The Strategic Dimensions of the U.S. Computer Exports to the USSR

Report Homeland Security

The Strategic Dimensions of the U.S. Computer Exports to the USSR

June 8, 1977 34 min read Download Report
Miles M.

(Archived document, may contain errors)

12 June 8, 1977 5 THE STRATEGIC DDENSION OF THE U. S. COMPUTER EXPORTS TO THE U.S.S.R L I. Introduction The ongoing controversy, in the U.S. government and in the media and in dustry, over the sale of one of the largest and most' advanced computer sys tems in the world to the Soviet Union focuses upon one of the most sensi tive aspects of U.S Soviet cmercial relations: the transfer of one of the most critical technologies and its products, both of which are of paramount importance in effective military systems and sophisti cated in telligence operations.

The Cyber 76 is a "super-computer" developed by the Control Data Corpora tion of Minneapolis, Minnesota. It is a scientific marvel, processing up to 40 million operations per second and estimated to be at least 40 times fast er than the best Soviet computer and some 12 to 15 years ahead of the best'indigenous computer system in the U.S.S.R. and its satellite countries The Cyber 76 serves as the rain center" of the Pentagon, of the U. S Air Force, of the super-secret National S ecurity Agency, of the ERDA Energy Research and Development Administration and of the National Aero nautics and Space Administration (NASA of the State Department and the Commerce Department port be thinking when granting permission for sale of Qber 76 to a state openly attempting to achieve military and technological superiority and eventual control over the U. S What, then, can those officials who favor the ex Those in favor of the deal insist that the Cyber 76 will be used by the Soviet communists only to process weather forecasting data cede, however, that the computer has a critical strategic capacity as well.

They do con Proponents of the export also argue that there will be a system of controls designed to make sure that the Soviet Union does not div ert' Cyber 76 i c y-3 fo military purposes. One such control would be the on-site monitoring by one or two Control Data technicians. CDC, however, so far has sold about 50 large' computer systems to the Soviet bloc, and only in one instance does w e a e e the system have on-site inspection. expressed skepticism,' however A typical statement comes from Ray Chapman director of the International Security Agency, ERDA, who remarked Because of 'the similarity between the meteorological and weapons modeliizg pro g rams in calculational 'characteristics, it is not enough to just monitor the ac tual program while they are going through the machine. It is necessary to look at the input and output data When confronted with Mr. Chapman's statement tie"CDC representative s said that they will try to spot-check the memory'and input and output data from the machine for possible Soviet violat'ions. They admit, however, that the t The majority of computer experts inside and outsidc the government have checks have to be only sp o radic, because there would be 3 mountain of tapes requiring enormous storage facilities which were not available at CDC Congressman Robert Dornan directed a question, to CDC's director .for public relations as to what the CDC intends to do if the company d iscovers that the Soviets are cheating whatsoever c for this the QC representative had no answer A Congressman Sam Stratton made an inquiry with the Fentagon regarding safe guards or any other advanced computer system He was informed that there are no ade q uate safeguards for Cyber 76 The computer experts contend that Cyber 76 will provide the Soviet military with critical strategic capabilities which they presently lack. For ex ample, the Soviets, with application of Cyber 76, stand to improve their sAp.I! s effectiveness and as a result could neutralize our entire B-52 bomber force nuclear strategic force capabilities as well as to penetrate our mi1ita.v and intelligence computer networks Also, Cyber 76.could assist them to vastly improve their Last fall, t hen, Secretary of State Henry Kissinger intervened in favor of ex port to the Soviet Union of Control Data?s Cyber 73 computer system overruled the objections by the Pentagon and ERDA.

Pentagon's and ERDA's objections to sale of two CDC Cyber 172 computer systems to Red China.

The objections by the Pentagon and ERDA were on the ground that both com puter systems weapons calculations, for anti-submarine warfare for large phased-array radar to track enemy ICEWs and for other military applications Today, the Department of Defense deploys close to6',000 of the so-called general purpose computers. The same type of computers, however, could be used in a wide variety of civil applications from sciences to RED in in dustry, to crime control and physchiatry He He a lso overruled the the Cyber 73 and Cyber 172, were suitable for nuclear B. rrz 2 r z P 3 c i Obviously, the national security implications of this trade are enormous.

Concern in the United States had led the Defense Department's Science nologies in which t he U. S. enjoys a clear lead over the U. S. S. R Six high-technology trade associations, however, vehemently disagreed with the position of the Defense task force. Their spokesman, Peter F. McCluskey president of the Computer and Business Equipment hlanuf a ctureres Association has demanded that the U. S. Congress relax export controls for strategic goods and eliminate the Pentagonls role as one of the participants in the export control process. Mr. McCluskey argued, in his testimony before the House Committ ee on International Relations In civilian government such as ours the control and administration must reside apart from the military."

Control Data's chairman and chief executive officer William Norris, puts it 3 more bluntly Board task force under the cha irmanship of J. Fred Bucy of Texas Instruments to recamend restrictions on the transfer of strategic tech to the communist superpowers and their satellites 4 c j Wur biggest problem isn't the Soviets, it's the damn Defense i. Department The aforementioned discussion makes it abundantly clear that we have a ser ious issue facing the executiveand legislative branches of our government.

The issue is dominated by deep disputes among concerned parties, and at the heart of the problem is the question of how computers relate to the vital national security interests of the United States.

