A Conversation with James Hannan

Jim Hannan is a professor who has lived an interesting life and one whose fundamental research in repeated games was not fully appreciated until late in his career. During his service as a meteorologist in the Army in World War II, Jim played poker and made weather forecasts. It is curious that his later research included strategies for repeated play that apply to selecting the best forecaster. James Hannan was born in Holyoke, Massachusetts on September 14, 1922. He attended St. Jerome’s High School and in January 1943 received the Ph.B. from St. Michael’s College in Colchester, Vermont. Jim enlisted in the US Army Air Force to train and serve as a meteorologist. This took him to army airbases in China by the close of the war. Following discharge from the army, Jim studied mathematics at Harvard and graduated with the M.S. in June 1947. To prepare for doctoral work in statistics at the University of North Carolina that fall, Jim went to the University of Michigan in the summer of 1947. The routine admissions’ physical revealed a spot on the lung and the possibility of tuberculosis. This caused Jim to stay at Ann Arbor through the fall of 1947 and then at a Veterans Administration Hospital in Framingham, Massachusetts to have his condition followed more closely. He was discharged from the hospital in the spring and started his study at Chapel Hill in the fall of 1948. There he began research in compound decision theory under Herbert Robbins. Feeling the need for teaching experience, Jim left Chapel Hill after two years and short of thesis to take a three year appointment as an instructor at Catholic University in Washington, DC. When told that renewal was not coming, Jim felt pressure to finish his degree.

💡 Research Summary

The article “A Conversation with James Hannan,” published in Statistical Science (2010), is an extensive oral history that intertwines the personal biography of James F. Hannan with a technical overview of his contributions to statistics, decision theory, and repeated‑game theory. Hannan was born in Holyoke, Massachusetts, in 1922, attended St. Jerome’s High School, and earned a Ph.B. from St. Michael’s College in 1943. During World War II he served as a meteorologist in the U.S. Army Air Force, a period that exposed him to weather forecasting, coded communication, and poker—activities that later informed his intuition about sequential decision making under uncertainty.

After the war he completed an M.S. in mathematics at Harvard (1947) and began doctoral studies in statistics at the University of North Carolina. A routine physical revealed a possible lung spot, leading to a year of treatment and a delayed start at Chapel Hill in the fall of 1948. Under the mentorship of Herbert Robbins, Hannan worked on compound decision theory, producing a 1953 thesis that contained a density central limit theorem for a generalized binomial distribution and exact and asymptotic distributions for the Kolmogorov statistic. These results laid groundwork for later empirical‑Bayes methods.

In 1953 Hannan joined the faculty of Michigan State University, where he shifted his focus to repeated games. He introduced what is now called a “Hannan strategy”: at each stage i the player uses a smoothed version of a component‑Bayes rule against the empirical distribution G_{i‑1} of the opponent’s past actions. Performance is measured by “modified regret,” the excess average risk relative to the best component‑Bayes rule evaluated at G_{i‑1}. A strategy is “Hannan‑consistent” if the limsup of modified regret is ≤ 0. Hannan’s 1977 paper “Approximation to Bayes Risk in Repeated Play” provides explicit regret bounds and demonstrates that his approach yields no‑regret learning in a broad class of games.

Although his work was largely unnoticed for two decades, the 1990s brought renewed attention from the computer‑science community. Researchers studying online learning and no‑regret algorithms (e.g., Hart and Mas‑Colell, 2001) recognized that Hannan’s earlier results were essentially the first formalizations of the no‑regret property. The term “Hannan consistency” entered the literature, and several later forecasting‑selection procedures—such as those by Foster & Vohra (1993) and FeDer et al. (1992)—were identified as specific instances of Hannan’s general strategy.

Beyond research, Hannan was a dedicated mentor. He supervised twenty‑two Ph.D. students, many of whom secured faculty positions at major universities (Penn State, Columbia, Michigan State, UC‑Santa Barbara, etc.). The interview recounts his modest demeanor, his reluctance to publish until results were polished, and his willingness to help students with both technical and personal matters.

The article also provides vivid anecdotes: Hannan’s poker games on army ships, his experience with the Mahalanobis Estate in Barrackpore (later the Indian Statistical Institute), and his collaborative textbook “Introduction to Probability and Mathematical Statistics” with V. Fabian (Wiley, 1985). These stories illustrate how his wartime experiences, early exposure to statistical manuals, and a lifelong habit of logical argumentation shaped his scientific perspective.

In summary, the conversation paints a portrait of a scholar whose early life in a small New England town, wartime service, and perseverance through health setbacks forged a unique blend of probability, decision theory, and game‑theoretic insight. Hannan’s “Hannan consistency” now underpins much of modern online learning, regret minimization, and ensemble forecasting, confirming that his once‑overlooked work has become a cornerstone of contemporary statistical and algorithmic research.