|
|
"A Traffic Jam Isn’t a Collection of Cars"A Review of Mitchel Resnick’s Turtles, Termites, and Traffic Jams: Explorations in Massively Parallel MicroworldsOne of the exciting characteristics of advances in computing technology is that they make possible avenues of scientific exploration that weren’t possible before. The development of massively parallel computers is one example. Conventional computers, those with a “von Neumann architecture” (after the Hungarian-born mathematician John von Neumann, who literally dreamed up this configuration), have a single processor and perform all tasks, no matter how complicated, serially, one instruction at a time. A parallel computer, in contrast, has more than one processor. A problem is divided into parts. Different processors simultaneously deal with each of these parts, and then their outputs are combined to create a solution. The introduction of massively parallel computers (ones containing thousands and, soon, millions of processors) yielded enormous increases in processing speed, because a lot of instructions could be calculated all at once. This is a technological development that is “on the curve,” albeit an exponential curve. However, new technologies have a way of introducing possibilities that lie off the curve as well. Massively parallel computers have made it possible, as never before, to study the habits of highly complex systems—ones consisting of thousands or millions of small parts. They have become, in fact, laboratories for the new science of complexity, a science that promises to yield important results in fields as distinct as economics, evolutionary theory, ecology, cellular automata and artificial life, morphology, neurology, molecular biology, fluid dynamics, materials science, meteorology, and cosmology.Mitchel Resnick, a researcher at the Media Laboratory at MIT, has written a fascinating, indispensable book on the the study of complexity in parallel systems that is at the same time accessible to any educated layman. That is not surprising, for Resnick, before taking his Ph.D. in computer science at MIT and establishing himself as a leading thinker in the field, made his living as a science journalist. On coming to MIT, he fell under the spell of Seymour Papert, the brilliant AI researcher and education theorist who co-developed Logo, the simple, intuitive computer program for teaching children geometric concepts by allowing them to manipulate objects (onscreen and off) known as turtles. Resnick developed a new version of Logo, called StarLogo, that enables young people to write programs to control the parallel operations of hundreds or thousands of turtles, a program that ran, initially, on the massively parallel computer known as The Connection Machine but has since been ported to UNIX and Macintosh systems. (These serial versions of the program run much more slowly, of course, and are hampered by their inability to run trials in parallel. However, they can be used to explore, in constrained fashion, the concepts that Resnick discusses in his book. A PC version of StarLogo is due to be released soon. To download a copy of the program for your own use to explore ideas presented in this article and in Resnick’s book, see http://www.media.mit.edu/~starlogo/.) Resnick’s desire was to create a means by which people without expert-level programming skills could nonetheless explore what are known as “self-organizing systems.” A self-organizing system is one in which elegantly organized phenomena emerge from the complex interactions, over time, of enormous numbers of small parts that are themselves relatively dumb and obey very simple rules. In his book, Resnick provides many superb examples. Consider the building skills of termites. One would expect such complicated behavior to be the result of an overall guiding force, of “top-down” command. However, consider an immense checkerboard containing thousands of virtual termites and thousands of pieces of virtual wood, each randomly scattered about. Start by giving the termites three simple rules: Move randomly about. If you bump into a piece of wood, pick it up. If you bump into piece of wood and you are already carrying a piece, drop the one you are carrying. Given the actions of large numbers of termites following these simple, local rules, an overall pattern emerges: Wood chips tend, over time, to be aggregated into piles, just as if there were some overall master plan for building mounds. Note the following characteristics of this scenario:
2. Each of the entities involved, that is, each of the termites, is itself very stupid. It follows a few simple, local rules, known only to itself. 3. The interaction of simple local rules and randomness yields an overall,
emergent design that is not inherent in the parts.
Resnick relates his studies of self-organizing, complex systems to a general trend—what geneticist Richard Dawkins would call a “meme,” the cultural equivalent of a gene that propagates through a population. This general trend is the tendency in modern culture toward decentralized modes of thinking and acting. Resnick points to the failure of centralized, planned economies, most notably that of the Soviet Union; to current management theories that favor flat organizations to hierarchical ones; to psychological theories such as object relations theory, which sees development as a matter of internalizing relationships in the world “as agents or objects within the mind”; to Postmodernist, reader response, and deconstructionist theories of literary criticism that replace the central text and the all-powerful author with multiple texts created by individual “readings”; and to the development of computing away from large, monolithic mainframes and toward decentralized PCs and palmtops connected in a vast, Weblike structure. In fact, according to Resnick, we live in “The Era of Decentralization.” What is exciting about the study of decentralized systems—made possible by the use of parallel computers to model the behaviors of enormous numbers of semi-autonomous primitives—is that it yields unexpected results. The beautifully organized emergent phenomena witnessed in these systems are surprising. One seems to be getting, in the words of one of Resnick’s high-school students, “something for nothing.” Given simple rules, a large degree of randomness, and no central designer, one would be surprised to find order, or design. Yet that is exactly what one finds. In fact, much of the order that we see around us, from buying patterns at delis to the “orderly” progression of a forest fire through a national park, is a result of just such massively parallel processes. Resnick’s work challenges us to do a lot of rethinking. Here is one example: in many parts of life where we automatically, unthinkingly, imagine that there is a centralized planner at work or that a centralized plan is necessary, neither is actually the case. No entity instructs ants to forage for food. Yet because of the local actions of individual ants, randomly following simple rules, the overall pattern of foraging emerges. Here is another: many of what psychologists refer to as “mental primitives,” the basic items or entities about which we have thoughts—traffic jams and flocks, for example—are not really entities at all. They are not even collections of entities. Cars move in and out of jams. The jams remain, emergent phenomena consisting not of individual entities but of interactions. Consciousness itself may be just such a phenomenon (although Resnick merely hints at this possibility in the introduction to his book), an emergent that can only be described as an interaction between agents in the mind but not of any particular agents. In this view, consciousness would not itself be an entity, any more than a traffic jam is. But it would be no less real. One delightful characteristic of Resnick’s book is that it mirrors the very phenomena that he describes. It is one of the clearest, simplest books that you will ever read, and yet its ideas are astonishingly powerful. Resnick’s ideas, like the turtles in his StarLogo program, are “things to think with,” in Seymour Papert’s apt phrase. Resnick’s ideas ramify, establishing connections among phenomena that one never before imagined were connected. Early in the book, Resnick quotes Gregory Bateson, in Steps to an Ecology of Mind: “Any study which throws light upon the nature of ‘order’ or ‘pattern’ in the universe is surely nontrivial.” Resnick’s book, by this measure, is nontrivial indeed. ReferencesBateson, Gregory. Steps to an Ecology of Mind. New York: Ballantine, 1990.Resnick, Mitchel. Turtles, Termites, and Traffic Jams: Explorations in Massively Parallel Microworlds. Cambridge, MA: MIT P., 1997. |
| Questions for Discussion and Review
The following questions are based on the preceding text. Clicking on a question will take you to the place in the text where the question is discussed. To return to these questions, simply click the "Back" button in your browser. 1. What is an "von Neumann architecture"? 2. How does a parallel computer differ from one with a conventional von Neumann architecture? 4. What is the subject of Resnick's book? 5. What is the name of Resnick's program, and what does it allow people to do? 6. What are the characteristics of a "self-organizing system"? 7. What rules do Resnick's termites follow in order to build wood piles? 9. In what way is the study of self-organizing systems related to the study of chaotic systems? 13. How might the
concept of emergence be related to consciousness?
|