The Proliferation Primer
International Security, Proliferation, and Federal Services Subcommittee
United States Senate Committee on Governmental Affairs
A Majority Report - January 1998
This chapter examines policies of the Clinton Administration that have led to the export to proliferators of high performance computers and their possible effect on American national security.
During the Cold War the wisdom of not exporting militarily useful goods to America's enemies was commonly accepted in the United States. This policy extended not just to technologies and systems whose sole application is military -- America's long-range ballistic missile manufacturers, for example, didn't try to sell these weapon platforms to the USSR -- but also to the sale of "dual-use" goods, technologies having both military and civilian applications.
Notwithstanding the natural tension in any free society between trade and export controls, the western export control regime COCOM -- the "Coordinating Committee for Multilateral Export Controls" -- was a success. At best it denied, and at least made more complicated and difficult, the transfer of significant dual-use technologies to the Soviet Union and its allies. Founded in 1949, COCOM, which consisted of Japan, Australia, and all the NATO countries except Iceland, acted on the basis of consensus among its members.
President Clinton won election in 1992 promising America's manufacturers he would make dual-use exporting easier. The Clinton campaign's policy paper -- Technology: The Engine of Economic Growth 1 -- "drafted by Apple [Computer] executives and others [in California's Silicon Valley]" 2 said, "[ e] xport controls are necessary to protect U. S. national security interests and prevent the proliferation of nuclear, biological and chemical weapons. Nonetheless, these controls are often overly restrictive and bureaucratic, creating a mountain of red tape and costing the U. S. tens of billions of dollars in exports -- while undermining the competitiveness of the hightech industries on which our national security depends." 3 It also said the United States should "[ f ]urther liberalize East-West export controls that are unnecessary given the end of the Cold War." 4 Less than one year after entering office, President Clinton assured the Chairman and CEO of computer manufacturer Silicon Graphics, Edward McCracken, he was "... currently engaged in seeking major reforms to COCOM, which should lead to significant liberalization of [export] controls on computers, telecommunications, and machine tools." 5
President Clinton's "major reform" to COCOM was its dissolution. COCOM's consensus approach, which gave the United States a veto over proposed dual-use exports of other nations, was apparently viewed as an "unnecessary" export control. This veto authority required COCOM member nations to harmonize controls on sensitive technologies and eliminated the competitive business pressures on countries to "interpret" COCOM's requirements in some convenient way to give their domestic industries an advantage over those of other nations. COCOM died on March 31, 1994, over two years before the establishment, on July 12, 1996, of its successor, the Wassenaar Arrangement (short for the "Wassenaar Arrangement on Export Controls for Conventional Arms and Dual-Use Goods and Technologies"). Unlike its predecessor, Wassenaar allows each member to determine for itself whether to allow an export to proceed. No member can veto another's exports. Where COCOM consisted of consensus before the fact, Wassenaar consists of reports after the fact.
The 1991 Gulf War ingrained the importance of advanced technology in the minds of civilian and military leaders the world over. Just as the United States intends to use supercomputers to "maintain and extend" its military technological lead, other nations are interested in using this key enabling technology to design and acquire weapons and platforms more advanced than currently in their inventories. As Richard Bernstein and Ross Munro note in The Coming Conflict With China, "China's overall economic strategy is also aimed at enhancing the acquisition of the most advanced Western technology, including 'dual use' technology that can be used for both civilian and military purposes." 8
Of course, as Deputy Assistant Secretary of Defense Mitchel Wallerstein pointed out in Senate testimony, "... the original designs for the first U. S. nuclear weapons were done on slide rules or on very primitive calculating machines." 9 Weapons can and have been developed without supercomputers, but according to Dr. Peter Leitner, a strategic trade advisor in the Pentagon's Defense Technology Security Administration,
The relationship of computers and advanced machine tools to the proliferation problem is often posed in simplistic terms: Since the U. S. did not need computers or computer-controlled machine tools to develop nuclear weapons and ballistic missiles, there is little need to control either technology for these purposes (emphasis in the original). The argument ignores the fact that computers and computer-controlled machine tools have become an essential tool for many activities that were previously accomplished either by secretly amassing dozens of Nobel laureates, supported by hundreds of top physicists, in the mountains of New Mexico for several years or by metalworking artisans fashioning unique parts for small lot production. Computers and computer-controlled machine tools have made themselves central by defining the very way technical goals are accomplished, and can substantially enhance the effectiveness of the limited pool of talent often available to a proliferant country while providing the capability for mass production of highly effective weapons systems. 10
Today's high performance computers can reduce a weapon system's development time, make it more capable, and increase the user's confidence in it.
