The Chronicle: Colloquy Live Transcript

Complexity Research and Its Challenge to Other Disciplines

Thursday, May 10, at noon, U.S. Eastern time

How can complexity theory reshape the way scholars examine evolution, economics, and many other fields?

What do the dinosaurs, the stock market, and British law have in common? The fate of those three may be governed by a similar process, called self-organized criticality -- the idea that systems don't evolve gradually but in avalanches of change. That theory is part of a sweeping new field called complexity research, which seeks to find broad patterns to explain the behavior of groups, whether atoms, people, or elephants. Complexity studies have earned a following among some mathematicians, physicists, and other scientists, but they have also drawn considerable criticism. Paleontologists, for example, have raised serious objections to the hypothesis that self-organized criticality can explain major extinctions in the earth's past.

» When Physicists Tried to Explain Evolution, Biologists Cried Foul (5/11/2001)

Stuart A. Kauffman is one of the founders of complexity research. He is a medical doctor-turned-theorist who is an external professor at the Santa Fe Institute, a nonprofit research organization. He is also founder, chairman of the board, and chief scientific officer of Bios Group, a complexity science consulting company. A recipient of a John D. and Catherine T. MacArthur fellowship, commonly called a "genius grant," he is the author of At Home in the Universe: The Search for the Laws of Self-Organization and Complexity and The Origins of Order: Self-Organization and Selection in Evolution (both from Oxford University Press). Dr. Kauffman will respond to comments and questions about complexity theory on Thursday, May 10, at noon, U.S. Eastern time. Advance questions are encouraged and may be posted now.

A transcript of the chat follows.

Richard Monastersky (Moderator):
Hello. Welcome to The Chronicle's online discussion of complexity research and how it is challenging other disciplines. I'm Richard Monastersky, a reporter here at The Chronicle. Our guest today is Stuart Kauffman, an external professor at the Santa Fe Institute and founder of Bios Group, a complexity consulting company.

Stuart Kauffman:
Thank you very much for having me on the chat.

Question from Richard Monastersky:
Dr. Kauffman, you've told me before that complexity research represents a turning point in science, that we are moving from taking things apart to putting them back together. To start off our discussion, could you give a quick definition of complexity?

Stuart Kauffman:
Unfortunately, there's no single definition of complexity. Seth Lloyd, now at MIT, when he was at the Santa Fe Institute, once listed 32 different definitions of complexity, Nevertheless, colloquially, and for myself, complexity involves systems with a large number of interacting parts, where the way the parts effect each other differs from part to part. The genetic regulatory networks in cells would be an example.

Question from R. Bertram, Bradley University:
Could you compare/constrast complexity theory with chaos theory.

Stuart Kauffman:
Chaos theory is a subset of complexity theory. Chaos theory typically deals with systems having continuous variables, but relatively few variables (3-10) and demonstrates the famous "butterfly effect," or sensitivity to initial conditions. Complexity theory typically deals with systems with hundreds or thousands of interacting parts and tries to understand the collective emergent properties of the dynamical behavior of such systems. Mathematical models of genetic regulatory networks or immune networks, or the behavior of the stock market, are typical examples.

Question from Tom Johnson, College of Communication, Boston University:
Hi, Stuart: Any thoughts or suggestions on sources, methods and readings to introduce complexity theory to liberal arts undergrads, esp. those who recoil from quantitative analysis? --Tom

Stuart Kauffman:
There are several good books that are introductions to complexity theory. Two books called Complexity, one by Mitch Waldrop, and the other by Roger Lewin, are both fine introductions to complexity and the Santa Fe Institute. John Holland's book Emergence is another fine book, and my own second book, At Home in the Universe, is also a very readable introduction to many of the topics central to complexity theory. None is overburdened with equations.

Question from Scott Sampson, University of Utah:
Do you see a major role for complexity theory in teaching scientists and the general public about the current global ecosystem crisis?

Stuart Kauffman:
Yes. There are grounds to believe that the global ecosystem is what physicist Per Bak and his colleagues call "self-organized critical." A number of us have made models of ecosystems which are in fact self-organized critical, and seem to capture the following phenomena:

A powerlaw size distribution of extinction events with many small and few large extinction events.
The powerlaw lifetime distribution of species and genera.
The powerlaw distribution of species per genus, genera per family, and so on.

If the biosphere's normal behavior is self-organized critical, then habitat fragmentation and other human impacts on the global ecosystem may be unleashing a major extinction event. Careful modeling and further scientific work needs to be carried out to explore the self-organized criticality hypothesis.

Question from Sherry, Florida community college:
I'm interested in how complexity theory can help explain the organizational structures and relationships that develop in educational institutions.

