Economic Cybernetics for Socialism

Discussion

Diana Kurkovsky West :

“To address the question of TPC planning, I must first situate the history of Soviet economic cybernetics as a cascading history – that is, as work that not only occurred inside the central institutes in Moscow, but which also ‘cascaded’ into regional and republic-based branches of the Academy of Sciences. I will therefore begin my inquiry with an overview of the history of economic cybernetics. The idea that the Soviet Union would be managed statistically had been the central premise of early Bolshevik thought, and was the vision behind the creation of Gosplan (the state central planning agency). With the advent of cybernetic theory, managing the entire economy from a central computer suddenly seemed possible. The appeal of cybernetic thinking for the command economy was evident in Soviet plans for connecting and regulating the economy through a cybernetic network called the OGAS, which Benjamin Peters has recently discussed in his book How Not to Network a Nation. The OGAS project, spearheaded by Viktor Glushkov, would have consisted of 20,000 data centres connected to 200 intermediate processing centres, and would have introduced a high level of self-governance, bureaucratic transparency, and even something we might consider a precursor to participatory governance and digital citizenship. As Peters argues, ‘although still hierarchical, acquiescent to Moscow as the center, [the OGAS] was openly worker-oriented, antibureaucratic, and decentralizing in principle’ (Peters, 2016: 113). Although the project never fully materialized – and Peters’ book uncovers compelling explanations for the bureaucratic stalemate that became the eventual death of the OGAS – it nonetheless offered a vision that was of a piece with the cybernetic ideals persisting in the Soviet discourse on governance. Glushkov, along with a number of eminent Soviet mathematicians like Leonid Kantorovich, Vasily Nemchinov, and others, put forth a vision of the economy that looked akin to managing a complex information network, where ‘all economic relations could be modeled, optimized, and managed with sufficient help from computers and their numerate keepers’ (ibid.: 67).

By 1960, mathematical models had gained prominent advocates in the new field of economic cybernetics. The reasons for this were as much political as they were pragmatic. Stalin had grown the Soviet economy at unprecedented rates, but by unconscionable methods, including famines that had starved millions of people in the Ukraine, and the use of prison labour for the construction of massive industries, canals, and dams in inhospitable climates. With Khrushchev’s rise to power, the political impetus to overturn Stalinism’s dark legacy led the new leadership to favour notions of profound economic reform. In this sense, the explosion in computing, cybernetic theory, and the changing political climate in the USSR were happily concurrent. Much of Kantorovich’s research into linear programming for algorithmically balancing complex and competing variables – research for which he received the Nobel Prize in 1975 – had already taken place in the 1930s, and therefore cannot be said to link directly to the advent of cybernetics. Rather, it seems that the complex demands of governing a vast command economy necessitated the kind of algorithmic thinking that was upheld and reified in cybernetic theory, and would be especially well-fitted for computer-aided analysis. In fact, a number of the key actors in economic cybernetics, among them Veduta, Nemchinov, Fedorenko, and others, articulated similar ideas about the mathematical and economic research of the 1920s and 1930s, and this, along with the pragmatic concerns of those ‘on the ground’ working in Soviet industries, conditioned Soviet planning in the direction of cybernetic thought.

With much support from Kantorovich, the economist-mathematician Vasily Nemchinov played a key role in the development of the Soviet school of economic cybernetics. In 1958, he founded the first laboratory of economic-mathematical modelling in Moscow, which lay the groundwork for the earlier mentioned 1963 establishment of the CEMI within the Academy of Sciences. Conjoining cybernetic methods with the task of algorithmic economic optimization, Nemchinov’s research built on the idea of inter-industry balances (known by their Russian acronym MOBs, or mezh-otraslevye balansy), which also owed a great debt to the Russian-American Nobel Prize-winning economist Wassily Leontief and his input–output model of inter-sector dependencies (Leontief, 1986[1951]).

Thanks in great part to the efforts of Kantorovich, Nemchinov, Nikolai Fedorenko, and others, a large network of institutes devoted specifically to research into economic cybernetics emerged, working mainly on the development of an optimal planning system based on mathematical analysis. Concurrently, the growing system of Inter-industry balances, building on Kantorovich’s linear programming concepts, ensured that the Central Statistical Office modelled demand trends and balance of labour. At the same time, the Soviets were concerned with global modelling trends; as Eglė Rindzevičiūtė’s research demonstrates, scientists from the Computer Centre and the Institute for Systems Analysis (VNIISI), both based at the All-Union Soviet Academy of Sciences in Moscow, were actively involved in international organizations such as the United Nations and the Institute of Applied Systems Analysis (IIASA) in Austria. Fedorenko, the founder of CEMI, was also a member of the Club of Rome, and attended the 1965 economic congresses in Rome (Rindzevičiūtė, 2015b). This ensured that leading Soviet scientists were familiar with the research conducted at Massachusetts Institute of Technology (MIT) by Jay Forrester in system dynamics, as well as with the widely known and heavily criticized Limits to Growth Report, the so-called ‘curve to Doomsday’ computer model that predicted global resource depletion by 2072. Authors of the computer model and the report, among them Donella and Richard Meadows and MIT’s Forrester, were invited to tour the Soviet Union, and Meadows made over 20 subsequent trips to the USSR (ibid.: 5). I will return to Limits to Growth later in this article, as researchers at IEOIP offered a cybernetic critique of the problem of resource depletion.