Confidence is critical for nuclear weapons, which are rarely, if ever, tested. The report used by the administration as its basis for liberalizing U. S. export controls on high performance computers, commonly referred to as the "Goodman Report," says, "... continued export controls will slow the exacerbation of existing nuclear threats. Control of HPC exports, by limiting those exports or imposing appropriate safeguards, to countries known to possess nuclear weapons will impede their development of improved weapons and reduce their confidence in their existing stockpile by limiting the opportunity to conduct simulations in lieu of live tests. Similar or more rigorous controls on HPC exports to countries with nuclear weapons development programs could impede their development of second-generation weapons." 11 This finding of the Goodman Report was no surprise: It followed a 1986 report entitled, "The Need for Supercomputers in Nuclear Weapons Design," in which the Department of Energy concluded, "[ t] he use of high-speed computers and mathematical models to simulate complex physical processes has been and continues to be the cornerstone of the nuclear weapons design program." 12
As valuable as supercomputers are for nuclear weapons, particularly in an era of diminished testing, they are also integral to the development of conventional weapons and delivery systems like ballistic missiles. The Goodman Report calls conventional weapons programs "... today's 'bread and butter' of high-performance computing applications in the U. S. national security community. The design and development of advanced conventional weapons (ACW) has developed a symbiosis with high-performance computing: programs are often defined on the basis of the current or projected state of the art in HPC, and new computer hardware and software are frequently developed in response to program requirements...." 13
The Defense Department's High Performance Computing Modernization Plan underscores this point, citing projects such as the Airborne Laser Challenge Project, B1B Radar Cross Section Prediction, Design of New Materials Using Computational Chemistry, and Modeling of Complex Projectile-Target Interactions as examples of the Pentagon's supercomputer "Challenge Projects" for fiscal year 1997. 14America and its allies have already faced their own technology in war. According to Dr. William Schneider, Under Secretary of State for Security Assistance, Science and Technology from 1982 to 1986, "[ t] he decontrol of advanced civil sector (' dual use') technology among the industrialized nations of the world was the enabling policy change which contributed to Iraq's indigenous capability for WMD and military missiles." 15 Dr. Stephen D. Bryen, who from 1981-1988 served concurrently as Deputy Under Secretary of Defense for Trade Security Policy and Director of the Defense Technology Security Administration, said, "[ i] n the case of China, we are transferring much more sophisticated technology than anyone ever sold to Iraq." 16 Bryen also noted, "China is seeking to enhance its nuclear weapons and their delivery systems... Supercomputers are important for China to achieve these goals," while pointing out that, "China can use supercomputers to enhance many other weapons programs." 17
Dr. Leitner summed up the danger presented by dualuse goods going to countries of concern when he testified, "[ a] t whatever stage of development, it is in the USG [U. S. government] interest to make a weapon of mass destruction (WMD) and ballistic missile delivery program as difficult, expensive, and unreliable as possible." 17 His statement applies equally to other military programs of nations with interests contrary to those of the United States. Moreover, supercomputers enable nations supplying rogues and others to upgrade more quickly the quality of their products.
The Clinton Administration has liberalized export controls on high performance computers in two stages. According to Kenneth Flamm of the Brookings Institute, who served from 1993-1995 as the Principal Deputy Assistant Secretary of Defense for Economic Security and Special Assistant to the Deputy Secretary of Defense for Dual Use Technology Policy, the Administration in 1993 first "... decontrolled computer exports to 500 MTOPS [Millions of Theoretical Operations per Second, a standard measure of computing capability], and imposed after- sale monitoring on only the most powerful computing machines," keeping the "strictest controls" on computers above 1,500 MTOPS. 19
On October 6, 1995, President Clinton dropped the second shoe, saying he was "... pleased to announce a major reform of our computer export controls that will adjust to the global spread of technology while preserving our vital national security interests." 20 This policy, which took effect on January 25, 1996, groups all nations into four "country tiers" and establishes high performance computer export licensing requirements and exceptions based upon their country of destination, as depicted below.
At the other end of the spectrum, Tier 4 countries -- the "terrorist nations" -- cannot legally receive any of these computers. According to William Reinsch, Under Secretary of Commerce for Export Administration, "... our limit [to Tier 4 countries] is 6 MTOPS, which eliminates everything." 21
Almost all of South America, Central America, the Caribbean, and Africa are in Tier 2, eligible for supercomputers capable of up to 10,000 MTOPS -- ten billion operations in a second -- before an individual validated export license is required. Tier 2 countries can transfer computers without U. S. government permission within Tiers 1 and 2.
The policy for the 50 Tier 3 countries, which includes China and Russia, is more complicated. Tier 3 policy requires an individual validated license granted by the Department of Commerce under only two circumstances. First, if the computer is capable above 2,000 MTOPS and is going to a military end-use or end-user, or second, if the computer to be exported is capable above 7,000 MTOPS. No individual validated export license is required for manufacturers vending supercomputers capable between 2,000 and 7,000 MTOPS to buyers in Tier 3 countries when there is to be a civilian end-use and enduser. The exporter, rather than the Department of Commerce, determines whether the purchaser's end-use and user representations are accurate. In transferring end-use and user determinations to industry, the Clinton Administration policy puts American exporters on the honor system. They police themselves supposedly to the point of denying themselves sales. This transfer of responsibility from government to industry is the embodiment of the Clinton-Gore campaign pledge to "[ s] treamline the current decision-making process for export controls." 22
According to Victor Mihailov, Russia's Minister of Atomic Energy, Russia obtained American supercomputers that will be used to simulate nuclear explosions and are "10 times faster than any previously available in Russia." 23 These exports clearly contravene the lenient new policy of the Clinton Administration. In fact, when IBM and Hewlett-Packard earlier requested specific individual validated licenses to export supercomputers to one of Russia's nuclear weapons labs, according to Secretary Reinsch, "[ w] e declined to approve those licenses." 24
The five computers Mikhailov spoke of are located in Russia's two premier nuclear weapons labs, Chelyabinsk-70 and Arzamas-16. 25 Four of these American supercomputers came from Silicon Graphics, the California company headed by Edward McCracken, while the fifth is an IBM. According to McCracken, it was his company's understanding that "the computers were for environmental and ecological purposes," 26 demonstrating how shifting the end-use and user determination burden from government to industry can work.
