A Theory of Almost Everything
HIDDEN ORDER How Adaptation Builds Complexity. By John H. Holland. Illustrated. 185 pp. Reading, Mass.: Helix Books/Addison-Wesley Publishing Company. $24. AT HOME IN THE UNIVERSE The Search for Laws of Self-Organization and Complexity. By Stuart Kauffman. Illustrated. 321 pp. New York: Oxford University Press. $25. ARE WE ALONE? Philosophical Implications of the Discovery of Extraterrestrial Life. By Paul Davies. Illustrated. 160 pp. New York: Basic Books. $20. FRONTIERS OF COMPLEXITY The Search for Order in a Chaotic World. By Peter Coveney and Roger Highfield. Illustrated. 462 pp. New York: Fawcett Columbine. $27.50.
IN 1993, Gottfried Mayer-Kress, a physicist at the University of Illinois, offered advice, in a press release, to American officials agonizing over the conflicts in Somalia and Bosnia: stop viewing the Somalian and Serbian warlords as "bad guys" and think of them instead as "parameters" within a "self-organizing complex adaptive system." In this way, Mr. Mayer-Kress suggested, the officials might come up with solutions to the crisis.
Mr. Mayer-Kress is no crackpot. He is a respected member of a scientific discipline that studies something called complexity. What is complexity? Good question. One practitioner has counted at least 31 different definitions, each of which has its adherents. These often involve other protean buzzwords like "chaos," "information," "self-organization" and "emergence." One common, and revealing, synonym for "complex" is "interesting." Complexity, it seems, is in the eye of the beholder.
Investigators of complexity -- let's call them complexologists -- may not agree on what they are studying, but they share a dissatisfaction with the "reductionist" science of the past. They are also imbued with an almost religious belief in the power of mathematics and computers -- or "macro scopes" as they are called by some complex ologists -- to transcend the limits of conventional science and reveal the order underlying the messiness of nature -- and even of human affairs.
This faith stems from a seductive syllogism: Computers, by following very simple rules, can generate astonishingly complicated patterns. Nature also generates astonishingly complicated patterns. Therefore, simple rules must underlie the patterns of nature, and, with the help of computers, scientists can root out those rules. Of course, simple rules do underlie nature. They are embodied in quantum mechanics, the theory of natural selection, molecular genetics. But complexologists insist that much more powerful rules remain to be discovered.
Scientists and journalists have been churning out books on complexity for almost a decade or so now. (My favorite is still "The Dreams of Reason," by the physicist Heinz Pagels, published in 1988.) Although complexologists have not really accomplished much yet, at least relative to their ambitions, the books have maintained their exuberantly optimistic tone.
The batch considered here is no exception. Perhaps the most modest is by John H. Holland, a computer scientist with joint appointments at the University of Michigan and the Santa Fe Institute, the self-anointed headquarters of complexity. More than 20 years ago, Mr. Holland invented genetic algorithms, which are pieces of computer code that can "evolve," that is, rearrange themselves at random in order to become more efficient at solving a problem.
Genetic algorithms, Mr. Holland has proposed, might lead to a general, mathematical theory that can illuminate a wide range of "complex adaptive systems," from immune systems to human brains to entire economies. In "Hidden Order: How Adaptation Builds Complexity," he moves briskly past the obligatory grand rhetoric and plunges with gusto into nuts-and-bolts descriptions -- complete with equations and flow charts and diagrams -- of how such a theory might work.
He has the heart of an engineer. He wants to solve complexity-related problems, and not just any problems but big ones: AIDS, mental illness, inner-city decay. He envisions computer-based "flight simulators" that would allow officials to foresee, for instance, how a low-income housing project might affect an inner-city neighborhood.
He only briefly considers whether a simulation of downtown Detroit would also apply to Los Angeles. But that is the crux of the problem. Would any official really trust a model general enough to apply to both cities, or to all cities and to human immune systems? Another question: Is a cure for AIDS more likely to emerge from mathematics and computer science than from the same kind of gritty, bench-top work that led to the discovery of the AIDS virus?
Mr. Holland's ambitions pale beside those of his colleague Stuart Kauffman, a full-time resident of the Santa Fe Institute and one of the most charismatic of the complexologists. "At Home in the Universe" is a condensed, passionately written version of Mr. Kauff man's 709-page magnum opus, "The Origins of Order" (1993). For decades, Mr. Kauffman has been performing computer simulations of abstract, interacting "agents," which can represent everything from molecules and genes to whole organisms and companies.
He also foresees practical consequences stemming from his work. His simulations, he suggests, may serve as a guide for managers of large, complicated systems, like the United States. He also devotes a chapter to the burgeoning field of directed evolution, in which chemists generate new enzymes, genes and other useful molecules by inducing them to "compete" among themselves.
BUT Mr. Kauffman once considered a career as a philosopher, and his hunger for deep truths has persisted. He believes his models have revealed "laws of complexity" at play in the universe that counter the inexorable drift toward heat death -- toward universal sameness -- decreed by the second law of thermodynamics. In his view, these laws of complexity, more than the random siftings of natural selection, made the appearance of life on earth and its subsequent evolution not only possible but probable. The refrain "we the expected" runs through Mr. Kauffman's book like a mantra.
