In 1985, in a Moscow office, a 28-year-old researcher sat before a computer that could not draw a picture. The machine, an Electronica 60, displayed only letters, numbers, and punctuation. So when Alexey Pajitnov wanted a small square to fall down the screen, he built it from two square brackets sliding down a well ten columns wide and twenty rows deep. The game he wrote that way would eventually sell more than 520 million copies.

He called it Tetris. And the machine on his desk was itself a deception. The Electronica 60 was an unlicensed copy of Digital Equipment Corporation’s PDP-11, built in Maynard, Massachusetts. The most famous program ever written in the Soviet Union ran on a computer the Soviet Union had copied from America.

Under Soviet law, the program did not belong to the man who wrote it. It belonged to the state. That outcome would have seemed unthinkable to the engineers who started it all. In 1950, in the grounds of a former convent at Feofaniya, outside Kiev, a small team had switched on a computer of their own, built from roughly 6,000 vacuum tubes, designed from first principles, copied from nobody.

For the next two decades, Soviet engineers built machines that were strange, stubborn, and sometimes brilliant. One counted in threes instead of twos. One completed its share of Apollo-Soyuz arithmetic faster than the Americans did theirs. The question the history demands is simple: how does a country go from building its own computers in a convent to writing its masterpiece on a copy?

The man who switched on that first machine was not a computer scientist, because in the Soviet Union of the 1940s no such field existed. Sergey Lebedev was born in 1902 in Nizhny Novgorod and trained as an electrical engineer. His specialty was power systems, the giant grids that carry electricity across continents. In 1939, he earned his doctorate for a theory of artificial stability, the problem of keeping those grids from tearing themselves apart.

It was unglamorous work, but it was an education in feedback and control, the exact ideas a computer is built from. The war took his mathematics to the front. Lebedev directed work on a stabilization system that kept a tank’s gun aimed while the vehicle rolled over broken ground, and on automatic guidance for missiles launched from aircraft. To solve the differential equations behind those systems, he built analog computers, machines that modeled problems with voltages and currents rather than numbers.

He came out of the war knowing two things: calculation could be automated, and the analog approach had limits he could already feel. The country he returned to had enormous needs and very few tools. The Soviet Union left the war with an atomic bomb program, a rocket program, and a vast territory to defend. Every one of those efforts ran on arithmetic, most of it still ground out by people working mechanical calculators, row after row, checking each other’s figures.

There was a second problem, and it was not technical. The science that described what Lebedev wanted to build was politically suspect. In 1954, an official Soviet philosophical dictionary called cybernetics a reactionary pseudoscience that had appeared in the USA after World War II. Lebedev did not argue philosophy.

He called his project a calculating machine, with the stated purpose of testing the principles of universal digital computers and meeting the government’s need for computation. He moved to Kiev to run a laboratory at the Kiev Institute of Electrotechnology, which found him a building outside the city: a two-story house that had once been the hostel of a convent at Feofaniya. The machine they built there was called the MESM, a Russian abbreviation for small electronic calculating machine. It used about 6,000 vacuum tubes, drew 25 kilowatts of power, stretched eight to ten meters along the walls, and stood about two meters tall.

On a good day, it could perform about 3,000 operations per minute. Six thousand tubes meant six thousand glowing filaments, each a heater that could fail, in a building never meant to house a machine. The team traced faults tube by tube by hand. On 6 November 1950, the MESM made its first test run.

On 4 January 1951, it performed its first useful calculations. On 25 December 1951, it passed official government testing. A country with no computer industry, no university course in computing, and a dictionary that called the underlying science a capitalist fraud now had a working stored-program computer built entirely by its own people. Lebedev did not stay to admire it.

In 1953, he became director of the Institute of Precision Mechanics and Computer Engineering in Moscow, a post he held for the rest of his life. That April, his first large machine, the BESM, was accepted into operation. Its successor, the BESM-2, went into series production and was used to calculate satellite orbits and the trajectory of the first rocket to reach the surface of the moon. A rival Moscow design called the Strela was built in a run of just seven machines for the most important state customers.