The purpose of this discourse is to shed light upon the problem by providing some grasp of camputer technology, of its role in technological competition between the United States and the Soviet Union, and of its impact.on the strategic balance between the two superpowers 11. Technological Competition With respect to national defense, the term "technological competition" re fers to the efforts of competing politico-economic systems to maintain, or to achieve,superiority in high-technology areas that are important in ef fective military systems. The history of such a competition between the United States and the U.S.S.R. dates back to 1943 when the Soviet Union be gan its effort to d e velop an atomic bomb. The unexpected orbiting of the Sputnik" earth satellite by the Soviets in 1957 shocked the United States and for the first time focused a broad public attention on the Soviet scientific and technological capabilities and objectives. T his event also resulted in a rapid development of our own space science I i I I I I I i I The competition between the United States and Soviet Union continued in all phases of med and unmanned space programs_and in the development of strategic weapon syst e ms logical strength is the key to our long-range survival as a nation In this era of unprecedented change, our technoI In his statement before the Committee on Appropriations of the U.S. House of Representatives Dr. Malcolm R. Currie, then Director of Def e nse Re search and Engineering, stated American security, like the American, economy, stands on a fotm defense technology First, because the openness of our society tells our adversaries what we are planning in military tedmo logy while their secrecy force s us to provide for many possibi lities. Second;in military operations we traditionally depend on superior quality to compensate for inferior nuhers. Third in order to interpret vital but fragmentary technical intelli gence information we must have extensi v e prior experience in the area.1 dation of techriological superiority. Ne need superiority in 3 The United States continues to hold a technological lead over the Soviet Union in most critical areas vital to our national security lead has been diminishing t he Soviets are ahead (e But that In some very important areas, it is gone directed-energy weapons based on 'laser beams or other charged particle I eams see Heritage Foundation Backgrounder #13 Moreover, the technology balance is dynamic technology balanc e and its dynamics, the qualified analysts agree that the U.S.S.R. has a very large and determined effort and that the Soviets are inexorably increasing their level of technology relative to ours and are in fact, seizing the initiative m important areas (e pressure weapons).

The technological development is molding future Soviet strategy. From z all indications, the future Soviet strategy will be world dominance, with technology as one of the key drivers.2 A crucial element in our strategy of deterrence is the maintenance of a margin of military advantage through possession of a number of sophisticated technologies In examining the current i already men tioned laser beam weapons surface-effect vehicles an anti-personnel I 1

11. The Revolution in Warfare: The Computer Impact Th ere is considerable confusion today about the strategic importance of computers. Many analysts point out that numerous other technologies are revolutionizing warfare, such as giros lasers avionics nucleonics 0 lThe 'kpartment of 'befense Program of Resear c h Development, Test, and Evalu ation, FY 1975, 93rd Congress,. Second Session; April 29, 1974 (U.S. Govern ment Printing Office 1974 2For a useful discussion of the subject see Hearings before the Conanittee on Armed Services, U.S. Senate, 94th Congress, S econd Session, on S. 2965 Parts 4, 6, and 11, U.S. Government Printing Office (Washington 1976 5 metallurgy and propulsion. This is true, yet in one way or another all emerging teclmologies, including computer technologies themselves, are de pendent on co m puters systems (e.g IUIAC IVY Qx: STAR-100, Texas Instnpnent's ASC and Good year's STARAN IV) were built with the help of several large computer systems. d 7- 5 For example, the world's most advanced computer L In short, today's emerging technologies are as dependent on computers as the technologies of the first industrial revolution were dependent on energy.

Computers multiply man's brainpower with the same force that the first in dustrial revolution multiplied man's muscle power lasers, and nucleonics are inter-related.

Furthermore, computers Corquters are as important and intrusive as the mathenlatics and the data processing systems required to conceive, build, and operate complex new machines and to create new materials cording and interpretation of en tire classes of observations. They make it possible to handle and quickly recall large nunibers of information bits.

They are part of communications and are needed for opcrations analysis.

They assume production and distribution jobs and perform a nearly limit less number of other vital tasks.

Without computersmdernweapons systems could not be built, integrated tested, deployed, kept combat-ready and operated. In fact, weapogs such as missiles, aircraft, tanks, high-performance satellite-based surveil la nce systems, ABM defense systems, and submarines incorporate computers as part of their armament. Avionics are intrinsically computer-linked as is missile accuracy. Helicopters used against tanks are provided with computers and computer links to obtain th e real time information needed for effective battlefield interaction They are indispensable for the re In .brief, there are no modern weapon systems that are not vitally dependent upon high-speed computers. A number of strategic missions are centered on hi gh-performance computers; e.g early warning systems, anti-ballistic missiles defense, command control-comications (C-3 anti-submarine warfare space operations and several branches of intelligence.

Corrputers are not just swift calculating machines which in clude, in addition to the computers themselves, internal and ex ternal memory stores, testing and correction mechanisms, and peripheral equTpment such as display units, input-output links communications, and software i.e the programing language and other a ids that assist computer users to avail themselves of the machine structures--for instance, a missile force or the meteorological and hydro logical services require several large general-purpose cowuters and special computers feeding the general-purpose - c hines plus field com puters on board mobile units such as ships aiqlanes missiles, and space vehicles. the achievement of superior military capabilities They are entire systems Big operational For these systems superior computer technology may also permit I I I 6 For example, in the Apollo program a fairly large computer was carried in the Saturn booster; one computer was housed in the comd spacecraft and'two computers were attached to the lunar module. The launch site had large installation. The Earth Res o urces Technology ERE) program would be useless without computers to handle and ltenhancell the inputs from the diverse sensors carried on the satellite C a large computer installation, the vast tracking system contained many smaller and several large comp uters, and mission control had still another ld To summarize, computer teclmology permeates all phases of the development production, operation, and support of modern nilitary systems mensions can be nonexhaustively distinguished and illustrated.