Stuart Kauffman:
The company that I founded to apply complexity theory to practical business and organizational problems, Bios Group Inc., has been making models of optimal organizational structure to preserve the capacity to adapt. We have tentative evidence that such systems should be poised on the boundary between order and chaos, such that small and large avalanches of adaptive change propagate through the system in a powerlaw distribution. If so, then it would seem sensible that the same general findings would apply to educational institutions, as their needs to adapt as scientific and other scholarly domains change, and as the cultural contexts of the institutions also change.

Question from Mike Nastanski, University of Sarasota:
What are the primary implications of complexity theory as it relates to business practices?

Stuart Kauffman:
As just mentioned in the previous question, I founded Bios Group precisely to explore these issues. We have now completed projects applying complexity theory to over 40 Fortune 500 companies, on problems ranging from supply chains for Procter & Gamble, to the sensitivity of market behavior to changes in market rules for Nasdaq, to product diffusion to optimal choices of a suite of products to make for a major auto maker, to scheduling problems for two major airlines, to work with the U.S. Marine Corps and the Joint Chiefs of Staff of the United States. In general, we are finding that agent-based computational models allow us to capture the fine-grained causal connections among the components of such complex systems. Using those computer models with thousands of "runs," we can map from policy space to performance space, seeking policies with optimal payoffs.

Question from tom abeles, sagacity, inc:
Where does "sustainability" and "complexity" come together or "part company?" Given the implications of Bak and Tainter, all societies, such as the United States are headed for collapse; humans are just a collapse away from the next biological transformation; and, perhaps hedonism should reign over the apparent futility of trying to build a sustainable world?

Stuart Kauffman:
You have hit upon a fundamental issue. I know of no work in complexity to date that simultaneously asks the question for the biosphere or the econosphere of what must be done to achieve sustainability, whether sustainability is consistent with ongoing economic growth, and other fundamental issues. Nevertheless, a global civilization is gradually emerging, we are in its heroic age when new transnational myths are needed to understand and sustain us. So there is much work to be done.

Question from Rick Chartrand, U. of Illinois at Chicago:
The Chronicle article quotes a "doctor-turned-theorist," physicists, and paleontologists, yet the subject at hand seems to be one of mathematics. What role do/should mathematicians play in the study and application of complexity theory?

Stuart Kauffman:
Mathematicians should hopefully become deeply involved in complexity theory. To date, the phycisists, biologists, chemists, economists, and computer scientists who are using computational models to study complex systems are ahead of the mathematicians who are just starting to enter the field. But it is largely the mathematicians who will prove the theorems and will provide analytic insight into the emergent properties that the rest of us are discovering.

Question from Jim, Oriental Medicine Journal:
My interest in acupuncture has led me to believe (correctly or not) that some Chinese medical models are prototypes of complexity theory. Can you comment or do you know of anyone doing research in this area? Thanks in advance.

Stuart Kauffman:
While I am personally interested in acupuncture, and believe that western medicine has much to learn from a oriental tradition that is thousands of years old, I am not aware of any work in complexity science that bears on these issues.

Question from Constant Beugre, Kent State University, Tuscarawas Campus:
How can complexity theory help explain how business organizations (and other types of organizations for that matter) evolve, mature, and/or decline and disappear?

Stuart Kauffman:
Funny you should ask. My colleagues and I, at this very moment, are studying a model in which firms are formed and evolve on a rugged and multi-peaked technology landscape. We believe that we can explain the well-known fact that after a major technological innovation, such as the airplane, at first there are many small firms making a high diversity of airplanes; for example, planes with one wing, two wings, three wings, up to seven wings; but as fitter technological designs are found, dominant firms and dominant designs emerge. There is a rapid die-off of the remaining firms and copycat firms that back-engineer the successful products begin to appear.

Question from Steve Hansen, University of St. Thomas, St. Paul, MN:
Has complexity theory been applied to the ideas of complexity theory/biology/physics? In other words, has complexity theory been used to explain vast and sudden changes in the way we look at the world (e.g. heliocentric solar system, Darwinism, etc.)

Stuart Kauffman:
You seem to be asking whether complexity theory has been used to try to do the historical sciences, namely, to ask whether or not complexity theory has been used to understand the changing world view that came with Darwin. To the best of my knowledge, it has not been so applied, but perhaps it can be. For example, consider Tiennemen Square in China. We know why the Chinese leaders killed the students, but we do not know what they thought they knew, which was that, presumably, alllowing the students to continue, might start a revolutionary process in China. If we could apply complexity theory to understand the sudden onsets of dramatic cultural changes, for example the fall of the former Soviet Union, then we could answer yes to your quesiton.

Question from Maria D. Ortiz, San Francisco State University:
The implications of complexity theory for the study of community organizing and campacity building seem very significant. Is there an ongoing effort to connect these two areas of research, and can you provide contact information? Thank you.

