(Review of 2010 online 2nd edition.)

Malinovsky (b1921) is a Russian/Ukrainian who began working on computers as a grad student in the 1950s in the USSR. His book is a mix of personal reminiscences, short biographies, primary documents & long quotations from memoirs, a diary contrasting '40s/'50s to his life in the '90s after a heart problem sent him to the hospital, and in this American edition a preface explaining the circumstances of an online release & appendix containing a few academic reviews of the English-translation manuscript.

As such, it is unique. The early American development of computing has been covered well & in detail by works such as Dyson's Turing's Cathedral, but Russian development is shrouded in obscurity. Before reading PoSS, about the only thing I knew about Soviet computing was that there wasn't much of it & that they had tried an interesting experiment in not binary but trinary or ternary-based computers, the Setun. Any attempt to give an overview of the history is bound to be interesting. It also vividly conveys the oppression that they worked under: blacklisting of people for trivial reasons like having an unusual Greek surname, discouragement of Jews, stringent security checks (why? given that no one in the world cared), difficulty in acquiring parts, expensive production, opaque bureaucratic decision-making about what projects to fund & the consequence reliance on military sponsorship to cut through red tape... (but also some of the benefits, like spies & industrial espionage of American projects).

That said, the informativeness is limited by the chaotic organization of topics, bouncing from person to person. This book would have benefited a good deal from some graphs or timelines to help one keep things straight, especially as PoSS spends a lot of time on the many overlapping projects in the '40s-'50s to develop varying flavors of computers. For example, I often found myself confusing Lebedev with other pioneers. (The confusing nondescriptness of many organizations' names also didn't help.) Malinovsky also deliberately limits the discussion to computer hardware, mentioning that "Beyond the scope of this book is the whole range of Soviet software developed during the Cold War and the distinguished scientists behind it, this including A.A. Lyapunov, M.R. Shura-Bura, A.P. Ershov, V.M. Kurochkin, E.L. Yuschenko, and others"; unfortunately, it is the software developments which would still be comprehensible & of interest to technical readers, whose eyes glaze over at the endless mentions of hardware details like one kind of semiconductor chip vs a slightly larger kind of semiconductor chip; worse, it is difficult to evaluate hardware achievements without information about the software which ran on it, since code & hardware are a continuum (anyone can design an ultra-fast computer which is a nightmare to write for; indeed, that has oft happened).

Paton writes "Lebedev suggested that his students prepare and publish materials about the formation and development of computer technology in the Soviet Union. 'In the West, they consider us to be worse than we really are. We have to change their opinion of us', he said. Unfortunately, his idea was not properly implemented at that time and only now has been embodied in this book." Indeed... In his attempt, Malinovsky omits perspective/context & is biased, which overall render the book more a source for future historians writing a history of Soviet computing than a history itself. Malinovsky patriotically protests

...the establishment and development of computer technology in the USSR advanced in the post-war years virtually without any contact with the Western scientists. The development of computers abroad was conducted secretly because at first, digital electronic computers were designated for military purposes. At the same time, the computer technology in the USSR evolved independently as well, led by top Soviet scientists.

Despite repeated quotes how they would avidly study American publications for any available details! If he cannot say a Soviet computer is faster, then it used less parts, or was more reliable, or was built quicker, or a cluster of 76 (!) was faster than an American supercomputer... In the biographies, each & every pioneer is hardworking, kind, modest, attentive, & loyal, & how each created computers in breathtakingly short times & how every computer seemed to operate perfectly & be competitive with the fastest American machines, & how many superlatives each super pioneer deserved (backed up by endless mentions of awards that they received, or occasionally, didn't receive due to bureaucratic sabotage). As the Abbate review notes, "Occasionally the prose takes on a heroic or patriotic tone that may be jarring to American readers (though quite common in its Russian/Ukrainian context)." More importantly, through the book Malinovsky damns following the IBM 360 paradigm rather than continuing domestic lines of development; the Slava review:

As a participant first-hand account, Malinovsky's book is both valuable & problematic. Like any other personal account, it is prone to certain biases. When Malinovsky touches upon controversial topics, he often provides only one side of the story. For example, the rivalry between the two first Soviet large-size digital computer projects, the BESM & the STRELA, is narrated largely from the viewpoint of the BESM camp. A historian would have written a more balanced account. Other topics that may require a historiographic commentary include the wide introduction of automated control systems actively promoted by the director of the Institute of Cybernetics in Kiev Viktor Glushkov (many observers claimed that this campaign led to inefficiency & waste) or the controversy over the decision to build the Unified Series of Computers that supposedly "copied" IBM 360 (Malinovsky claims that this decision directly led to the "demise" of the Soviet computer industry). In both cases, Malinovsky covers one side of the story in great detail but gives little voice to Glushkov's critics or to the supporters of the Unified Series, who claimed that Unified Series computers were no copies of IBM but were only software-compatible with IBM & had high performance characteristics. Anne Fitzpatrick's explanatory comments are very helpful; & it would be very beneficial for the reader if she could also address controversial historiographic issues, either in the endnotes or in the Introduction.
The translator should be complimented on having done an excellent job in conveying the style of the original Russian text. This style, however, may sound a bit too heavy for an American reader, for it carries some of the typical features of Soviet-era formal discourse: too many nouns, the abundance of passive voice, overblown epithets, etc. Adjusting the style for an American audience would make the book much more readable.