Six di y 1 2 3 4 5 6 Research and Development. Computers permit a major saving in time and resources aircraft, missiles, and new warheads laboratory study can be partially displaced, with the pre sumed result that a better device is achieved, although the temptat' i on to over-engineer is rarely resisted This is evident iri the design of Prototyping and Production and quality control improves the product, mininuze waste and lead to systems less apt to malfunction in an opera tional environment. The effect is to incre ase effective ness by having more units operational.

Support and Maintenance. Electronic data processing again enhances ekfectiveness by providing a higher percentage of machines in an operationally ready state at any time, pre ceding both commitment to op erations and recomnitment after sortie recovery, at least for aircraft Computer-aided design and production processes Onboard Computers. These devices penit one machine to do each of several missions better than a mix of simpler mission-specialized machin e s. On board computers may per mit 'targeting not otherwise possible, such as the redirec flight of a missile to a target acquired during that flight. Certainly such computers lead to improvements in CEP (the missile accuracy Tactical Fragging. Eff ectiveness increases when the time ok the cycle--target acquisition, designation, force com mitment, ordnance loading, routing, comnunications, re covery--is decreased while its precision is increased.

That is, forces not comnitted on a timely basis are in ef fect tqorarily useless 3. It is evident that if a CS system deploys in real time can perform damage assessment, determine I sensor systems which 7 residual enemy force posture, provide empty-hole informa tion, perform boost-phase and midcourse trackin g to de termine own forces at risk, evaluate the evolving enemy main battle.plan, exercise fingertip control over own forces, and reoptimize plans, then strategic force effec t.iveness increases because of more effective applica tions and less waste of com b at capital IV. Comwter Technolow The term "computer technology" is often used to mean only the hardware aspects of digital computers. In this reportcomputer technology will be more broadly defined to encompass analog computers 1 (still widely used in the S oviet Union) and digital computer software. Including the latter accentuates the evident fact that any computer, no matter how capable is of little use without equally capable software. Indeed, it is now recognized that in most large-scale computer applic a tions, software de sign and production are far more difficult and expensive than hardware development and procurement fort in software than the entire computer industry does on development of hardware For example, IBM spends more on R 6 D ef In this compa r ative analysis of computer technology in the United States and in the Soviet Union, reference will be made to various types of computer systems and architectures, computer and component generations, and computer characteristics terms There are several dim e nsions for classifying computers the environment in which they are designed to operate, there are This section will define and briefly describe these it On the basis of Commercial computers--manufactured for use in benign environ ments which are establish e d to satisfy the computer's require ments for floor space, temperature, humidity, floor stability, and the like Military computers--manufactured to be used in environments that are only partially controllable and consequently may be adverse to comercia1 c o mputers. These systems may have to operate under a wide range of climatic conditions, in moving vehicles, unattended, and may be e'jcposed to damaging nuclear radiation 1An analog computer is a computer that operates with numbers represented by directly m easurable qumtities (as voltage, resistance, or rotations).

ZKosy, D. W., Air Force Command and Control .Information Processing in the 1980's: June 1974 Trends in Software Technology, The Rand Corporation, R-1012-PR,b Both commercial and military computers may be further categorized as general-purpose computers, which are designed to handle a wide variety of computational tasks reasonably efficiently, and s ecial- u ose computers tasks. computers which are designed to optimize the computation o a spec1 ic class of Many militarized computers are special-purpose digital or analog 7 The performance of a computer system, especiallyits processing speed, is a function of the architecture of the computer system and its hardware.

One criterion for architectural cla ssification of computers is the degree of parallelism in computation, expressed in terms of the number of concur rent instruction streams and data streams that the system can handle. N though there is a continuous, evolution of architectural concepts, sev eral architectures are sufficiently well established and widely used to warrant their description. Table1 on page22 lists these architectures and repre sentative U.S. computers.

The ability to design and manufacture mini- and microcomputer systems re presents an important milestone in the advancement of computer technology and warrants making a distinction in the subsequent analyses between "con ventional" computers and mini- and microcomputers. The mdifier 'konven- tionall e.g conventional computer, conventional uniprocessor) will be used whenever the intention to exclude mini- and microcomputers is not clear from the context of the discussion.

Computer hardware and computer syst ems are often discussed in terms of I components generations and computer generations. The former refers to the hardware used (components and their packaging qd the latter de notes both the architectural and software aspects of computer systems.

On page 29 Table 2 illustrates one set of definitions of component and computer generations. As in any classification system, there are ex ceptions generation software technology to implement third-generation architectures and software capabilities.

Finally, while the above classification dimensions set a framework for comparisons of computer technology in the United States and the Soviet Union, it is also useful to indicate more specific quantifiable computer system characteristics. ter hardware characteristics T h us, the well-known CDC 6600 and Cyber 76 computers use second The following are important descriptors of compu Processing speed. The raw maximum possible) computing speed in terms of instructions processed per second for a particular mixture of short and long instructions. Processing speed is a function of component speeds the algorithms or instructions used in the mix, and the processor and memory architectures. Usually expressed in terms of MIPS (millions of instructions per second).

