end use is the
whole -- I mean, the Chinese have been very difficult about it, as is well
known, because they don't want us looking at their operations.
But I think you have to assume that any machine that you let go to a country
like China is going to be used for whatever national security priority the
Chinese may have, whatever defense priority they may have, including the
nuclear programs.
HUNTER: Do you think the benign use theory that we operate under is invalid?
BRYEN: Yes, sir. I think it's a mythology. And the whole differentiation
between military and civilian in a country like China is meaningless.
HUNTER: OK. Maybe Mr. Hoydysh could tell us whether or not he trusts them and
thinks everything's fine or maybe we should...
HOYDYSH: Let me answer some of your questions. But
first I'd like to say you had complimented Dr. Bryen for working against the
tide when he was in the administration. I'm here to tell you that Dr. Bryen
was the tide.
(LAUGHTER)
The rest of us were working against the tide. I guess I would like to clarify
just one thing. In fact, if you can leave here with one understanding, what we
are talking about, the reforms that we want implemented have nothing to do with
supercomputers. Supercomputers are a thing you talked about last week. We
took a flatbed truck to drive it from wherever it was to California into that
warehouse. That is a supercomputer. And I think Steve agrees with me.
Supercomputers have hundreds of processors. They are massively parallel
machines that are used for specific number crunching. These are not
supercomputers that we are talking about.
Now maybe we have a problem with how do you define that or whether we transfer
MTOPS into something else, but I think if you get beyond the differences of
definition and
maybe the misunderstandings, that there is absolutely no disagreement between
what Steve has said and between what I've been saying here this morning -- that
eight processors and below, however you measure them, are basically commercial,
mass market systems. Above that we're into a different realm and different
rules would have to apply, and we'd have to look at how they are actually used.
HUNTER: How do you define a supercomputer in terms of MTOPS?
HOYDYSH: I would define a supercomputer starting from the top down. I would
look at the machines, the 500 most powerful machines in the world. And we're
talking about MTOPS in the hundreds of thousands of range.
HUNTER: Well, that's how you would define it. If it's not hundreds of
thousands, then it's not a supercomputer.
HOYDYSH: If it's -- if you have to use MTOPS, I would say that the
supercomputers begin somewhere above 50,000, 100,000, something in that
category. There's no precise boundary
for it.
HUNTER: In that case, why is that the Japanese have been established -- the
Japanese have a policy of not selling computers of certain performance to
China? What is it that the Japanese feel it's not in their interest to sell to
China?
HOYDYSH: I don't know. I don't know as a fact.
HUNTER: (OFF-MIKE)
HOYDYSH: I would guess that it does not apply to the boxes that we are talking
about.
HUNTER: Let me tell you, Mr. Hoydysh, and I know I've got to go. But we've
heard what is known -- what I call the old gym socks argument, made on the
House floor a lot of times. It's always that all we want to do is let old gym
socks be sold to these foreign countries, and these dumb old congressmen won't
even let old gym socks be sold to these foreign countries. And it always ends
up that
it ain't old gym socks that they want, because they obviously can make them
themselves or get them lots of other places.
Inevitably, what they want is something just as is the same in all human
experience, that you can't get somewhere else. So, my impression of this thing
is that one thing that I've always felt to be a fairly solid proposition is,
you don't have to give people the absolute cream of the crop, top of the line
to help their military operations. We just made that mistake with Pakistan and
India. We've discovered now they've detonated mid-level nuclear weapons using
our published material, because we were so stupid, we never thought somebody
would want to make a crude nuclear weapon. We thought they'd only want to make
a top of the line nuclear weapon.
So, I think this is a fairly complex area with respect to whether or not the
system that you said,
hey, it's old gym socks, no big deal, can appreciably help the Chinese nuclear
weapons complex produce early on more effective killing capability on systems
that'll be aimed at American cities. I don't think you've done away with that
argument by showing us this nice simple device down here on the table.
I remain to be convinced.
HOYDYSH: I'm not suggesting that computers cannot be used for anything that you
want to use them. But your question is, at what level does it make sense to
devote government resources to chasing these things as opposed to concentrating
on things that are really important? If you have a million single-processor
PCs that are over 2,000 MTOPS, is it worthwhile to chase those, or are you
going to focus on the things that Steve is concerned about, the massively
powerful things, to either not
sell them at all, or to put such safeguards on them that they cannot be used
for anything other than what they're supposed to be used for? And that's the
question before us, I think.
HUNTER: Well, I agree with you. I want to hear your answer. Where you think we
should have limitations.
HOYDYSH: Well, we're here right now to suggest that the proposal to raise it to
6,500 MTOPS, which is the parameter that we're stuck with at this point, which
would free personal computers, it would free this kind of two- and
four-processor system, which is available worldwide in the millions of units,
we think that that limit is appropriate for today's technological and
competitive reality, and we're here to support that.
HUNTER: And does Japan transfer those?
HOYDYSH: I am absolutely that Japan will sell two-, four- and eight-processor
computers to China.
HUNTER: OK. I want to get -- and Mr. Chairman, I'll stop asking questions
here, but
I want to get the absolute fact on that; what level, and for the record or
whatever, what level Japan presently sells as a matter of policy to China, OK?
Thank you.
MILHOLLIN: I'd just make a couple of comments here. We also need I think a --
there's a question of perspective. The performance -- I'm sorry. The two or
three countries, which are the only ones affected by the NDAA, represent five
percent of the market for high- performance computers -- five percent. If all
of the supercomputer companies, all of our computer companies and all the
Japanese companies just didn't sell at all to that market, it's hard for me to
see how the destiny of any individual company would be greatly affected by
that. We're talking again about five percent of the market.
