John H. Holland 1929-2015

John Henry Holland, a pioneer in the study of complex adaptive systems and of what became known as genetic algorithms, died in August at the age of 86.

Holland, a longtime professor of computer science and engineering and professor of psychology at the University of Michigan, Ann Arbor (where he founded and led the Center for the Study of Complex Systems), also was a Santa Fe Institute (SFI) professor and external professor for many years and, at the time of his passing, a member of the Institute’s Board of Trustees and Science Board.

He had been interested for six decades in what are now called complex adaptive systems, starting with his early work at IBM in the 1950s on computer simulations of Hebb’s theory of cell assemblies. He formulated genetic algorithms, classifier systems, and the Echo models as tools for studying the dynamics of such systems.

Holland was born in Fort Wayne, IN, in 1929. He studied physics at the Massachusetts Institute of Technology and received a B.S. degree in 1950, then studied mathematics at the University of Michigan, receiving an M.A. in 1954. In 1959, he received the first computer science Ph.D. from the University of Michigan.

In 1975, Holland published the book Adaptation in Natural and Artificial Systems, intended to be the foundation for a general theory of adaptation, introducing genetic algorithms as a mathematical idealization to develop his theory of schemata in adaptive systems. Genetic algorithms eventually became widely used as an optimization and search method in computer science.

Holland’s Learning Classifier System statistical learning system incorporated a reinforcement learning algorithm for non-Markovian environments. His 1989 book Induction, co-written with psychologists Keith Holyoak and Richard Nisbett, applied his ideas about classifier systems to induction in cognitive science.

An early champion of interdisciplinary approaches to science and engineering, many of Holland’s lesser-known projects anticipated trends and advances in modern computer science. He trained generations of graduate students, many of whom have taken inspiration from his interdisciplinary approach to make seminal contributions of their own.

Holland received the MacArthur Fellowship in 1992, and was a fellow of the World Economic Forum. He received the Louis E. Levy Medal from The Franklin Institute in 1961, and the MacArthur Fellowship in 1992.

Stephanie Forrest, a Distinguished Professor of computer science at the University of New Mexico in Albuquerque, observed while Holland had been her Ph.D. advisor, "he was so much more.

"In 1979, when I walked into the Computer and Communications Department at the University of Michigan looking to take a computer programming course, I was sent down the hall to talk with John, at that time the associate chair of the department. Forty-five dizzying minutes later, I walked out with an application form for the Ph.D. program, and the rest is history. As I recall, our conversation ranged from Godel’s Incompleteness and Undecidability results to flying gliders to some weird ideas I couldn’t parse about combining genetics and computers. Since that time, John never failed to be interested and encouraging of my ideas, and over the years he offered up great, if nonstandard, career advice. I credit most of my own limited success in the academic world to John’s great example and inspiration (and the many, many letters of recommendation he wrote on my behalf)."

Melanie Mitchell, a professor of computer science at Portland State University, recalled that when she started graduate school, "I didn’t know anything about the CS department. It was my great luck that I ended up taking John Holland’s class ‘Adaptation in Natural and Artificial Systems’, [which] completely changed my perspective on what computer science was (and should be) about."

Holland, she said, "became my co-advisor at Michigan, and did so much to support and encourage me in my work. He let me know when I had done enough for my Ph.D. (and told me it was time to get my ‘union card’, as he called it). He recommended me for the Michigan Society of Fellows, and then invited me to join the famous BACH group during my fellowship. He also invited me to visit SFI, first for a summer, and then asked me to direct SFI’s Adaptive Computation program. This led to my faculty appointment at (and hopefully lifelong engagement with) the Institute.

"I was fortunate to become one of John’s close-knit group of former Ph.D. students, all of us, including John, meeting every now and then to talk about everyone’s research projects and to speculate on big questions. Our last meeting was in fall 2014; John, in spite of his illness, was in great spirits, and regaled us with his new ideas and enthusiasms. In addition to his great intellect, John was perhaps the most enthusiastic, cheerful, and lively person I’ve ever known. I’ll miss him greatly."

"An Interview with John H. Holland" appeared in ACM SIGEVOlution, the newsletter of the ACM Special Interest Group on Genetic and Evolutionary Computation, in 2008. It may be accessed in the Digital Library at http://dl.acm.org/citation.cfm?id=1562109.
        —ACM Staff