Data processing rate . The product of the processor word length in bits and the processor cycle time (usually the time for short instructions 9 Expressed in bits per second, this measure removes the variations due to different word lengths, but hides the .precision of the res u lts Random-access (and mass) memoxysize in terms of bits, bytes, or words, the maxinnm data transfer rate of the memorv device. and the access time of a request to the memory device to obtain These characteristics depend on the type of memory device and i ts architectural features.

Viabilit Aspects of viability are: reliability (in probabilistic terms dime between failures, MI'BF availabilit when needed; mafntain ability when applicable; and rn edness d en subjected to substantial en vironmental variations or hazar s word 0; a block of words.

Physical attributes requirements of the computer system.

The size, weight, power consumption, and cooling For the input-output peripherals, the principal descriptor is the maximum data rate terms of bits per second, that they can produce or accept.

Software characteristics are more difficult to quantify. They depend on the system's architecture (available instructions and other capabilities that are implemented in hardware or firmware, e.g using microprogramma ble con .tro1 units users' capabilities that sharing, interactive terminals, security and be supported (e.g real time input from other systems or data collection devices It is also important to note that'.it is the software efficiency that determines what frattio n of the potential hardware speed of a pro cessor will actually be achieved, and that the lack of software reliabi lity is a major reason why systems seldom meet their plmerational capability dates .l One very crude but comn descriptor which reflects the e n eral level of operational computers .2 computer applications in a country is the total number o Installed and V American Computer Technolow-Computer technology in the United States made substantial advances into the fourth generation--nearly all new U.S. a nd Soviet Camp uter Technologies i A kyber 76 and its software are a very good example of a computer with ware the processing speed can reach 40 MIPS 2 Such possesses only if one takes into account the difference in performance characteristics or the effi c iency of their use I given capacity to process 10 to 12 MIPS, and in fact with special soft I f totals are useful for assessing the computing cxGability a countryc 10 conventional, mi- and'multiprocessors can support resource sharing by multiple users fro m remote terminals, provide hardware features fqr im plementing virtual memory and virtual processor operating systems 1, use microprogramming techniques and use advanced semiconductor integrated circuits and memory units The minicomputers of the late 1960 ' s have become minicomputer systems with complete sets of terminals, auxiliary storage in the form of magnetic tape cassettes and rtfloppy" disks (inexpensive, phonograph-record-like storage units and software that includes cortpilers and operating system T he cost has decreased steadily while performance has improved. For ex ample, minicomputer kits may be bought for $300 and hand-held scientific slide-rule" calculators for 90 a llicro2rocessor. chips are being manufactured in vast quantities for in clusion into other types of systems; stand-alone microcoquters are also emerging I The very large capacity computers whose design was begun in the late 1960's e.g& ILLIAC IVY CEC STAR-100) are now in operation and have been joined by o ers that are commercially m arketed (e.g Texas Instruments' ASC and Goodyear's STAIiAN IV include over 40 disparate computer systems (includSng the ILLIAC IV) that are connected by landline, radio, and satellite comnunications links.

Several other computer networks have been established and are being expanded.

By 1976 the United States already operated about 150,OO general-purpose computers 80% of which are third- and fourth-generation systems The DARPAZ computer network ?xis expanded to I Soviet Computer Technology By contrast, the U.S.S.R. is believed to have had by 1976 about 16,000 computers (80% of which are first- and second generation machines virtually all of which were allocated to the mili- tary and am industry, and to some extent to scientific institutions Of course, witho u t a substantial cagability 'Soviet equipment could not have reached the moon, Venus, and Mars; and the Soviets could not have developed a MIRV capability large number of Soviet computers are obsolete and obsolescent; Soviet com puter memory devices are in f erior; peripheral equipment is poor; software is inadequate; time-sharing is poorly organized; and parallelism and multi processors are in developing stages But there are strong indications that a Computer technology in the Soviet Union is virtually entir e ly imported from the West. In the early 1930's, the U.S.S.R. was importing unit re cord equipment manufactured by Powers ahd Hollerith. The business machine plant was established in hbscow in 1932 with Western assistance, c Virtual memory is a particular h ardware-implemented memory addressing system; a virtual processor is a software capability in which each user may be served by a separate operating system. the .Defense Advanced Research Projects Agency.r f e 1 11 and it began to produce copies of Western machines. This practice of copying earlier models of Western machines continues to the present time although Soviet modifications of the Western models are now more fre quent, so that the copies are not as faithfil as they once were The list of computer t e chnological advances pioneered in the West that made their appearance in the Soviet Union after a lag of from three to twelve years is rather lengthy all significant technological innovations in computer technology have occurred in the West. Advances in S oviet computer technology have been without significant exceptions, direct transplantations this may not always be the case.