We're talking about -- the most recent figures I have from the Commerce
Department showing exports to
Tier Three countries of computers that were considered by the Commerce
Department to be high performance computers, which at that time I think was
over 2,000 -- we're looking at five percent of the market. And so, I can't see
why our companies are so desperate to crack that market if it means that the
cost will be that we'll be helping their military capability. It just seems to
me that the money isn't worth the risk.
It's also possible to think about changing the system so that for certain
countries we wouldn't recognize the difference between civilian and military.
And I think, Mr. Hunter, that's what you're suggesting. In the case of China,
perhaps there are countries where there should be an exception to the existing
differentiation in the regulations between civilian and military. But it's
whatever the level is for high performance computer exports for civilian and
for military, they ought to be the same for China, and
perhaps for other countries.
CHAIRMAN: Mr. Skelton? No more questions? Mr. Bateman? Mr. Hunter, do you
have any more questions?
HUNTER: Yes, just one follow-up, Mr. Milhollin. When you say five percent of
the sales go to Tier Three countries, Mr. Hoydysh is, as I understand his
position is that these, what he would consider be a low-level technology
computer. That's right, Mr. Hoydysh? Is a fairly substantial -- I take it the
reason we're here because that's a fairly substantial and promising market in
these Tier Three countries. If they're not going to buy any, this whole thing
is...
HOYDYSH: Congressman, for some of our companies, it's considered more than five
percent. It's up to 15 percent. It's a very rapidly- growing market.
HUNTER: Define what level computers those are that make the 15 percent.
HOYDYSH: Well, I'm
talking about overall sales of all of our IP products into the Tier Three
countries is on the order of, well, maybe from five for some companies to 15
percent. All I'm suggesting is that we cannot afford to forego five percent of
the market unless we have a compelling reason not to do so. I mean, any CEO
who would give up five percent of his revenues would be out in three hours from
his job. I mean, this is important -- revenue is important to keep our
companies healthy. And unless it can be shown that there is a compelling
reason why we should not compete in that market, we ought to be allowed to
sell.
HUNTER: I understand. And if every time we do a post-mortem of it, whether
it's Saddam Hussein or somebody else, we find out that American companies and
other Western companies have ended up creating part of a military apparatus
that
destroyed our soldiers, your neighbors, on the battlefield. And generally
speaking, when we err, we err on the side of having liberalized these exports
too much. And if the standard is always that you want to see a direct link
between a military operation and a technology, many times we've missed that.
The binders (ph) that we sold the Soviet Union made $12 million bucks on the
sale in 1972, ended up accurizing their SS-18 missiles, nuclear warheads aimed
at the United States. Now those were going to be used ostensibly for a
domestic project in the Soviet Union.
So, we make a lot of mistakes, but I think -- and that's why I think it's
important for us to not underestimate the value to a military operation of even
what I would call mid-level computer technology. We just made a massive
mistake with
respect to India and Pakistan because they used open American literature to
build nuclear weapons. We never dreamed they'd want to make a mid-grade
nuclear weapon, but they did, and now we may have to pay the consequences.
HOYDYSH: Well, Congressman, I agree with everything you say. All I can say is I
take great comfort in the fact that Dr. Bryen and I are on the same page at
least on this one. And the Commerce Department, I always said -- I always
thought, well, you know, we could push for anything we want, because he would
never let us sell what was truly strategic. And if he's on the same page with
eight processor systems, then I feel confident that we can go ahead and do it.
HUNTER: Dr. Bryen, are you heavily sedated today, or is that true?
(LAUGHTER)
BRYEN: I may need to be after this.
(LAUGHTER)
What I
said, Mr. Hunter, is that you can easily fix this problem. It won't satisfy
everybody in the computer industry, but it'll protect our national security.
And the way you do that is you distinguish between parallel processing machines
and these desktop machines, which are not parallel processing machines. And
you can simply say in the regulations that anything less than eight processors
is exempted from the controls, because first of all, they are commodities, and
they're hard to control. And you know, if you want to keep it out of India,
you won't succeed, or out of Pakistan or any of those places. It won't work.
And then let's get some tough controls on the supercomputers, because that's
where we're going to get hurt. And I don't think our controls right now are
tough. I think they stink, actually. I don't think they're doing the job, and
we need to control the software, and we need to get a national security impact
study that's real. I wouldn't go along with raising the threshold one point
without a national security impact study that looks at what our adversaries
might do.
So that's my position. And I think we can help out industry and solve a
national security problem at the same time, and I hope the committee will take
it up.
HUNTER: OK. Thank you.
CHAIRMAN: Gentlemen, thank you very much. There are a couple of questions from
staff and other members. If you would submit them for the record, we'd
appreciate it as rapidly as possible.
This committee will stand in recess, and we will reconvene at 2:00 with a third
panel. The meeting stands in recess.
(RECESS)
(JOINED IN PROGRESS)
CHAIRMAN: (OFF-MIKE)
REINSCH: Yes, sir. I'm aware of its difficulty. For example, a phone company,
we have to go
visit them again. If our investigators were permitted to exercise their
discretion, we could focus our efforts directly on any problematic locations we
might identify, adjusting the MTOPS level upwards for these inspections -- and
we appreciate the Congress' action in the defense authorization in getting that
-- will be helpful. But I must say, it will be temporary. It will only be
temporary relief, as chip speed increases and commodity-level PCs and laptops
grow increasingly powerful. This is going to become more important in the near
future as several manufacturers place tens of thousands of chips with speeds
greater than 2,000 MTOPS on the market, we expect to see single chips with
speeds greater than 3,000 MTOPS enter mass production before the end of the
calendar year.
