Pacific Connection(英語)

Keeping the R&D Payoff In-House

この記事を読むのに必要な時間:およそ 10.5 分

Everyone, including Xerox, seems to have learned the lesson of Xerox PARC

You already know this story, but it still bears repeating. One day in December, 1979, Steve Jobs was visiting Xerox's Palo Alto Research Facility (PARC) when he saw an early version of the now ubiquitous graphical user interface. Xerox PARC invented the GUI, but being principally a photocopier company, didn't really have the drive to bring it to market. Instead, it took Job's unique blend of crazed cheerleader and daring entrepreneur to transform the GUI into a viable product in the form of the Apple Macintosh.

This story has become an industry legend-an instructive tale of how even the most stellar research and development effort may wind up being someone else's opportunity. The golden egg hatches in the research lab, but either the corporate structure, company culture, or just plain shortsightedness fails to bring the idea to market. The Xerox PARC story is famous in the industry because the lab invented not only the GUI, but the local area network and object oriented programming-none of which have brought great wealth to Xerox Corporation. And of course PARC is not the only example. At one time, Apple Computer itself had at least three operating system projects in operation simultaneously, including its Taligent project with IBM and Copeland. Despite the obvious importance to Apple of upgrading its vintage 1980s OS, none of these research efforts paid off. Jim Carlton, in his recent book on Apple, blamed it on a corporate culture which placed too much emphasis on consensus, and in which too many engineers wanted to add too many features, too late in the design process. Not only did Apple squander millions of R&D dollars with these unfocused efforts, but it eventually had to purchase a third-party OS-NeXTStep.

And then there's the relational database. IBM researcher Ted Codd described the relational data model in the 1970 article, Relational Database Technology. But it was a brash kid from Chicago, Larry Ellison, who beat IBM to market with the first relational database product. Today Larry Ellison's Oracle Corporation is easily the marketshare leader in relational DBMSs and Ellison is a multi-billionaire.

If there's a moral to all this, it has not been lost on Burt Sutherland, the research director of Sun Research Laboratory at Sun Microsystems. Sutherland once ran the Systems Sciences lab at PARC and was in the room when Jobs saw the GUI demo. Now he is in a position to make corporate research pay off. "We're strongly focused on value for the company," he says. "While we've had some far out developments, including the project that turned into Java, most of our efforts are aimed clearly at opportunities for the company." Sutherland is part of a new generation of corporate researchers who, while still reaching over the horizon, retain a strong pragmatic streak. "A researcher needs to sell me on the plausible value of an idea, because we are not going to pursue an irrelevant direction," he says. When a project is ready for development, the lab has a set procedure for moving it into the company."

One way Sun is ensuring that R&D costs don't spiral out of control is by limiting the size of the lab. While Sun has grown three times since 1990, SunLabs' personnel has expanded from 100 to just 120 researchers. The rest of Sun's R&D budget goes to Sun's hardware and software operating companies, where-in one of the world's most competitive industries-it provides some margin of error.

"If a project can get started in a development group, that's the place for it," Sutherland says. "I don't like to do things that the rest of the company understands or is willing to do, so I avoid the duplication that way. To start a new project, I need a plausible idea with relevance, but most importantly, I need a champion. Creativity is a very individual thing, so I need the fairly senior person who will stand up and say 'yes, I think this is a neat problem and I'm going to devote the next couple of years of my career into making this work.' And then I have to get enough resources to get to a critical mass, because there is a great tendency to try to do too many things at half speed. Progress at that rate is almost zero.

"One of the things we try and do is leverage the smarts. When we started the lab, Bill Joy said a wonderful thing-that innovation happens all over the world, most of it not at Sun. So another role of this research laboratory is as an intellectual import station. We bring new thoughts into the company, and lots of them come from outside. That's very much in keeping with Sun's philosophy as an open company. We say our competitive advantage comes from running like hell. I've shot and wounded the tiger-you skin him-and I'll go back and find another one."

Sutherland says that SunLabs is engaged in much more near-term activity than Xerox PARC ever was-with a few notable exceptions. "We've had some few, far out crazy things in the lab, one of which turned into Java, which at the time was not directly coupled to Sun's existing business. But by and large, most of our efforts range pretty close to three to five years on some kind of company opportunity."

Sun, of course, is hardly alone. One of the reasons Microsoft has conquered the world is that its R&D effort is, by and large, disciplined and focused. The company founded its R&D arm-Microsoft Research-in 1992, and plans to expand it from about 230 to 600 researchers. Most research is done in Redmond, WA, with satellites in the Bay Area and Cambridge in the U.K. Areas of research include speech recognition, decision theory, and 3D graphics and animation.

To make certain that ideas don't go astray, the lab's research chief, Nathan Myhrvold, reports directly to Bill Gates. "Nathan answers to Bill only. It's a very short chain," says Dan Rosen, Microsoft general manager, new technology. "And much of the research projects are to some degree motivated by what Bill sees as the future of the company and the industry. That's not to say that product agendas direct the research. But if you came up with the next great gene splicing technology, the likelihood of pursuing it here is very small."