Writing about the Soviet use of cybernetics in social sciences in the 1970s, David Holloway noted that the period of initial optimism about creating a highly centralized computer system for managing the economy had been short-lived, and researchers at the various institutes (including the Siberian IEOIP) had realized that the processing power necessary for the central computer to deal with the amount of information faced by Gosplan was orders of magnitude greater than what was plausible at the time. This further advanced the cybernetic steersman model of decentralized decision-making among cybernetic economists. On the one hand, central planning would be ‘required in order to make structural changes in the economy, to determine its basic proportions, and to fix the parameters which regulate the behavior of lower-level units’ (Holloway, 1976: 115). On the other hand, however, a great degree of self-governance on the lower level was necessary in order not to overwhelm the central planning apparatus. This interest in using cybernetics not to control the Soviet economy at all levels, but rather to set the general parameters for economic governance, became an important and largely unstudied feature of late Soviet planning. This sentiment is evident in the TPC optimization research, which advocated the creation of territorially optimized sub-units within the broader system of Soviet production.”

…

Conclusion: Planning for a cybernetic future

A late Soviet textbook on economic planning captured the challenge of the command economy as follows: ‘When approached broadly, the problem of territorial organization of production, besides the economic and social aspects, also touches technical, ecological, planning, architectural, and other questions’ (Kistanov, 1981: 5–6). For this reason, it argued, the problem of territorial organization was always a priori macroeconomic; the optimization of the entire Soviet economy, in turn, was a cybernetic challenge to meet the macroeconomic development goals. TPC optimization grappled continually with a range of local and national challenges and imperatives, working toward a system that would be capable of optimizing any configuration of factors at a given point in time. Thus, a marriage of economic cybernetics and economic geography was key for maintaining this kind of nationally solvent, yet locally alterable system. Unlike some of the earlier ideologically motivated publications on cybernetics for socialism, TPC planning did not advocate a specific teleology, posit eternal ideals, or lay claim to ultimate scientific methods. Instead, it anticipated modifications, a host of changing economic, social, and ecological factors, and sought a high degree of administrative autonomy from national ministries, councils, and central planning organs. Not geared toward a single, optimized, and perfected end goal, TPC optimization models were systems for continual input–output analysis, which would be reconfigured based on new information and research. In time, these systems would also learn, evaluating new inputs against the larger statistical body of past information received.

The work on mathematical modelling and system optimization coming out of the Academy of Sciences points to the existence of an important late Soviet internalization of cybernetic ideas in Soviet economics. This work not only resonated with the research agendas of CEMI and other Moscow-based institutes, but also aimed to integrate a cybernetic agenda into the core of Soviet regional planning. It also points to a more emergent cybernetic body of operations, which not only enhances our understanding of dynamic systems within late Soviet planning, but also points to the way in which Soviet researchers tried to embed the concept of entropy into central planning through dynamic mathematical modelling.

In conclusion, I would like to return to Nikolai Veduta’s conceptualization of the cybernetic command economy as a massive, steerable whole composed of dynamic moving parts. The work of IEOIP took that challenge to another level, presenting calculations for a total vision of the Soviet economy depicted in spatial terms. The factories, railways, dams, settlements, and entire production complex of the vast Siberian territories had to be ‘hard-wired’ into the landscape, and unlike abstract economic concepts, the built environment was not easily alterable once those elements were in place. In this sense, projects like the OGAS played second fiddle to the actual tasks of computerization within each industry, region, or factory complex, which simply could not wait for a centralized network to emerge. Critical of over-determining and over-centralizing Soviet economic cybernetics, Veduta focused on loose coupling, arguing that loosely coupled systems were much more desirable than the tightly coupled systems of feedbacks usually coveted by cyberneticians. He wrote that if, in a tightly coupled system, the output was not desirable, the system would be far more difficult to change than if there was sufficiently loose coupling to allow for intervention. Thus, in building a computerized system for steering the Soviet economy, he insisted that the network be loosely coupled, localized, and distributed. ‘Practice’, he wrote, ‘always leaves room for uncertainty [sic] in the outcome of any experiment’ (Veduta, 1971: 58). For Veduta working at the Academy of Sciences in Minsk, as well as for the researchers at the Siberian IEOIP, the optimization of the Soviet economy entailed the emergence of complex industrial systems, which, though computerized, would retain a great degree of autonomy. Simply put, Soviet TPC planning was to establish general laws for the cybernetized command economy of the future, where entropy was a positive feature allowing for superior, localized governance.”

(https://journals.sagepub.com/doi/full/10.1177/0952695119886520)