JOHN D.C. LITTLE (1928-2024)

BY STEPHEN C. GRAVES AND
RICHARD C. LARSON

JOHN DUTTON CONANT LITTLE, an Institute Professor at the Massachusetts Institute of Technology (MIT) and professor of management science in the MIT Sloan School, died Sept. 27, 2024, at the age of 96.

John was born Feb. 1, 1928, in Boston, the son of John D. and Margaret J. Little, and grew up in Andover, Massachusetts. John attended Andover’s elementary and middle public schools and, for high school, earned a scholarship to Phillips Academy in Andover, graduating in 1945. He then spent the next three years at MIT in its wartime accelerated program, earning an S.B. degree in physics in 1948. While at MIT, John served as editor-in-chief of its venerable Voo Doo magazine, “MIT’s only intentionally humorous publication.” After graduation, he hitchhiked around the country for 10 months, doing various odd jobs to support himself along the way.

Tiring of life on the road, he joined the General Electric Company (GE) as an engineer. There he met his future wife, Elizabeth (Betty) Alden, a graduate of Wellesley College. In 1951, both John and Betty were accepted into MIT’s doctoral program in physics. They married in 1953. Betty earned her Ph.D. in 1954; John completed his in 1955.

John’s Ph.D. advisor was the renowned MIT physicist Philip M. Morse (NAE 1985), who led the U.S. military’s operations research efforts during World War II. Morse is regarded as the founder of operations research (OR) in the U.S., and John was the first doctoral student in the field. Like Morse, John approached OR with the mindset of a physicist who is both an experimentalist and a theoretician. He was guided by Morse’s definition of OR: ‘‘Operations Research is a scientific method of providing executive departments with a quantitative basis for decisions regarding operations under their control.’’¹

In his work, John began with observations and data collection in the real world and then returned to his office to model the phenomena – often collaborating with students – iterating between real-world observations and theoretical models until the results aligned. This approach defined his work across a range of OR subfields, including traffic signal synchronization, queueing theory, management processes, and marketing.

His doctoral thesis focused on the use of storage water in a hydroelectric system – an example of multi-period decision-making under uncertainty. The question was how much water to release from a reservoir in each period to generate power, given the seasonal uncertainty of river inflows. He modeled the problem using stochastic dynamic programming, one of the earliest applications of the method. John implemented his algorithm on the Whirlwind computer, one of the world’s first digital computers, and validated his findings using data from the Columbia River and Grand Coulee dams.

After a two-year stint in the Army, John began his academic career in 1957 at Case Institute of Technology (now Case Western Reserve University). His first major contribution to OR was the general proof of the now-famous queueing formula, L = lW. Under broad assumptions about system stability, the formula states that the time-average number of customers in the system (L) equals the average arrival rate of customers accepted into the system (l) multiplied by the average time each spends in the system (W). He was inspired to investigate the general truth of the relationship by Morse, who noted – while writing the first textbook on queueing theory – that this curious formula consistently held true for every queue whose steady-state behavior could be derived using traditional methods.²

John devised his proof during a summer family vacation on Nantucket and had it published the following year.³ He succinctly motivated the paper with the observation that the relationship “…is of interest because it is sometimes easier to find L than W (or vice versa) in solving a queuing model.” The paper was significant not only for establishing the result under very general conditions but also for offering a completely new perspective on how to prove it. He had a novel insight based on the physics of a queueing system: In his words, “…at the same time that a customer is standing in line and so can be counted, he or she is also accumulating minutes waiting...”⁴ He used this insight to develop his arguments for the proof, and his approach later evolved into the sample path methods that have become a fundamental tool in probabilistic analyses.

The formula is now universally known as “Little’s Law” and remains a central construct in a wide range of application domains, including manufacturing systems, service systems, computer architecture design and engineering, epidemiological modeling of disease prevalence, and “stock and flow” models in economics, environmental science, and system dynamics.

Shortly after publishing the Little’s Law paper, John turned to the infamous traveling salesman problem (TSP). In a 1963 paper with coauthors, he developed a novel implicit enumeration search procedure for solving the TSP.⁵ At the time, the best methods were based on deterministic dynamic programming and could solve problems with up to 13 cities. John’s algorithm allowed solutions for much larger problems—up to 25 or even 40 cities. The paper also introduced the term “branch and bound” (B&B) to describe the methodology and was among the first, if not the first, implementation of B&B to solve integer programs.

In 1962, he returned to MIT with a faculty appointment in the Sloan School of Management, where he remained for the rest of his career. One of his first areas of focus at MIT was the synchronization of traffic signals – specifically, timing signals across a network of streets to create maximal green waves. He provided the first formulation of this problem as a mixed integer linear program (MILP) and demonstrated how to solve it using his B&B algorithm. He and his research colleagues defined the new state of the art in the field by producing a set of algorithms called MAXBAND. Compared with earlier efforts, this formulation accounted for real-world complexities such as variable driving speeds, restrictions on light cycles, queue clearance times, and left turns.