When Microsoft sees potential for one of its research projects, it is quick to jump. For example, the company identified natural language recognition as a promising science and hired world class researchers to pursue it. But it didn't hurt that early on the researchers identified a practical application for the technology-and what began as pure research project resulted in the grammar checker that is now a part of the Microsoft Word word processor.

In an interview in New Scientist magazine Myhrvold himself maintains that Microsoft now has the greatest collection of computer graphics experts in the world. More recently, the company has been hiring mathematicians-people who are not only interested in theory, but in how that theory can change the world. But Myhrvold also thinks product-oriented research can go too far. "One of the great mistakes that people make is directing research to be too applied," Myhrvold said. "It ceases to be research. It becomes development, and often loses the spark to make it worthwhile."

As for R&D payoffs, they have ranged from zero to astronomical. For example, the company is still working to bring its speech recognition products to work, a project that has gone on for six years now. On the other hand, Microsoft Windows operating have had had their origins with the group-a business worth billions of dollars to the company

U.S. Japan-pragmatic research

Because companies are under so much pressure to transform their research investment into revenues, most are taking a far more pragmatic approach, and most research leans toward the applied sciences. Moreover, this trend is taking place around the world, including, of course, Japan. "Only virtual monopolies can afford the luxury of a true research center-one doing university-style research with no expectation of a time limit to deliver product," says Eleanor Westney, Sloan Fellows Professor of Management, MIT Sloan School of Management. "Perhaps the only exception was in the late 1980s, when some Japanese companies set up basic research labs with ambitions to move to the forefront of their respective fields. But the downturn in the Japanese economy has put a lot of pressure on this effort. I wouldn't say they are thinking shorter term, but the Japanese have become more demanding of their R&D organizations in seeing more deliverables come out the pipeline."

Westney, whose academic focus is on Japanese R&D, does see some difference in how the two countries go about it. "One is that in the United States, the big research universities remain very powerful models for research. While it's less than in the past, there's still there is a link between universities and the big industrial research labs. For example, biotechnology as a field was largely created in universities, and many of the leading biotech firms have been set up by university professors on the basis of their research." She says that the cluster of technology companies near Boston known collectively as "Route 128" was largely created by entrepreneurs with very close links to MIT.

"By contrast, in Japan, the most prestigious research universities are national universities in which the professors are, at least formally, civil servants and therefore highly restricted in the kinds of relationships they can have with private industry. Japan has been making an effort in the last decade to loosen some of those restrictions, but they are still much more significant than in the U.S." Another difference: the U.S.'s tenure system drives younger researchers to do cutting-edge work, or lose their job, a system known as "publish or perish."

On the other hand, Japanese R&D, with its reliance on corporate labs and existing technology, has surpassed the U.S. in making incremental improvements that add to major breakthroughs, even through they are done step by step. Westney says that's especially an advantage for process-related industries, like consumer electronics, autos, and telecommunications equipment.

"In the United States, you still have a strong tendency for people who develop technology inside large corporations to take it out and set up a startup. And corporations, who are often reluctant to bet directly on the technology themselves, are often quite willing to have them do this. This means that you have a lot more entrepreneurial spin-offs-people developing technology and then going out and setting up their own companies. In Japan, you are much more likely to have the large companies following through on that technology to create a new division, or a corporate sponsor to spin off, where the company itself sets up the new business area."

One of the more unusual corporate research centers in the Silicon Valley straddles both of these worlds. Xerox FX Palo Alto Lab is an autonomous U.S. company that is 100 percent funded by Fuji Xerox, and provides contract work in the area of software development. (Fuji Xerox is a joint venture between Rank Xerox, which is fully owned by Xerox Corp., and Fuji Photo Film) James Baker, the lab's president and CEO, is well aware of Xerox PARC's legacy and it determined to learn from it. Part of his game plan is maintaining the lab's focus, which is to affect how software is developed at Fuji Xerox. "We also pay a lot of attention to how Fuji Xerox works. In Japan, for example , planning has a much stronger function within the company than it does in the U.S. That means it's important that we are plugged into Fuji Xerox's planning organizations. We're linked into the right technical people, so that we get joint projects and prototypes going in Tokyo, and ultimately make certain we see the technology at work."

The approach taken by companies like IBM, Sun and Apple is not for everyone. Cisco Systems, for example, has obtained most of its innovative ideas by buying up smaller companies-so that a savvy program of mergers and acquisitions has largely substituted for in-house R&D. These smaller companies of course are at their embryonic stages, strictly R&D machines. The very definition of a startup is a company working on a single technology that it hopes to bring to market. If the company succeeds, it may well be acquired. The founders make their millions, and sometimes, they go out and found another startup. That's the Silicon Valley model. Fueled by venture capital funding, the Silicon Valley itself might be considered one big R&D laboratory.

An interview with James Baker, president and CEO of Xerox FX Palo Alto Lab Based near Stanford University, Xerox FX Palo Alto Lab is an autonomous company that does contract work for Fuji Xerox. The lab itself has some 50 people from all over the world, including a small number of researchers from Fuji Xerox itself, who each spend two to three years at the facility. Much of the development work takes place under Windows NT, and while several programming languages are used, Java predominates.