The Soviet Union now had computers in the plural. One of the people sitting at a Strela, a young military engineer named Anatoly Kitov, was beginning to think these machines should not merely calculate trajectories. He thought they should run the country. Kitov had something Lebedev lacked: a talent for argument.

As the first user of the Strela inside the Soviet military, he had seen what a computer could do for an institution, and he knew the word cybernetics was still a liability. In 1955, with two mathematicians, he published an article titled “The Main Features of Cybernetics,” laying out the ideas plainly for a Soviet audience. The ban did not survive it. By 1961, the field once dismissed as American pseudoscience was being promoted in an official volume titled Cybernetics in the Service of Communism.

Kitov had won the argument about ideas. He then made the mistake of trying to win an argument about power. In the autumn of 1959, as director of the Ministry of Defense’s Computing Center No. 1, he sent a report to the Communist Party’s Central Committee and to Nikita Khrushchev himself.

It became known as the Red Book. In it, he proposed a single national automated management system, a network of computer centers serving both the armed forces and the civilian economy, sharing one infrastructure. He also included sharp criticism of how slowly the Ministry of Defense was adopting computers. The generals did not want to share their machines with civilian planners, and they did not enjoy being criticized by a subordinate.

The commission accused Kitov of pursuing private interests and disgracing the ministry’s leadership. By the middle of 1960, he had been expelled from the Communist Party and dismissed from his post. He was not imprisoned. He kept working, moved into civilian research, and decades later chaired a department of computing at the Plekhanov Institute in Moscow.

The first proposal for a national computer network in the Soviet Union cost its author his career, and he went back to work. While Kitov was being disciplined in Moscow, two of the most unlikely engineers in the Soviet Union were rising in Leningrad. They were Americans. Joel Barr was born in New York in 1916 and studied electrical engineering at City College alongside a classmate named Julius Rosenberg.

In 1940, he went to work at the Army Signal Corps Laboratories at Fort Monmouth, New Jersey. In 1941, a Soviet intelligence officer recruited him as a spy under the code name Meter, and Barr brought in his friend Alfred Sarant, who worked on secret military radar. Together they handed Moscow more than 9,000 pages of documents covering more than 100 American weapons systems. When the Rosenberg network unraveled in the summer of 1950, both men ran.

Barr vanished from his Paris apartment the day after David Greenglass was arrested, leaving his belongings behind. Sarant fled through Mexico, Guatemala, and Poland. By 1951, they were together in Prague with new names: Joel Barr became Joseph Berg, and Alfred Sarant became Philip Staros, with an invented Canadian past. In Czechoslovakia, they led the design of an automated anti-aircraft fire control system, and their handlers rated them the most productive members of the old ring.

In 1956, they were moved to Leningrad and put in charge of a military electronics research institute. They were exiles, traitors to one country and outsiders in the other. But they were also two engineers trained in American laboratories, carrying an American instinct for small teams, fast prototypes, and miniaturization. What they needed was a patron strong enough to let them build something enormous.

In the spring of 1962, he walked through their door. Khrushchev toured the institute, and the Americans made an audacious pitch. The Soviet Union, they argued, needed its own version of the cluster of semiconductor companies growing around Stanford in California: a whole city devoted to microelectronics, where research, design, and production lived side by side. Khrushchev approved.

The city already existed, barely. A new town northwest of Moscow had been founded in 1958 with plans for textile production. In 1962, those plans were scrapped and the town was handed to electronics. It was named Zelenograd, the Green City, and it became the Soviet Silicon Valley, a closed town where photography was restricted and foreigners were not welcome.

