<<

International Perspectives Seymour Goodman

The Origins of Digital Computing in Europe Retracing the paths of influential, but often isolated, pioneers.

he mainstream of the Cold War. For over two decades, The MESM was based on an early history of comput- he actively combined a position as original design that used a fixed- ing is generally ascribed chief designer and CEO-equiva- point representation and a three- T to British and American lent in a superpower country, a address command format. At efforts during the 1940s–50s. A long-term achievement unique acceptance, its average speed was number of European pioneers among early computer scientists. 50 operations per second, with a were disconnected from this primary component base of mainstream by time, most singu- approximately 6,000 larly , or by a vacuum tubes; it was combination of geography, poli- improved and used con- tics, and language. The latter tinuously until 1956. As include Konrad Zuse, now was the case with the ENIAC recognized for building the and other early machines, in first binary, digital com- addition to its in-house users, puter anywhere, and a teams of scientists working on number of other Euro- problems of national importance peans who built the first would visit to use large quantities in their respective of computing time—appreciative countries [4]. users included many world-class S.A. Lebedev, whose centenary mathematical scientists of the era. was last November, was one of the In 1956, the MESM was moved most accomplished, but least to the Kiev Polytechnic Institute known, of these “isolated” pio- Ukraine and the MESM where it was used for three years neers. Between 1947–51 he The MESM (Small Electronic to train young programmers. It designed and built the MESM, Calculating Machine) was was later scrapped for parts, and the first operational all-electronic, designed and constructed only a few pieces, easily fitting digital, stored-program, general- by Lebedev and a small group into a shoebox, remain today. For purpose computer on continental of coworkers at the Institute more extensive details on the tech- Europe, and did so under extraor- of in nology and history of the MESM dinarily difficult circumstances. Kiev. On November 6, 1950, it and other Lebedev machines, see He was responsible for more than solved its first simple problem [2, 6]. 15 subsequent models that were (Maurice Wilkes dates May 6, Perhaps the most incredible used to work on many high-prior- 1949 for the British EDSAC), aspect of the MESM is that it was ity problems in the Soviet Union, and accepted for full operation successfully built at all. No all- including the machine that came by a high-level commission of electronic computer was ever built closest to closing the “computing the Academy of Sciences in under more difficult conditions.

PAUL WILEY gap” with the West during the late 1951. By the time construction of the

COMMUNICATIONS OF THE ACM September 2003/Vol. 46, No. 9 21 International Perspectives

Lebedev was a leader of the computer scientists strongly favoring the development and use of an architecture of their own for an upward-compatible family.

MESM began, Ukraine had been of what their British and Ameri- However, things got really tough hit particularly hard by Stalin’s can counterparts had. Then there for Zuse when the Allies started forced collectivization during the were the problems with the quality winning, and bombing destroyed 1930s, when millions died from (reliability and tolerances), quan- his program-controlled relay starvation and terror. Recovery tity, and the actual delivery of computer in 1944. To avoid the had not gone far when Hitler’s Soviet-made vacuum tubes and bombs and approaching Allied armies arrived in 1941. The Nazi other components. Lebedev also (especially Soviet) armies, Zuse military pounding, a savage occu- had to make a successful case took his nearly completed general- pation, the reconquest by the Red “from below” to the various gov- purpose relay computer, the , Army, and Stalin’s brutal reasser- ernment, Academy, and Commu- on a refugee’s trek from , to tion of political control from 1945 nist Party authorities for the Gottingen, to a small town in All- until his death in 1953 left support of a new technological gau/Bavaria, and finally to the aca- Ukraine in terrible shape. initiative. demic haven of the ETH in Lebedev arrived in Kiev in Of all the early computer Zurich, , where he got 1946 and decided to build a digi- builders, only Konrad Zuse’s tra- it working and into use by 1950. tal computer based on vacuum vails might be comparable to tubes. He assembled a small, Lebedev’s. Zuse built electro- Russia and the BESMs young, group of 20–30 people mechanical computers in Hitler’s In late 1948, the Institute of Pre- (about 25% women) [6]. For Berlin [1, 4, 5], notably the , cision Mechanics and Computer many, their backgrounds would the world’s first operational binary Technology (ITMVT) was cre- read something like: undergradu- computer. Berlin was thriving just ated in Moscow. In 1953, Lebe- ate education as a radio engineer before and in the early years of the dev moved to Moscow as the finished or interrupted in 1941; war, when Zuse first worked on ITMVT Director, a position he spent the next four years surviving his machines. His primary prob- held with distinction until he the war. lem was that he could not make died in 1974. By early 1949, they were ready the case for financial and other At ITMVT, Lebedev led the to start building their computer, support to the Nazi government, development of approximately 15 using the abandoned former Feo- which apparently thought the war computer models [3, 6], several of faniia monastery outside of Kiev as would end quickly and favorably which went into production and their laboratory. The monastery and did not want to expend formed the core of his country’s was severely damaged during the resources on Zuse’s longer-term high-performance computing for war, and could not be reached via ideas. He started building com- many high-priority users into the a paved road. More generally, the puters where he could, including 1990s. Although he never again support infrastructure––heat, elec- his parents’ living room, but con- had to work under the conditions trical power, phones, and con- ditions were still much better than at Feofaniia, designing, building, struction materials––fell far short they would be later at Feofaniia. and serially producing a computer