It seems that Silicon Graphics should have been able to determine that selling supercomputers to Russia's nuclear weapons labs went beyond the new supercomputer export decontrols. According to Professor Gary Milhollin, Director of the Wisconsin Project on Nuclear Arms Control, "[ i] n a memorandum dated January 15, 1997, which Silicon Graphics sent to the Commerce Department, Silicon Graphics admitted that it sold the computers to the 'All-Russian Scientific Research Institute for Technical Physics (VNIITF), ' which is the official name for Chelyabinsk-70." The World Wide Web homepage giving the mission of Chelyabinsk-70, found by searching for "VNIITF," is shown below. 27 Moreover, in testimony before the House National Security Committee, Professor Milhollin noted that, "[ i] n May 1995, the Commerce Department's Bureau of Export Administration published The Russian Defense Business Directory, a guide to acquaint American exporters with Russia's military sites. The guide listed Chelyabinsk-70's 'product line' as the 'development of nuclear weapons. '" 28
According to its web page, the "Russian Federal Nuclear Center -- All-Russian Research Institute of Technical Physics (RFNC-VNIITF)... is under the authority of the Ministry of Atomic Energy of the Russian Federation...." If that doesn't suggest nuclear activity, other entries should. The first of the RFNC-VNIITF's "main scientific directions" is "the development of techniques and equipment provided to record explosion processes, including nuclear ones." Another of its "main scientific directions" is "investigation, development, and test of powerful explosives." If these clues aren't sufficient to put an exporter on notice, the last sentence on the page should be: "At present about 50% of the investigators, engineers, and workers are oriented to solve pure peaceful problems."
The New York Times, citing "American Government officials who requested anonymity," said, "[ t] he supercomputers sold to China would allow the country to significantly improve its nuclear weapons by processing huge amounts of data from very small underground nuclear weapons tests. These tests are currently banned by international treaty, but the high-performance computers would allow the Chinese to conduct weapons tests with explosions so small that they would be undetectable by outsiders...." 33
While some of its scientific and technological institutes are not working for the People's Liberation Army (PLA), the Chinese Academy of Sciences' participation in defense programs and its status as a military-end user is not a closely held secret. Indeed, when asked in a Subcommittee hearing about the Chinese Academy of Sciences' involvement in upgrading China's nuclear weapons and missile technology, Commerce's Reinsch responded, "... that is something that the intelligence community has looked into in considerable detail. We have information on that, but it is classified and I can't provide it to you in open session." 34 Thus, PLA generals seeking new hardware with which to challenge America's interests in Asia in the next century will benefit directly from American technology freely exported by Americans.
But there are end-users with less obvious backgrounds than Chelyabinsk-70, Arzamas-16, or the Chinese Academy of Sciences. In June of 1997 the Director of Central Intelligence (DCI) sent Congress a report entitled The Acquisition of Technology Relating to Weapons of Mass Destruction and Advanced Conventional Munitions. 34 The unclassified report covers the period July through December of 1996 and levies serious proliferation charges against, among others, Russia and China. According to the DCI, "[ m] any Third World countries -- with Iran being the most prominent example -- are responding to Western counterproliferation efforts by relying more on legitimate commercial firms as procurement fronts and by developing more convoluted procurement networks." 36 U. S. exporters are not capable of determining (and shouldn't be), except in the clearest instances, which purchasers are "procurement fronts." It is wrong to place this burden on industry. That is what America's intelligence agencies are paid to do. As Senator Glenn said on the Senate floor, "[ t] here are significant limitations in the extent to which the Government can delegate export control responsibilities to the private sector. Companies simply do not have the capabilities of U. S. intelligence agencies. That is the reason why licensing is such a good idea. It is the best known technique for making efficient and effective use of the resources of our Government... to assess the proliferation risks in certain exports." 37
PLA generals seeking new hardware with which to challenge America's interests in Asia in the next century will benefit directly from American technology freely exported by Americans.
According to Commerce Under Secretary Reinsch, if a company has a question about the legitimacy of a prospective purchaser, it "... can always consult with us...." 38 But, what about a company that seeks information on its own about a buyer and finds no military connection? What reason will it have under Administration rules to consult the Department of Commerce? The most thorough search by an exporter with limited resources may not expose a prospective purchaser's military-related activities if they are buried in a U. S. intelligence agency compartment or if a front company is being used to procure the computer.