In particular I would call your attention to the decision of at least one
company, one U.S.
company, not to market laptops and PCs with performance above 2,000 MTOPS in
Tier Three at this time because of NDAA requirements. We do not believe it is
possible for PCs and laptops sold in the thousands in retail stores to wait 10
days for interagency review. If these company sell these commodity computers
in Tier Three, the NDAA review system would be overwhelmed.
More to the point, we must be clear that the NDAA process does nothing to
prevent militaries in Tier Three countries from obtaining single-chip PCs and
laptops with performances above 2,000 if they want them, since these are
inexpensive, highly portable and widely available from retail outlets around
the world.
In conclusion, Mr. Chairman, computer export policy has broad ramifications for
the health of the entire American economy and for U.S. national security.
We're the world leader in
information technology and in computer manufacturing, and that industry's
future depends on exports. Ill-advised export controls could put this vital
sector at risk with little benefit to national security.
In the short term, U.S. companies would lose sale opportunities to foreign
firms. In the long term, U.S. companies would suffer significant and perhaps
irreversible loss of market share in foreign markets. The loss of export
revenue would adversely affect the ability of U.S. companies to fund R&D on next-generation technologies, thereby harming their ability to maintain
product lines on the cutting edge of technology, including products with
strategic military applications.
Thank you, Mr. Chairman.
CHAIRMAN: Thank you. I was just wondering, you were talking about the 2,500
notifications, 205 were converted to licenses and only nine were denied. Of
the nine that were denied, did anyone appeal that decision or
contest the fact that they were denied for good reason?
REINSCH: Looking at the list, no, sir. There were no appeals of those.
CHAIRMAN: And only nine were turned down, so of all the ones that were
requested, it seems like a good many of them -- would that indicate that the
process was working, or what's the problem?
REINSCH: Well, there are -- let me just give you two more numbers for the
record, just to flesh it out, Mr. Chairman. Nine were denied; 28 were
approved. That means the application, after it came back, after an objection
was approved; 139 were returned without action. Some of those may have
reappeared as separate, new applications. We have no way of tracking that.
Some of them disappeared. But when you keep in mind also that a lot of these
notifications are on spec if you will, and put in in anticipation of deals
being closed that ultimately don't close, the number of computers
actually shipped is not necessarily related to the number of notifications we
receive.
Because of the visit requirement, we actually keep two sets of data. We keep
data on notifications, obviously. In order to perform visits, of course, we
want to visit computers that are actually there, not mythical computers that
were not shipped. So we keep track of computers actually shipped. And that
number is probably between, depending on what time period you look at, between
one-third and one- fourth of the notifications. So a lot of these fall by the
wayside for commercial reasons that have nothing to do with us, and I think I
need to make that point.
In terms of what works and what doesn't work on the NDAA, the part that clearly
does not work the most is the visit part, and I can discuss that further if you
want. But you've asked about the notification report. I think at
low volumes, the notification process is functional. And we have been able to
-- the test for us is, you know, can we do it with existing resources and
manpower and maintain the ten-day limit that it provides for, or are we
beginning to, you know, fumble around at the edges, either in the sense that
we're taking longer or that we're starting to see sort of blanket rejections
without any review because people simply can't handle the volume.
At low volumes such as we've had until this summer, the system is manageable.
What happens when you start getting into the hundreds each week or more, which
is what's happening essentially since the end of September, late August, where
our curve has started to go up, which is exactly what we've all been
predicting, then the system becomes -- rapidly becomes unmanageable. Because
we have had at least one company voluntarily decide not to market in these
countries and therefore not give us any notifications, that's made the growth
curve somewhat slower than it otherwise would have been. But really what we're
worried about.
Now, if we go to 6,500, what will happen at that point is the system will ramp
down again, and the number of notifications will be much lower, because we'll
only be taking them above that level, and pressures will ease and the system
will be manageable again. If we have to wait till January to get there, I
foresee a lot of difficulties in November and December if you look at our
growth curve.
CHAIRMAN: I don't think anyone is probably advocating lifting alternate
controls, are they?
REINSCH: Well, we haven't advocated that either. I think what we feel is that
6,500 is an appropriate level to go to, given the state of technology now, and
I'd be happy to discuss that if members want to. We felt that it was
inappropriate, obviously, since we conducted our study and our review, and the
president made that decision and he was prepared to do it as of that point, if
Congress has given itself six months to conduct your study, we think -- we
would encourage you to do it sooner than that. And I would simply say in
response to your question, I think if we wait until January 23rd, which would
be the 180th day, if you will, for the time period to expire, I think the
system would be in serious trouble by that point, and we're rapidly getting
there now. But I could not tell you that it's, you know, that it was
nonfunctional from day one. It was not, and it worked effectively, and it does
work effectively at low volumes.
CHAIRMAN: Without this system that was set up, those nine that were denied
would have gone?
REINSCH: Well, I don't think so, Mr. Chairman. I mean, that -- to discuss that
would necessitate getting into specific end users, which the law prohibits me
from doing in public session, as you know. I gather one of your
private-sector witnesses this morning did not feel constrained by the law,
which I think is unfortunate, but I am. It's our belief that in those
particular cases, the end users -- I'm sorry, the exporter probably would have
applied for a license from day one because of the nature of the end user. But
in a sense, the notification process means they don't really have to make that
decision, because your process says you have to notify anyway, and that
effectively then puts on the government the burden of deciding when a license
is needed or not.
I think it is not really correct or fair to the companies to assume that in the
absence of your process they would never file any license applications. We did
have cases that you're well aware of in Russia where we thought they should
have and they didn't, you know, so I can't tell you that the companies are 100
percent, you know, pressing us on this, but neither I
think is it fair to assume that their compliance would be zero if they didn't
have the notification process.