1962 was the year everything seemed to move at once. That year, a young engineer named Yuri Osokin produced the first Soviet integrated circuit. In Kiev, a mathematician named Victor Glushkov founded the Institute of Cybernetics of the Ukrainian Academy of Sciences and picked up Kitov’s buried idea, making it bigger. Glushkov proposed a national automated system for computation and information processing: a three-tier network with a main computer center in Moscow, up to 200 mid-level centers in other cities, and up to 20,000 local terminals linking the planned economy in something close to real time.

In Washington, people noticed. Arthur Schlesinger Jr. , special assistant to President John F. Kennedy, warned the president that the Soviet commitment to cybernetics gave Moscow a tremendous advantage and that without an American program to match it, “we are finished.

”

This was the pivot. The Soviet leadership had decided at the very top that computers and microelectronics were a strategic priority and was pouring money, land, and people into them. There was a city, an institute in Kiev, a network plan, and a premier who believed. But there was one more detail worth setting down.

Alexander Shokin, the official in charge of Soviet electronics, had his own preferred method for catching up with the Americans: get hold of an American chip and copy it. For the moment, it hardly mattered whose method won, because Staros and Berg had the one thing that mattered in the Soviet system: they had Khrushchev. They would keep him for about two more years. In October 1964, Khrushchev was removed from power by his own colleagues.

For most Soviet institutions, it was a change of leadership. For Philip Staros and Joseph Berg, it was the end of their protection. Their patron was gone, and with him went their status, their budget, and their position at the center of the microelectronics program. Over the following years, their proposal for a compact chip factory was turned down again and again.

That same year, on the other side of the world, a company made a bet that would quietly decide the Soviet future too. On 7 April 1964, IBM announced the System/360. It was six processor models and 54 peripheral devices, developed over four years at a cost of about $5 billion, a sum so large it put the whole company at risk. The engineering inside each machine was less important than the promise across all of them: software written for one model would run on every other model in the line.

For the first time, a customer could buy a computer as a platform and keep every program they had paid to write. The value of a computer was moving out of the hardware and into the software. The Soviet Union was not allowed to buy into that world. The Western Allies and Japan ran an export control committee that embargoed more than 500 categories of goods from the Soviet bloc, and advanced computers sat near the top of the list.

Whatever Soviet engineers wanted from the West, they would have to build themselves, acquire through third countries, or steal. Then the two halves of the Cold War ran an experiment in networks almost side by side. On 29 October 1969, a computer at the University of California, Los Angeles, sent its first message to a computer at the Stanford Research Institute. It was the first link of the ARPANET, the American defense research network that would grow into the internet.

Eleven months later, Victor Glushkov’s request for funding for his national network reached its final hurdle. His plan had been estimated to cost no less than 100 billion rubles. It threatened every ministry that controlled its own data, and above all it threatened the Ministry of Finance, whose head saw the network as a rival to his own authority. On 1 October 1970, the funding was denied.

Twenty thousand terminals had been planned. None of them were built. The Americans had started with two computers and a phone line, and they had kept going. The Soviet Union had planned the entire system at once, and the bureaucracy had killed it whole.

On paper, Soviet hardware was still holding its own. In a Moscow institute, Sergey Lebedev’s team was finishing the computer that would prove one last time exactly what Soviet engineers could build when nobody told them to copy. The BESM-6, the last great machine of Lebedev’s life, had its design completed in 1965 and production began in 1968. It had a 48-bit word, memory of up to 32,768 words, and a clock of around 10 MHz.

It ran two separate instruction pipelines, one for the control unit and one for the arithmetic unit, so the machine could prepare the next instruction while still finishing the last one. The result executed about one million instructions per second. Britain’s Science Museum places it in the same class as the CDC 6600, the American supercomputer built by Seymour Cray. It was not a desktop object.

A BESM-6 installation drew 30 kilowatts, occupied between 150 and 200 square meters of floor space, and cost 530,000 rubles. Over 19 years of production, roughly 350 were built. Its finest public hour came in July 1975 during the Apollo-Soyuz test project, when an American and a Soviet spacecraft docked in orbit. Both sides processed the mission’s data on the ground.