To obviate even the occasional inquiry from industry, the Department of Commerce proposes publishing a list of prohibited end-users. The administration and some in the computer industry insist that the solution to industry's difficult problemlies in a published list of suspect end-users. Sales to listed entities would not be prohibited by law but an individual validated license would be required. This acknowledges that American supercomputers shouldn't be in Russia's and China's nuclear weapons design labs, among other places, but the implementation of the proposal would do more harm then good. Any published list would necessarily be incomplete, for a complete list inevitably would compromise U. S. intelligence sources and methods. According to Commerce Under Secretary Reinsch in Senate testimony, "We have not done it [publish a list] extensively so far. There are intelligence sources and methods issues that come up frequently, as well as some other considerations." 39 Mr. Reinsch went on to say, "... frequently we don't publish for that reason, even though we have identified someone that, for other reasons, ought to be published." 40
Senator John Glenn observed problems with this policy when he said on the Senate floor, "... such a listing could be quite useful to a proliferant country or group, effectively amounting to free market research for the proliferators." 41 Any published list would be easy to manipulate by both purchaser and exporter if they are not devoted to the honor system. If Chelyabinsk-70 is on the list of suspect locations, can a "Chelyabinsk-71," not on the list, receive U. S. exports of high performance computers? An exporter could decide Chelyabinsk-70 does nuclear weapons work, but "Chelyabinsk-71" conducts only environmental research. This type of list might even increase exports to those who ought not to have them.
The Department of Commerce has published such a list, most recently on October 1, 1997, consisting of 15 locations in China, Russia, India, Pakistan, and Israel. 42 Called by Commerce the "Entity List," it alerts the public that "[ t] hese end users have been determined to present an unacceptable risk of diversion to developing weapons of mass destruction or the missiles used to deliver those weapons. Publishing this list puts exporters on notice that any products sold to these end users may present concerns and will require a license from the Bureau of Export Administration. While this list will assist exporters in determining whether an entity poses proliferation concerns, it is not comprehensive." 43
Only a casual inspection reveals its inadequacy. On this list are Chelyabinsk-70 and Arzamas-16 in Russia and parts of the Chinese Academy of Sciences, all of which are currently using American supercomputers. Because of this list, America's computer exporters now know that they need a license to ship a high performance computer to any of these entities. But where is the Chinese company, really part of China's government, which shipped the specialized ring magnets to Pakistan for use in its nuclear program? What about the Chinese company, or government entity, that shipped M-11 missiles to Pakistan and now, according to press reports, is helping Pakistan build a factory for the indigenous manufacture of M-11 missiles? Why isn't that entity on the list? What about the Russian companies or government entities which are helping Iran upgrade its nuclear and ballistic missile programs? Why aren't they on the list?
This list only adds to the confusion. It is not a solution to the problem.
The legislation was based on an amendment, introduced by the chairman and ranking minority member of the National Security Committee of the House of Representatives, Congressmen Floyd Spence and Ronald Dellums, which passed the House by a vote of 332-88 on June 19, 1997. 48 While similar legislation 49 was offered by Senators Thad Cochran and Richard Durbin and defeated in the Senate after weeks of lobbying by the Administration and the computer industry, the CochranDurbin amendment acquainted Senate conferees on the bill with the problem and led to an agreement in the conference report on the issue.
The legislation, which was signed into law by the President, applies only to U. S. supercomputer exports to Tier 3 countries and contains the following provisions:
1. Exporters must submit for review any proposed Tier 3 sale above the Clinton Administration's 2,000 MTOPS threshold. The review (the government's enduse and user determination) is conducted by the Secretaries of Commerce, Defense, State, and Energy, and the Director of the Arms Control and Disarmament Agency within ten days of submission. The exporter must obtain an Individual Validated License (IVL) to export the supercomputer if, during the ten-day period, any of the five finds the proposed sale outside the "license exception" policy announced by the President in October of 1995.
2. Any change by the President to the 2,000 MTOPS threshold for Tier 3 can take effect only 180 days after a report to Congress justifying the change.
3. Any change to the composition of Tier 3 only takes effect 120 days after the President reports to Congress on the reasons for the adjustment.
4. Post-shipment verification (PSV) -- checks by U. S. government officials after a supercomputer export is completed to ensure, among other things, that the computer is being used for the purpose originally represented by the purchaser and has not been diverted to some other location -- for all Tier 3 exports must be performed, with a report submitted annually to Congress explaining the results of these PSV's.
Senators have described the rationale for this legislation in these ways:
Testifying on technology transfer in June of 1997 before the Joint Economic Committee, retired U. S. Army Lieutenant General Robert L. Schweitzer said soldiers are "... grateful when Congress acts ahead of time to bar technology transfers, not only the simple ones... but the more serious, albeit subtle ones, which can affect the outcome of battles and wars." 58
Secretary Reinsch couldn't be more correct in saying, "...[ t] here is a lot you can discover with prelicense checks." And if Mr. Reinsch's Bureau of Export Administration conducted more of them, not to mention post-shipment verifications, Congress would have greater confidence that the Department of Commerce is as concerned about national security as it is about promoting exports.