CHAIRMAN: I understand. Mr. Skelton?
SKELTON: Let me ask you a similar question to the panel this morning. In your
opinion, how much time is required for Congress to conduct a good review?
REINSCH: Well, I speak from the perspective of someone was up here for 20 years
in various capacities doing this kind of thing. My personal view is that given
what I think what you ought to do as part of that, I think it can be done in 30
days. Now, I think, you know, if the committee in its wisdom wants...
SKELTON: What happens if the clock starts ticking the day before Thanksgiving?
We're not around.
REINSCH: Well, I think there's two -- that's a two-part question, Mr. Skelton.
SKELTON: Just answer that
question. Just that question.
REINSCH: The staff is around, and I think the people that probably would
conduct your study and review are around and able to do it. The question is,
if your conclusion was that we made a mistake, will you be in a position to do
anything about it 30 days after the day after Thanksgiving? And the answer to
that is probably no. You wouldn't be around to do anything about it.
SKELTON: Well, 30 days would be adequate.
REINSCH: Thirty days would not be adequate for you to take an action. I do
think that 30 days would be adequate for your committee staff and members that
were accessible even come to the conclusion...
SKELTON: Mr. Reinsch, some of them do take off for Thanksgiving and Christmas
and New Year's as well.
REINSCH: Well, I'm delighted to hear that. I must say, I'm not sure because we
(inaudible).
SKELTON: Well, I do my
best to keep them here, but some of them just insist.
(LAUGHTER)
Let me ask you this. There was some testimony this morning about the term and
measure of MTOPS not being the best way to measure. Could you address that so a
country lawyer can understand it, please?
REINSCH: Yes, sir. I'm aware of this morning's discussion. I had several
people alert me to this. We have -- each time we've reviewed this we had
looked at other ways of doing it. And I've said before, publicly and
privately, that we're not wedded to MTOPS as an answer. MTOPS is an artificial
construct that is not used in marketing these things, and the industry doesn't
particularly like it.
SKELTON: You're right. Do you have a better idea?
REINSCH: Not yet. I know the idea that was suggested this morning was to go by
the number of processors.
SKELTON: That's correct.
REINSCH: And let me say two
things about that. One, it's a very interesting idea. We considered it
before. We did not exactly reject it. In fact what we did was embody that
thought in MTOPS' formula: 12,300 is eight times one Pentium III. That's how
we got to 12,300, by making the same assumption that I understand Mr. Bryen
made this morning, that eight processor and below models were not -- could not
effectively be controlled.
So we -- what we simply did is rather than say eight processor, is we said,
well, what's eight times the current, you know, performance capability, and
we'll pick that MTOPS number.
I would just observe...
(AUDIO GAP)
... by processors rather than MTOPS, you start out at the same place that we
are, but you end up with a significantly greater liberalization, because eight
processors with a Pentium III comes out to be, you know, 12,300 MTOPS.
Eight processors with the MERCED (ph) -- now called the Itanium with -- Intel's coming out with next summer --
will be over 40,000. But it's still eight processors. So if you are prepared
to accept that kind of ramp-up in capability, even though the number of
processors hasn't changed, controlling by processors is -- is I think an
equally effective way to go.
And I think it's a fair point to say that if your assumption is that you can --
you can do eight processors without substantial vendor support, they're widely
available, they meet all those criteria, that's going to be true whether it's
an Itanium or a Pentium or whatever it is.
But, you know, the committee should understand that that -- that embodies
within it a substantial increase in capability that might or might not be where
this administration will end up.
SKELTON: Thank you so much.
CHAIRMAN: Mr. Hansen. Mr.
Hansen's not here?
Mr. Hayes.
HAYES: Thank you, sir.
REINSCH: I'm sorry, I didn't see you over there.
HAYES: Let me preface my question by saying my technological capabilities are
better suited for this than some of the equivalents you're talking about.
For my edification, based on the retail marketplace, what can be purchased by
anyone, anywhere, over the counter? What is the ceiling that presently exists
-- the ceiling under which people that we do not want to give aid and comfort
to of any kind of military hardware or software, what is the ceiling out there
now that they can go to the marketplace and purchase equipment that we're
talking about.
REINSCH: Well, for -- as a general matter we concluded that that level is
really reached at 6,500 MTOPS, or to use the concept that I was just talking
about with Mr. Skelton, you know, that comes with that level, is really reached
at eight processor
models.
Now, if you're an Iraqi, you know, or an Iranian, that number is probably lower
because particularly in the case of Iraq, you're subject to a UN embargo and no
one's supposed to trade with you. And so that -- there are special
circumstances like that.
Saying that an item is uncontrollable at 6,500 MTOPS -- which is what we say --
is not the same thing as saying every single individual in the world can walk
down the street and buy one. You know, it's logistically more complicated than
that.
But what it does mean is that people -- and particularly governments that are
determined to get them -- can do so relatively easily at that level.
HAYES: That wasn't quite as clear as I hoped it would come out.
The gentlemen who have spoken already, and every one in the room and those not
present share the same belief that we should do nothing to help our enemies or
potential enemies around the globe. So
I think unanimity there is concrete. There's no question about it.
In another presentation -- again talking about computers and software -- some
private providers outside of the defense context were talking about similar
terms, and I got the impression that you could buy in personal computers retail
market 100,000 -- 200,000 MTOP capability equipment without any real
difficulty. Am I mistaken? Did I miss something there?