At the helm is James Baker, who came to the lab after working in research at Honeywell and Schlumberger. He holds a Ph.D. in mathematics from Iowa State.

What's a research lab for Fuji Xerox doing in California?
First of all, Fuji Xerox has a significant research presence in Tokyo-8,600 people. They have historically done a lot of very good work for Xerox in bringing about new generations of copiers. But as Fuji Xerox looks to the future, they see software as a very big deal. In order to prepare themselves for this they want a strong technology base, which means research. And if you are going to do software research, there is hardly a better place than Silicon Valley. Also, being here in Silicon Valley, we have access to Xerox PARC, to the universities nearby, and to other software startup companies in this area.
Who comprises your lab? People from Japan? People from the U.S.?
Mostly the people are from the U.S. and we recruit from all over the world. We have three resident researchers from Japan, as well as people from the U.S., Taiwan, England, Denmark, and Holland-among others. We have a very heterogeneous group here. But we are run as an American company.
What areas are you specifically researching?
I recently saw a statistic that said that 70 percent of the revenue from the computer industry comes from products that are less than two years old. In other words, things are changing rapidly. With this as a background, we're trying to create an evolving vision of what the work experience for the year 2005 will look like.
Some themes that have emerged include multimedia and how it will be used in the workplace. Another theme is mobile computing. We see the desktop as we know it going away; a lot of the facilities people will be taking it with them. We see collaboration as being a very big deal, including shared virtual spaces-which is a strong theme for us-as well as how people collaborate person-to-person. For example, we built a state of the art conference room that is providing a testing ground for our research. The fourth theme is knowledge science, which is going to be a big deal for both Fuji Xerox, as well as Xerox.
The year 2005 seems pretty far out.
Determining how far out to look is a real challenge. If you go too far out and put your stake in the ground-and you're not strong enough to drive the whole industry in that direction-the industry may take a different turn and go around you. The year 2005 is about as far as we can go.
Do you think that this vision will be parallel both in the U.S. and in Japan, or do you see differences between the two countries?
I see a difference. In the end it will be about the same, but things are happening in the U.S. much faster than they are happening in Japan. You just have to look at the adoption of the Internet and intranet to see that. It is not taking place as fast in Japan now.
What about cultural differences between the U.S. and Japan. Do they affect your work?
Sometimes. For example, we have a collaborative space that we use in our laboratory. It's built with a base of some commercial software with some applications on top. In this space, people represent themselves as avatars, and we noticed that the Japanese absolutely will not use cartoon avatars-it must be a photograph. I don't know if this is because the cartoon implies a lack of seriousness, but that's the case.
On a wider scale, we know that Japanese meetings are much more structured than ours and have a different style of interaction. That's important to notice when you are thinking about how a meeting is conducted in virtual space.
Is R&D itself a different perceived process in Japan than it is here?
Absolutely. We both profess the same goals-shorter cycle times for product development, for example. But once that's done, we diverge in implementation. Our experience in Japan is that people want to specify in great detail how this is done. Whereas in the United States, you get some of the $3 billion venture funding that was spent in the Silicon Valley last year. You bring your prototype to market, sell it to either the public or to a company. All this takes place in a much more fluid way than the identified process we see in Japan.
But its important to note that Fuji Xerox is very supportive of us doing this different ways. They really challenge us to do that.
So the put the lab in California because they were intrigued by the model?
Right. And they are saying we should not do it the Japanese way. The most criticism that I personally have had is when people pick up that I might have adopted some Japanese things that I happen to think are very good.
Tell me about the researchers from Japan.
We have roughly one researcher from Fuji Xerox in each of the themes we pursue. The researchers are here for two to three years. Fuji Xerox sends us very good people and they like to have these very good people back being a little better.
That seems like a low number, comparatively.
Not really. We have three to four people out of 50, which is about 8 percent of our research staff. Don't forget, we are bringing Japanese over here to do things in a way that's different from what they're used to. That's harder to do if you have a large number. So we have our own style that we're developing here, our own environments. We're bringing in a rich culture from people all over the world, and I think they're beginning to pick this up.


Bart Eisenberg

Bart Eisenberg's articles on the trends and technologies of the American computer industry have appeared in Gijutsu-Hyoron publications since the late 1980s. He has covered and consulted for both startups and the major corporations that make up the Silicon Valley. A native of Los Angeles and a self-confessed gadget freak, he lives with his wife Susan in Marin County, north of San Francisco. When not there, he can sometimes be found hiking with a GPS in the Sierra, traveling in India, driving his Toyota subcompact down the California coast, or on the streets of New York and Tokyo.


1980年代後半より,『Software Design』や『Web Site Expert』などの雑誌に,アメリカのコンピュータ業界のトレンドと技術に関するレポートを執筆しています。シリコンバレーで,スタートアップ企業から大企業まで幅広い分野でコンサルタントを務めました。

ロサンゼルス生まれで,自称ガジェットフリークです.現在,妻のSusanとともに,サンフランシスコ北部のMarin County在住。また,SierraのGPSを携えてハイキングしたり,インドを旅したり,カリフォルニア海岸をドライブしたり,NYや東京の街中を歩いたりしています。