If computer technology in the U.S.S.R receives the required technological input from the West, and if the Soviet authorities decide to elevate it high on the scale of priorities and that appears to be the case--it is reasonable to expect that there will be independent Soviet contributions to this technology There are several significant features of Western computer technology that th e Soviets have failed, thus far, to acquire is the technique of large-scale mass production of high quality computer components, subsystems, and systems better quality Soviet computer is a custom-made item. A second element that the Soviets have yet to mas t er is reliability engineering and quality control into their computer industry the kind of creative dynamism so characteris tic of the Western computer scene c The point to be made is that literally I I I Obviously The first of these For all practical pur p oses, each Third, but certainly not least, the Soviets have yet to inject In computers the Soviet Union is 10 to 12 years behind the U.S. in de veloping its own hardware techndlogy, and 10 to 15 years in software This is a narrower gap than in the past, b u t they remain particularly weak in the technologies of mass storage, microelectronics, and in systems design and software are only 6 to 8 years behind the U.S. due to the confusion in our export control legislation which has permitted the Soviets to legal ly obtain U.S software exports also due to the fact that they were able to obtain the necessary techno logy and know-how from the West, but they are still about 8 to 10 years behind the leading edge of U.S. technology.

However: interesting Soviet work on t he theory of automatic programing may be, it has not contributed' to software that would economize program ming time and make computers more accessible to more users Almost precisely 'thirteen years ago, one of the highest officials in the USSR Academy of Sciences in ?Oscow leaked to a Western scientist the Soviet decision to proceed with development of a family of third-genera tion computers closely patterned after the'Iw360 The working name for the project was RYAD which is the Russian word for series. I t should be noted that KGB agents, in the middle 19601s, succeed- covert means in procuring several IBhV360. series computers, their manuals, blueprints In certain areas of software, however, they Soviet integrated circuits appear to be improving 12 and sp ecifications from IBM facilities in West Germany served as the models for production of RXO series.

The Soviets and their COMECON partners decided that the RYAD hardware and software should be compatible with IBI and most other Wesm co nputers This provided them with the benefit of IBM-compatible hardware and soft ware around the world to aid their own computer effort.

Several years went by before any mentions of the ambitious undertaking were seen in the Soviet technical literature the Unified System US) development history places the date of decision to proceed in December 1969 the fact The 1969 reference point actually refers to the signing of a multilateral agreement between the U.S.S.R, Bulgaria, Hungary, East Germany GDR Poland, a n d Czechoslovakia to cooperate on the ES (RYAD project. lbre recently Cuba has been included, but it is believed-t Cuba's role is highly limited. hbst probably, Cubawill be accorded "most favored nation" status in purchasing ES equipment These systems The ' qofficial" version of Clearly, this is at least five years after During the first two years of the 1971-75 five-year plan (1971-1972 there were many indications that announcement of the ES computers was imminent. At the same time, there were persistent re 2 orts of problems and delays; at least one firm deadline had to be pushed back. By the end of 1972, considerable dissension had cropped up among the East Euro pean participants, and it. is possible that some members were on the verge of bolting, preferring to turn to Western imports in order to satisfy critical needs for modern computers.

By this time, the ES-1010 and Es-1020 computers had al-ready been announced.

However, they were back door" announcements, brief statements of fact rather than a' formal u nveiling of a complete family of machines. These two machines reportedly went into production in 1972 Also in 1972, the Czechs successfully tested a prototype of the 1021 machine (referred to at that time as the ES-1020A), and the Soviet ver sion of the 1 030 was approved for production. The Polish ES-1030 may not even yet be in production.

There were reports several years ago that the formal announcement of the series would be made with considerable fanfare. A massive display of the entire range of units i n operation would be lavish, fully descriptive color brochures, and computer experts from all over the world would be invited i. r The actual event took place in May 1973 with little advance publicity, no special fanfare, no color brochures , and very little of the displayed equipment in operation,(A notable exception was the East German display which dazzled visitors with wilirling tapes, blinking lights, and on-the spot horoscopes the ES-1050 and the ES-1060, were missing. The Soviets openl y admitted that they were (and still are bbst significantly, the top machines in the series far from completion. Computer and Year BESM-6 (1967 300,000 MIPS Years Since Appearance of American Computer at Least as Powerful Number of Times the Most Powerful A merican Contem orary Computer is %re owerful f 2 (IBN 360/75) 1.5 MIPS 16 (IBM 360/90 1967 3 (a 6600) 3 MIPS 40 (cIx= Cyber 76 1968 13 According to the intelligence sources, the Soviet nuclear weapon design and development facilities are still waiting for arrival of their first ES-1050 and ES-1060 computer systems. All the evidence at this point sug gests that the Soviets have failed in their effort to develop and mass produce reliable high-speed third-generation computer systems.

The best Soviet computer produced in series so far is BE%-6, which was introduced in 19

67. The BESM-6 boasts 32K (50 bits) of 0.8 sec core storage and has 16 registers operating at 300 nano-seconds. It uses two instructions per word and reportedly has five levels of instruction "look ahead It is capable of simultaneously processing several programs and makes extensive use of overlapping various operations, such as storage accessing, arithmetic, and 1/0 (input/outprt) control. These features give the BESM-6 a capacity of 300,000 t o 500,000 operations per second Some of these machines have been.sold to India, Pakistan, Afghanistan and Eastern European countries, which lack hard currency to buy more ad vanced equipment from the West or Japan. One must stress, however, that the Sovie t capacity to produce BESM-6 is limited to about 35 machines per year, that demand for them within the U.S.S.R. and its bloc considerably exceeds swly, and that at least 14 different machines in the United States,, the United Kingdom, Germany, France, and9 Japan in 1967 were vastly superior.