REINSCH: Well, we wouldn't agree with that. I'd like to have a conversation
with those people and learn a little bit more about what they are talking
about. I don't think -- certainly that wouldn't be true for a personal
computer. We are at the point right now where PCs are getting into the 2,000,
3,000 range. And the items that the notification process really works for are
-- historically have not been PCs, although that's beginning, but servers and
workstations -- the server
being the kind of thing that runs your office e-mail system, that would be a
server; or a workstation on which you would do computer-aided design and things
like that for designing anything from a house to, you know, a military item.
Those -- it's the workstations and the servers that are really caught in the,
you know, more than 2,000 and less than, you know, less than 10,000 or 20,000
market. And those are individual items, but they're not the kind of thing that
you would just walk out to a store and buy. It's a little more complicated
than that.
And at the 100,000 to 200,000 level, I would dispute anybody who said that it's
that simple.
HAYES: OK, let's go down another road. You were referring to the screening
process -- and I don't think anyone has any problem with the screening process.
We all think it's wise to do that. But you said under the present system,
2,500 screenings
resulted in nine rejections.
It would appear that this present screening process is not operating at maximum
efficiency. People are possibly -- and again correct me if I'm wrong -- tied
up examining things that they really don't need to examine. Twenty-five
hundred versus nine -- what -- in an ideal world, what should that ratio be to
show that we are examining the right kinds of products -- if that's a
legitimate question.
REINSCH: That's a very interesting question, and we haven't really asked it
ourselves -- asked it of ourselves, Mr. Hayes, because the law doesn't give us
that flexibility.
You use it for a point if all those, that's the number we're coming out, you
know, what's of it. I can say that, to put it in a slightly different
perspective though, the number that were objected to, if you will, by a single
agency, was higher than the nine. The number that was objected to by somebody
were 189
out of that 2,500. Some of them are still pending, so that -- 189 were objected
to. Now what happens when there's an objection is that means that the
applicant has to come back and submit a license application. A lot of them
don't do that -- a lot of them just don't do that. And it sort of vanishes
into the ozone.
Of the ones that did, nine were denied, which is a very small number.
HAYES: One more real quick question, Mr. Chairman -- these answers are more
complex than my questions. Am I hearing you say that 6,500 is a very
reasonable, safe, common sense number based on today's marketplace and national
security?
REINSCH: Yes.
HAYES: Thank you.
REINSCH: How's that for a shorter answer?
CHAIRMAN: Mr. Smith.
SMITH: Thank you. First of all, I want to thank the administration for what
they've done. I happen to think you are moving in the right direction in
making
sense out of a policy that was really outdated. And I think updating it to
match the technology is the correct way to go. I may be in the minority on
that viewpoint on this committee, but I do agree with it.
Also I have some questions on the 30-day versus six-month notification process.
As I understand it, that clock has already started ticking when the decision
was made. I think you said it runs out on January 24th, is that correct?
REINSCH: Twenty-third, yes.
SMITH: Twenty-third. So we're currently in that six-month notification
process. In fact, one of the things that I want to have submitted for the
record is a copy of maybe the amendment that the chairman and the ranking
member got passed back in -- for the FY '98 budget requiring the review of
these exports, which requires a
report.
That report has already been submitted. It was mentioned this morning by one
of the people who testified that this hadn't even been studied yet. That's not
true. The administration has sent us over and I have attached a letter with a
copy of that report for everybody to have a look at.
So in essence, the process has already started. I guess maybe I'm missing
something. But if we were to come in now and say that it ought to be 30 days
-- actually I should ask permission of the chair to submit that for the record,
sorry. Bad timing on my part. I'll get back...
REINSCH: I think he'll probably say yes.
SMITH: OK. Never mind.
My question is: If we came in right now and said, OK, it shouldn't be six
months, it should be 30 days, arguably hasn't that
already passed? Or do they have to start from when we would say it would be 30
days?
REINSCH: Well, since we've asked for an amendment, I mean, you could write it
either way. My view would be, Mr. Smith, that it's already passed. We're in
the 94th day -- or whatever day we're on, it's about the 94th day -- and I
would say if you change to 30 days, time's up.
SMITH: Yes. And this -- and I'm sorry Mr. Skelton had to run to another
meeting. I'll talk to him and his staff about it. But I think, you know, the
Thanksgiving, Christmas issue really isn't an issue for us, because -- I mean,
I would hope since this process has already started, that there are people on
the national security committees who are reviewing the report that has been
submitted and are already moving forward to try to determine whether or not
your decision is wise or not.
REINSCH: Well, if I may, I'm glad you brought that up again, because I wasn't
-- I wasn't quick enough to make the more important point to Mr. Skelton, and
perhaps if you can relay it I would appreciate it.
What the president said in this regard was that we have committed to review
this issue every six months. Now what that means is July and January. And we
have now put ourselves on the schedule that we are determined to meet, which
would preclude the day after Thanksgiving being a problem. We expect these
things -- if we have a change to make, we expect to make it in January and we
expect to make it in July. July, you may run afoul of the August recess, I
suppose, but I think that's a more flexible issue than the question of end of
session. So I don't think that would be a problem.
SMITH: And I think it's fairly
important that members know that because I think you're right. If we don't
update this -- and this is something I've been working on.
And, Mr. Chairman, I'd ask -- I've got a letter here with a copy of the
administration report -- if I could get that submitted for the record, with
your permission. Thank you.
I think it's really important that we do that because the backlog that comes in
November and December is going to be a major problem, and I hope we step up to
that before we go home, hopefully in the next week or so.
In other -- not really so much a question, just I unfortunately had meetings
this morning and didn't have an opportunity to make a statement or ask
questions of the earlier panel -- but I really think this policy makes sense.