Malinovsky never really justifies his claims, and one wonders. The IBM 360 was a landmark design, successful in the market for all sorts of purposes, and in general, the computing market has been unkind to any attempts to take alternate paths from the current leading contender (the Lisp machines being an example), as by doing so, one cuts oneself off from an entire world of innovation & Moore's law. (Vigoda: "In practice replacing digital computers with an alternative computing paradigm is a risky proposition. Alternative computing architectures, such as parallel digital computers have not tended to be commercially viable, because Moore's Law has consistently enabled conventional von Neumann architectures to render alternatives unnecessary. Besides Moore's Law, digital computing also benefits from mature tools & expertise for optimizing performance at all levels of the system: process technology, fundamental circuits, layout & algorithms. Many engineers are simultaneously working to improve every aspect of digital technology, while alternative technologies like analog computing do not have the same kind of industry juggernaut pushing them forward.") Isn't it more likely that Soviet computing could have gone down a dead end & stagnated permanently?

Indeed, there are many signs that Soviet computing could easily have disappeared up its own navel. For example, the parts dealing with Glushkov's grandiose plans to turn the Soviet economy into a centrally-computer-planned cybernetic program by the 1970s - this sounds like complete idiocy to the modern mind, aware of the full complexity of a modern economy & how inefficient Soviet management was & how centralization inevitably fails & of the incredible computing power needed to efficiently run even a small chunk of the economy like Walmart or Amazon - & yet Malinovsky, even after the fall of the USSR & complete discrediting of centralized economies, seems to think it was a great idea killed by politicians & could have saved the USSR & Glushkov was a prophet rather than a dreamer! It's no surprise that the politicians were not eager to spend 20 billion rubles on a plan with no guarantee of working. And even has the chutzpah to claim "And now a huge information network - the Internet - is stretching across the Commonwealth of Newly Independent States and around the world, fulfilling Viktor Mikhailovich's dreams and predictions of forty years ago." The Glushkov sections also exemplify Malinovsky's willingness to claim credit for Soviet software achievements but not discuss any of the details, many of which sound like awful ideas or meaningless, leading one to wonder if he doesn't understand what he's talking about or just is bad at describing them eg he quotes Glushkov as writing:

What was the difference between Mir & other computers? We considerably upgraded the machine language. However, back then the popular point of view was that machine language must be as simple as possible & the rest would be done by software. We were even mocked for our efforts to develop different computers. The majority of computer scientists in the world believed that it was necessary to develop computer-aided programming, that is, to create software that would help produce other programs.

Yes, that was the popular view then & still is, because it's right. RISC is still the dominant view of Western computer scientists as baroque CISC architectures are always left in the dust. Glushkov was dead-wrong, but no mention is made of this. Or,

In designing the Mir machines, we had tackled a daring problem - to match the machine language as close as possible to the human language, and here I mean mathematical nonverbal language, though later we made attempts with normal human language. So, we created 'Analytic,' a special mathematical language, supported by an internal interpretation system. Mir computers were used in all regions of the Soviet Union. Their creation became an intermediate stage in research aimed at the development of artificial intelligence, since the intelligence realized in them was still fairly primitive. It also looked very impressive when a machine quickly solved independent and dependent integrals, while not many professors of mathematics were able to solve them. In addition, the machine found substitutions, not just the easy ones from tables, but the difficult ones as well...the Mir computer family was quickly developed and put into serial production, receiving high marks from its users. Its creation was a giant step in the development of artificial intelligence in small computers.

In what sense? Solving integrals isn't much of an accomplishment. What does it mean to "match the machine language as close as possible to the human language"? I'm not aware of any important work in AI stemming from USSR research. Or:

Glushkov proposed a macro-conveyer principle based on the idea that each processor was given a separate task during every step of the computing process, which allowed it to work independently for a long time without the interference from other processors. In 1959, at the Soviet All-Union Conference on Computer Technology in Kiev, Glushkov spoke about the idea of a brain-like computer structure that could be realized when the designers were able to integrate not thousands, but billions of elements with practically limitless connections between them, into a single system. There would also be a confluence of memory and data processing, a system in which data would be processed throughout the memory with a highest possible degree of parallelism in all operations...only the development of new non-Von Neumann computer architecture...would solve the problem of creating a supercomputer with unlimited growth in productivity and progressively more sophisticated hardware. Unfortunately, further research showed that a comprehensive realization of the construction principles of recursive computers and brain-like structures was beyond the level of electronic technology at that time.