Table 3 A COMPARISON OF THE ARITHMETIC CAPABILITY OF SOVIET BESM-6 aOI*PUER WI'ZH "AT OF EMLIER AND MIRE POVERFUL AfilERICAI4 COMWTERS The Soviet Union has only a few domestic minicomputer models in production a situatio n similar to that in the Wited States in 1965-1966 No capa bility to produce microcomputers outside the laboratories is evident from the open Soviet computer literature The officially rated capacity for BES4-6 is 1 kllion MIPS; however, be cause of its re l atively small and not expandable 32K word storage the BESM-6 is able to achieve only about 300,000 to 500,000 MIPS. I L 14 BESM-6 computers (introduced in 1967) will continue to be the highest speed Soviet computers until the WAD ES-1060 model becomes ava ilable possibly in 1978) or the more~l-8 (or BESM-X) is produced. Al though the establishment of statewide ne&orks of computers is a major Soviet objective, none is known to exist computer systems have been made, and some are in use.

To summarize, the Sovi et Union's new general-purpose computing systems WAD) lag by at least a generation behind the new systems in the United States, but they nevertheless provide a step-function improvement of Soviet computer capabilities. In microcomputers and very high-spee d computers, the Soviets have nothing that can be compared with U.S. capa bilities Experiments with local multi See Figure 1, page 15).

The testing and deployment of Soviet MIRVs suggests an improvement of the U.S.S.R's computer capability boosters carried on-board computers for the first time, but what general-purpose computers were used to back up the MIRV program is not hOWIl starting with 1973 tests, Soviet The important point is that, so far as we know, all presently deployed Soviet ICBYs were built o n a computer technology corresponding to the U.S. technology of the early 1960s. They are tied to ground-controlled guidance systems based on relatively low capacity computers will certainly be upgraded, but there are limits to such improvements especially if the new computers still are not date and may not be available in sufficient numbers The system VI. Technology Transfer Mechanisms In view of the fact that it is unable--by relying on its own resources to bridge the computer gap between itself and the West, the Soviet Union is intensifying its efforts to obtain large computer systems, miniaturized computers and computer manufacturing technology from the West. The KYAD system itself is, of course, an example of adopting a Western computer design. At present, Western manufacturers are more than willing to sell their wares in the Soviet Union and other Soviet bloc countries, make licensing agreements, install complete manufacturing plas, and launch development efforts jointly with tlese governments. Th e only real ob stacles to these efforts, no matter how many loo?holes were in the system were the export controls (under the old Trading with the Enemy Act and Export Control Act) placed on computer systems of' certain size and' associ ated equipment by Co Com (Coordinating Committee, composed of the United States, NATO nations, and Japan) and the U.S. Export Administration Of fice at the Department of Commerce with its interdepartmental comittees.

As might be expected, the current export control policies ar e considered too restrictive by the U.S. computer and electronic industry; on the other hand, the export control procedures were brought into jeopardy'under I i- f 1000 100 10 n m 2 c w al a a W I-I a U 7 a 6 0 U W 1 0.1 0.01 1 15 f IBM 360/195 Cyber 76 I IBM 360/85 f CDC 6600 x Ryad 3 0 r XRyad UNIVAC LARC M-MOO *Ryad Y Minsk-32 Ryad ES-106 ES-105 ES-103L ES-102 0 1 d Ural-1 I I ,Ryad ES-1010 9 hg Year i Figure 1 Comparison of U.S. and Soviet computer speeds L 16 Kissinger's policy of transferring sophist icated te&nology and its pro ducts to the COMECON nations.

Three'aspects of technology transfer are involved in the export control question 1) transfer of products 2) transfer of manufacturing capa bility (e.g manufacturing and testing equipment or entire factories The third forms a base on which the receiving country.can build its own research, development, and manufacturing capabilities are involved in exporting computer technology to the' Soviet Union and other connnunist-dominated countries IO and (3) t ransfer of design and manufacturing. technology and know-how. i All three aspects In the first type of technology transfer--exporting complete products such as computers, peripherals, and components--the CoCom export restrictions and the relevant U.S. leg i slation (Export Administration Act) resulting in the export control procedures until a year ago applied to computers above a particular processing capability, semiconductors and computer testing and producing equipment. Officially, no computers. in the IB M 360/50 class and larger could be exported to the communist-dominated countries. However exceptions were made as a part of Henry Kissingei's detente policy. For example, about 50 CDC scientific computers have been exported to the Soviet Union and other aB E CON countries. Among them were CDC 6200 and Qx: 6400 computer systems, CDC Cyber 73, CDC Cyber 172 and other computer systems. From IBM, systems 360/50 and 360/65 have been exported, as well as the fourth-.generation machines such as IBM 370/145, 370/155, and 370/158 computer sys tems The export of large computer systems will not only provide the Soviet Union with additiunal computing capacity in the military and nonmilitary spheres but also will enhance Soviet strategic capabilities across the board.