And it disturbs me a great deal that a lot of members on this committee don't.
I frankly don't
understand their argument.
I understand the first part of it, and the first part of it is:
high-performance computers -- or, hell, just computers -- can have national
security implications. No question. As you mentioned in your testimony,
computers with as little as 500 MTOPS have helps us develop some of our most
advanced weapons systems. The F-117 -- if my reports are correct -- was
developed with computers with 100 MTOPS. You know, if we're talking 2,000,
obviously there's defense implications. There's no question.
But the other part about it is, because of where the technology is going, and
maybe the industry hasn't gotten there as fast as they thought they would, but
they're getting there pretty fast. We now have a situation where, you know,
it's not a server or a workstation. You can buy a PC or laptop with over 2,000
MTOPS capability. You can buy it retail, as somebody who testified this
morning, you can buy it from six or seven different companies outside of the
United States of America, over the Internet with 24 hours delivery.
And one quick point about Iraq -- you mentioned that Iraq can't do that because
they have sanctions against them. Theoretically Iraq can't buy a spiral
notebook from people because of those same sanctions. So it's not peculiar to
computers. I mean, we have sanctions against them, period.
REINSCH: Correct.
SMITH: So it's there, and the computers are out there and available just
retail. And you know, if you walk into store and want to buy a computer over
the counter and have to go through the Department of Commerce or Department of
Defense doesn't make any sense whatsoever.
So they're out there. I guess I would ask -- the people who aren't here --
where do you disagree with that? I mean, are these computers not available or
what? They're available. They're out there.
And I guess the final
point -- which is the one we never get to because we get stuck on the fact of
"how dare you export something that could have national security implications" -- is the point that you made very, very well in your testimony, that us --
being the United States of America -- maintaining the leadership in this
technology area is critical, not to our commercial industry, not just to our
commercial industry, but to our national security. If we cease to be the
leaders in technology and somebody else moves past us, it is precisely our
national security that will suffer because of the working relationship that the
Defense Department has with U.S. companies that they don't have with German or
Japanese or Chinese companies. So it is a national security issue.
We seem to be thinking that as long as we err on the side of not selling this
stuff, we have then erred on the side of national security. That is dead
wrong. We are erring in a direction that because of the changing world is
going do the most to jeopardize national
security.
So I applaud the administration moving in the correct direction, and I for one
would like to work with you precisely for national security reasons.
Thank you, Mr. Chairman.
CHAIRMAN: Thank you. Mr. Jones.
JONES: Thank you, Mr. Chairman. And for somebody who has less than a minimal
amount of knowledge about this issue, I'm totally fascinated with it because, I
mean, it's a fairly interesting area of national security. And I want to make
sure that I understand what you said in response to the last short
question/short answer of Mr. Hayes.
But basically, are you saying that any computers who have 6,500 or less MTOPS,
basically there's no national security problem and they should be available to
anybody?
REINSCH: The best answer to that I think was Deputy Secretary Hamre's when he
was asked that question, which is: We can't say that there's no national
security
problem, because they have military applications below that level. What we can
say is that at that level and below, we can manage the national security
problems that might occur.
JONES: One thing that's always bothered me about this, particularly in lieu of
a discussion last spring, I guess, I understand there are national security
implications and obviously we've got to be concerned about that. Mr. Smith is
absolutely right.
But by the same token, when we turn around and we put our business hats on, are
we being real presumptuous that we're the only people in the world that
manufacture those computers, or are they available out there to every one of
our -- two or three countries -- from other sources? What's your answer to
that? And how do we deal with that?
REINSCH: Well, were the only ones, and we could deal with it. And 30 years ago
on a host of technologies, we were the only ones. In this case -- as I think
my testimony
indicated -- of the 170 companies, we found that account for three-quarters of
the world's supply of computers, 120 of them are outside the United States; 28
of them are in countries that don't adhere to the multilateral control regimes
that control -- or the regime that controls these items.
So we're talking about a substantial number of producers who don't have
multilateral obligations, don't necessarily -- and in fact don't empirically --
do what we want them to do, as far as sales are concerned.
I can give you perhaps one of the most interesting examples. And, you know, a
nice visual would be to bring a laptop here and, you know, call down an
Internet web page. But there's a company in Taiwan, a Taiwanese computer
company, which is not only marketing for a processor Pentium III computers --
which means computers that are under 6,500, but well over 2,000 -- they not
only are marketing them,
but they are selling them in manufacturing facilities in China.
So this is what is happening with this technology. It is not -- the technology
is not ours alone. Where we are relatively unassailable -- except for the
Japanese -- are with the economics. We can make these things better, faster
and cheaper than anybody else. And as result, we dominate the world and I hope
and think make a lot of money.
People don't try to make the high end to compete against us because they can't
do it as cheaply. If we drop out of the market, or if we take ourselves out of
the market with export controls, the Koreans, Italians, Germans, a number of
others will step in and they will step in and make the products that they don't
make now because without us in the marketplace, they can make -- they can make
those products and they can make money out of those products.
If you ask me, frankly, what the biggest challenge we have with China is --
which has been a concern of this committee, I
know, and a well-taken one, a concern for all of us -- in the computer area,
it's not an export at all. It's indigenous Chinese production. You know, more
than half of the PC market in China now comes from indigenous Chinese
companies. And the American companies -- I don't think anybody asked them that
this morning -- but the American companies are losing market share in China at
the low end, the PC end.
Now if you look at any other industry, if you look at textiles, if you look at
steel, if you look at the history of what happens in developing countries is
they move up the value-added chain. They start out with the cheap stuff and
they start making more expensive stuff. That's going to happen in computers in
China, just like it's happened in
all the other things they make. And they've taken over the low end in their
own country. Give them a year or two, they'll be selling workstations and
servers of their own. And that's what we ought to be looking at.
