Пятнадцать байтов на стек от конца

In 1986, the Soviet Union had slightly more than 10,000 computers. The Americans had 1.3 million. At the time of Stalin's death, the Soviet Union was the world's third most proficient computing power. But by the 1960s, the US-Soviet computing gap was already years long. Twenty years later, the gap was undeniable and basically permanent. Why did this happen? The Soviet state believed in science and industrial modernization. Support for research & development and the hard sciences were plentiful. They had the country’s finest minds. Goodness gracious, they launched Sputnik! They landed on Venus! How did it come to this?

ITMVT and the Calculator Girls

In 1948, news arrived in the Soviet Union about the ENIAC - the first general purpose, programmable computer - having been built two years earlier in the United States. The Soviets knew little about how the ENIAC worked, other than that it used vacuum tubes. And many dismissed the news as simply propaganda. But they could not help but notice that the West was putting a lot of resources in the space. The Soviet government did need to improve its calculating workflows. For instance, the nuclear weapons effort required 100+ young calculator girls working daily 8-hour shifts doing nothing but calculations. Administrators were concerned about finding enough people to do error free calculations of things like differential equations. Differential equations are ... just kidding, guys! No math here. I still wake up at 3 AM from calculus nightmares. Anyway. In 1948, the Soviet Union set up a new institute within the Soviet Academy of Sciences, the Institute of Precision Mechanics and Computer Technology or ITMVT to explore this. However, the ITMVT's first efforts were towards mechanical, analog calculating machines. This was due to the political influence of its director - the powerful General Nikolai Bruevich - who thoroughly believed that reliable electronic general computers in the near-future were impossible. A number of Soviet scientists believed this direction to be wrong. In the Stalin era however, voicing such concerns and going against the Party directive was … dangerous, to say the least. But over in Ukraine, a new development gave them the headroom to try.

MESM

In 1948, Sergey Lebedev, the director of the Kyiv Electrotechnical Institute in Ukraine, learned about the American efforts to build digital computers. Lebedev was a trained electrical engineer and war hero. During World War II, he developed a tank gun stabilization system that allowed tanks to aim and fire properly at the enemy whilst on the move. This won him a medal and the Order of the Red Banner of Labor. Realizing the computer's importance to his country, over the next two years, Lebedev built his own "little calculating machine" - the Malaya Elektronno-Schetnaya Mashina or MESM. Ironically named since it had about 6,000 vacuum tubes and took up 60 square meters of space. On December 1951, MESM solved its first real calculation: Solving for 585 possible values of a probability distribution function for an artillery weapons calculation table. This required 250,000 calculations and 2.5 hours. In late 1949, the device was shown to a group of top political leaders including future Soviet leader Nikita Khrushchev - at the time Secretary of the Ukrainian Communist Party. It apparently impressed him. The renegade scientists at the ITMVT reached out to Khrushchev, who gave the go-ahead to depose the once-powerful General Nikolai Bruevich and allow the ITMVT to move forward with a new, more powerful digital computer.

BESM and STRELA

The bolshaia elektronnaia schetnaia mashina, or BESM-1. The name literally means "large calculating machine", and it was capable of 8,000-10,000 operations each second - or 8-10 kiloFLOPS. The BESM-1 would have been one of the world's fastest computers. Certainly the fastest in Europe. To compare, in 1951 the UNIVAC I was one of America's leading edge computing machines and it was capable of only about 1 kiloFLOP. But ITMVT struggled to deliver the BESM. That’s because the Ministry of Machine and Instrument Engineering had previously started a new institute - called SKB-245 - for making differential analyzers. These are mechanical analogue computers that use wheels and disks to solve differential equations. They were quite popular in America during the 30s and 40s. However, SKB-245 soon went against this mandate and proposed to build their own digital high-speed computer - the STRELA. STRELA was computationally inferior to the BESM-1 - just 2 kiloFLOPS where the BESM did 8-10. But its supporters argued that 10 kiloFLOPS was thoroughly unnecessary when the STRELA could perform in 10 hours what would take a single mathematician 100,000 days. A bizarre argument. It reminds me of the “28 nanometer chips are just fine” credo. With Stalin's blessing, the two institutes competed, which hurt the BESM. Near the end of its development, the ITMVT needed a special part to bring their machine up to spec - cathode ray tubes for memory storage. However, only one Ministry in the country made such things. And unfortunately, that was the Ministry of Machine and Instrument Engineering - SKB-245's ministry. They refused to supply tubes for the BESM-1, forcing the team to adopt mercury storage tubes - an inferior part. A 1951 state commission eventually chose the STRELA over BESM-1 - perhaps because of its flashy lights. But actual users soon discovered serious issues. First, it broke down too often. And secondly, it could not solve the complicated quadratic equations needed for fission development. This requirement was fairly important. The head of Soviet nuclear development effort requested BESM replacements. By then, it was 1955. The BESM-1 would remain Europe's fastest computer for a few more years but the effort illustrates future Soviet issues in building a competitive computer.

