Engineering good decisions: Reflections on the life and work of Ronald A. Howard

This story is adapted from an article published in OR/MS Today. See the original article here.

The 2025 INFORMS Annual Meeting marked a special session dedicated to the first anniversary of the passing of Ronald A. Howard, the late Professor of MS&E, a founder of the field of decision analysis, and a pioneer of operations research (O.R.). 

We'd like to reflect on and share the life and work of our dedicated colleague.

An extensive biography of Howard's career was published in Profiles in Operations Research [1] by co-author Jim Matheson. We will provide further reflections about his legacy and multiple contributions, including a background on his formative years, connections between his interests, and how interactions with his academic advisors, Philip Morse and George Kimball, helped to catalyze the development of decision analysis that we know today.

Background and influences

Howard is recognized for his breakthrough work in dynamic probabilistic systems, including the introduction of policy iteration [2]. He is widely celebrated for his contributions to and leadership in transforming statistical decision theory into the modern discipline and practice of decision analysis [3]—a phrase he coined to describe methods and practices for engineering good decisions.

Howard pursued his work on probability, decision theory and decision analysis at Stanford University and the Stanford Research Institute [4]. For decades, he delivered influential classes in decision analysis at Stanford and served as advisor to numerous doctoral students. He co-authored The Foundations of Decision Analysis with Ali Abbas [5], a comprehensive text in the area. Beyond the pursuit of maximum expected utility, Howard thought deeply about values, organizing classes and seminars on ethics and writing on an ethical code to guide decisions [6, 7]. 

Howard was passionate about communicating key ideas to colleagues and students. He shared his thoughts widely in many fora, including publishing his reflections over the years in OR/MS Today [8, 9]. Driven by a belief that decision analysis should be made available to young people, he co-founded the Decision Education Foundation, an organization aimed at adapting professional methodologies in decision analysis to people without expert training. He provided guidance to the organization for many years.

Childhood through undergraduate education

Howard grew up during World War II in Rosedale, Queens, near his parents' candy store. Across the side street, he discovered the Rosedale public library—"where you didn't even have to pay," as he would recall. He shared that he read as many as six books a week during that time. Howard entered the Massachusetts Institute of Technology (MIT) as an electrical engineering student but wanted to ensure that he included the human interaction element in his education. He sought out a program at MIT in engineering and economics that required an additional year of course work, leading to a second bachelor's degree. Through this program, Howard took courses in probability under Robert Solow (future Nobel laureate) and labor relations under George Schultz (future U.S. Secretary of State).

While taking a probability class at MIT, Howard did not find the homework problems challenging enough. He engaged with the professor and offered to compose homework problems and solutions for future classes, in exchange for being excused from the standard assignments. The professor agreed. Howard graduated in 1955 with B.S. degrees in both electrical engineering and engineering economics.

Doctor of Science at MIT and policy iteration

Howard remained at MIT for his D.Sc. work. Phillip Morse was Howard's official thesis advisor, with George Kimball serving as his de facto advisor. Morse led the first Operations Research group in the U.S. during World War II and was head of the new MIT interdepartmental Operations Research Center, the ORC. His colleague, George Kimball, was a chemical physicist from Columbia University. Morse and Kimball wrote the first seminal book on O.R., Methods of Operations Research [10].

Howard often mentioned that Kimball and Morse had different perspectives on O.R. Morse pursued mathematical elegance, writing equations and working on innovative solutions. Kimball sought applied solutions to practical problems. Howard enjoyed both perspectives, which provided him with a balanced lens on both the theory and practice of O.R.

Howard worked closely with Kimball on projects with the Operations Research Group of Arthur D. Little, Inc. At MIT, Howard also met with Bob Sittler (a student of Bill Linvill) and Richard Bellman, who had invented dynamic programming. During that time, Howard was interested in long-run optimal policies within dynamic programming. He invented the policy iteration algorithm during his doctoral work, leading to the dissertation, "Studies in Discrete Dynamic Programming." Later, Bellman commented, "Of these methods of successive approximation, the most important one is due to R. Howard" [11]. Ten years later, Howard published the two-volume set Dynamic Probabilistic Systems [3].

Influences of Jaynes and Laplace

Howard's views on probability—specifically, that probability is subjective and reflects an individual’s state of knowledge rather than an objective frequency—were influenced by the writings of Edwin Jaynes [12] and Pierre-Simon Laplace. It was Myron Tribus, then dean of engineering at Dartmouth, who gave Howard a manuscript written by Jaynes. Jaynes used inferential notation, which conditioned every probability assignment on the state of an individual's information. This notation was later adopted by Howard in many writings, where the state of information is represented by an ampersand (&).