One would expect that both PLC's and PSV's would be routinely required of countries which have proven by past behavior not to be trustworthy. Instead, they are the exception. According to Secretary Reinsch's testimony in November of 1997, "[ t] hus far in 1997, BXA [Commerce's Bureau of Export Administration] has requested 22 post shipment verifications of HPCs exported to Tier 3 countries. Seventeen have been completed, all with favorable results. The remainder of the requests are pending at our embassies." 61 Of the 1,437 supercomputers exported from the United States from the inception of the President's latest decontrol in January of 1996 through some point in March of 1997 (the latest period for which the Commerce Department has provided data), 91 -- or 6.34% -- went to Tier 3 countries. Of these 91, 47 went to China and ten to Russia, making these two countries the recipients of approximately 63%, nearly twothirds, of Tier 3 exports for the first 14 months of the President's most recent decontrol. 62
Astonishingly, not a single supercomputer exported to China had either a pre-license check or a post-shipment verification. In fact, according to Mr. Reinsch on December 11, 1997, "No formal postshipment verifications have yet been requested to China for the 47 systems exported." 63 For Russia, not one of its ten American supercomputers received a pre-license check and three post-shipment verifications were conducted, though none of these postshipment checks were for the high performance computers obtained by Chelyabinsk-70 or Arzamas-16. 64
As, per Secretary Reinsch, "[ t] he Russian government allowed U. S. Embassy officials to conduct the [three] requested post-shipment verifications," 65 the Clinton Administration has failed even to request post-shipment verification access to the ill-gotten American supercomputers at Chelyabinsk-70 and Arzamas-16. Thus, of 57 American supercomputers shipped to China and Russia between January 25, 1996 and some point in March of 1997, there were no prelicense checks and three post-shipment verifications (5.3% of the 57).
In countries with histories of proliferation and diversion of computers to improper end-users, no supercomputers should be shipped without a prelicense check, and all should have a post-shipment verification. According to Mr. Reinsch, "BXA has opened four investigations on HPC exports: two involving China and two involving Russia. All these investigations are in the hands of the Department of Justice." 66 As the Commerce Department Inspector General notes in his recent report, "[ a] PSV, conducted after an export has occurred, is used to determine whether the licensed item or technology was received by the party named on the license or shipper's export declaration (SED) or was diverted to an unauthorized end user. The check is also used to verify whether the commodity is being used in accordance with the license provisions." 67 It stands to reason that supercomputers shipped under the license exception policy (without an individual validated license) to a country of concern should also receive these post-shipment verifications.
Two conclusions from the Commerce Department Inspector General's report are worth noting. First, though the report was not specific to supercomputers but covered all dual-use goods, its conclusion that the "[ q] uality and utility of end-use checks should be improved" 68 is unavoidable given the failure of Commerce to conduct any end-use checks on supercomputers in either China's or Russia's nuclear weapons complexes. Second, Secretary Reinsch "... disagreed with our [Commerce Inspector General] recommendation to screen all parties to export license applications against the TECS [U. S. Customs Service database]. Specifically, their response argued that 'to refer all license applications to Customs for review in TECS effectively makes Customs a referral agency. This would not be consistent with the President's goal to streamline the export licensing process.... '" 69 The Commerce Department's Inspector General took exception to Mr. Reinsch's response, stating, "BXA's argument, however, is not convincing." 70
Because of the casual attitude toward end-use checks by the Commerce Department's Bureau of Export Administration, Congress included a provision in Public Law 105-85 requiring the Secretary of Commerce to "... conduct [a] post-shipment verification of each digital computer with a composite theoretical performance of more than 2,000 millions of theoretical operations per second (MTOPS) that is exported from the United States..." 71 to a Tier 3 country. A statute should not be necessary for the Bureau of Export Administration to conduct comprehensive enduse checks in countries like China and Russia.
The statute requires the government to determine end-use and user -- and apply the law. Nothing else. The composition of the four tiers, the MTOPS thresholds for each tier, and the policy allowing for the export of supercomputers without IVL's based upon the license exceptions established by the President are unaffected.
The issue is broader than placement of U. S. supercomputers in Russian and Chinese nuclear weapons labs, although State Department Spokesman James P. Rubin has at least said of the Russian labs, "... we believe they [the U. S. supercomputers] are at locations precisely where we didn't want them to be." 72 The United States should not be helping Russia and China to upgrade the quality of the weapons and technology they are proliferating. Director of Central Intelligence George Tenet's report to Congress confirms that "... countries of concern continued last year to acquire substantial amounts of WMD-related equipment, materials, and technology, as well as modern conventional weapons. China and Russia continued to be the primary suppliers, and are key to any future efforts to stem the flow of dual-use goods and modern weapons to countries of concern." 73
In spite of the weight of the evidence supporting reform of current policy, the President criticized the provisions of the 1998 Defense Authorization Act relating to supercomputer export controls, saying they were an attempt "... to severely limit the President's flexibility to conduct foreign policy...." 74 But America's Constitution does not give the President absolute authority to conduct foreign policy. Article I, Section 8, says, "[ t] he Congress shall have Power...[ t] o regulate Commerce with foreign Nations...." 75
Prior to the President's signing the defense authorization bill, Secretary Reinsch also criticized its supercomputer provisions. In testimony before the House National Security Committee, he said, "[ t] he requirement to conduct post shipment checks will become an extraordinary resource burden, is unadministrable and unnecessary [emphasis in the original]. 76 Remember, from implementation of the President's policy on January 25, 1996, through some point in March 1997, only 91 of 1,437 supercomputers, or roughly 6% of the total, were exported to Tier 3 countries. 77 Conducting 91 post-shipment verifications -- an average of 1.5 per week over that 14 month period -- doesn't seem to be "an extraordinary resource burden" or "unadministrable."
Administration officials also criticized the 2,000 MTOPS threshold in the legislation, though this level was set not by Congress but by the Administration's policy. But the legislation allows for the 2,000 MTOPS level to be raised 180 days after the Administration provides justification to Congress.