Lack of Incentive

The Soviets had succeeded in developing a first-generation computer. They did it as fast as the West could, and their computer was arguably better. Another Soviet success, right? Well, no. At the leading edge and on the newspaper headlines, it might seem like the Soviets and the Americans were neck to neck. But at the production level the story was very different. In 1959, the Soviet Union produced computers worth a cumulative $59 million. Impressive by itself, but that same year the United States produced $1 billion of computer hardware. By 1964, the Soviet Union produced 200 million semiconductor pieces. A great improvement, but still far behind the 1.36 billion the United States produced that same year. This is where the battle was lost. Far more relevant than any hostile ideology - I have omitted here a long, meandering section on Soviet cybernetics - the Soviet computing industry failed because it could not produce enough good computers, fast enough. In the Soviet command economy, the computer failed to tap its full potential. The Soviet military had some demand for computers, but only so much. In the West, computer-aided automation revolutionized industrial processes like steel-making and manufacturing. This drove titanic commercial and industrial demand for computers - far larger than the demand from the military. But the Soviet steel maker only needs to make enough steel for their quota. Their quota says nothing about how good that steel needed to be nor how efficient they had to be in making it. Soviet computer makers had little incentive to produce productivity improving computers. Just as critically, they also had no incentive to provide the necessary training and customer service so that their customers can even properly use their products. As a result, ministers failed to see the computer's full potential - and thus were reluctant to risk their quotas by allocating limited resources to computer development. So factories frequently failed to receive funding, materials, and talents they needed to produce. For instance, in 1958 the Penza Computer and Analytical Machine Plant - one of only two in the Soviet Union at the time - requested 50 computer experts, but received just 7. This was a systemic problem. The country has long lagged behind its peers when it came to distributing new, revolutionary technologies across the country. For example, the Soviets were pioneers in steel-making and synthetic fibers. 30 countries - including Japan - purchased the rights to the Soviet steel casting patent. Despite this, the Soviets universally lagged their foreign peers in pushing those innovations across their own economies.

The Second Generation Computer

In 1959, the West began an industry transition from vacuum tubes to transistors - second generation computer technology. The transistor's solid state reliability and smaller size turned the computer from a scientific curiosity into a powerful industrial tool. The IBM 1401 - the Model-T of computers - sold over 12,000 units as they replaced older punched-card technologies. This spurred the development and commercialization of new peripherals - things like cathode ray displays and external storage - as well as new paradigms of programming. The Soviets worked to develop their own second-generation computer too. While the Ural series of computers - built by the aforementioned Penza Computer Plant - were notable for their unified nature, the true Soviet 2nd generation flagship was the BESM-6. The BESM-6 was the Soviet Union's first supercomputer capable of performing over a million operations each second. It entered production in 1967 and would be extensively used across the country's computing centers. BESM-6's wide adoption led to the development of new software and multi-program operating systems. Generally, it showed great progress. But by then, the Americans were already moving onto the next big thing.

The Third Generation

The Soviets managed to deliver a first and second generation computer two or so years behind the West. Behind but not too far behind. But that changed with the third generation computer - devices powered by integrated circuits. The first mentions of a computing gap between the United States and the Soviet Union begin here. In 1964, IBM announced the System/360 - a complete line of computer hardware with both backwards and forwards compatibility. Customers who invested in writing expensive software for their second generation computers can use the System/360 without fear of losing that investment. The 360 revolutionized the computer industry and became one of IBM's greatest commercial successes. Over in the Soviet Union, it became clear that demand for future computers had a seemingly limitless ceiling. And the domestic industry - with its disparate parts - needed to step up. In 1964, the Izvestia - a daily broadsheet and the Soviet Union's newspaper of record - levied the first major criticism of the country's computer efforts. It pointed out that the current industry effort was too disjointed and needed some centralization. Two years later, Pravda published an article asking whether the Soviet computer industry was yet ready to meet the national economy's computing requirements. The author writes: "Unfortunately, we must state [that] the situation is far from optimistic."

Technology Transfer

Now, Russia has been importing foreign technologies going as far back as Peter the Great. And in many cases, this has helped them acquire expertise in disadvantaged industries. For instance, rocketry. Soviet rockets gained much of their early expertise from transferred German scientists and war booty acquired after World War II. Another example would be in the chemical industry. I have also covered other cases of countries importing foreign technologies and successfully improving upon them to dominate the market. China from Germany in the solar panel space. Japan/Taiwan/South Korea from the United States in the early semiconductor space. So on. However, there is always a balance. If the technology transfer process fails to take hold and innovation cannot be self-generated, then what might result is an uncontrollable and irreversible addiction to the imported technology.