This same idea enlivens a new book by the Australian physicist Paul Davies. In "Are We Alone?" Mr. Davies considers not complexity per se but the likelihood that there are other conscious beings in the universe. He warns at one point that we cannot assume life like ours is probable simply because we exist; he compares such reasoning with that of the lottery winner who thinks winning is easy.
But then Mr. Davies indulges in this very fallacy. Pointing to the work of Mr. Kauff man and others, he bravely declares that not only life but also intelligence -- intelligence capable of discovering mathematical laws -- is inevitable. There must be other conscious beings in the universe. He reiterates the message of his previous books, notably "The Mind of God" (1992): the fact that we can discern highly mathematical laws ruling the cosmos suggests that it has a purpose or plan, one in which we play a role.
Mr. Davies is a writer of great charm and lucidity, and his message has been well received. Earlier this year he won the million-dollar Templeton Prize, which is bestowed annually on someone who has "advanced public understanding of God or spirituality." But one wishes he would take up the only theological question that really matters: if there is a plan to the universe, why does it require so much human suffering? Even if there are laws of complexity that made us inevitable, how is that fact supposed to comfort a Bosnian woman who has seen her daughter raped and slaughtered?
Given the extraordinary claims, both worldly and spiritual, being made for complexity, one yearns for a hard-nosed appraisal of the field. "Frontiers of Complexity" is hardly that. Peter Coveney, a physicist, and Roger Highfield, a journalist, both British, echo and even amplify the claims of Mr. Holland, Mr. Kauffman and Mr. Davies. In the view of Mr. Coveney and Mr. Highfield, complexity may lead to the creation of artificial brains and ecosystems, to the elimination of disease and warfare and pollution, and to the building of "a bridge between science and the human condition."
But they are good reporters. Buried in the hype are nuggets of important criticism. They point out, for example, that a concept called the "edge of chaos," which Mr. Kauffman and others have adopted as their definition of complexity, has been studied and rejected as flawed by several of Mr. Kauffman's own colleagues at the Santa Fe Institute.
The British authors note that John Holland's genetic algorithms are not as clever in practice as they are in theory; so far, the algorithms have proved useful only for limited applications. One catch is that the programmer must intercede to insure that they evolve in the right direction; the "watchmaker" (to borrow Richard Dawkins's phrase) is hardly blind. Moreover, because it is usually impossible to trace the steps that lead genetic algorithms to a solution, programmers may not entrust them with important problems.
Mr. Coveney and Mr. Highfield also provide overdue criticism of a popular candidate for a theory of complexity: self-organized criticality. The theory's paradigmatic system is a sand pile; as one drops sand on a pile, it "organizes" itself into a "critical" state, in which each additional grain of sand is increasingly likely to trigger an avalanche down the sides of the pile. In the same way, say adherents of self-organized criticality, many systems -- brains, economies and so on -- tend to approach a state in which a small disruption can alter the entire system.
STUART KAUFFMAN has become a proselytizer for self-organized criticality, as has Vice President Al Gore. In "Earth in the Balance," Mr. Gore wrote that self-organized criticality had helped him understand not only the sensitivity of the environment to potential disruptions but also "change in my own life." But experiments done at the University of Chicago and elsewhere have shown that self-organized criticality does not even provide very good descriptions of sand piles.
Finally, Mr. Coveney and Mr. Highfield review developments in the theory of mathematics and computation that cast doubt on the very notion that there is a general theory of complexity. The mathematician Gregory Chaitin, for example, has suggested that as science seeks to provide mathematical descriptions of increasingly complex phenomena, it must keep expanding its base of axioms; to know more, in other words, we must assume more. "Our efforts to create a grand theory of complexity may be built on sand," Mr. Coveney and Mr. Highfield comment.
In fact, this is one of the most peculiar aspects of complexity: according to its own tenets, its lofty goals may be unreachable. As most complexologists freely acknowledge, their mathematical models offer only probabilistic descriptions of nature. "We must give up the pretense of long-term prediction," Mr. Kauffman writes in one of his rare lapses into humility. "We cannot know the true consequences of our own best actions." But as the philosopher of science Karl Popper pointed out, prediction is our best means of distinguishing science from pseudo-science.
The history of 20th-century science should also give complexologists pause. Complexity is simply the latest in a long line of highly mathematical "theories of almost everything" that have gripped the imaginations of scientists in this century. Previous candidates included cybernetics, information theory, catastrophe theory and, of course, chaos. All have gone through a boom-bust cycle, in which great expectations were inevitably followed by only limited applications and hence disappointment.
It is hard not to empathize with the dissatisfaction that underlies complexity. Modern science, for all its power, is still pitifully inadequate when confronted with the problems that really matter, such as poverty, racism or conflict between states. Nor has science answered the deepest of philosophical questions, namely: Were we inevitable, or just a fluke? Complexity enthusiasts think they can find the answers that have eluded their "reductionist" predecessors. Maybe they will. But their own work, and the record of their predecessors, raises another possibility: maybe these answers are beyond the reach of science.