Bottoms Up


Snazzy and overarching scientific paradigms are notoriously tough to corral, but in his fine new book, Steven Johnson gets a good grip on one of the most important of late: “the eerie, invisible hand of self-organization.” As Johnson explains with brainy but convivial clarity, self-organization describes systems, like slime molds or computer simulations, that generate rich and complicated global behavior without being controlled through hierarchical “top-down” commands. Instead, their behavior emerges, as if by magic, from the “bottom-up” interaction of a mess of relatively simple agents pursuing their own narrow agendas, with little or no concept of the whole.

Longtime fans of edgy pop science can be forgiven for feeling initially underwhelmed by Johnson’s topic, since the scientific and social implications of self-organization have been explored rather thoroughly by others. But Johnson’s originality lies in gingerly applying the concept of emergence to familiar systems—like cities, media, and software—that derive from the culture side of the nature/culture divide. With a digital denizen’s eye toward practical applications, Johnson also focuses on how self-organizing systems learn and respond to the environment around them. The former editor of the now deceased Feed, Johnson even turns out some of those rules of thumb that dominate business books and self-help manuals: Berkeley-bumper-sticker admonitions like “More Is Different,” “Encourage Random Encounters,” and “Look for Patterns in the Signs.”

In his prose, Johnson brings these rules to life, drawing sometimes wayward connections between science, technology, and culture in a smooth but omnivorous manner reminiscent of James Burke’s marvelous BBC shows. For example, in his quest to prove that ant colonies are “the exact opposite of command economies,” Johnson begins his first chapter with a journalistic portrait of a Stanford ant researcher in her lab. Then he leapfrogs to steam-age Manchester, and then on to Alan Turing, homosexual urban enclaves, Jane Jacobs’s cybernetic vision of New York, and finally, coming round again, to computer models of ant paths.

Johnson isn’t just noodling here. You start to recognize self-organization as a tendency or “law” only when you trace its behavior inside different zones of reality and then superimpose those maps on one another. This means that, in writing about self-organization, analogies transcend their usual role as pop science bric-a-brac and become tools of analysis. “Start by taking the analogies literally,” Johnson advises us, in the midst of a Manuel De Landa-esque comparison between the sudden rise of medieval cities and the “phase transition” H20 undergoes between liquid and ice. To follow the drift, in other words, you have to look for patterns in the signs.

Johnson is especially good at plucking the patterns from our urban experience. Noting that “we are a species of city dwellers now,” Johnson treats the city both as an abstract machine and as a glorious, meaty mess. He even provides an alternative to the usual complaints about L.A.: The city is superficial not because of the boob jobs and sun but because “the potential for local interaction is so limited by the speed and the distance of the automobile that no higher-level order can emerge.” This higher-level order is the key to great cities, a product of the diversity and density of connections.

Many progressive critics still reach for their revolvers when they hear organic language invoked to characterize urbanism, since they fear that such talk tends to “naturalize” the human and institutional power relations that are really driving the show. Johnson notes that the human mind has an equally problematic tendency to assume the existence of “pacemakers”: powerful individuals that ultimately call the shots. Johnson makes his case with a marvelous Engels quotation describing the philosopher’s experience of Manchester during its boom times. Though Engels knows that the city’s growth patterns are totally unplanned, he still can’t help but see the conspiring hand of the moneyed class behind the layout of the city. Johnson’s point is that our whole concept of control—political, technological, even psychological—is in need of some serious revision.

This is juicy territory for a writer who, like a wired Malcolm Gladwell, is charting a path between techno-scientific punditry and literate cultural criticism. But Johnson’s enthusiasm for emergence, his own tipping point, can grow a bit giddy at times. In one instance, he compares our growing ability “to capture the power of emergence in code” to “the revolution unleashed when we figured out how to distribute electricity a century ago.” But his examples leave something to be desired: smart ads and some cool computer games. Nonetheless, Johnson is on target when he suggests that political and cultural forces will increasingly self-organize in the third millennium. Brilliantly analyzing the organizational properties that drive, the great online community news service, Johnson argues that emergence accounts for the growing power of digital “neighborhoods” that draw together self-selecting cultural consumers. Emergence also lies behind the antiglobalization movement, which Johnson describes as a pacemakerless organization that does politics like a slime mold, or a swarm.

As we shift into a new world disorder, however, Johnson’s analysis of emergence politics takes on a more ominous light. “For any movement that aims to be truly global in scope, making it almost impossible to rely on centralized power, adaptive self-organization may well be the only road available.” Of course, Osama bin Laden is an old-school hierarchical leader. But the highly distributed and semiautonomous network he has cultivated also exploits the logic of the swarm—even the removal of Bin Laden and his senior command is unlikely to put the kibosh on his cells and sleepers. We may yet discover that emergent properties achieve their most powerful cultural effects not when they run riot, but when they remain loosely coupled with good old command and control.