These days, a handful of engineers are trying not only to simulate such "emergent" behavior but to actually create it by designing robots that swarm together to produce unforeseen results. Wei-Min Shen, director of the Polymorphic Robotics Laboratory at the University of Southern California, proudly showed video clips recording the first steps of his creature: a set of identical frog- sized robots that can reconfigure themselves on the fly, joining up to form a lizard (in order to climb a hill) that then rips off its own legs to slither away like a sidewinder. With no central brain, decisions made by the robots arise as a collective property of the whole system. Such machines could be useful for exploration, search and rescue, and even combat, says Shen, who is also adapting the technology for a 10-kilometer orbiting array of solar panels planned by NASA.
Neo Martinez, a San Francisco State University ecologist wearing a Hawaiian shirt and Converse sneakers, presented his network models of evolving food webs. He showed a series of slowly rotating nest- like structures on his laptop, each one a tangle of thousands of colored spheres representing species, connected by links showing who eats whom. As the model played out, one species would bulge as it sucked the biomass from another, only to shrink again as its predators became more numerous. Designed to match the deluge of data culled from real-world aquatic ecosystems, these models helped him generate artificial food webs to check his intuitions about which feeding strategies lead to equilibrium, and which to oblivion.