22 September 2003/Vol. 46, No. 9 COMMUNICATIONS OF THE ACM in the USSR was never easy. maintenance, and 8% on unex- chief executive. In the USSR, this The most important of the gen- pected problems, including the existed on a larger scale than else- eral-purpose computers bearing time required to repeat lost calcu- where, sometimes including exten- Lebedev’s signature include the lations. sive facilities, hundreds or BESM, the BESM-2, the M-20 By the mid-1950s, the USSR thousands of people, and dealing and the transistorized versions M- had a need for a substantial num- with the outside world in ways 220/222, and the BESM-6, a ber of computers. In addition to that resemble a U.S. national labo- machine with a remarkable history Bazilevskii’s Strela, it was decided ratory. Some of the best known of that still deserves to be told in full to have a BESM production these institutions were built [2, 7]. He was also responsible for model. Two competing models around I.V. Kurchatov (nuclear), several machines made specifically arose, the BESM-2 and the S.P. Korolov (rocketry), M.V. for military applications. M-20, both of which were in serial Keldysh (applied mathematics), The BESM (High-Speed Large production by 1959. Ultimately and A.N. Tupolev (aviation). Electronic Calculating Machine) the Strela and BESM-2 faded Lebedev at ITMVT was such an was a floating-point computer that away, but a large number of M- “institution” and in this way had was extensively tested in 1952, 20s and its transistorized succes- great influence as a designer, and formally accepted by a state sors the M-220/222 were teacher, and developer of technol- commission in April 1953. Its ini- produced into the late 1960s. ogy that went out into the real tial speed was only about 1,000 After what effectively amounted world, including to most of the operations per second, largely to experimentation with a few institutes just noted. because of Lebedev’s inability to operational models (BESM-3, In some ways, Lebedev and the obtain a sufficient supply of CRTs, BESM-4, Vesna), Lebedev and his BESM-6 mark the end of an era. which were going in greater num- team developed the BESM-6 dur- As was the case in the U.S., Great bers to his competitor Bazilevskii ing 1964–67. It went into serial Britain, Germany, and other coun- and his Strela computer. In 1954, production in 1967; about 350 tries, the early development and Lebedev forced a formal direct were built during more than two application of much of the digital comparison of the BESM vs. decades, and quite a few were still computing in the USSR was dri- Strela, which he readily won. This in use in 1991, the year the USSR ven by defense and other govern- result secured the necessary CRT abolished itself. The speed of the ment needs. As long as the supply, and BESM performance BESM-6 approached that of the Americans were similarly driven, jumped to 7,000–8,000 opera- then-contemporary U.S.-made ITMVT and other Soviet R&D tions per second. CDC 6400, winning Lebedev and and production facilities were able In 1955, Lebedev presented the ITMVT a 1969 State Prize (for- to adequately meet national needs. BESM to an international com- merly Stalin Prize), at least partly The East-West computing puter science conference in Darm- because of the comparison with “gap” opened rapidly when a stadt, West Germany, where it was the U.S. greatly expanded range of com- apparently recognized (including In the USSR, some leading aca- mercial and other applications by U.S. intelligence) as one of the demicians effectively built institu- took over as the primary drivers of most powerful indigenously built tions around themselves. The the U.S. computer industry, accel- computers on the European conti- institute director was an active sci- erating development and the infu- nent. By 1956, it had an experi- entist, led a group of other scien- sion of better technology into both mental ferrite core memory. The tists and students, and determined the civilian and national defense BESM was often used 24 hours a the main line of work, procured sectors. As the conditions of these day, with an uptime of approxi- funding, space and other sectors in the USSR and Eastern mately 72%, 20% in preventive resources, and functioned as a Europe became of increasingly

