Cheaper than printing it out: buy the paperback book.

One can imagine the future shape of companies by stretching them until they are pure network. A company that was pure network would have the following traits: distributed, decentralized, collaborative, and adaptive.

Distributed -- There is no single location for the business. It dwells among many places concurrently. The company might not even be headquartered in one place. Apple Computer, Inc., has numerous buildings spread thickly over two towns. Each one is a "headquarter" for a different function of the company. Even small businesses may be distributed within the same locality. Once networked, it hardly matters whether you are on the floor below, or across town.

Open Vision, based in Pleasanton, California, is an example of a rather ordinary, small software company, molded in the new pattern. "We are operating as a true distributed company," said CEO Michael Fields. Open Vision has clients and employees in most US cities, all served on computer networks, but "most of them don't even know where Pleasanton is," Fields told the San Francisco Chronicle.

Yet in this stretch toward ultimate networks, companies will not break down into a network of individuals working alone. The data collected so far, as well as my own experience, says that the natural resolution of a purely distributed company coalesces into teams of 8 to 12 people working in a space together. A very large global company in the pure network form could be viewed as a system of cells of a dozen people each, including minifactories manned by a dozen people, a "headquarters" staffed with a dozen, profit centers managed by eight and suppliers run by ten people.

Decentralized-How can any large -- scale project ever get anything done with only ten people? For most of the industrial revolution, serious wealth was made by bringing processes under central control. Bigger was more efficient. The "robber barons" of yesteryear figured out that by controlling every vital and auxiliary aspect of their industry, they could make millions. Steel companies proceeded to control the ore deposits, mine their own coal, set up their own railways, make their own equipment, house their own workers, and strive for self-containment within the borders of a gigantic company. That worked magnificently when things moved slowly.

Now, when the economy shifts daily, owning the whole chain of production is a liability. It is efficient only while the last hours of its relevancy lasts. Once that moment of power recedes, control has to be traded in for speed and nimbleness. Peripheral functions, like supplying your own energy, are quickly passed on to another company.

Even supposedly essential functions are subcontracted out. For instance, Gallo Winery no longer grows the specialized grapes required for its wines; it farms that chore out to others and focuses on brewing and marketing. A car rental company subcontracts out the repair and maintenance of its fleet, and focuses on renting. One passenger airline subcontracted its cargo space on transcontinental flights (a vitally important profit center) to an independent freight company, figuring they would manage it better and earn the airline more than it could itself.

Detroit automobile manufacturers were once famous for doing everything themselves. Now they subcontract out about half of their functions, including the rather important job of building engines. General Motors even hired PPG Industries to handle the painting of auto bodies -- a critical job in terms of sales -- within GM's factories. In the business magazines this pervasive decentralization by means of subcontracting is called "outsourcing."

The coordination costs for large-scale outsourcing have been reduced to bearable amounts by electronic trading of massive amounts of technical and accounting information. In short, networks make outsourcing feasible, profitable, and competitive. The jobs one company passes off to another can

be passed back several times until they rest upon the shoulders of a small, tightly knit group, who will complete the job with care and efficiency. That group will most likely be a separate company, or they may be an autonomous subsidiary.

Research shows that the transactional costs needed to maintain the quality of a task as it stretches across several companies are higher than if the job stayed within one company. However: (1) those costs are being lowered every day with network technology such as electronic data transfers (EDI) and video-conferencing, and (2) those costs are already lower in terms of

the immense gains in adaptability -- not having to manage jobs you no longer need, and being able to start jobs you will need -- that centralized companies lack.

Extending outsourcing to its logical conclusion, a 100 percent networked company would consist solely of one office of professionals linked by network technology to other independent groups. Many invisible million-dollar businesses are being run from one office with two assistants. And some don't have an office at all. The large advertising firm of Chiat/Day is working on dismantling its physical headquarters. Project team members will rent hotel conference rooms for the duration of the project, working on portable computers and call-forwarding. They'll disband and regroup when the project is done. Some of those groups might be "owned" by the office; others would be separately controlled and financed.

Let's imagine the office of the future in a hypothetical Silicon Valley automobile manufacturer that I'll call Upstart Car, Inc. Upstart Car intends to compete with the big three Japanese automobile giants.

Here's Upstart's blueprint: A dozen people share a room in a sleek office building in Palo Alto, California. Some finance people, four engineers, a CEO, a coordinator, a lawyer, and a marketing guy. Across town in a former warehouse, crews assemble 120-mpg, nonpolluting cars made from polychain composite materials, ceramic engines, and electronic everything else. The hi-tech plastics come from a young company with whom Upstart has formed a joint venture. The engines are purchased in Singapore; other automobile parts arrive each day in bar-coded profusion from Mexico, Utah, and Detroit. The shipping companies deal with temporary storage of parts; only what is needed that day appears at the plant. Cars, each one customer-tailored, are ordered by a network of customers and shipped the minute they are done. Molds for the car's body are rapidly shaped by computer-guided lasers, and fed designs generated by customer response and targeted marketing. A flexible line of robots assemble the cars.