The Soviet team working on a BESM-6 completed its data processing half an hour before the Americans finished theirs. The second machine is stranger and its story sadder. In 1956 at Moscow State University, the mathematician Sergey Sobolev asked a young engineer named Nikolay Brusentsov to lead the design of a small computer for the university. The machine they built was called Setun, after a small river near the university, and it did not use binary.

Every computer you have ever owned counts in twos, each digit either zero or one. Setun counted in threes, in a system called balanced ternary, where each digit can be minus one, zero, or plus one. That choice was not a stunt. In balanced ternary, negative numbers need no separate sign because the sign is built into the digits themselves, and rounding can be done by simply cutting off the trailing digits.

Donald Knuth, the American author of The Art of Computer Programming, would later call balanced ternary “perhaps the prettiest number system of all. ” Setun was cheap, and it worked. Fifty were built at the Kazan mathematical plant between 1959 and 1965, and thirty went to Soviet universities. Then the plant refused to continue.

The reason recorded is almost unbelievable: the machine’s sale price was too low for the factory to bother with it. The binary computer that replaced it performed about equally well and cost two and a half times as much. The original Setun prototype was destroyed, and Brusentsov’s laboratory was moved into rooms in a student hostel. The third machine came at the end of the golden years.

Elbrus-1, completed in 1979 under the designer Vsevolod Burtsev, was a superscalar computer with out-of-order execution techniques that let a processor work on several instructions at once and rearrange them for speed. In the West, those ideas would become the basis of mainstream microprocessors in the following decades. Three machines, three original ideas, built by engineers working under secrecy, embargo, and chronic shortage. And yet during the very same years that the BESM-6 was entering production, a different argument was being settled in ministry offices in Moscow.

The question on the table was whether to keep building Soviet computers at all. Anatoly Kitov, by then long out of favor, had a view too. His son would later recall the way his father put it: if you copy something, you always go behind, and as a consequence, you fall behind. The designers lost.

At the end of the 1960s, the Soviet government chose to build its next generation of mainframes as copies of the IBM System/360. Work on the clones began around 1968, and the project was carried out with the Warsaw Pact partners. The program was called the Unified System, and its machines were known as Ryad, meaning row or series. Bulgaria, Hungary, Poland, Czechoslovakia, Romania, and East Germany built computers for it.

The operating system was a modified copy of IBM’s own OS/360. According to one history of Soviet technology smuggling, the first Soviet machine modeled on the IBM design was based on a System/360 Model 40 brought into the country through Poland. The first Unified System machines appeared in the early 1970s. There are two honest ways to read that decision.

The first is that it was rational. IBM had spent years and billions of dollars building not just a family of machines but a library of software for them, and Keldysh understood how enormously difficult it was to build an operating system from scratch. The Soviet economy was short of programmers, short of software, and short of time. IBM had published a detailed description of the System/360 architecture, which made a compatible copy possible.

Copying the most successful computer family in the world promised instant access to a mountain of proven programs, one standard across the whole socialist bloc, and an end to every institute building its own incompatible machine. The second reading is that it was a surrender. Funding and talent flowed toward the clones and away from the schools that had built original machines, and both Lebedev’s institute and the Elbrus line were badly damaged. The country tied its future to a design that was already several years old and would never be its own.

Copying requires an original, which means the copy always arrives after it. Each Unified System model followed its IBM counterpart by about eight years. During the five-year plan from 1971 to 1975, the industry produced only ten to fifteen percent of the Unified System computers it had planned. Eventually, more than 15,000 of the machines were built, which made the program a production success and a strategic dead end at the same time.

Sergey Lebedev stepped down as director of his institute in 1973 and died in Moscow on 3 July 1974 at the age of 71. The man who had built the MESM by hand died just as his country decided that it no longer needed people like him to design its computers. And the method that won in the mainframe ministries did not stay there. It moved down into minicomputers and into the chip itself.