COMMUNICATIONS OF THE ACM September 2003/Vol. 46, No. 9 23 International Perspectives

critical concern, the political and pioneers are in order. Arguably and influence elsewhere, Zuse did governmental leaderships of these the two most important criteria not. At least he did not suffer the countries (Bulgaria, Czechoslova- for consideration should be origi- fate of many German scientists kia, East Germany, Hungary, nal technological contributions and engineers who were rounded , and Romania—most of and influence on others. up by the Soviets, although one which had started building their Babbage’s originality stands in could speculate on what might own computers in the 1950s) con- magnificent isolation. The extra have happened if he had been and cluded that their own institutions century and color he gave to the assigned to Lebedev. Lebedev was could not deliver the kind of com- roots of computing guaranteed an outstanding mentor and unbi- puting they needed. The bottom- everyone interested in its history ased manager of bright, hard line decision was to try to would embrace him. It also helped working, young scientists. A lot substitute IBM’s solutions for considerably for him to be English, would have depended on Zuse’s what was lacking indigenously. and to have a young female profes- ego and attitude toward working From the standpoint of Lebedev sional consort who liked to write with and under relatively inexperi- and this column, this meant that about his ideas. His substantive enced Russians and Ukrainians. large parts of the Soviet and East influence is another matter, and it Lebedev has no claim to high- European computing communi- is not clear whether the other pio- profile worldwide “firsts,” ties were mobilized, in some cases neers—isolated or not—knew of although there was some paral- against resistance, for this effort, him in their early working stages lelism in his early designs that known as the Unified System (Zuse explicitly denied it) or might should merit further attention. (Edinaia Sistema, ES), based on have been only minimally influ- Furthermore, his legacy for cre- the functional duplication of enced since his designs and compo- ativity suffers from Western suspi- IBM’s S/360 architecture [3]. nent technologies were not cions that, as was likely the case Lebedev was a leader of the seriously studied until the 1970s with other Soviet technological computer scientists opposed to [4, 7], and were necessarily so dif- achievements ranging from the this decision, strongly favoring the ferent from theirs. atomic bomb to the “Buran” space development and use of an archi- Zuse’s highest profile claims to shuttle, his achievements benefited tecture of their own for an technical firsts are for the binary from the fruits of Soviet intelli- upward-compatible family. They electromechanical computers built gence collection. Soviet intelli- lost this battle for the mainframes, in Berlin by the early 1940s. His gence did collect against computer and ultimately for minicomputers work was so confined with so little technology in the U.S. and West- as well. He died as the ES pro- impact that Allied intelligence did ern Europe and Lebedev would gram was ramping up into full not seem to notice it during the have been an obvious recipient, swing, perhaps taking some satis- war, nor become interested after- but we do not know what he faction in its initial difficulties. ward. Zuse and his central Euro- received. ITMVT scientists continued to pean supporters had to campaign Architectural and implementa- have license to pursue their own later to get recognition for his tion details aside, the composite designs for supercomputers after originality. He continued to work concept of a binary, all-electronic, Lebedev died; the best known of as a computer scientist, and digital, stored-program, general- these were the El’brus machines started a not-too-successful com- purpose computer using a sequen- under V.S. Burtsev [8]. puter company, but his later tial fetch-execute cycle with I/O to efforts were not very influential. the outside is hardly so self-evi- Places in History? Although a few other creative Ger- dent that any isolated person Some observations on the signifi- man scientists like von Braun wanting to automate arithmetic cance in history of the isolated went on to post-war achievements calculations would independently