Stuart Kauffman:
The problem sounds interesting, but I know of no complexity work in this area, with the exception of efforts at the Santa Fe Institute to build an agent-based model to account for the Chacoan civilization in the southwest, and its collapse.

Question from Steve Hansen, University of St. Thomas, St. Paul, MN:
Are you aware of research applying complexity theory to the evolution of computer/software systems?

Stuart Kauffman:
The closest cousin that I know is the study of the statistical structure of the web itself. Recent work suggests that the structure of the internet is a "small world network." Small world networks typically have a power-law distribution of node connectivities with many nodes having one connection, fewer nodes having two connections, still fewer nodes having three connections, but a long tail to the right such that some nodes are extremely highly connected. In a log log plot of such a histogram one obtains a straight line sloping down to the right, hence a power-law.

Question from Eduardo A.M. Koutsoukos, Petrobras-Cenpes, Rio de Janeiro; University of Heideberg:
According to complexity theory what would be the main driving force(s) in the self-organization of complex natural systems? Is there any theoretical reason to expect that co-evolution should lead to greater stability of ecosystems?

Stuart Kauffman:
There are now mathematical models which constitute well-formulated hypotheses about systems such as genetic regulatory networks and ecosystems. In the models of genetic regulatory networks, spontaneous order arises as a self-organized property of an enormous class of networks characterized by a few simple parameters. Recent analysis of regulated genes is demonstrating that real cells meet the parameters that cause orderly dynamics. In models of co-evolution, a number of authors have found that such systems go to a "self-organized crytical state" with a power-law distribution of extinctions events, that is, many small extinction events and relatively few huge extinction events. In my own second book, At Home in the Universe, I desribe one such model in which the entire stability of the ecosystem increases over time as the coevolving ecosystem partners gradually tune the ruggedness and coupling of their fitness landscapes.

Question from Clyde Smith, Cultural Research:
Complexity theorists, with the notion of "power laws," seem to be claiming that because something can be modeled mathematically that the mathematical equation then becomes a cause for the behavior. Am I correct in this assumption and isn't that kind of a bizarre perspective on reality? Do scientists really think like this?

Stuart Kauffman:
Mathematical models are meant to capture a description of the causal connections among the components in a system. Then study of the mathematical model reveals whether or not the modeled causal connections appear to account for the behavior of the system in the real world. Newton, with his Three Laws of Motion and Law of Universal Gravitation and the resulting Celestial Mechanics, is an example of just this pattern of reasoning.

Question from Ron DeGray, Saint Joseph College, Connecticut:
The size of avalanches in the sand pile model are said to occur by self-organization. Are the influences of gravity excluded?

Stuart Kauffman:
No, the influence of gravity is central to the sand pile model because it's precisely gravity that causes the toppling of the first grain of sand, which in turn triggers the falling of the remaining grains of sand in the avalanche.

Question from David Lee, Emperor's College:
What organic or natural models most influence your economic models of complexity?

Stuart Kauffman:
First, models of rugged, multi-peaked fitness landscapes now appear to account for the phenomena of learning curves in economics. Consider a truck factory. Every time the number of trucks produced in the factory doubles, the cost per truck falls by a constant fraction, typically 10%. Adaptive hillclimbing walks on rugged landscapes have exactly the same statistics. Beyond this, and my answer to a previous quesiton, models of coevolotion of species and self-organized crytical ecosystems are simultaneously models of firms coevloving with one another and the small and large avalanches of extinction events of old goods and services and the emergence of small and large numbers of new goods and services.

Question from Steve Hansen, University of St. Thomas, St. Paul, MN:
Has complexity theory looked at the effect that the number of dimensions has in the organization of systems? For example, some biological units of organization are more two-dimensional than three dimensional (terrestrial plant communities, non-arboreal mammals). Others have a three-dimensional component (fish and forest birds, perhaps). Some systems currently in existance could be considered to have more than three dimensions (the world-wide web, for example). Does the addition of "extra" dimensions affect the predictions of complexity theory? As an example, how would complexity theory scale a recent article's suggestion for crowd-control (irregularly placed columns), if the the crowd was operating in three or more dimensions?

Stuart Kauffman:
I do not know the answer to your last question, but Geoffery West and his colleagues have published articles and a recent book by the Santa Fe Institute entitled Scaling Laws in Biology that accounts for the fractal dimention of such things as tree branching, pulmonary bronchial branching, and similar phenomena.

Richard Monastersky (Moderator):
Thank you, Dr. Kauffman, for taking the time to participate in this discussion. And thank you in the audience for your thoughtful questions.

Stuart Kauffman:
Thank you all very much for your questions. I hope that you have found the exchange useful; I have.