Wher e they will serve the. civilian sector, they will free domestic computers for the military sector. The more serious computer technology transfer starts with licensing the Soviets to manufacture components, peripherals memories, and computers, since this i n volves all the infomation for .the manufacture of these devices (such as drawings, manufacturing techniques tooling specifications as well as assistance in setting up production lines production machinery as well. In other cases, a great deal of technical information is provided to the Soviet ministries by the Western fimq in bidding for a contract and negotiating trade avangements ment has little information about what goes on under such agreements In some cases, such a licensing agreement involves the sa l e of The U.S. govern Another mechanism for transfer of technology is the technical exchange agree ment, involving both the government and private sectors in the West with appropriate ministries in the U.S.S.R. For example, the Soviet Ministry of Science a nd Technology concluded a ten-year agreement with Control Data Corporation of Minneapolis, Minnesota, in October 19

73. The agreement covered cooperation in a wide range of projects, including joint development of a new super comp uter, a joint peripheral manufacturing venture, and a nationwide time-sharing service and software development in the U.S.S.R.c r 17 i Examples of the current rush to make arrangements with the Soviet Union and other COEECON governments include the follow i ng Competition by several sets of U.S. and Western European companies for the contract to install computerized air traffic control systems in the Soviet Union are Sperry-Rand Univac and Raytheon, IBM and Thomson CSF of France, Lockheed Electronics and I'I T, and Selenia of Italy and Saab of Sweden. Univac has been reported to be one of the finalists, with a proposal to install Arts-3 automated radar terminal systems now used in the United States. The other finalist is the Selenia and Saab team.

IBM proposed to use triplex 360/65 computers as a part of its system Competition to install electronic telephone exchange system in the Soviet Union by IT CIT of France, Ericson of Sweden, Siemans of West Gemany, and the Nippon Electric and Hitachi team of Japan. The ITT pro2osal would first sell the ITI' 3200 processor, then license the Soviets for eventual production ITT has also signed a contract to install an electronic telephone exchange in bbscow; the ex port license is still pending Broadlyworded agreements on s cientific and technical co operation with the Soviet State Committee for Science and Technology have been signed with some 30 to 40 U.S. firms Among these are Lockheed Corporation (navigation systems, oceano Participants graphic apparatus, air traffic con t rol systems In (teleconummications systems, electronic and electro-mechanical components General Dynamics (telecommunications, computer operated microfilm equipment CDC (a possible joint venture for the development of an advanced computer and operation of a com puter communications network. The computer would be based on advanced Soviet design and the pro totype would be built in the Soviet Union with CDC assistance Singer Business Machines (exchange of information and joint development in computers and el e ctronic instruments Sperry-Rand (to market Univac products in the Soviet Union e..i 1s Joint development agreements can be signed by a U.S. fimi without approval but the actual exports must be approved A number of U.S. firms have made arrangements to prov i de the Soviet Union with licenses to manufacture Western equipment or with actual turn-key manufacturing plants or are waiting for export licenses CDC has signed an agreement with Romania to produce 1200-card min readers ZSO-card/min punches, and ZOO-line / min line printers and production equipment, and CIX will provide expertise and technical assistance Romania will provide .the capital, plant facilities CDC has an agreement pending to manufacture.100-megabit disk memory units in the Soviet Union Fairchild Camera and Instrument Corporation's agreement to set up a MX, p- type semiconductor micrb-circuit &ufacturing plant in Poland is waiting for an export license. These circuits are regarded as inadequate for high-speed computers, but the con tract also woul d include transfer of any new technology developed in the next five years Westinghouse Electric is constructing a factory in Poland for the production of semiconductor rectifiers Dataproducts, Inc has made an arrangement with Videotron of Hungary to supply line printers; parts of printers would be assembled in Hungary, and Videotron would later be licensed to mimufacture some of the parts.

The adoption by the Soviets of the IBM System 360 design for their Ryad family of computers represents another form of design information trans United States of expertise in transferring Western designs into their own systems--it is rumored that IBM System 370 designs will be implemented in the Soviet wad-2 computers that are now being planned. fer. Such information is re a dily available from manufacturers in the Indeed, the Soviets may be developing a considerable level The United States component and computer industry is very interested in the Soviet and Eastern European iwket, which generates pressure on the U.S. Office of Export Administration to relax restrictions or make ex ceptions. At the same time, there is concern over transfer of technolog cal Soviet military capabilities. That this concern was shared by the U.S.

Congress was evident in the 1974 extension of the E xport Administration Act and in the Defense Authorization Act for FYJ975, both of which con tain provisions authorizing the Secretary of Defense to recomend against progosed exports to a carnrmnist nation if it would increase that nation's military capabi l ities, If the President failed to follow the Secretary's expertise and know-how that, in the long run, may contribute to I recomnendation, Congress, by majority vote within 30 days, could overrule the President tension further strengthened the export cont r ol role of the Defense Department However, the 1976 extension of the Export Administration Act, which was vetoed by former President Ford in September of 1976, already demonstrated a change in the mood of Congress regarding the eqorts to the cornnunist do m inated countries Thus, at the time, the Export Administration Act ex VII. Ineffectiveness of U.S. Controls The U.S. control system is also ineffective because it lacks overall policy guidance. The current U.S. laws and regulations covering technology tran s fer to comist nations amplify the conflict between the need for control and the traditional American free trade posture trolling our exports is the Export Administration Act of 1969 as amended by the Equal Export Opportunity Act of 1972 partment of Comme r ce to control commercial exports to any nation -for reasons of short supply, foreign policy, or national security. The man date affecting exports to comist nations calls for promotion of trade and technology transfer to the maximum degree consistent with n ational security considerations The basic U.S. law con This law authorizes the De U.S. trade policy is generally oriented to trade promtion and minimum replatian, as evidenced by the Trade.Zkpansion Act of 1969 and, prior to that, the so-called Kennedy bu n d negotiations. The United States Senate seems to be encouraging this trend with respect to comist nations recently passed Senate version of the Expoxt Administration Act, S. 69 would eliminate the distinction between comunist countries and non-com munist countries in our export control policy, seeking instead to focus on the national security implications of exports regardless of the country to which they are exported This conflict between ow traditional free trade approach and our need for controls for n a tional security purposes is resolved by the Office of Ex port Administration, Commerce Department, on a case-by-case basis in today's environment. The decision as to whether to grant a license for export of an item on the Comdity Control List is generally subject to interagency deliberations.