In the mid-1960s, John began working with the M&M Mars candy company on allocating its advertising budget. Widely regarded as a pioneer of marketing science, he published research on a broad range of modeling and decision-support issues, including models of individual choice behavior, adaptive control of promotional spending, and marketing mix models for consumer-packaged goods. His first paper addressed the adaptive control of advertising expenditures. Later, through work with Nabisco, he developed and implemented decision-support models for budgeting and media allocation.⁶

This body of work, along with consulting for Coca-Cola, led to his groundbreaking Decision Calculus paper.⁷ In it, he offered a revolutionary idea to marketing and management scientists: they should build models that are actually used, not just publishable academic work. He proposed six criteria for such models: 1. Simple, 2. Robust. 3. Easy to control, 4. Adaptive, 5. Complete, and 6. Easy to communicate with. The full expression of Decision Calculus and market response served as the foundation for a marketing mix planning model called BRANDAID.⁸

John was one of the earliest researchers to recognize – and then demonstrate – how to extract marketing-relevant insights from supermarket scanning data. This work helped establish logit modeling as a cornerstone for understanding market response and sparked a large body of research on probabilistic models of market response.⁹ His work was instrumental in shifting the focus of consumer purchasing analysis from aggregate to disaggregate models.

He successfully applied his ideas in the private sector as well. John co-founded a marketing models firm, Management Decision Systems, which was acquired in 1985 by Information Resources, Inc. At the time of the merger, the firm had 210 employees and offices in Boston, New York, Chicago, San Francisco, Toronto, Los Angeles, London, Paris, and Sydney. Its principal products included EXPRESS decision-support software for financial and marketing applications, along with widely used packages such as ASSESSOR, BRANDAID, PERCEPTOR, and PROMOTER.

At MIT, John held several leadership positions: director of the Operations Research Center (1969-75); head of the Management Science Area in the Sloan School (1972-82); and head of the Behavioral and Policy Sciences Area (1982-88). He also served as a longtime steward of Sloan’s undergraduate program, which has consistently ranked among the world’s best.

John was an inspirational leader in his professional societies as well. He served as president of the Operations Research Society of America (ORSA) in 1979-80, and The Institute of Management Science (TIMS) in 1984-85. He then played a key role in merging ORSA and TIMS to create the Institute for Operations Research and the Management Sciences (INFORMS), serving as its founding president in 1995. In recognition of his foundation contributions to marketing science, the INFORMS Society for Marketing Science established the John D.C. Little Award, given annually to the best marketing paper published in Marketing Science or Management Science.

In 1989, he was named an MIT Institute Professor, a title reserved for a small number of scholars of special distinction. The honor is initiated by faculty and jointly awarded by the faculty and administration.

John and his wife lived in Lincoln, Massachusetts, where they raised four children. For more than 30 years, John and Betty hosted Thanksgiving dinners for international graduate students—an experience many still recall as a highlight of their time at MIT.

John, who was predeceased by his wife, Betty, and his two sisters, Margaret and Francis, is survived by four children—Jack, Sarah, Thomas, and Ruel—along with eight grandchildren and two great-grandchildren.

__________________________________
¹Morse PM, Kimball GE. 1951. Methods of Operations Research. The MIT Press.
²Morse PM. 1958. Queues, Inventory and Maintenance. Wiley.
³Little, JDC. 1961. A proof for the queuing formula: L= λW. Operations Research 9(3):383-7.
⁴Little, JDC. 2011. OR FORUM—Little’s Law as viewed on its 50th anniversary. Operations Research 59(3):536-49.
⁵Little, JD, Murty KG, Sweeny DW, Karel C. 1963. An algorithm for the traveling salesman problem. Operations Research 11(6):972-89.
⁶Little JDC. 1966. A model of adaptive control of promotional spending. Operations Research 14(6):1075-97; Little JDC, Lodish LM. 1969. A media planning calculus. Operations Research 17(1):1-35.
⁷Little JDC. 2004. Models and managers: The concept of a decision calculus. Management Science 50(12-Supplement):1841-53.
⁸Little JDC. 1975. BRANDAID: A marketing-mix model, part 1: Structure. Operations Research 23(4):628-55; Little JDC. 1975. BRANDAID: A marketing-mix model, Part 2: Implementation, calibration, and case study. Operations Research 23(4):656-73.
⁹Fader PS, JM Lattin, Little JDC. 1992. Estimating nonlinear parameters in the multinomial logit model. Marketing Science 11(4):372-85; Guadagni PM, Little JDC. 1983. A logit model of brand choice calibrated on scanner data. Marketing Science 2(3):203-38.