JONES: Is there any control in place that would be able to track or if needed
prohibit the sale of a supercomputer over the counter -- whatever the maximum
MTOP you can buy over the counter -- to somebody who then takes it to a
friendly country, and it goes from that country to one of our Tier Three
countries? Is there anything to control or track that in place?
REINSCH: Well, yes, our rules will prohibit that. Our rules would provide that
if a computer were exported legally to the first place, if they wanted to
re-export it to, you know, an unacceptable location, it would be subject to the
same requirements as if it were going directly from here. That's what our
rules say. I mean, the reality, as you can imagine, is that when
you're in the third country, you know, the items there -- we are here. We
don't always find about those things.
We get -- up to a point, we get good cooperation in some cases from European
friends and allies, and we get very good cooperation on what we refer to as
"the terrorist states" -- Iran, Iraq, Libya, North Korea, Sudan, Syria, Cuba -- countries in that
category.
We get much less cooperation on China because they don't regard China as a
proliferation threat in the same way that we do, and that's a significant
difference between us and our allies.
JONES: And I don't guess there's any practical way to control the sale of
over-the-counter to somebody who exports it to a friendly country and they
maintain a terroristic operation with the use of that computer in that friendly
country, and I know we're bound to have some of that out there.
REINSCH: Well, we have a -- we have -- we have a
rule that was promulgated in the Bush administration, which is no one has
absolutely enhanced proliferation control initiative, and it contains what's
called a catch-all provision. And we have prosecuted people under this
provision. This is not by any means a dead letter.
What it says is that if you know or have reason to know that your customer,
your end-user, is engaged in proliferation activities, you have an obligation
to come in and seek a license, which we can then say yes or no to. And
companies do that. This is not something that is honored in absentia.
And as I said, we prosecuted people here on this successfully. So it works. Is
it perfect? I doubt it. I mean, that's asking me to sort establish the
negative, you know: I don't know what I don't know.
But it works, and our enforcement people -- Congress has been good to us, not
as good as I'd like,
but good to us in giving us additional resources to beef up enforcement --
spend a lot of time figuring out who questionable end-user's arm and tracking
this sort of thing down.
JONES: Thank you.
CHAIRMAN: Thank you.
We are here, of course, looking at this whole matter to see where we go from
here. A lot of thought there was justification for the law that was passed
back then. And in that law, provision was made for just what is happening
today. In addition, notification, licensing and all the rest of those kind of
things, and provision was made for the president to be able to increase the
threshold and we would have a chance to come back and disagree with that if we
wanted to. And so we understood that there would be these kind of situations
arising.
Mr. Skelton and I have sent a letter to the president, as you know, asking
about 12 questions, which we hoped he could get to us by the first of November.
And I understand you're
probably the ones who are getting all these answers up for the president.
REINSCH: Most of them. My people are hard at work, even as we speak.
CHAIRMAN: And I was just wondering how far along you are on that.
REINSCH: I hope we meet your deadline, Mr. Chairman. I've been out of town so
I can't say for certain. We got the questions. We're working on them. Most
of the ones that ask for data are ours. Some of them, of course, are policy
questions and that answer will come from the White House.
CHAIRMAN: I understand. The deadline of January 23rd, I guess...
REINSCH: Yes, sir.
CHAIRMAN: ... means that we're going to be out of session until about that time
after we leave here, and, so the (inaudible). I doubt if he'll have a chance
to respond one way or the other. So things are going to go on from there.
But the bottom line of it is that we are trying to see if sometime what can we
achieve if we leave to make, if any, and I don't know that anyone has said that
we should just eliminate the law; go back to how it was before when no
controls, set up Tier One and Tier Two and Tier Three countries, and Two or
Three countries (inaudible) MTOPS the threshold.
And of course, that is being increased because of the president's actions, he's
entitled to do.
What is the -- what would you recommend at this point that we do to change the
law, modify it, or you said you need more people to handle the applications.
Is that part of more people? Or do you need more time? (OFF-MIKE)
REINSCH: Well, let me say -- you're obviously making a good faith effort to
grapple with a hard question, Mr. Chairman, and I appreciate that. I think
it's appropriate now a couple of years down the road to, you know, to do that
and look
at it. And obviously it would be simple for me to say it would be -- you know,
we'd like it to go away. But I think that, you know, I'm happy to make a good
faith effort to work with you to take a look at this.
Clearly, I think that on the notification side, the process works as long as we
are adjusting the levels appropriately upwards, that allows us to stay sort of
even with the flow of notifications coming in. If we get behind, then we get
swamped and there's a lot of people disadvantaged. So as long as we can
maintain equilibrium through up- ticking the standards, I think that's all
right. I think where we are right now is that the six-month waiting period
threatens to disrupt that equilibrium and that's really why we have recommended
a shorter period.
Now I think it would be fair for you to come back and say, well, if that's so,
why didn't you have
brains enough to do this in March? You know, so the six-month period would be
up now. And that's a fair point and we didn't -- it takes a while for us to do
this sort of thing, too, and we wanted to gather all of our data.
I think we're a little bit behind the curve in terms of us having the
marketplace. And there's only two ways to I think get up -- catch up. One is
if you shorten the waiting period and make things a little bit simpler, or we
effectively would be forced to take bigger, bigger bites, bigger leaps in
January, because if we do in January isn't going to be effective until July, we
have to anticipate that and take a bigger bite, and in July, take another
bigger bite. Whereas if we thought it was going to be effective sooner, you
know, we might come to a different conclusion. So we might
just think about that piece of it as we work through this.