The negative impact of today's export controls on commercial trade is not so much due to overly restrictive controls on the transfer of technology and critical products. Instead it is caused by lack of clearly defined objectives and by a control list administration that is excessively concerned with sglitting hairs over prdduct performance specifications and end-use statements overloading the administrative staff, excessive delays in processing The This current practi ce results inlicenses, ambiguities between the U.S. and its CoCom allies, and an almost inpossible burden on enforcement agencies.

Over the past five years, the Department of Defense has not assimed ade demands a cmbersome assessment by selected individual s on an ad hoc basis from Defense Research and Engineering, the service comands, and the intelligence community tirely historical in its perspective, with little definitive policy and few guidelines. As a consequence, there is no coherent policy for contr o l ling current technology. su*resources to define U.S. control objectives. Instead, each review The experience base developed is almost en Clearly, a well-understood national technology transfer policy is necessary to prevent tliese decisions from being m ade on an ad hoc basis without re gard for the overall: impact on the U.S. Such a policy must clarify the primary inportance of technology transfer rather than product transfers.

This change in focus from the present approach would do much to resolve the i nherent conflict between the desire for trade and the need or con trol in a manner consistent with both objectives VIII. Conclusion The Cyber 76 affairs raises the question: are the present administration and the National Security Council comnitted to cur b ing the growing Soviet strategic advancement place a cap on further Soviet strategic expansion, it becomes obviously counterproductive to provide them with a strategic item which will help the Soviet Union to bridge the qualitative gap which exists betwee n them and the United States If the present administration is determined to In 1921, Lenin made the statement, The capitalist countries...will supply us with the materials and technology we lack and will restore our military industry, which we need for our future victorious attacks upon our suppliers.

In other words, they will work hard in order to prepare their suicide."

In 1973, Soviet Consnunist Party boss Leonid Brezhnev defined detente in the following statement made to the members of his Politburo an d commu nist leaders of the Warsaw Pact countries We comrmnists have got to string along with the capitalists for a while. We need their credits, their technology, and their agriculture. But we are going to continue a massive military buildup, and by the m iddle 1980s we will be in a position to return to a much more aggressive foreign policy designed to gain the upper hand in our relationship with the West L.0- r 3 In its study Detente in Soviet Strategy, The United States Defense Intel ligence Agency info rmed the American President in one of the paragraphs 21 Y

1. The Soviets needed Western trade, capital and technology notably advanced US technology not only for economic reasons.

They value economic strength largely for its contribution to Soviet politic al and military power. Thus, the detente policy seeks to facilitate building a powerful operating base from which Soviet foreign policy goals can be pursued The proclaimed objective of eventual victory for "socialism" on a world scale makes it clear that a ny accomodation with the Western powers is a purely temporary ?base while the Soviet Union and its satellites around the world build up their..strength in the advancement of that standing ob jective. It is paradoxical for the West to assist the military m ight of a regime dedicated to the destruction of its creditors.

By Miles M. Costick Consultant r Ii 22 Architecture Cniprocassor l4in iconpu ter ic roconpu tr r Pipeline Processor Array Processor Associative Array I'rocessor z Eiiiltiproccssor Fr J i. r a L L. d Cum pu t e r tiys tclns CoeipuCcr Networks r Table 1 COI4FU'TER ARC t I ITECTLrRES Number of Concurrent Instruction Streams Single Single Single Single Single Single Several Seve ra 1 Several Data S t reams Single Single Single Several Many p Elany Several Sevc r a1 Several 3 Characterfstics The conventional computer organization: iastructiun and data look-ahead features; multiple spscializcd cxocu tion units. Examples: CDC 7600, IKI Hodels 360 and 370 A subclass of uniprocessors characterized by lo u cost small size, simple instruction set. short word lengcli. and small random-access memory. Recently minicomputers have becorrie more sophisticated and ttie distinction between a small "conventional" uniprocessor and a minicomputer I ies more in price t han in perforniance. Example: Dixital Equipment Corporation's PI)I' 11 A "bare-burlt!s" Iirocessor on il single circuiL card or even on o single chip. Sinplc iiistructinn set, shorr -Lord length, very low price. r\\ yplisations In tiniid-lielcl calcu lators. Elfcroconputer capah11 i ty is also inprovfng riid npproaches rliat of early minicomputers. Example: Intel aoao A subclass of uniprocessors fu which specialized execution units can he arranged in taiidrm (pipeline) for the appli cation of a sequence of o: \\

ralions to a data streirn. T*.o or three pipelines may be in operation concurrentl Ex amples: CDC STAR,.Tcxas Instrument'' AX A large array of rehtive1.y ifmple processors exeiuting the same instruction on many data streams; various data flu patterns bet ween r.lumeiits Examples: 1I.I.IAC IV (66 ele ments Bell Laboratorie'' PEFE (16 elencrits A subclass of array''processors in which ~liu N processors arc simple serial-by-bit wits associated wltti ttie (super words of an associative memory. Example: ST,\\ Kr\\RI\\:I IV Identical multiple processors operate concurrcntly on in structions and data from a comrnun main memory. Examylcs Burroughs B 6700, UNIVAC 11


Miles M.