Robot repair and improvement is outsourced to a robot company. Acme Plant Maintenance Service keeps the factory sheds going. Phone reception is hired out to small outfit physically located in San Mateo. The clerical work is handled by a national agency who services all the other groups in the company. Same with computer hardware. The marketing and legal guys each oversee (of course) the marketing and legal services which Upstart also hires out. Bookkeeping is pretty much entirely computerized, but an outside accounting firm, operating from remote terminals, tends to any accounting requests. In total about 100 workers are paid directly by Upstart, and they are organized into small groups with varying benefit plans and pay schedules. As Upstart's cars soar in popularity, it grows by helping its suppliers grow, negotiating alliances, and sometimes investing in their growth.

Pretty far out, huh? It's not so farfetched. Here's how a real pioneering Silicon Valley company was launched a decade ago. James Brian Quinn writes in the March-April 1990 Harvard Business Review :

Apple bought microprocessors from Synertek, other chips from Hitachi, Texas Instruments, and Motorola, video monitors from Hitachi, power supplies from Astec, and printers from Tokyo Electric and Qume. Similarly, Apple kept its internal service activities and investments to a minimum by outsourcing application software development to Microsoft, promotion to Regis McKenna, product styling to Frogdesign, and distribution to ITT and ComputerLand.

Businesses aren't the only ones to tap the networked benefits of outsourcing. Municipalities and government agencies are rapidly following suit. As one example out of many, the city of Chicago hired EDS, the computer outsourcing company Ross Perot founded, to handle its public parking enforcement. EDS devised a system based on hand-held computers that print out tickets and link into a database of Chicago's 25,000 parking meters to increase fine collection. After EDS outsourced this service for the city, parking tickets that were paid off jumped from 10 percent to 47 percent, raising $60 million in badly needed income.

Collaborative -- Networking internal jobs can make so much economic sense that sometimes vital functions are outsourced to competitors, to mutual benefit. Enterprises may be collaborators on one undertaking and competitors on another, at the same time.

Many major domestic airlines in the U.S. outsource their complex reservation and ticketing procedures to their competitor American Airlines. Both MasterCard and Visa credit card companies sometimes delegate their vital work of processing customer charges and transactions to arch-competitor American Express. "Strategic Alliances" is the buzz word for corporations in the 1990s. Everyone is looking for symbiotic partners, or even symbiotic competitors.

The borders between industries, between transportation, wholesaling, retailing, communications, marketing, public relations, manufacturing, warehousing all disappear into an indefinite web. Airlines run tours, sell junk by direct mail, arrange hotel reservations, while computer companies hardly even handle computer hardware.

It may get to the point that wholly autonomous companies become rare. The metaphor for corporations is shifting from the tightly coupled, tightly bounded organism to the loosely coupled, loosely bounded ecosystem. The metaphor of IBM as an organism needs overhauling. IBM is an ecosystem.

Adaptive -- The shift from products to service is inevitable because automation keeps lowering the price of physical reproduction. The cost of copying a disk of software or a tape of music is a fraction of the cost of the product. And as things continue to get smaller, their cost of reproduction continues to shrink because less material is involved. The cost of manufacturing a capsule of drug is a fraction of the cost it sells for.

But in pharmaceutical, computer, and gradually all hi-tech industries, the cost of research, development, stylizing, licenses, patents, copyrights, marketing and customer support -- the service component -- are increasingly substantial. All are information and knowledge intensive.

Even a superior product is not enough to carry a company very long these days. Things churn so fast that innovative substitutions (wires built on light instead of electrons), reverse engineering, clones, third party add-ons that make a weak product boom, and quickly shifting standards (Sony lost badly on Beta VCRs but may yet prevail with 8-mm tapes) all conspire to bypass the usual routes to dominance. To make money in the new era, follow the flow of information.

A network is a factory for information. As the value of a product is increased by the amount of knowledge invested in it, the networks that engender the knowledge increase in value. A factory-made widget once followed a linear path from design to manufacturing and delivery. Now the biography of a flexibly processed widget becomes a net, distributed over many departments in many places simultaneously, and spilling out beyond the factory, so that it is difficult to say what happens first or where it happens.

The whole net happens at once. Marketing, design, manufacturing, suppliers, buyers are all involved in the creation of the successful product. Designing a product concurrently entails having marketing, legal, and engineering teams all design the product at once, instead of sequentially as in the past.

Retail products (cans of soda, socks) have communicated their movement at the cash register to the back office since the 1970s when the UPC bar code became popular in stores. However in a full-bore network economy, the idea is to have these items communicate to the front office and customer as well by adding weak communication abilities. Manufacturing small items with active microchips instead of passive bar codes embedded into them means you now have hundreds of items with snail-minds sitting on a shelf in a discount store by the thousands. Why not turn them on? They are now smart packages. They can display their own prices, thank you, easily adjusting to sales. They can recalculate their prices if the store owner wants to sell them at a premium or if you the shopper are carrying a coupon or discount card of some sort. And a product would remember if you passed it over even after seeing the sale price, much to the interest of the store owner and manufacturer. At least you looked, boasts the product's ad agency. When shelf items acquire awareness of each other and themselves and interact with their consumers, they rapidly erupt into a different economy.