To understand John Holland you have to start with Herman Hesse--in particular, Hesse's
Das Glasperlenspiel, or The Glass Bead Game. Set in the distant future, the
book tells the story of Joculator Basilensis, who designs a game for an elite order of
intellectuals. "He invented for the Glass Bead Game the principles of a new language,"
Hesse writes, "a language of symbols and formulas, in which mathematics and music
played an equal part, so that it became possible to combine astronomical and musical
formulas, to reduce mathematics and music to a common denominator, as it were."
Computer scientist and psychologist Holland is also quietly building a new language for
his community of scholars. It is a language to study cognition, evolution, and learning; a
language of mathematics, philosophy, biology, and psychology. Instead of beads,
however, he is using a computer.
The common denominator for both men is mathematics. Since graduate school, Holland,
66, has been working on ways to meld math with the "softer" sciences. He is a leader in
"machine learning," an area of artificial intelligence (AI) incorporating a symbolic
approach to computer learning by creating categories and rules, much like natural
language. Holland is also the inventor of "genetic algorithms," a computational tool based
on the elements of genetics: Using DNA-like building blocks that combine and recombine,
he has managed to develop a metaphor for problem solving and, ultimately, create
"complex adaptive systems" that evolve much like life. Taking genetic algorithms a step
further, he has spawned "classifier systems," computer programs that literally learn from
their mistakes. The upshot of all this: nothing less than artificial life.
Holland's work goes far beyond exotic intellectual gamesmanship. As the nexus of
mathematics, biology, physics, and advanced computing, artificial life extends the
examination of real life. Complex adaptive systems such as ecosystems, human societies,
operations of the brain and immune systems, economics, geoclimates, and more are
daunting to study. Artificial adaptive systems offer a new model for exploring the world
as it is--as well as how it could be. According to Holland's 1995 book, Hidden Order:
How Adaptation Builds Complexity, several perplexing problems of contemporary
society -- inner-city decay, AIDS, mental disease, and deterioration -- are likely to persist
until we understand the dynamics of these systems. Such understanding, he notes, will help
us resolve a whole range of problems from an imbalance of trade in the global economy to
viruses on the Internet to birth defects in an embryo.
Prior to graduate studies in the early fifties, Holland was a member of IBM's logical design
group building the first large-scale computer. He may have been the first person in the
United States to receive a Ph.D. in computer science. He is the person who
conceived the idea of parallel computers. More recently, Holland has been an integral part
of the Santa Fe Institute, a research center investigating complex adaptive systems. At
Santa Fe, he has helped promote computer modeling for analyzing complex systems, and
often collaborates with economist Brian Arthur, Nobel laureate Murray Gell-Mann, and
others. In 1993, he won a MacArthur "genius" grant.
Officially, Holland is part of the Department of Psychology at the University of Michigan,
although he holds another appointment in the School of Electrical Engineering and
Computer Sciences. But he also meets regularly with a bootleg group of Michigan
professors, known as "BACH," to discuss evolution and adaptation. BACH, the acronym,
comes from its four founders: philosopher and computer guru Arthur Burks; political
scientist and author of The Evolution of Cooperation Robert Axelrod; political
scientist and organization theorist Michael Cohen; and Holland. For Holland, the group is
a local and much smaller version of the Santa Fe Institute.
Holland lives in Ann Arbor with his wife, research librarian Maurita Peterson Holland.
Their house sits on a forested hill above the Huron River. Conducting much of his
research from a spacious study on the second floor, Holland has a view of a pond and the
small mill town of Delhi. What's not glass windows is bookshelves. The book titles take
you back to a time before the Internet and e-mail, before laptops and multimedia, before
color TV, for that matter. Surveying the spines, it becomes easier to see the scope of
Holland's career and follow the thinking that took him from his first encounter with a
computer, calculating physics on MIT's Whirlwind, to attempting to build entire
ecosystems on his Power Book 540-C.
With the question of retirement looming, the youthful Holland is becoming more
possessive of his time. He's cut back on teaching and is hesitant to accept speaking
invitations. The extra hours allowed him to finish Hidden Order and begin work on
another book and on a "seed machine." "When you get to be 66, you begin to think, `I've
got files of notes; now I want the time to do the things I know how to do well,'" he says.
John Holland is not close to concluding his own glass bead game.
On Day Two: OUT OF THE WHIRLWIND