Putting Ideas to Work: The Case of Xerox PARC
Xerox Parc was always known for innovative ideas. The trouble was, someone else always commercialized them. That has changed. Here's how Xerox did it.
Management of technology is a critical issue, particularly in the computer and communications industries, where hardware and systems change rapidly. Critical choices on which technologies to develop and which to ignore can determine the shape and fortunes of entire industries.
Executives in Silicon Valley, with the considerable benefit of hindsight, agree that two such choices stand out from the infancy of the personal computer era. One was the International Business Machines Corporation's decision that it was a computer, not a software, company. That enabled the rise of Bill Gates and the Microsoft Corporation to dominate operating systems and applications. The other was the Xerox Corporation's decision that it was a copier, not a computer, company. That led to the abandonment of the Alto, the world's first practical personal computer, which had been developed by the company's Palo Alto Research Center (Parc).
The use of icons, windows, point-and-click commands, local area networks and other features that are now part of the foundation of the personal computer industry were developed by Parc, but exploited by others. Indeed, even Xerox admits that "whole companies have been built on inventions born at Parc."
One of the distinguishing characteristics of Xerox is that, as a corporation, it still believes in the value of research. This is in contrast to many other manufacturing companies that have largely eliminated research in favor of incremental development of existing products. While Xerox may have fumbled the dawn of the personal computer era because top management was preoccupied with defending its core copier business from a Japanese invasion, current Xerox leaders are determined to capitalize on the research flowing from Parc and its other laboratories. These include new methods of capturing and displaying high-resolution images and new connections between the oceans of electronic information and the harried user. Other attractive technology that is not of much use in the printing and copying business is spun off to small companies that are operated independently, but remain in the Xerox fold.
A few years ago the director of research for a major American corporation, whose laboratories had produced Nobel Prize winners, articulated this changed attitude when he said that he was not interested in inventing anything new. What he was looking for, he said, were developments that would augment the company's 13 existing lines of business.
Keep Pushing The Outer Limits
"We didn't go short term," said Mark Weiser, the computer scientist who is chief of technologies at Parc. "It was not one of these things where we were told 'don't invent anything that we don't know exactly how to use immediately.'" He said the instruction from corporate headquarters was for the scientists to keep pushing the outer limits of technology related to handling information while the company pondered ways of extracting benefits from their innovations.
One thing the Alto incident appears to have embedded in Xerox's corporate culture is how easy it is for even an outstanding idea to get trampled in a more than $17 billion company whose main business is still copiers and printers. Xerox spends $1.6 billion a year on research and development, although the vast bulk of that goes into the printers and copiers that are in the midst of a transition from stand-alone machines using optical technology to networked systems that process and transmit documents digitally.
"Generally, the technologies closest to the company's core-business focus are the ones that receive attention and funding," the company has conceded. "Companies have pursued varying avenues to capitalize on their total intellectual property, but few have been successful."
One approach to developing new ideas was to emulate the venture capitalists whose investments have fueled the fabulous growth of Silicon Valley. The company established Xerox Technology Ventures in 1989 to invest in promising new technologies. Though financially successful, X.T.V. was an outward-looking process that did little to move internal research to the market. Seeking a better method, in 1996 the company folded X.T.V. into a new entity, Xerox New Enterprises.
"Venture capitalists come in late in the game," said Andrew R. Garman, vice president for strategy and new business development for Xerox New Enterprises. "They want to be in the market within six months and they care more about the senior management team than developing technology."
In fact, many technological entrepreneurs have complained about what they call U-shaped capital for new ventures. Funding can be found for basic research, they say, and venture capital can take promising products to market. Finding support, however, for developing research into a product can be extraordinarily difficult.
Within the Xerox context, Xerox New Enterprises tries to deal with this problem by taking technologies not directly applicable to today's printing and copying business and using it as the basis for semi-independent companies. Typically, a team of researchers joins the company to continue technical development while outside executives with relevant experience are recruited to manage the business functions.
