7 The Soviet Optimist

Learning Objectives

Understand Kantorovich’s independent discovery of linear programming in 1939
Appreciate the political context that suppressed and delayed his work
Understand the transportation problem and its historical significance
Connect the Soviet planning context to the modern interpretation of LP duality

7.1 A Problem from the Plywood Trust

In 1938, Leonid Kantorovich was a twenty-six-year-old professor at Leningrad University, already recognized as one of the leading mathematicians in the Soviet Union. He was asked by the Laboratory of the Plywood Trust — a Soviet state enterprise — to help with a practical problem: how to distribute the workload across their machines to maximize output, given that different machines were better suited for different tasks and different raw materials.

The problem was, in modern terms, a linear program. Kantorovich recognized this immediately. More importantly, he recognized that it was one instance of a general class of problems — the class we now call linear programming — and that the same mathematical structure appeared in many different economic and industrial contexts. He wrote up his results in a 1939 monograph titled Mathematical Methods of Organizing and Planning Production, which was published in a small edition by Leningrad University Press and promptly disappeared.

The monograph disappeared for a reason. Its subject was dangerous.

7.2 The Political Calculus

Soviet economic planning in the 1930s was not, in principle, opposed to mathematics. Stalin’s industrialization drive required technically trained personnel, and the Soviet educational system invested heavily in science and engineering. Mathematics was respected, even celebrated.

But economic theory was another matter. The mathematical theory of optimal allocation of resources — the idea that prices, or shadow prices, or any quantitative signals could guide economic decisions — was ideologically fraught. Marxist economics held that prices under capitalism were instruments of exploitation, not coordination mechanisms. The idea that a planning economy needed price-like signals to function efficiently was, to doctrinaire Marxists, a concession to bourgeois economics that bordered on treason.

Kantorovich’s results implied, inescapably, that optimal resource allocation produced shadow prices — the Lagrange multipliers of the planning problem. These multipliers had the same mathematical structure as market prices. Publishing this observation prominently in Stalin’s Soviet Union was not obviously wise.

Kantorovich navigated this carefully. His 1939 monograph framed the results in terms of “resolving multipliers” rather than prices, and emphasized practical computation rather than economic interpretation. He knew what the multipliers meant. He was careful about saying so.

7.3 The Transportation Problem

The most elegant problem in Kantorovich’s 1939 monograph was the transportation problem. Suppose there are $m$ sources, each with a known supply of some commodity, and $n$ destinations, each with a known demand. The cost of shipping one unit from source $i$ to destination $j$ is $c_{ij}$. Find the shipping plan that minimizes total cost while satisfying all supply and demand constraints.

This is a linear program with $mn$ variables (the amounts shipped on each route) and $m + n$ constraints (supply at each source, demand at each destination). Its special structure — the constraint matrix has a particularly simple form, with each variable appearing in exactly two constraints — makes it far easier to solve than a general LP.

The transportation problem appears throughout logistics, supply chain management, and economic geography. It is also the mathematical backbone of the optimal transport theory developed by Gaspard Monge in 1781 and Kantorovich in 1942, which has become one of the most active research areas in modern mathematics and machine learning (the Wasserstein distance between probability distributions is an optimal transport problem).

7.4 Rediscovery, Suppression, and the Nobel

Kantorovich continued working on linear programming through the 1940s and 1950s, developing what he called the method of resolving multipliers — equivalent to the simplex method, though he arrived at it by different reasoning. His 1942 paper on optimal transport was published during the siege of Leningrad. He had, by that point, been working on optimization methods for more than a decade, with almost no contact with Western mathematics.

When Dantzig’s work became known in the Soviet Union in the early 1950s, Soviet mathematicians quickly recognized that Kantorovich had priority on the fundamental ideas. This was a politically sensitive situation: acknowledging a Soviet mathematician’s independent discovery was a point of national pride, but engaging seriously with Western academic work required institutional permissions that were difficult to obtain.

Kantorovich eventually published a comprehensive treatment of linear programming in Russian in 1959. The English translation appeared in 1960. Western economists, encountering his work for the first time, were struck by two things: its mathematical quality, and the care with which he had avoided discussing the ideological implications of shadow prices.

In 1975, Kantorovich shared the Nobel Prize in Economics with Tjalling Koopmans (who had independently developed linear programming in the Western context) for “contributions to the theory of optimum allocation of resources.” The Nobel committee’s phrasing was careful. Kantorovich had spent thirty-five years being careful too.

7.5 Summary

Leonid Kantorovich independently discovered linear programming in 1939 — eight years before Dantzig — while working on a practical industrial problem in Leningrad. The political context of Stalinist Soviet Union suppressed the publication and dissemination of his work; its mathematical content implied the existence of price-like signals in planned economies, a politically dangerous observation. Kantorovich developed the transportation problem and the method of resolving multipliers, equivalent to the simplex method, and eventually received the Nobel Prize in Economics in 1975. His story is a reminder that mathematical ideas do not develop in political vacuums.

7.6 Further Reading

Kantorovich (1960) is the English translation of his major work. For the political and intellectual context, Aron Katsenelinboigen’s Studies in Soviet Economic Planning (Sharpe, 1978) is informative. Cédric Villani’s Optimal Transport: Old and New (Springer, 2009) traces Kantorovich’s optimal transport work to its modern consequences in mathematics and machine learning.