24 September 2003/Vol. 46, No. 9 COMMUNICATIONS OF THE ACM think of it. From the standpoint of engineering. His subsequent government essentially said it did historical assessment, Lebedev had machines demonstrate this was not not believe he and his fellow the disadvantage of starting after a fluke. He was an exceptionally Soviet computer scientists could the Anglo-Americans, whose work capable computer scientist who deliver what was most needed any- was being described in places like deserves recognition for a string of more. c an article on the ENIAC in the technical achievements—perhaps May–June 1946 issue of the the longest and most sustained of References and Notes American Army Ordnance maga- any pioneer—although they were 1. ACM. The Machine that Changed the World. Five-part video series produced by ACM and zine, something likely to have geographically confined. He shown on PBS during 1990–91. been collected by Soviet intelli- labored for a distinctive, long-term 2. Crowe, G.D. and Goodman, S.E. S.A. Lebe- dev and the birth of Soviet computing. IEEE gence. There were also reports on national capability, and fought Ann. 16, 1 (Jan.–Mar. the stored-program ideas and against the wholesale adoption of 1994), 4–24. machines (EDVAC, EDSAC). We technology from abroad. 3. Davis, N.C. and Goodman, S.E. The Soviet Bloc’s unified system of computers. ACM may never know if any of this Finally, it might be noted that Computing Surveys 10, 2 (1978), 93–122. influenced Lebedev’s initial grasp all three men tried to build com- 4. IEEE Annals of the History of Computing has published many articles on Babbage, Zuse, of the composite concept. Given puters to serve recognized national and the history of computing in several East how hard he worked to make a needs, having no choice but to and West European countries. case for the MESM, it is clear pursue their aims through their 5. Lee, J.A.N. Konrad Zuse. In People and Pio- neers; ei.cs.vt.edu/~history/people.html. that, in contrast to other technolo- governments, with mixed results 6. Malinovskii, B.N. Materiali Pro Stvorennia. gies like the atomic bomb or and all ultimately ending in rejec- Copies of historical material collected for guided missiles, the computing tion. Babbage received initial Lebedev’s centenary commemoration, Kiev, Nov. 14–15, 2002. Malinovskii has written effort was not pushed “from funding for his difference engine, prodigiously on Lebedev and other Soviet above” in response to pressing but he did not deliver a working computer scientists. See www.icfcst.kiev. ua/museum/Lebedev.html. competition from abroad. It is machine, and his ambitions for 7. Swade, D. Private communications, Nov. possible that Lebedev thought of the went beyond 26–Dec. 4, 2002. Swade has worked exten- the concept on his own, saw the what the British Admiralty would sively on Babbage, and obtained a complete BESM-6 for the National Museum of Science information on Anglo-American support. Zuse built some small and Industry in London in 1993, “with a view efforts, and used it to make his machines such as a special-purpose to restoration to working order.” 8. Wolcott, P. and Dorojevets, M.N. The Insti- case with the authorities. When it relay computer for a factory mak- tute of Precision Mechanics and Computer came to defense-related technol- ing remote-controlled bombs, but Technology and the El'brus Family of High- ogy, an argument to the effect that the German government would Speed Computers. IEEE Ann. History of Com- puting 20, 1 (Jan.–Mar., 1998), 4–14. “the Americans are doing it” often not support his more advanced Wolcott has also studied other post-BESM-6 helped get attention and support. ambitions. Although it was an high-performance computers in the USSR. None of Lebedev’s designs was uphill battle all the way, Lebedev based on close copying of foreign was remarkably successful with his machines and, given some funda- government—surprisingly so since Seymour Goodman (goodman@cc. gatech.edu) is a professor of International Affairs mental differences (such as working in general it was the least support- and Computing at the Georgia Institute of with Soviet-made components), ive of lower level people showing Technology. what he might have gotten from initiatives—for more than a quar- Permission to make digital or hard copies of all or part of abroad would have been of limited ter century, but ultimately was to this work for personal or classroom use is granted without use. Anyone who appreciates what be rejected in a way that probably fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this it took to build the MESM around greatly disappointed him. In the notice and the full citation on the first page. To copy oth- erwise, to republish, to post on servers or to redistribute 1950 in Lebedev’s circumstances cases of the first two men, the to lists, requires prior specific permission and/or a fee. must respect the effort as a world- rejection was largely of personal class achievement of computer ambitions; in Lebedev’s case his © 2003 ACM 0002-0782/03/0900 $5.00

COMMUNICATIONS OF THE ACM September 2003/Vol. 46, No. 9 25