At a factory in Voronezh, from 1978 until 1991, the Soviet Union produced the Electronica 60, the unlicensed copy of Digital Equipment’s PDP-11 that would one day sit on Pajitnov’s desk. The 1980s made the gap visible to everyone. The West tightened its export controls again in 1979 and again in 1981, down to rules so fine that an 8-bit microcomputer could be exported to the Soviet bloc, but a 16-bit one could not. Inside the bloc, a cheap personal computer was seen as a political risk as much as a technical goal.

Poland’s propaganda ministry warned that the mass spread of computers was creating a variety of ideological endangerments. The Soviet answer to the personal computer arrived late and expensive. The Agat, introduced in 1983, was built around the same processor family as the Apple II, aimed at schools, and cost as much as twenty times the average monthly salary. The cheaper Electronika BK-0010 followed in 1985, a 16-bit machine costing more than four months’ pay for a computer whose keyboard was a flat panel of switches with no key caps.

In March 1985, Mikhail Gorbachev became general secretary, and one of his first announced initiatives was a program to bring computer literacy into Soviet schools. By then, the military already understood how far behind the country was. Marshal Nikolai Ogarkov, chief of the Soviet general staff, had put it plainly in 1983: modern military power is based upon technology, and technology is based upon computers. The country that had planned 20,000 network terminals in 1962 was now trying to buy enough desktop machines for its classrooms.

Then the Soviet Union itself came apart, and the industry went with it. After the breakup in 1991, nearly every Soviet computer manufacturer ceased operations. Institutes lost their budgets overnight. Engineers who had spent their lives on classified machines found their skills were valuable mainly to the companies they had once been told to copy.

In May 2004, Intel announced a research agreement with two Russian design firms, Elbrus and Unipro. Key researchers from both companies moved into Intel’s Moscow and St. Petersburg operations, and Boris Babayan, the man who had led the later Elbrus designs, became an Intel fellow. The descendants of the Soviet supercomputer program were now working for Silicon Valley.

The final twist came almost two decades later. Modern Russian processors, including the Elbrus and Baikal chips, were designed in Russia but manufactured by TSMC in Taiwan on a 16-nanometer process. Russia’s own factories could reach only about 90 nanometers. When sanctions followed the invasion of Ukraine in 2022, that supply line was cut.

The country that once built its own computers from scratch now could not manufacture the processors it designed. And yet the machines themselves have outlived the state that commissioned them. In 1992, one of the last surviving BESM-6 machines was acquired by the Science Museum in London, where it stands today in the information age gallery. A few of the MESM’s original vacuum tubes survive in the archive of the Kiev House of Scientists.

The Setun prototype is gone. The home computers found a second life their designers never planned for: in the 1990s, the Electronika BK-0010 became a favorite of the demo scene, where programmers compete to make old hardware produce graphics and music it was never meant to manage. And Tetris escaped. Pajitnov’s game spread through Moscow’s computing centers, and then a 16-year-old programmer named Vadim Gerasimov, working with Pajitnov, ported it to a Soviet IBM-compatible personal computer, adding color and a scoreboard.

From there it copied itself floppy disc by floppy disc across the Eastern bloc. Under Soviet law, the rights went to the computing center, and Pajitnov received nothing from the early commercial versions. Only at the end of 1995 did the rights revert to him, and the Tetris Company was formed to manage them. The version for Nintendo’s Game Boy alone sold 35 million copies.

Everything around those brackets on Pajitnov’s screen was borrowed, licensed, or confiscated. The idea inside them was his alone. That is the part the ministries never understood. The Soviet Union was never short of original minds.

Sergey Lebedev built the MESM when there was nothing to copy. Nikolay Brusentsov built Setun in threes when the rest of the world counted in twos. Lebedev’s team built the BESM-6 to stand in the same class as the fastest American machines. Victor Glushkov designed a national computer network before the ARPANET sent its first message.

Then the decision came down to copy, and the copies were always eight years behind the originals. The Soviet Union did not lose the computer race to Silicon Valley. It stopped running and started copying.