New Technology Needs Different Market Research
Xerox executives say they need to set up speculative ventures such as these because new technology does not lend itself to standard market research techniques. "We have to find a market beachhead," Mr. Garman said. "Business school focus groups can find out whether people want blue or green speckles in their detergent, because they already know what detergent is," he said. "But how do you do it with something completely new?"
Typically the parent company provides seed funding and retains 80 percent of the stock. The remaining 20 percent is reserved for the managers of the new companies, who stand to profit handsomely if the company is successful and can float an initial public stock offering in a few years.
Xerox executives at Palo Alto note that years of backbreaking work followed by an initial public offering and instant wealth is an accepted practice in Silicon Valley, although perhaps less familiar to managers in Rochester, N.Y.
"We have to reorient the managers of these companies to a new way of doing business," Mr. Garman said. "They have to abandon the Xerox pension, the profit share and the stock options. They have to adjust to life outside big, steady Xerox and bet on the upside of an IPO."
Outside Regular Business Structure
The New Enterprises companies (10 of them so far) report to a New Enterprises board at headquarters and ultimately to Xerox's chief financial board. This is to make it clear that the new companies are a corporate undertaking and are not a part of the company's regular business structure.
Xerox's willingness to experiment with new forms of organization, whose combined sales don't amount to much more than a decimal point in total revenues, stems from a realization that the company was not getting an adequate return from its investment in research, said Robert A. Burgelman, a professor of management at Stanford University's school of business.
Professor Burgelman, who consulted for Xerox in the late 1980's and early 1990's, but who has had no recent connection, said Xerox managers now have "a more sophisticated understanding of the limitations of the business to pursue undertakings beyond the core business." He said the company has many functional managers immersed in the details of its reprographics operations, but few general managers with responsibility to look afield.
This, Professor Burgelman said, is why Xerox has had difficulty transferring technology from Parc to operations outside the basic business. (Xerox officials, though, are quick to point out that one of their hottest products, the high-speed Docutech printer, is the result of Parc research.)
Entrepreneurial Energy
Because the X.N.E. companies stay a part of Xerox, with access, for example, to the company's worldwide marketing network, the outcomes for them may be more varied than the sale-or-shutdown scenario facing most venture-capital-based companies. Even if there is an initial public offering, Xerox is expected to retain an interest in the companies as a way of participating in their future results.
In fact, if a company is successful and develops a product line that fits into the core company, Xerox may reabsorb it, buying back the stock that had been purchased by employees. Because there is no public market yet for the shares, each of the new companies is given a valuation each year to determine what price the shares would command.
Mr. Garman describes X.N.E.'s approach as "a new model to bring entrepreneurial energy to technology." He doesn't argue that it is similar to that of the Thermo Electron Corporation, which has partly spun off 21 publicly traded companies over the past 15 years.
Mr. Garman said it will take three to five years to determine if the X.N.E. companies are suitable for an initial public offering, should be returned to Xerox or need to be closed down. It is inevitable, he said, that some of the X.N.E. companies will not be able to make it on their own, given the uncertain and changing markets for their products. "If we do not end a few it will be a signal that we haven't taken enough risk," he said. "The venture capitalists have a 50 percent failure rate, but we should be able to do a lot better than that."
Flat-Panel Displays And Radiation Sensors
One of the more notable of the Xerox start-ups is a company called DpiX (pronounced depicts), which uses the capitalized X to call attention to the connection with Xerox. The president of the two-year-old company, Malcolm Thompson, came from Parc, but most of the other senior managers have backgrounds in the commercial electronics industry. The company is marketing high-resolution, flat-panel displays and sensors to the military and medical electronics industries, with plans to enter new markets as the product line is expanded.
The displays and sensors are based on thin-film transistor-on-glass technology originally developed for inclusion in Xerox products. Since the military has considerable interest in replacing cathode ray tubes in warplanes with flat panels, the Federal-government-backed research has boosted the quality of the displays to near that of photographic film.