--- title: "The Soviet Optimist" --- ::: {.callout-note} ## Learning Objectives - Understand Kantorovich's independent discovery of linear programming in 1939 - Appreciate the political context that suppressed and delayed his work - Understand the transportation problem and its historical significance - Connect the Soviet planning context to the modern interpretation of LP duality ::: ## A Problem from the Plywood Trust In 1938, Leonid Kantorovich was a twenty-six-year-old professor at Leningrad University, already recognized as one of the leading mathematicians in the Soviet Union. He was asked by the Laboratory of the Plywood Trust — a Soviet state enterprise — to help with a practical problem: how to distribute the workload across their machines to maximize output, given that different machines were better suited for different tasks and different raw materials. The problem was, in modern terms, a linear program. Kantorovich recognized this immediately. More importantly, he recognized that it was one instance of a general class of problems — the class we now call linear programming — and that the same mathematical structure appeared in many different economic and industrial contexts. He wrote up his results in a 1939 monograph titled *Mathematical Methods of Organizing and Planning Production*, which was published in a small edition by Leningrad University Press and promptly disappeared. The monograph disappeared for a reason. Its subject was dangerous. ## The Political Calculus Soviet economic planning in the 1930s was not, in principle, opposed to mathematics. Stalin's industrialization drive required technically trained personnel, and the Soviet educational system invested heavily in science and engineering. Mathematics was respected, even celebrated. But economic theory was another matter. The mathematical theory of optimal allocation of resources — the idea that prices, or shadow prices, or any quantitative signals could guide economic decisions — was ideologically fraught. Marxist economics held that prices under capitalism were instruments of exploitation, not coordination mechanisms. The idea that a planning economy needed price-like signals to function efficiently was, to doctrinaire Marxists, a concession to bourgeois economics that bordered on treason. Kantorovich's results implied, inescapably, that optimal resource allocation produced shadow prices — the Lagrange multipliers of the planning problem. These multipliers had the same mathematical structure as market prices. Publishing this observation prominently in Stalin's Soviet Union was not obviously wise. Kantorovich navigated this carefully. His 1939 monograph framed the results in terms of "resolving multipliers" rather than prices, and emphasized practical computation rather than economic interpretation. He knew what the multipliers meant. He was careful about saying so. ## The Transportation Problem The most elegant problem in Kantorovich's 1939 monograph was the transportation problem. Suppose there are $m$ sources, each with a known supply of some commodity, and $n$ destinations, each with a known demand. The cost of shipping one unit from source $i$ to destination $j$ is $c_{ij}$. Find the shipping plan that minimizes total cost while satisfying all supply and demand constraints. This is a linear program with $mn$ variables (the amounts shipped on each route) and $m + n$ constraints (supply at each source, demand at each destination). Its special structure — the constraint matrix has a particularly simple form, with each variable appearing in exactly two constraints — makes it far easier to solve than a general LP. The transportation problem appears throughout logistics, supply chain management, and economic geography. It is also the mathematical backbone of the *optimal transport* theory developed by Gaspard Monge in 1781 and Kantorovich in 1942, which has become one of the most active research areas in modern mathematics and machine learning (the Wasserstein distance between probability distributions is an optimal transport problem). ## Rediscovery, Suppression, and the Nobel Kantorovich continued working on linear programming through the 1940s and 1950s, developing what he called the method of *resolving multipliers* — equivalent to the simplex method, though he arrived at it by different reasoning. His 1942 paper on optimal transport was published during the siege of Leningrad. He had, by that point, been working on optimization methods for more than a decade, with almost no contact with Western mathematics. When Dantzig's work became known in the Soviet Union in the early 1950s, Soviet mathematicians quickly recognized that Kantorovich had priority on the fundamental ideas. This was a politically sensitive situation: acknowledging a Soviet mathematician's independent discovery was a point of national pride, but engaging seriously with Western academic work required institutional permissions that were difficult to obtain. Kantorovich eventually published a comprehensive treatment of linear programming in Russian in 1959. The English translation appeared in 1960. Western economists, encountering his work for the first time, were struck by two things: its mathematical quality, and the care with which he had avoided discussing the ideological implications of shadow prices. In 1975, Kantorovich shared the Nobel Prize in Economics with Tjalling Koopmans (who had independently developed linear programming in the Western context) for "contributions to the theory of optimum allocation of resources." The Nobel committee's phrasing was careful. Kantorovich had spent thirty-five years being careful too. --- ## Summary Leonid Kantorovich independently discovered linear programming in 1939 — eight years before Dantzig — while working on a practical industrial problem in Leningrad. The political context of Stalinist Soviet Union suppressed the publication and dissemination of his work; its mathematical content implied the existence of price-like signals in planned economies, a politically dangerous observation. Kantorovich developed the transportation problem and the method of resolving multipliers, equivalent to the simplex method, and eventually received the Nobel Prize in Economics in 1975. His story is a reminder that mathematical ideas do not develop in political vacuums. ## Further Reading @kantorovich1960mathematical is the English translation of his major work. For the political and intellectual context, Aron Katsenelinboigen's *Studies in Soviet Economic Planning* (Sharpe, 1978) is informative. Cédric Villani's *Optimal Transport: Old and New* (Springer, 2009) traces Kantorovich's optimal transport work to its modern consequences in mathematics and machine learning.