Despite being a programmer interested in AI, I have no idea what any of it means. This culminates in idiotic boasting: "Unfortunately, the potential of the Mir computer line was never fully realized. During my 1979 presentation in Novosibirsk on the integration of artificial intelligence into computers, I heard the academician Andrei Ershov criticize the Institute of Cybernetics by saying: 'If you had not stopped upgrading the Mir family, the USSR would have had the best personal computers in the world.'" No, there was 0 chance. Not in a system as pathological/impoverished/repressed as the USSR was - there were no opportunities for the economies of scale which power microchip development, & if there had, PCs would never have been allowed outside of a few restricted roles. The whole point of the PC revolution in America was that anyone, including little kids who would grow up to be great programmers & entrepreneurs, could access cheap unrestricted computing power for the most trivial of reasons & create whatever they wanted to without friction.

Nor was Glushkov alone. No matter how much dead, he'll still hold out hope that a dead end is not a dead end. "To this day, Brusentsov maintains that the trinary system is superior to binary, but only time will be able to tell whether or not he is correct" - how long should we wait, exactly? Or from the Setun article, we read that its programming language, DSSP, "was not invented. It was found. That is why DSSP has not versions, but only extensions. Forth is created by practice. DSSP is created by theory. It is not a word." This is pathological linguistic mysticism, one of the delusions of the 20th century - the idea that language is terribly important & that a better purer language would unlock wasted powers & enable undreamed-of productivity. If we could invent a more logical & compact language, if we could strip out the illusions built into language, if we could come up with a better one, we would solve AI / create world peace / become geniuses etc. What's the stock trope for becoming superintelligent in 20th century SF? Your own language in which you can convey concepts more efficiently & fast; we see this in Heinlein's Speedtalk, Anderson's Brain Wave, even Chiang in (and anything to do with that nebulous cluster of Californian stuff called ), or enthusiasm for conlangs like Loglan/Lojban... it's why Russian fascists intently studying Ithkuil feel like such an anachronism. It is the fallacy that strong Sapir-Whorf is correct, that languages powerfully shape thoughts rather than channel trivialities like color-name choices. The truth is that specialized languages & notations are indeed powerful, but they always succeed innovation & insight, not precede it: they codify powerful insights & choices, & can only be created after having had the insight they embody. to design a language before the powerful ideas it embodies is to put the cart before the horse. To go from Leibnizian calculus notation to say 'Lojban will make your life more awesome' is to ignore the specialization that gave the notation power. There are no general powerful insights you can embody in a language to turn its users into geniuses, although you can take the insights of past geniuses in statistics & design a specialized statistics language which is far better than ordinary language. Learning Ithkuil won't give you access to any ideas or heuristics you didn't have before, because natural language is already general & flexible. (Would Newspeak actually work? Consider Gene Wolfe's counter-example, "Loyal to the Group of Seventeen's Story - The Just Man" or the Darmoks of Star Trek).

The politics of Soviet computing are interesting. There remains a great deal of lingering guilt & doubt around the Manhattan Project - whether it was really a good thing. scientists working on the SDI missile defense program are even more prickly about whether their work was harmful in destabilizing the precarious peace. One wonders about Russian counterparts: did they regret endeavoring mightily to put atomic bombs in the hands of a psychopath like Stalin? Or assisting bomb & ICBM development to ensure that all of humanity would live under a Damoclean sword? Or how about the environmental consequences, far from limited to Chernobyl. But there is no such doubt in the people Malinovsky quotes: "In retrospect, the rush was justified: possession of such missiles gave our country weapons parity with the United States."; 'Once, one of Sergei Alexeevich's daughters asked him: "Why do you make computers for the military?" He replied: "To avoid a war."'; etc. Indeed, the worse the USSR treated its researchers, the more loyal & devoted they seemed to become. For example, Rameev saw his grandfather expropriated, his father fatally purged under Stalin & his great invention stolen from him, & Rameev's conclusion? "a stern voice warned him: 'Live quietly & don't contact us ever again!' At that moment, Rameev understood that he had to do something unusual, outstanding, & very important for his people & nation in order to give his life meaning." Is that so? Or in the story of the researcher Akushsky who was threatened with summary execution because a plane went down, & who cleverly saves himself by proving it was the pilots' fault; very amusing, & chilling. Malinovsky blandly remarks at one point, "Things did not go smoothly at first because some Communist leaders overseeing the project remembered that Kisunko was the son of a repressed kulak."