One of DpiX's goals is the replacement of X-ray film with faster, easier-to-handle radiation sensors. In a medical imaging device, the sensor, which is less than an inch thick, captures the X-ray radiation after it passes through a person's body and converts the image to digital signals that can be read almost immediately on a computer monitor. This eliminates the delay involved in chemically processing X-ray film and does away with the need to maintain a film room at hospitals to store old images. Because the information is stored digitally, it can be transmitted to remote locations for additional consultation.
"This could be a big market for us," said Mr. Thompson. Analysts have estimated that medical sensors could be a $100 million-a-year business by the year 2000 because of the need for faster diagnostics. Because DpiX operates independently from Xerox, except for periodic financial and progress reviews, Mr. Thompson said he is free to enter into alliances without assistance or interference from headquarters.
For example, to get display panels into the lucrative but demanding military market, DpiX has formed a joint development, manufacturing and marketing agreement with Planar Advance Inc., the largest supplier of military displays in North America and Europe. As part of the deal, DpiX will design and fabricate the panels, which Planar will modify for military requirements, and then use them as replacements for conventional instruments in airplane cockpits. Because of the ability to display large amounts of information, one panel can replace multiple instruments, saving valuable space in crowded airplanes. Although the end of the Cold War has reduced spending for new aircraft, Mr. Thompson said there appears to be a large opportunity to retrofit existing planes with the advanced panels.
DpiX plans to broaden its product line into offerings for the graphic arts and prepress industries, as well as digital photography, video and multimedia. Central to its strategy is the development of a highly flexible manufacturing facility, which will be capable of producing short runs of custom-designed displays and sensors. Because of its need for military products, the Department of Defense is underwriting the conversion of a research fabrication plant into one capable of commercial production.
Mr. Thompson said the profitable opportunities in the display and sensor business appear to be at the top end of the market, not in the mass-produced but lower-resolution liquid crystal displays that are part of laptop computers.
"Our customers want us to design and produce specialized products and it appears that this can be an attractive and profitable business," he said. The commodity display business, he added, has been beset with overexpansion. "Our friends in Asia have not been very profitable" producing displays for laptops, he said.
With enough flexibility in manufacturing, Mr. Thompson said, DpiX can produce a wide variety of products and make money at the same time. "If you use essentially the same process to do all those different products, it can be cost effective."
Despite his background in research, Mr. Thompson brings an unusual appreciation for the practical details of making products in the real world. He is a founder and is chairman of the United States Display Consortium, a government-industry organization that is working on the development of a flat-panel manufacturing base in the United States.
Mr. Thompson said the plans being developed for DpiX would probably not be suitable for the other Xerox spin-offs. "That's the reason you give companies the freedom to run their businesses individually," he said. "You are probably going to have a unique business model for each one."
More User Friendly
Xerox's development of point-and-click commands, icons, pull-down windows and other picture-like controls are what the industry calls the graphical user interface. Although those developments dating back to the 1970's helped create the personal computer industry by freeing it from the complex written commands that previously prevailed, scientists at Parc say a new generation of interfaces is needed to cope with today's explosion of information. That is the thinking behind one of the newer spinoffs, InXight (pronounced insight), which is marketing what its officials call knowledgeware to producers of commercial software.
InXight, like DpiX, plans to stay in the background as a supplier of a key component to more visible companies that will do the sales and marketing. Customers include Microsoft, Oracle, Comshare and Verity. Nevertheless, its objectives are bold. Clicking from screen to screen was fine while the amount of electronic information available was relatively small. Now, InXight executives say, so much is available that existing search mechanisms are being overwhelmed.
"A window's interface can handle about 50 items before it breaks down," said Mohan Trikha, the president of InXight. "What we need is a system that can handle thousands of objects."
The other problem with existing systems is that users get lost as they go from one screen to another. One Parc scientist described the process as looking at the world with paper-towel tubes attached to a person's eyes. There is no peripheral vision. "People are tired of clicking hundreds of pages," he said. "You don't remember where you went or what came beforehand."