I think the other (inaudible) might consider from a drafting perspective is
that the way the law is drafted, you have given yourself six months to review
the decision. And presumably, the implication of course is that that gives you
time if you don't like it to take appropriate action.
What you have not given yourselves is the opportunity to say positively that
you do approve of what we've done and that you want to terminate the waiting
period. There's no mechanism for doing that. And you know, there are things
that led the president to suggest a shortening of the waiting period is that we
don't have another device to, you know, deal with a situation in which you
might be -- you might be if you did the right thing. And if you think we did
the right thing and we think we did the right thing, why do we have to wait for
six months? -- you know, which I think is a fair point. But the law doesn't
come with flexibility there.
And then finally, as I've said before, and I
want to tell you a little story, if I may. I think the visit issue has been
the most problematic for us, mainly from a resource point of view. But if you
think about it, it's actually in some respects I think pushed us further from
the goal. The Chinese have -- and we'll talk about China specifically -- have
actually increased their cooperation with us significantly in this area
recently. And we've done quite a number of visits. In fact, we have someone
out in a different part of China this week doing an additional number, and I'd
be happy to provide you with a -- well, we're going to give you a report next
month, I think, and that will be complete numbers. But you won't see one in
the China column. You'll see a bigger number.
But in the
process we've discovered something that we hadn't expected, which is now we
have to ask them for every one -- to visit every one. Now you know and I know
and they know that we don't have enough people or money to visit every one, and
they don't have enough people and money to visit every one either. So by
having to ask them to visit every one, what we've effectively done is put the
decision on which ones get visited in their hands, not in our hands, because
they pick and choose. And they'll say: Well, we'll do these 10, you know, this
week, and two weeks from now we'll do these five over here. And they'll say
you can't do the rest because we don't have enough time and money. And we
don't know. I mean, we can't say that. And so we chalk up 15, if that's the
number, and say we've done 15.
If we could pick and choose ourselves, you know, my people are professionals at
this. They
can pick out the ones that are iffy -- the ones that are problematic are the
ones we like to look at. And we could ask the Chinese to do those. And then
if they said no to those, you know, then we'd know something, you know. We'd
know they were hiding something and we could do something about that.
But in the current situation, we don't know anything when they say no, because
we've made an unreasonable request from their point of view and we know they're
going to say no to some number of them. And what we've essentially ceded to
them is the opportunity to make the choice of which ones we see and which ones
we don't see. And our investigators don't get to make that choice anymore.
And I would assume, not having seen the ones we visited, I assume we're
visiting all the ones that check out. And these all -- and in fact, they all
have checked out in China, and we've done a lot more.
So we all want to think about that -- that this is one of those cases where
I think I would encourage you to have confidence in our professional
investigators who are, you know, they all go down to the FBI place in Georgia
to get training. These are professional people. They've got the same kind of
training as other law enforcement officers. They know what they're doing. And
I urge you to have confidence in their ability to select the ones that are
problematical and not force us to spend -- not only spend a lot of time and
effort doing ones that don't matter, but really to give effectively either to
the Chinese the opportunity to make the choice.
So we urge you for flexibility there. You already gave us some by letting us
up-tick the standard, and we really appreciate that.
CHAIRMAN: (inaudible) also aware that two of the (inaudible) are being
investigated. Are those in China?
REINSCH: No, they are not. And I'd be pleased to tell you where they are
privately, Mr. Chairman, but I'm not
going to do that publicly.
CHAIRMAN: I understand. (inaudible) of advocating that we go back and abolish
the law completely, are you?
REINSCH: Well, I have in the past, Mr. Chairman, and...
CHAIRMAN: (inaudible)
REINSCH: ... in the interests of comity because you've been so nice to me
today, I won't say that again today.
CHAIRMAN: You mean you were in favor of shipping to two or three countries
where it (inaudible) interest in national security.
REINSCH: I'm in favor of working with you to develop a law that will allow us
to do our job better and make you feel that our national security is fully
protected.
CHAIRMAN: Well, I appreciate all you're doing, the effort you're spending, and
also the work you're doing in getting these questions answered for us because
we'll come back with them later on, I guess, after we come back from the break.
REINSCH: We'll get them to you as
soon as we can, Mr. Chairman.
CHAIRMAN: I appreciate that. And you've been a great help to us, and the work
of our committee and I appreciate it. I can't say that we're all going to
agree, as you well know from the past, but we're trying anyway, and this whole
law is an effort to do something that we thought needed to be done. And as I
said, we provided what is in the law to have these things happen; maybe we do
need to tighten a few screws and unloose a few screws, or do something to make
it more acceptable to people in facing the realities of life that's brought us
to where we are today.
But I appreciate what you're doing to help us and we'll be back in touch, we
say around here. Thank you.
REINSCH: Thank you.
CHAIRMAN: Any other questions? The meeting will be adjourned.
END
NOTES:
Unknown - Indicates speaker unknown.
Inaudible - Could not make out what was being said.
off mike - Indicates could not make out what was being said.
PERSON:
FLOYD SPENCE (94%); BOB STUMP (57%); DUNCAN L HUNTER (57%); JOHN R KASICH (57%); HERBERT H BATEMAN (56%); JAMES V HANSEN (56%); CURT WELDON (55%); JOEL HEFLEY (55%); CHRIS JOHN (54%); LEE TERRY (53%); ROSCOE G BARTLETT (53%); JOHN N HOSTETTLER (51%); JOE SCARBOROUGH (50%); SAXBY CHAMBLISS (50%); VAN HILLEARY (50%);
LOAD-DATE: November 1, 1999