Operations Research has one of the clearest origin stories in applied science: it was born of wartime necessity, flourished in the peacetime economy, and eventually became so embedded in industrial and commercial practice that its presence became invisible. This article traces that arc from the first operational analysis groups of World War II through the formalization of the discipline and its explosive post-war growth.
Military operations in modern industrialized warfare generate staggering coordination problems. Moving troops, supplies, and equipment across continents; allocating scarce resources across competing theaters; timing offensives to exploit fleeting windows — these decisions had always been made by experienced commanders relying on judgment and precedent. By 1939, the scale and technological sophistication of warfare had exceeded what intuition alone could handle.
The immediate trigger was the German Blitz and the Battle of Britain (1940). The Royal Air Force needed to understand not just aircraft and pilots, but systems: how radar stations should position their search beams, how interceptor controllers should be trained, how incoming raid data should flow from observers to command. These were not questions that engineering or military tradition could answer directly.
The response was to assemble small groups of scientists — physicists, mathematicians, biologists, psychologists — and embed them with operational commanders. The most famous was the group assembled around Patrick Blackett (later Lord Blackett), the Nobel Prize-winning physicist, at RAF Fighter Command and then Coastal Command. Blackett's team was nicknamed Blackett's Circus by the military officers who initially regarded them with skepticism.
Their method was empirical and quantitative. They collected data on real operations, built simple mathematical models of what was happening, and made recommendations based on analysis rather than tradition. Early results were dramatic:
Similar groups were established across the Allied militaries. In the United States, the first OR group was formed in 1942, working first on antisubmarine warfare and later on bombing strategy, logistics, and ground operations. The term "operations research" (or "operational research" in British usage) was coined during this period.
As the war progressed, OR groups tackled increasingly ambitious problems:
The single most important post-war development was the formalization of linear programming by George Dantzig in 1947. Dantzig, working for the U.S. Air Force, devised both the mathematical framework (optimizing a linear objective subject to linear constraints) and the simplex method for solving it efficiently. This gave OR its first general-purpose computational tool.
Within a decade, LP was being applied to:
The U.S. Air Force's Project SCOOP (Scientific Computation of Optimum Programs), for which Dantzig developed LP, aimed at mechanizing military logistics planning. It demonstrated that what had been a weeks-long manual planning exercise could be automated.
Almost simultaneously, the Russian mathematician Leonid Kantorovich had independently developed a similar theory for central economic planning (published in 1939 but largely unknown in the West until much later). Kantorovich would share the 1975 Nobel Prize in Economics with Tjalling Koopmans for this work.
The discipline formalized rapidly:
Universities began offering OR courses and degrees. Consulting firms sprang up to apply OR methods to industrial clients. The discipline attracted some of the best mathematical minds of the era.
The 1950s produced a burst of foundational theory:
The availability of mainframe computers transformed what OR could accomplish. Problems that had required weeks of hand calculation could be solved in hours, then minutes. This made OR economically viable for routine business decisions, not just occasional large-scale planning efforts.
The airline industry became the first major commercial adopter of large-scale OR:
Manufacturing followed:
The late 1950s saw two independent developments of network-based project scheduling methods:
Both methods model projects as networks of activities and find the critical path — the sequence of tasks that determines total project duration. PERT added probabilistic time estimates to handle uncertainty. These methods became standard practice in construction, aerospace, and defense and remain widely used today.
Two developments transformed OR capabilities from the 1980s onward:
Interior-point methods: Narendra Karmarkar's 1984 polynomial-time algorithm for LP demonstrated that the simplex method, while practically fast, was not the only approach. Interior-point (barrier) methods proved competitive or superior for very large problems and are now standard in commercial solvers.
Metaheuristics: For problems too large or complex for exact methods, heuristic approaches flourished:
These methods find good-but-not-necessarily-optimal solutions to problems where exact algorithms are computationally infeasible.
Solvers and modeling languages: Commercial LP/IP solvers (CPLEX, Gurobi, XIMSS) and algebraic modeling languages (AMPL, GAMS, JuMP) made OR accessible to practitioners without deep algorithmic knowledge.
The rise of the internet created entirely new OR application domains:
Modern practice increasingly combines OR with machine learning:
The boundary between OR and AI has become genuinely blurry. Both Google DeepMind's protein structure prediction work and Amazon's supply chain systems combine techniques from both traditions.
| Person | Contribution | Era |
|---|---|---|
| Patrick Blackett | Founded wartime OR, Coastal Command analysis | 1940s |
| George Dantzig | Linear programming, simplex method | 1947 |
| Leonid Kantorovich | Independent LP development; Nobel Prize 1975 | 1939/1975 |
| Richard Bellman | Dynamic programming | 1957 |
| L.R. Ford & D.R. Fulkerson | Max-flow min-cut theorem | 1956 |
| Edsger Dijkstra | Shortest path algorithm | 1959 |
| Ralph Gomory | Integer programming cutting planes | 1958 |
| Narendra Karmarkar | Interior-point LP algorithm | 1984 |