Cheaper than printing it out: buy the paperback book.

In mechanical communities, or ecosystems, some machines are more likely to associate with certain other machines, just as red-winged blackbirds favor nesting in cattail swamps. Pumps go with pipes, furnaces go with air conditioners, switches go with wires.

Machines form food webs. Viewed in the abstract, one machine "preys" upon another. One machine's input is another's output. A steel factory eats the effluent of an iron-mining machine. Its own extrusion of steel is in turn eaten by an automobile-making machine, and fashioned into a car. When the car dies it is consumed by a scrapyard crusher. The crusher's ejected iron cud is later swallowed by a recycling factory and excreted as, say, galvanized roofing.

If you were to follow an iron particle as it was dug out of the ground to be passed up the industrial food chain, it would trace a crisscrossing circuit for its path. The first time around the particle may appear in a Chevrolet; the second cycle around it may land in a Taiwanese ship hull; the third time around it shapes up as a railroad rail; and the fourth as a ship again. Every raw material meanders through such a network. Sugar, sulfuric acid, diamonds, and oil all follow different routes, but each navigates a web that touches various machines and may even cycle around again to its elemental form.

The tangled flow of manufactured materials from machine to machine can be seen as a networked community -- an industrial ecology. Like all living systems, this interlocking human-made ecosystem tends to expand, to work around impediments, and to adapt to adversity. Seen in the right light, a robust industrial ecosystem is an extension of the natural ecosystem of the biosphere. As a splinter of wood fiber travels from tree to wood chip to newspaper and then from paper to compost to tree again, the fiber easily slips in and out of the natural and industrial spheres of a larger global megasystem. Stuff circles from the biosphere into the technosphere and back again in a grand bionic ecology of nature and artifact.

Yet, human-made industry is a weedy thing that threatens to overcome the natural sphere that ultimately supports it. The crabgrass character of industry sparks confrontations between advocates for nature and apologists of the artificial, both of whom believe only one side can prevail. However, in the last few years, a slightly romantic view that "the future of machines is biology" has penetrated science and flipped a bit of poetry into something useful. The new view claims: Both nature and industry can prevail. Employing the metaphor of organic machine systems, industrialists and (somewhat reluctantly) environmentalists can sketch out how manufacturing can repair its own messes, just as biological systems clean up after themselves. For instance, nature has no garbage problem because nothing becomes waste. An industry imitating this and other organic principles would be more compatible with the organic domain around it.

Until recently the mandate to "do as nature does" has been impossible to implement among isolated and rigid machines. But as we invest machines, factories, and materials with adaptive behavior, coevolutionary dynamics, and global connections, we can steer the manufactured environment into an industrial ecology. Doing so shifts the big picture from industry conquering nature to industry cooperating with nature.

Hardin Tibbs is a British industrial designer who picked up a sense of machines as whole systems while consulting on large engineering projects such as the NASA space station. To make a remote space station, or any other large system, utterly reliable requires steady attention to all the interacting, and at times conflicting, needs of each mechanical subsystem. Balancing several machines' opposing demands, while unifying common ones, instilled a holistic attitude in engineer Tibbs. As an avid environmentalist, Tibbs wondered why this holistic mechanical outlook -- which stresses a systems approach to minimizing inefficiencies -- could not be applied to industry in general as a way to solve the pollution it generated. The idea, said Tibbs, was to "take the pattern of the natural environment as a model for solving environmental problems." He and his fellow engineers were calling it "industrial ecology."

The term "industrial ecology" was a metaphor resurrected by Robert Frosch in a 1989 Scientific American article. Frosch, a scientist who runs GM's research laboratories and was once head of NASA, defined this fresh perspective: "In an industrial ecosystem...the consumption of energy and materials is optimized, waste generation is minimized, and the effluents of one process... serve as the raw material for another process. The industrial ecosystem would function as an analogue of biological ecosystems."

The term industrial ecology had been used since the 1970s as a way to think about workplace health and environmental issues, "stuff like whether you have mites living on dust particles in your factory or not," says Tibbs. Frosch and Tibbs expanded the concept of industrial ecology to include the environment formed by and among a web of machines. The goal according to Tibbs was "to model the systemic design of industry on the systemic design of the natural system" so that "we could not only improve the efficiency of industry but also find more acceptable ways of interfacing it with nature." In one daring step, engineers hijacked an age-old metaphor of machines as organisms and put the poetry to work.

One of the first ideas born out of the organic view of manufacturing was the notion of "design for disassembly." Ease of assembly has been the paramount factor in manufacturing for decades. The easier something was to assemble, the cheaper it could be made. Ease of repair and ease of disposal were almost wholly neglected. In the ecological vision, a product designed for disassembly would combine the tradeoffs of efficient disposal or repair as well as efficient assembly. The best-designed automobile, then, would not only be a joy to drive, and cheap to assemble, but would also easily break apart into common ingredients when dead. These technicians aim to invent devices that adhere better than glues or one-way fasteners, but are reversible, and materials as sound as Kevlar and molded polycarbonate, but are easier to recycle.