Jaynes was a scholar of the work of Laplace. Howard noted that whenever someone brought up a criticism of Laplace, Jaynes would go back to the original source and find that Laplace had been misquoted. In this spirit, Howard obtained a copy of Laplace's original work in French, "Théorie analytique des probabilités," and learned French to ensure that he captured the precise nuances of Laplace's thought without translation.

Stanford and decision analysis

After his postdoctoral work at MIT, including numerous interactions with Howard Raiffa at Harvard University, Howard joined the new department of Engineering-Economic Systems (EES) at Stanford University in 1965. Bill Linvill, then department chair, advocated that all Ph.D. students gain practical experience through internships at companies. Linvill and Howard focused their attention on having students exposed to the work at Stanford Research Institute (SRI) (now SRI International). SRI was closely associated with Stanford University, though there were few faculty connections at the time. The EES program at Stanford was established with seed funding from two SRI departments.

In early 1965, Howard had invited Jim Matheson to take a consulting position with SRI. A year later, the team at SRI created the Decision Analysis Group (DAG), headed by Matheson, with Howard as the key academic member. Their vision for DAG was to be the "teaching hospital" for the advancement of decision analysis. In this role, Howard actively collaborated with Matheson, and they compiled and edited the "Blue Books" [4], a collection of articles referred to by generations of Howard's students. The DAG team became a hub of international work on the practice of decision-making, attracting visitors such as Tribus.

Stepping back, decision analysis is built on the foundations of statistical decision theory. While giving an executive seminar to General Electric, a participant asked Howard whether what he was teaching could help them with a decision about nuclear waste. Howard replied, "Of course," and he spent that night wondering how his teachings could help with the practicalities of such a massive, complex problem. He later coined the term "decision analysis" to bridge the gap between theory and practice [2]. Since that time, numerous tools and practices have been developed for decision analysis.  

Along with Howard and Matheson, Tribus became a champion for the practice of decision analysis. The DAG consulted on projects when Tribus served as Undersecretary of Commerce for Standards and Technology, including the "Project Stormfury" study on whether to seed hurricanes [13] to diminish their destructiveness. When Tribus joined Xerox, he continued to be a major client of the DAG team, where another member, Carl Spetzler, had the opportunity to work closely with Tribus for many years. They, alongside Carl-Axel Staël von Holstein, Amos Tversky, and Daniel Kahneman, conducted workshops on probability assessment aimed at minimizing bias.

Strategic Decisions Group

In 1981, Howard, Spetzler, Matheson and Jeff Foran formed the Strategic Decisions Group (SDG), a decision-analysis consulting business. SDG attracted outstanding staff, including many of Howard's students and alumni from the SRI DAG teams.

Tribus continued to play a strong role as client and supporter. He became a convert to Deming's "total quality management" and challenged leaders at SDG on their approach to practice. Although decision analysis centered on reaching clarity on the best action under uncertainty, Tribus helped establish the "decision quality" (DQ) paradigm. With DQ, practitioners recognized that clarity of intent is distinct from commitment to action. In large, complex organizations, getting true commitment to action requires significant participation and conflict resolution to gain alignment among actors.

Ethics

Howard believed that ethical difficulties tended to remain unaddressed because it is much easier to avoid complex discussions about sensitive values and trade-offs than to reflect and converge with clarity on points of view. He taught that ethical difficulties can be avoided by following three practices: (1) decline being part of organizations that have ethical codes and behavior inconsistent with your own, (2) avoid participating in ethically objectionable activities and (3) treat all people as you would those you care about.

He developed the first ethics course in the Stanford School of Engineering and titled it "The Ethical Analyst." During the class, Howard would provide examples and distinctions to help students form their personal ethical code. Many derivatives of this work were later published, including Ethics for the Real World (co-authored with Clint Korver (PhD '94, MS '90)) [6] and what is believed to be Howard's last work, "A Hippocratic Oath for Technologists," co-authored with Abbas [7].

Decision analysis for all: Decision Education Foundation

In 2001, the early pioneers of decision analysis, including leaders at SDG and several of Howard's students (David Heckerman, Eric Horvitz, Tom Keelin, and many others) reflected that the power of decision analysis and decision quality should extend beyond the corporate world. How could the insights, practices and tools be leveraged more universally? This thinking led to the formation of the Decision Education Foundation (DEF), with Howard as its first president. Howard continued as a DEF director until his passing. It was this initiative that led to numerous OR/MS Today articles, including the prospect and promise of teaching decision skills to youth and troubled teens [8, 9].