Unified System

In 1967, the Communist Party of the Soviet Union followed through and issued a decree - Resolution 1180/420 - to consolidate its disparate computing industry. The product would be a series of unified third-generation computers for every problem: The Edinaia Sistema project or ES, literally meaning Unified System. The resolution grew the computer manufacturing budget ten-fold to 3 billion rubles in 1975. It directed the country to build 28 new computer plants, expand 22 existing ones and finally produce 20,000 computers by 1975. To build the ES, the two ministries in charge of development opted to copy an American IBM 360 model's design to make the ES compatible with 360 software. This decision was controversial. Some argued that this would make the USSR an intellectual colony of the West. A few influential scientists felt that this was an opportunity to pioneer a new Soviet computing experience, built from the ground up with modern Soviet standards in mind. One memo read: "Copying foreign work excludes the possibility of utilizing our own collective experience of computer research, and in the immediate future, will hinder our ability to employ new principles. This will bring the development of computer technology in our nation to an end". This assessment wasn't entirely incorrect. Furthermore, at the time the decision was made, the US Government forbade IBM from exporting 360 systems to the Soviet Union. This meant the Soviets would not have access to adequate documentation and have to acquire genuine systems through espionage or intermediaries.

A Momentous Decision

There were other options on the table. ICL - the United Kingdom's biggest computer-maker - leapt at the chance to tap the massive Soviet market and overtake their rivals at IBM. In 1968, the computer giant offered to share the technology and documentation from their 360-compatible System-4 line of computers. ICL was even willing to commit significant funds and send specialists for future co-development. I mentioned the System-4 in a previous video I made about ICL. ICL had the System-4 because one of its predecessors - English Electric - acquired a license from RCA. This was a promising alternative approach - one where the Soviets can have their 360 software compatibility, learn British programming methods, and also retain some industrial sovereignty. However, working with ICL would also push back the project's delivery date by at least one and a half years. Such a prospect at the time would be politically difficult to swallow considering how wide the Soviet-American computer gap already was. This was especially the case in software, with the domestic situation so dire that even Brezhnev knew about it. The Communist Party of the Soviet Union had made this a high political priority, and the penalties for missing those are not pretty. Furthermore, the East German Government had already started an effort to study and copy the IBM 360 computer. Telling them to change directions and work with ICL would have had political consequences. In other words, pissed them off. This copying decision - documented in meeting notes as it was being made - had tragic, long lasting consequences on the Soviet computing industry. Several prominent computer scientists resigned their posts in protest. Computer legend Sergey Lebedev attempted to reverse the decision later on, but failed. These efforts are said to have hastened his death in 1974.

Building the ES

Over 100 organizations, 46,000 scientists and 300,000 workers across Poland, East Germany, Hungary, Romania, and Czechoslovakia would help develop the ES. Over 15,000 pieces were created. Many of the ES systems’ hardware were created through reverse engineering. The peripheral situation - as was often the case in the Soviet computing industry - fell short of Western alternatives. Soviet disk storage solutions were particularly iffy. Software-wise, the ES's 360-compatibility did accelerate Soviet capabilities. Contemporary CIA estimates in the 1970s believed that the Soviet computing gap shrank from the 10-15 years it was in the mid-60s to about 4-6 years. However, the short term gain soon led to a long term loss. Starting in the 1980s, the Soviet industry struggled to reverse engineer increasingly complicated Western computers, widening the gap once more. This time, the Soviets had lost the native industrial capacity to develop anything new on their own - making the gap permanent. Industrial productivity struggled as the computers of the 70s aged and frequently broke down. By 1989, a quarter of all the Soviet Union's 13,613 general purpose computers were ES-1022s made in 1974 but copying a technology first released in 1965 - 24 years old. The linkage to foreign technology also caused issues when the United States and the West levied export bans on their advanced technologies during certain international crises.

Conclusion

In the United States, policymakers debated about the merits of allowing the Soviets access to American technologies. Hawks complained that the US was selling them the capability to close their own gap. And in some ways they were right. Looking back at it, with their human capital shortages, the Soviets could never have created anything close to the IBM 360 software. But even with the transfers the gap still remained, and the Soviets were never able to fully close it. The fall of the Soviet computing industry was a story of wasted talents, lost progress, foolish short-term strategic choices made by politicians over scientists, and the limits of centralized planning. It's a lesson for all to study.