This 180 day review period, criticized by Mr. Reinsch as "... mak[ ing] no technological sense," 78 is an integral part of the legislation precisely because past Administration projections on supercomputers -- projections that formed the basis of President Clinton's October of 1995 announcement decontrolling high performance computer exports -- have been wrong.
A White House "Fact Sheet" accompanying the President's October 6, 1995, announcement says, "... we conservatively judged that computers up to 7000 million theoretical operations per second (Mtops, a standard measure of computing performance) will become widely available in open commerce within the next two years." 79 And it was this "conservative judgement" that was instrumental in determining the Tier 3 export control policy.
The judgement is not supported by the facts. In testimony before the House National Security Committee on April 15, 1997, Mr. Harold J. Johnson, Associate Director for International Relations and Trade Issues in the Security and International Affairs Division of the General Accounting Office, said, "... the United States or Japan [are] the only countries currently producing computers above that [3,500 MTOPS] level." 80 Secretary Reinsch agreed with the GAO when testifying before the Subcommittee on June 11, 1997, saying, "I cannot, at this point, make a convincing case that that is wrong, Mr. Chairman.... For purposes of this discussion, I am happy simply to assume that that is correct." 81
This "wide availability" to the 7,000 MTOPS level suggested by the White House in October of 1995 was based not on foreign availability, but on the ability of U. S. manufacturers to produce such powerful machines. This future production capability appears to have driven the President's policy, not the ability of rogues or other nations to acquire these machines outside the United States. Although GAO's Johnson noted Japan's over-3,500 MTOPS capability, Japan's export restrictions on supercomputers are tougher than those of the United States. Kenneth Flamm, an architect of the Clinton Administration's supercomputer decontrol, even noted that Japan opposed the scope of the decontrol, saying the Administration "... proposed the strictest controls on exports of computers above 2,000 MTOPS (though it was able to negotiate only 1,500 MTOPS in bilateral discussions with Japan....)" 82
Industry also has been overly optimistic in its predictions. Officials of IBM and Intel, for example, have provided the Subcommittee with conflicting estimates of when a personal computer capable of 2,000 MTOPS will be available. And it is the Intel official -- whose company is "... the world's biggest maker of microprocessors, with its chip in 85 percent of all personal computers" 83 -- who says the 2,000 MTOPS personal computer will not be on the market nearly as quickly as claimed by IBM. 84
This "wide availability" to the 7,000 MTOPS level suggested by the White House in October of 1995 was based not on foreign availability, but on the ability of U. S. manufacturers to produce such powerful machines.There is, of course, room for disagreement on when specific thresholds will be crossed. But if the Administration presents a factual case, the flexibility in the legislation signed by the President in November of 1997will make it possible for Congress to agree with proposed threshold changes.
As documented in previous chapters, China and Russia are constantly involved in sales of technology, components, and delivery systems for weapons of mass destruction, as well as sales of highly-capable advanced conventional weapons and other critical military technologies to other nations. The facts support President Clinton's describing the proliferation of weapons of mass destruction and the means of delivering them "... as an unusual and extraordinary threat to the national security, foreign policy, and economy of the United States," and his declaring "a national emergency to deal with that threat." 85 The President has subsequently reaffirmed this executive order annually, most recently on November 12, 1997.
The facts underlying the necessity for this executive order are not based on isolated past examples. A flood of proliferation is reported daily in America's newspapers. Israel has found this threat to be so severe that their leaders are publicly warning America of the dangers. Prime Minister Benjamin Netanyahu, for example, recently said, "Iran is unseen, unperturbed and undisturbed... building a formidable arsenal of ballistic missiles, actually inter-continental ballistic missiles...[ that] they actually plan to [be able to] reach the eastern seaboard of the United States, Manhattan." 86 America's own intelligence agencies warn of these problems, too. According to the DCI's June of 1997 report to Congress, "China was the most significant supplier of WMD-related goods and technology to foreign countries. The Chinese provided a tremendous variety of assistance to both Iran's and Pakistan's ballistic missile programs. China also was the primary source of nuclear-related equipment and technology to Pakistan, and a key supplier to Iran during this reporting period. Iran also obtained considerable CW-related assistance from China in the form of production equipment and technology." 87 It also said, "Russia supplied a variety of ballistic missile-related goods to foreign countries during the reporting period, especially to Iran. Russia was an important source for nuclear programs in Iran and, to a lesser extent, India and Pakistan." 88
Trade is vital to the United States, but not its only interest. National security cannot be subordinated to trade. While a balance must be struck between national security and exports, U. S. national security interests dictate that there are some goods which must not be sold in some markets. The willingness of some western European countries to help Libya construct a chemical weapons production complex, for example, does not justify the involvement of U. S. companies in similar ventures.
Nations which threaten the security interests of the United States should not be armed by America, nor should America help them arm themselves. America's government should be reducing the likelihood that the world's foremost proliferators are engaging in this activity with the assistance of the United States. The fight against proliferation must include self-discipline at our own borders.