Some of the researchers who developed the original user interface have been working on this problem for a long time. In fact, Mr. Trikha said, some of the solutions have been on the shelf at Parc for a decade. The proposed solution is a new browser for the Web that shows where a user started and all the intermediate steps to the page being viewed. Called the Hyperbolic Tree, the program uses perspective to keep previous pages in sight, but reduced in size, as the user searches through the maze of information.
InXight is also offering a product that searches data bases using common-language terms rather than the stylized requests needed for conventional searches. Called LinguisX, the system can analyze not only words, but root relationships. For example, Parc officials say it can recognize that "buying" and "bought" are different forms of the same word. The system can also automatically prepare summaries of documents by studying the basic concepts contained in them.
"All language can be specified as data," Mr. Trikha said. "We have the capability of 13 languages now and within two years we expect to have 50."
Mr. Trikha said the older search processes were designed on the assumption that computers were expensive and humans cheap. "Now we assume that computers are cheap and we need to adapt them to human cognitive ability."
Top Xerox executives realized that they needed to find a way to commercialize these developments without trying to compete directly with established software companies, Mr. Trikha said. "The corporation understood that new innovations need a new type of management. They did not want the technology to disappear."
Arm Wrestling Results In Draw
In "Fumbling the Future," their classic study of how the Xerox Corporation lost the opportunity to dominate the personal computer industry, Douglas K. Smith and Richard C. Alexander made much of the cultural clash between the irreverent, informal scientists at the Palo Alto Research Center (Parc) and the managers based at the corporate headquarters in Stamford, Conn., and the operating headquarters in Rochester, N.Y.
"Parc was floating around in free space," they quote a Xerox executive as saying. "Parc was a head. But a head to which body? Who was going to pick up from whatever was done at Parc and do all of the rest of the hard work to make a business of it?"
Parc researchers in the 1970's, as the concepts for the personal computer were being developed, were young (many in their first jobs beyond academia) and had little respect for corporate executives who did not know the then-arcane terms of computer science. At the forefront of technology, they admitted to arrogance toward pin-striped managers who did not understand what they were doing. "When we felt sometimes that someone was not worth talking to, we sometimes told them that," one scientist of the era conceded.
Relations between Parc and the corporation have improved because they had to and because of societal change, argues Mark Weiser, Parc's chief technologist. "Young people coming here now are interested in commercial development of their research, which was not true in the past," said Mr. Weiser, who is 45. "When I was starting out, Albert Einstein was my hero. To the younger people, Bill Gates is their hero."
He said some of the stylistic differences that caused irritation in the past -- pony-tailed, sandaled scientists meeting with executives in blue suits -- have eased over the years. "When we are visited by a group from Stamford, they seldom arrive in suits. And more and more people from Parc own suits."
Still, no one is likely to confuse Parc with the accounting department at Xerox. Mr. Weiser's resumé notes that he is the drummer for a rock group called "Severe Tire Damage," which bills itself as the first band on the Internet. It also notes that he was awarded a Ph.D. by the University of Michigan without benefit of an undergraduate degree.
Mr. Weiser said the lab is closely tied to the corporation, with its director, John Seely Brown, reporting directly to the chairman of Xerox, Paul A. Allaire. In addition, Mark Myers, the head of corporate research and technology, is a member of the six-person management committee that runs the company.
One notable characteristic of Parc is that it employs social scientists, psychologists and anthropologists to study how people relate to machines in the real world. They help curb the tendency of experts to load up devices with features that are maddening complications for ordinary users.
"We are leaning toward the 'big green button' philosophy," Mr. Weiser said, explaining that there should be a single control, even for complex machines, that simply needs to be turned on to make the device work.
Parc is also trying to tie its research more closely to product development. When customers expressed a desire for faster laser printers, Parc officials linked that with the physics laboratory's development of a laser diode that can emit two beams of light at once. Two beams means the printer can be twice as fast.
"We transferred the technology from Parc to a laser diode manufacturing company that, in turn, sells the diodes to Xerox," Mr. Weiser said. "It's a complicated transaction, but we're doing what it takes to get research into a product."
Reprint No. 98108
Authors
John Holusha, John Holusha has covered business and technology for The New York Times. |