A legacy extending into the era of AI

Howard has mentored many students all over the world, largely within statistics and decision-making. In the 1980s, several of his students became leaders in the blossoming area of leveraging probability and statistical decision theory to solve artificial intelligence (AI) challenges. At the time, the work of David Heckerman, Eric Horvitz and colleagues [14] was considered countercultural; the dominant paradigm for building systems had become focused on logic-based methods, often skipping over the critical importance of probabilistic representations and reasoning under uncertainty. Howard's mentorship and support had an important influence on the growth of uses of decision-theoretic methods and concepts in AI, adapting decision theory and tools designed for human decision-making to the challenge of automated reasoning under great uncertainty and incompleteness. A branch of AI became dominated by probabilistic graphical models and decision-theoretic optimization, and AI evolved to more centrally rely on probability and decision theory, including the wide use of "reinforcement learning," which builds directly on the work of Richard Bellman and Howard's contributions to policy iteration.

Talks given by several of Ron’s former students at his retirement celebration provide a sense for the influence that he has had on the lives of colleagues and students around the world [15, 16, 17].

Concluding reflection

We hope that these reflections about the life and legacy of Ronald A. Howard will shed light on the influences on his early work, the founding of the vibrant field of decision analysis and the ripple effect that his mentorship had on students, colleagues and the world. Howard's legacy will continue to inform the way we approach the most difficult decisions as we move into the future.

References

1. Matheson, J., 2025, "Ronald A. Howard," Profiles in Operations Research, New York: Springer.
2. Howard, R. A., 1966, "Decision Analysis: Applied Decision Theory," Proceedings of the Fourth International Conference on Operational Research, New York: Wiley-Interscience, pp. 55-71.
3. Howard, R. A., 1971, Dynamic Probabilistic Systems (Vols. I and II), New York: John Wiley & Sons.
4. Howard, R. A. and J. E. Matheson, eds., 1983, Readings on the Principles and Applications of Decision Analysis (Vols. I and II), Menlo Park, CA: Strategic Decisions Group.
5. Howard, R. A. and A. E. Abbas, 2015, Foundations of Decision Analysis, New York: Pearson.
6. Howard, R. A. and C. Korver, 2008, Ethics for the Real World: Creating a Personal Code to Guide Decisions in Work and Life, Cambridge, MA: Harvard Business Press.
7. Abbas, A. E., M. Senges and R. A. Howard, 2019, "A Hippocratic Oath for Technologists," Next-Generation Ethics: Engineering a Better Society, Cambridge, U.K.: Cambridge University Press, pp. 71-80.
8. Abbas, A., C. Spetzler, N. Hoffmann and R. A. Howard, 2007, "Teaching Decision Skills to Troubled Teens," OR/MS Today, Vol. 34, August.
9. Abbas, A. E., D. Reiter, C. Spetzler and S. Tani, 2004, "Teaching Teens How to Make Good Decisions," OR/MS Today, Vol. 31, No. 4, August.
10. Morse, P. M. and G. Kimball, 1951, Methods of Operations Research, Cambridge, MA: MIT Press.
11. Bellman, R., 1961, Adaptive Control Processes: A Guided Tour, Princeton, NJ: Princeton University Press.
12. Jaynes, E., 2003, Probability Theory: The Logic of Science, Cambridge, U.K.: Cambridge University Press.
13. Power, B. A., R. W. Kates, R. A. Howard, J. E. Matheson and D. Warner North, 1973, "Seeding Hurricanes [Letters]," Science, Vol. 181, No. 4075, pp. 744-747.
14. Horvitz, E., 2008, "Artificial Intelligence in the Open World," Presidential Lecture, Association for the Advancement of Artificial Intelligence, July, https://erichorvitz.com/AAAI_Presidential%20Address_Eric_Horvitz.pdf.
15. Video lecture from the Howard retirement celebration at Stanford University by Eric Horvitz, Oct. 27, 2018: https://www.youtube.com/watch?v=2hnZQ7hxbSQ.
16. Video lecture from the Howard retirement celebration at Stanford University by Ali Abbas, Oct. 27, 2018: https://www.youtube.com/watch?v=MW6fxmq4Yu4.
17. Video lecture from the Howard retirement celebration at Stanford University by Jeannie Kahwajy, Oct. 27, 2018: https://www.youtube.com/watch?v=Y9MHPC1yveQ.