There is no reason to believe the Clinton Administration intended dual-use decontrol to endanger America's national security. According to Brookings' Kenneth Flamm, "[ n] o one involved in the ongoing [supercomputer export decontrol] policy reform effort -- and I know, because I was part of it -- had any intention of handing America's military adversaries greater access to more powerful computers." 89 However unintended the results, supercomputers are now in places they shouldn't be. As Dr. William Schneider said, the Clinton Administration has "... liberalized export controls on dual use technology, equipment, and services that has had the unintended consequence of facilitating the process of proliferating WMD and their means of delivery as well as advanced conventional weapons." 90
The Clinton Administration would do well to heed the warning of John Fialka, Wall Street Journal reporter and author of War By Other Means, who said, "[ n] ations that take their technological edge for granted have a great deal to lose." 91
6 Department of Defense, High Performance Computing Mod-ernization Plan, March 1997, p. 1. Hereafter cited as HPCM Plan.7 Ibid. 8 Richard Bernstein and Ross H. Munro, The Coming Conflict With China (New York, Alfred A. Knopf, 1997), p. 141. 9 U. S. Congress, Senate Governmental Affairs Subcommittee on International Security, Proliferation, and Federal Services, hearing on Proliferation and U. S. Export Controls, June 11, 1997, p. 22. Here-after cited as Hearing, Proliferation and U. S. Export Controls. 10 Dr. Peter Leitner, prepared statement before the Joint Economic Committee, 105 th Congress, 1 st Session, June 17, 1997, p. 9. 11 Seymour Goodman, Peter Wolcott, and Grey Burkhart, An Examiniation of High- Performance Computing Export Control Policy in the 1990s (IEEE Computer Society Press, Los Alamitos, California, 1996), p. 57. Hereafter cited as the Goodman Study. The Goodman Study is deficient in many areas, as detailed in a report prepared for the Committee on National Security of the House of Representatives, An Assessment of the Impact of the Export of Advanced Computers and Computation Technology on the Proliferation of Conventional Weapons, Weapons of Mass Destruction, and Their Means of Delivery, by Stephen Bryen, Willliam R. Graham, Phil Marcus, and William Schneider, Jr., July 15, 1997. However, the Goodman Study does have a chapter which clearly describes the uses of supercomputers. 12 U. S. Department of Energy, The Need for Supercomputers in Nuclear Weapons Design, January 1986, p. 34. 13 Goodman Study, p. 56. 14 HPCM Plan, p. 19. 15 Hearing, Proliferation and U. S. Export Controls, p. 47. 16 Ibid., p. 41. 17 Ibid. 18 Dr. Peter Leitner, prepared statement before the Joint Eco-nomic Committee, 105 th Congress, 1 st Sess., June 17, 1997, p. 9. 19 Kenneth Flamm, "Decontrolling the Uncontrollable", The Brookings Review, Winter 1996, p. 24. 20 White House Press Release, "Statement By The President," October 6, 1995. 21 Hearing, Proliferation and U. S. Export Controls, p. 17. 22 Clinton- Gore Campaign publication, Technology: The En-gine of Growth, September 18, 1992, p. 20. 23 Owen Matthews, "Supercomputer Sale Under Investigation," Moscow Times, February 18, 1997. 24 Hearing, Proliferation and U. S. Export Controls, pp. 18- 19. 25 Ibid., and John Fialka, "U. S. Investigates Silicon Graphic's Sale of computer to Russian Weapons Lab," Wall Street Journal, Feb-ruary 18, 1997. 26 Bill Gertz, "Supercomputers From U. S. Sold to Russian Nuke Site," Washington Times, February 19, 1997, p. A3. 27 "RFNC- VNIITF's Mission," www.ch70.chel.su/vniitf/mission.html. It took 3 seconds to find this page after putting "VNIITF" into the "Excite" World Wide Web search engine. 28 Professor Gary Milhollin, prepared statement before the Committee on National Security of the U. S. House of Representatives, 105 th Congress, 1 st Sess., April 15, 1997, p. 2. 29 Jeff Gerth, "China Buying U. S. Computers, Raising Arms Fears," New York Times, June 10, 1997, p. 1. 30 Professor Gary Milhollin, prepared statement before the Com-mittee on National Security of the U. S. House of Representatives, 105 th Congress, 1 st Sess., April 15, 1997, p. 5, citing "information published by the [Chinese] Academy [of Sciences]." 31 HPCM Plan, pp. 17- 19. 32 Professor Gary Milhollin, prepared statement before the Com-mittee on National Security of the U. S. House of Representatives, 105 th Congress, 1 st Sess., April 15, 1997, p. 5, citing "information published by Silicon Graphics." 33 Jeff Gerth, "China Buying U. S. Computers, Raising Arms Fears," New York Times, June 10, 1997, p. 1. 34 Hearing, Proliferation and U. S. Export Controls, p. 19. 35 Director of Central Intelligence Report to Congress, The Acquisition of Technology Relating to Weapons of Mass Destruction and Advanced Conventional Munitions, July- December 1996, June, 1997, Hereafter cited as DCI Report. 36 Ibid., p. 6. 37 U. S. Congress, Congressional Record, 105 th Congress, 1 st Sess., p. S7099. 38 Hearing, Proliferation and U. S. Export Controls, p. 17. 