The world is too loud. Read what matters.

Asianometry

Soviet computing lost not to inferior engineering, but to transistors without a commercial market

The Ural family was genuinely mass-produced with modular design, but chronic software shortages and inadequate transistor capacity sealed its fate; the system pivoted to cloning IBM System/360 instead.

Soviet computingCold War technologySemiconductor industryIBM System/360Mass productionSoftware shortage
An obscure but complete line through Soviet industrial history: from the son of a purge victim to chief designer, then to an entire product line killed by political decision, with solid detail but flat delivery—suitable for background listening.

The argument · tap a timestamp to hear it

2:03

The son of a purged enemy became Soviet computing's chief designer

Rameev's father was imprisoned as a political prisoner due to overseas study and liberal views, and died in the camp ten days before scheduled release. Years later, the Soviet government rehabilitated this Stalin-era injustice, but Rameev himself, as ‘enemy of the people's son,’ was forced to leave Moscow Power Institute and struggled to find formal work for years. This repressive and low-key background shaped his quiet, publicity-avoiding personality, but didn't prevent him from becoming one of the Soviet Union's most important computer designers.

6:11

STRELA led once, then lost to political favor for its competitor

STRELA won the Stalin Prize in 1954 and was approved for mass production, briefly outpacing its competitor BESM. But BESM had been stripped of its specialized cathode-ray storage tubes for political reasons; once it obtained the technology, its computing speed jumped to 8000-10000 operations per second, overtaking STRELA and becoming a first-rate European computer. The victory of one technical approach over another often came down to struggles over resource allocation, not pure engineering merit.

8:17

Ural's designer rejected raw speed as a meaningful performance metric

Ural-1 ran at only 100 operations per second—an order of magnitude slower than STRELA and BESM. But Rameev argued in a 1956 paper that measuring raw speed in operations per second was misleading; the real bottleneck was data entry and program loading. So he focused on automating data input and expanding the instruction set, designing repetition instructions like modern for loops. By his estimate, this cut solution time by 20-40%.

10:22

Mass production exposed unreliability: machines failed every 18 hours

Ural-1 was the first computer widely accessible to Soviet civilians; by the early 1960s it was deployed in 128 cities. But reliability was abysmal: a 1962 record shows one system with about 73% uptime, failing roughly every 18.5 hours; another consumed 3096 vacuum tubes and 5700 diodes in two years. The gap between production volume and engineering reliability haunted the entire Soviet computer industry.

17:42

The Soviet-American gap was transistor capacity, not computer design

Bell Labs published the point-contact transistor in 1948; the Soviet Union replicated it within a year by 1949. But it never translated lab results into mass production. In 1957, US companies sold 28.7 million transistors; the Soviet Union produced only 2-2.7 million. By 1964, Soviet output had risen to 200 million, but American semiconductor production had reached 1.36 billion. This gap partly stemmed from the Soviet lack of a commercial market large enough to amortize costs, and never truly closed.

21:52

Computers shipped without operating systems—there was no one to write them

Penza Factory had only dozens of programmers handling operating systems, the ARMU compiler, and utility software—severe understaffing. When a Ural-14 arrived at a computing center in East Germany, it arrived with almost no system software and sat unused; the locally-made Robotron 300 proved more useful. An East German chemical company used an American CDC computer for years; in 1968 it was ordered to switch to the ‘socialist’ Ural-14, only to find it lacked both operating systems and high-level programming languages, had crude peripherals, and by 1972 was scrapped entirely.

23:57

Soviet planners chose to clone IBM System/360 rather than finish Ural

Ural was the only Soviet system that genuinely attempted family-series design—different models sharing architecture and components. But by 1966-1967, when the Unified Computer System (ES-EVM) committee decided, Ural's high-end models weren't finished and Penza Factory was overwhelmed, unable even to participate in the debate. Ural's ARMU language was only partially portable, not truly compatible. The committee chose to replicate the mature IBM System/360 ecosystem; a key reason was the 8-bit byte—adopting this international standard meant they could copy the entire software library wholesale rather than rebuild it themselves.

28:10

The chief architect left, and the Ural line ended with him

In 1968, Rameev left Penza for Moscow to lead ES-EVM software development, and never returned to Ural. Penza secured Soviet State Bank funding in 1971 to design Ural-25, compatible with both ES-EVM and old Ural architecture. But the project dragged on; by 1972, when real ES-EVM hardware (ES-1020, ES-1030) shipped, the Bank reconsidered and saw no need to keep old Ural software. The project was cancelled, and the product line ended.

In their own words · checked verbatim

If we could make five to seven such computers, the Soviet Union would be in great shape!

Visitor6:11

once and I was blown away by the grandeur of it. It's bigger than anything that I've seen before. It's as big as a 6600 and it certainly has more lights ...

Charles Simonyi13:35

The tube was sticking out from a unit, and next to it there was a light that was indicating the state of the flip-flop, and the tubes were of course glowing constantly in a very nice orange color, and then the little light was blinking.

Charles Simonyi13:35

And if you can think of a thousand of these in glass cabinets, or rather, cabinets with a window in the doors ... as the machine was running the whole room was flashing in intricate patterns, very much like the cheap science fiction movie

Charles Simonyi13:35

bring the development of computer technology in our nation to an end

Rameev and Glushkov26:07

Figures

STRELA production76:11
BESM operations per second8000-100006:11
Ural-1 operations per second1008:17
Ural-1 uptime (one system, one year)~73%; failure every 18.5 hours10:22
Ural-1 total production (1957-1961)18311:28
Ural-2 operations per second500012:34
Ural-2 production (five years)13912:34
1964 semiconductor production: Soviet vs. US200 million vs. 1.36 billion17:42

Glossary

ATsVM
Automated Electronic Digital Computer—first Soviet universal digital computing system, proposed by Rameev and Brook
STRELA
Soviet high-speed digital computer of the 1950s; lost competition to BESM
ferrite core memory
Random-access memory using magnetic cores; the early standard before semiconductor RAM
ES-EVM
Unified Computer System (Ryad); Soviet decision to replicate IBM System/360 architecture wholesale
ARMU
Ural-series automatic code; high-level programming language designed for portability across Ural models

How to listen

Who it's for

Readers interested in Cold War industrial history, semiconductor industry history, or how organizational decisions kill technical roadmaps.

Skip

The opening section on Rameev family background and early ATsVM proposals has lower information density; you can skip it.