06/23/97 INSIDE THE SANTA FE INSTITUTE

Return to main story

INSIDE THE SANTA FE INSTITUTE

The former back bedroom of a house overlooking Santa Fe, N.M., is humming with neuronal and silicon energy. In one corner, artificial-life guru Chris Langton demonstrates how hundreds of Anasazi families--simulated on a Sun Microsystems Inc. workstation--might gather into settlements in the American Southwest. Across the room, Dubliner Barry McMullin is creating simulated ''beings'' capable of replenishing their parts--one key to helping machines catch up with the simplest living creatures. And at two other desks, researchers are fine-tuning a software package, dubbed Swarm, that is already being used to model such complex systems as forests and factories.
From the outside, the 13-year-old Santa Fe Institute looks exactly like what it once was: the brown-stucco home of a prominent Santa Fe diplomat. But behind its blue doors ''is an atmosphere of unbridled boldness,'' says Steven N. Durlauf, a professor at the University of Wisconsin at Madison who co-directs the institute's economics program. At any given time, some 25 to 50 scientists (most of them visiting for a few weeks to a year) are working at desks and computers crammed into rooms, hallways, even former garages, or swapping ideas in a sunlit inner courtyard.
Their ambitious overall goal: to discover how individual things, be they viruses, trees, or people, interact and adapt to create complex systems such as epidemics, forests, or cities. ''We are trying to understand how patterns emerge from total randomness,'' explains the institute's president, immunologist Ellen Goldberg. From the patterns, Santa Fe scientists argue, will come new theories and hints of solutions for some of the world's most difficult problems.
BREAKING THE CYCLE. Getting there requires a new approach to science, say the lab's founders. Since its creation in 1984, the institute has plucked researchers from a variety of disciplines and thrown them together. The effort has proved painful but fruitful. When economist Durlauf first ventured to Santa Fe, for instance, he found the physicists so arrogant that ''I didn't come back for six years,'' he recalls.
Now, ideas from physics have helped Durlauf generate new notions about race, out-of-wedlock births, and other pressing issues. Just as the ''spins'' of individual atoms align in a magnetic field, he explains, so do people conform to community norms. Traditional economics holds that people's behavior flows from estimates of the costs and benefits of decisions. But Durlauf's equations emphasize the key role played by conformity in influencing, say, a teenager to drop out of school or have a child.
What's more, his mathematical model shows that a big boost in perceived benefits, such as the payoff of graduating from high school, can override--and ultimately change--the social norms. Durlauf now must see if the math actually represents the real world. But his model offers the tantalizing suggestion that affirmative action and other policies that boost opportunities are vital to breaking the cycle of inner-city decay.
That's a long jump from computer-science research, but Santa Fe isn't just about digital science. Durlauf himself doesn't even use a computer for his models. And other research at the Santa Fe Institute ranges far afield. Institute scientists plan to begin testing ideas of childhood brain development by intervening in local day-care centers, then looking for changes in both cognitive ability and brain structure.
The Santa Fe Institute is most known for using computer simulations to study complex systems. ''The computer amplifies the brain's capacity to recognize patterns,'' explains Santa Fe's John L. Casti. At Los Alamos National Laboratory, Casti helped put together a $50 million computer representation of every house, street, and car in Albuquerque to search for pollution patterns.
NEW MATH? Now, Casti is preparing to start a similar detailed analysis of a supermarket. It's a pet project of British grocery giant J. Sainsbury, one of several dozen companies (others include DuPont and Xerox) tapping into Santa Fe's ideas. The supermarket model, Casti explains, could help pinpoint the best shelf arrangements, or suggest shifts in offerings to lure higher-spending shoppers. Ultimately, he hopes the models will lead to a new type of mathematics, just as gambling provided the impetus for probability theory. ''I hope this will lead to a Newton of complex systems,'' he says.
Other computer programs at the institute have mimicked the stock market, estimated how long drugs must suppress the AIDS virus before the body can rid itself of the infection, and offered a new explanation of why evolution proceeds in fits and starts. The most concrete tool to come from the Santa Fe Institute so far, however, is probably the Swarm software package.
A few years ago, Langton realized that the art of simulating thousands of independent agents was still primitive. So he set out to build what he calls a ''virtual Lego kit.'' The program allows users to fashion a collection of such agents, equip them with rules for behavior, then turn them loose and watch patterns arise. Developing the software required solving some ''deep'' computer-science problems, Langton says: ''But we're finally at the point where we can stop working on the tools and apply them to real-world problems.''
In the case of the Anasazi, for instance, anthropologist Timothy A. Kohler of Washington State University has input data about everything from nutrition and crop productivity to soil type and trading rules. The hope is to understand why settlements occurred in certain patterns. So far, the simulation doesn't mimic the real world, Kohler and Langton admit. But that's partly beside the point. ''We may never end up with a model of the world,'' says Langton. ''But in the process of building it, we will think of the world in a very different way.''
By John Carey in Santa Fe, N.M.

Return to main story