39 Ibid., p. 18. 40 Ibid. 41 U. S. Congress, Congressional Record, 105 th Congress, 1 st Sess., p. S7100. 42 Department of Commerce, Entity List, Supplement No. 4, Part 744, October 1, 1997. 43 Ibid., p. 1. 44 Public Law 105- 85, sections 1211 through 1215. 45 "Commerce Seeking Quick Implementation of New Com-puter Export Curbs," Inside U. S. Trade, December 5, 1997. 46 "New Export Controls on Supercomputers," The Risk Re-port, Special Bulletin, November 21, 1997, p. 1. 47 The President's policy amounts to what is virtually a license-free zone for Tier 1 countries. While these countries are only sup-posed to reexport American supercomputers subject to U. S. licensing policy, it is a fact that many European nations -- France and Germany are noteworthy, but not the only, examples -- take a more casual approach to dual- use export controls than does even the Clinton Administration. Of the five supercomputers in Russia's nuclear weapons complex, the one not provided by Silicon Graphics was an IBM machine transferred from a nation in western Europe. 48 U. S. Congress, amendment #186 to the National Defense Authorization Act For Fiscal Year 1998, H. R. 1119, 105 th Congress, 1 st Sess., 1997. 49 U. S. Congress, amendment #420 to the National Defense Authorization Act For Fiscal Year 1998, S. 936, 105 th Congress, 1 st Sess., 1997, as introduced on June 19, 1997, U. S. Congress, Con-gressional Record, 105 th Congress, 1 st Sess., p. S5991- 96. 50 U. S. Congress, Congressional Record, 105 th Congress, 1 st Sess., p. S5992. 51 Ibid., p. S5993. 52 Ibid., p. S5995. 53 Ibid., p. S6023. 54 Ibid., p. S6900. 55 Ibid., p. S7053. 56 Ibid., p. S7099. 57 Ibid., p. 7101. 58 Lt. Gen. Robert L. Schweitzer (Ret.), prepared statement before the Joint Economic Committee, 105 th Congress, 1 st Sess., June 17, 1997, p. 1. 59 Hearing, Proliferation and U. S. Export Controls, p. 23. 60 U. S. Department of Commerce, report #IPE- 9524 from the Office of the Inspector General, Export Application Screening Process Could Benefit From Further Changes, September 1997, p. 7. Hereafter cited as Commerce Department Inspector General Report. 61 William Reinsch, prepared statement before the Committee on National Security of the U. S. House of Representatives, 105 th Congress, 1 st Sess., November 13, 1997, p. 2. 62 Table prepared by the Department of Commerce/ Bureau of Export Administration, sent to Senator Cochran on June 25, 1997, as amended by William Reinsch via letter to Senator Cochran on Sep-tember 4, 1997. 63 Letter to Senator Thad Cochran from William Reinsch, December 11, 1997. 64 Ibid., and, conversation between Subcommittee staff and senior Bureau of Export Administration official, December 12, 1997. 65 Letter to Senator Thad Cochran from William Reinsch, Decermber 11, 1997. 66 William Reinsch, prepared statement before the Committee on National Security of the U. S. House of Representatives, 105 th Congress, 1 st Sess., November 13, 1997, p. 2. 67 Commerce Department Inspector General Report, p. 7. 68 Ibid., p. 24. 69 Ibid., p. 22. 70 Ibid., p. v. 71 Public Law 105- 85, Section 1213( a). 72 "IBM Says Cooperating in US Probe of Computer Sale to Russia," Associated Press, October 29, 1997. 73 DCI Report, p. 6. 74 White House Press Release, "Statement By The President" November 19, 1997. 75 U. S. Constitution. 76 William Reinsch, prepared statement before the Committee on National Security of the U. S. House of Representatives, 105 th Congress, 1 st Sess., November 13, 1997, p. 3. 77 Table prepared by the Department of Commerce/ Bureau of Export Administration, sent to Senator Cochran on June, 25 1997, as amended by William Reinsch via letter to Senator Cochran on Sep-tember 4, 1997. 78 William Reinsch, prepared statement before the Committee on National Security of the U. S. House of Representatives, 105 th Congress, 1 st Sess., November 13, 1997, p. 4. 79 White House Press Release, "Export Controls on Supercomputers," October 6, 1995, p. 1. 80 GAO testimony before the HNSC, Harold J. Johnson, April 15, 1997, p. 5 of prepared statement (GAO/T-NSIAD-97-128). 81 Hearing, Proliferation and U. S. Export Controls p. 16. 82 Flamm, "Decontrolling the Uncontrollable", p. 24. 83 "New chip boosts computer's power and storage," Associated Press, September 17, 1997. 84 Meeting with Subcommittee Staff and Aaron Cross, Director of Public Policy, IBM, June 24, 1997, and phone conversation with Subcommittee Staff and Intel official, December 5, 1997. 85 Executive Order 12938, "Proliferation of Weapons of Mass Destruction." 86 "Netanyahu: Iran Might Target U. S.," Associated Press, No-vember 17, 1997. 87 DCI Report, p. 5. 88 Ibid. 89 Flamm, "Decontrolling the Uncontrollable", p. 22. 90 Hearing, Proliferation and U. S. Export Controls, p. 48. 91 John Fialka, prepared statement before the Joint Economic Committee, 105 th Congress, 1 st Sess., June 17, 1997, p. 8.
The Proliferation Primer
International Security, Proliferation, and Federal Services Subcommittee
United States Senate Committee on Governmental Affairs
A Majority Report - January 1998