The incentive for these inventions is imposed by requiring the manufacturer, rather than the consumer, to be responsible for disposal. It pushes the burden of waste "upstream" to the producer. Germany recently passed legislation that makes it mandatory for automobile manufacturers to design cars that dismantle easily into homogeneous parts. You can buy a new electric tea kettle featuring easy-to-dismember recyclable parts. Aluminum cans are already designed for recycling. What if everything else was? You couldn't make a radio, a running shoe, or a sofa without accounting for the destination of its dead body. You'd have to work with your ecological partners -- those preying upon your machine's matter -- to ensure someone took on your corpses. Every product would incorporate its engineered offal.

"I think that you can go a long way with the idea that any waste you can think of is a potential raw resource," Tibbs says. "And any material that might not have a use right now, we can eliminate upstream by design so that that material is not produced. We already know, in principle, how to make intrinsically zero-pollution processes. The only reason we aren't doing so is because we haven't decided to do it. It's a matter of volition rather than technology."

All evidence points to ecological technology being cost effective, if not shockingly profitable. Since 1975, the global conglomerate 3M has saved $500 million while reducing pollution 50 percent per unit of production. By reformulating products, modifying production processes (to use less solvents, say), or simply by recovering "pollutants," 3M has made money by applying technical innovations to its internal industrial ecology.

Tibbs told me of another example of an internal ecosystem that pays for itself: "In Massachusetts a metal refinishing plant had been discharging heavy metal solutions into the local waterways for years. And every year the environmental people were raising water-purity thresholds, until it got to the point where the plant would either have to stop what they were doing and farm out the plating to somewhere else, or install a very expensive state-of-the-art full-scale water treatment plant. Instead the refinishers did something radical -- they invented a totally closed-loop system. Such a system did not exist in electroplating."

A closed-loop system constantly recycles the same materials over and over again, just as Bio2 does or a space capsule should. In practice small amounts leak in and out in industrial systems, but overall, the bulk of mass circles in a "closed loop." The Massachusetts plating company devised a way to take the tremendous amounts of water and toxic solvents demanded by the dirty process and recycle them entirely within the walls of the factory. Their innovative system, which reduced pollution output to zero, also paid for itself in two years. Tibbs says, "The water treatment plant would have cost them $1/2 million, whereas their novel closed-loop system cost only about $1/4 million. They saved on water costs by no longer needing 1/2-million gallons per week. They reduced their chemical intake because they now reclaim the metals. At the same time they improved the quality of their plating product because their water filtration is so good that the reused water is cleaner than the local water they bought before."

Closed-loop manufacturing mirrors the natural closed-loop production in living plant cells, which internally circulate the bulk of their materials during nongrowth periods. The same zero-pollution closed-loop principles in a plating factory can be designed into an industrial park or entire region. Add a global perspective and you up the scale to cover the entire planetary network of human activity. Nothing is thrown away in this grand loop because there is no "away." Eventually, all machines, factories, and human institutions will be members of the greater global bionic system that imitates biological manners.

Tibbs can already point to one ongoing prototype. Eighty miles west of Copenhagen, local Danish businesses have cultivated an embryonic industrial ecosystem. About a dozen industries cooperate in exploiting "wastes" from neighboring factories in an open-loop which is steadily "closing in" as they learn how to recycle each other's effluent. A coal-fired electric power plant supplies an oil refinery with waste heat from its steam turbines (previously released into a nearby fjord). The oil company removes polluting sulfur from gas released by the refining process which can then be burned by the power plant, saving 30,000 tons of coal per year. The removed sulfur is sold to a nearby sulfuric acid plant. The power plant also precipitates pollutants from its coal smoke in the form of calcium sulfate, which is consumed as a substitute for gypsum by a sheetrock company. Ash removed from the same smoke goes to a cement factory. Other surplus steam from the power plant warms a biotech pharmaceutical plant and 3,500 homes, as well as a seawater trout farm. Hi-nutrient sludge from both the fish farm and the pharmaceutical factory's fermentation vats are used to fertilize local farms, and perhaps someday soon, also horticulture greenhouses warmed by the power plant's waste heat.

Yet, to be realistic, no matter how cleverly manufacturing loops are closed, a tiny fraction of energy or unusable stuff will be wasted into the biosphere. The impact of this inevitable entropy can be absorbed by the organic sphere if the mechanical systems that generate it run at the pace and scope of natural systems. Living organisms such as water hyacinth can condense dilute impurities in water into a concentration with economic value. In '90s lingo, if industry interfaces well with nature, biological organisms can carry what minimal waste the industrial ecosystem generates.

The bugaboos in larger versions of this optimistic vision are highly variable flows of material, and decentralized, dilute concentrations of reclaimable stuff. Nature excels in dealing with variance and dilute being, while human artifacts do not. A multi-million-dollar paper recycling plant needs an unvarying stream of constant quality old paper to operate; it cannot afford to be down a day if volunteers tire of bundling their used newspapers. The usual solution, massive storage centers for recycled resources, burns up its slim profitability. Industrial ecology must grow into a networked just-in-time system that dynamically balances the flow of materials so that local overflows and shortages are shuttled around to minimize variable stocks. More net-driven "flex-factories" will be able to handle a more erratic quality of resources by running adaptable machinery or making fewer units of more different kinds of products.