"Engineering is--and has to be--a creative profession."
By Richard Cole
The book The Soul of a New Machine by Tracy Kidder was published in 1981 and became an improbable--but instant--best-seller. Describing the development of a new 32-bit minicomputer at Data General, Kidder was able to create a lively and suspense-filled narrative out of a highly technical subject by clearly explaining not only complex technology but also the equally complex people who developed it. At the center of this story was Tom West, head of the engineering team that developed that new machine.
Even now, 15 years later, Tom can still find himself recognized as "that manager in the Kidder book," and he has to admit that "the book has changed my life, in some ways, forever." He feels that Kidder did a thorough and accurate job of reporting, although he does admit that it was "a bit disarming" to find Kidder interviewing his children at the breakfast table.
It was also strange for Tom to see himself as a character in someone's book. On the one hand, he was portrayed as an inspiring leader. As Kidder put it, "West . . . had a way of making ordinary things seem special; in this case a 32-bit Eclipse was being transformed into the occasion for an adventure." On the other hand, his character remained an enigma--sometimes intriguing, sometimes even vaguely threatening--to even his closest coworkers. One engineer is quoted as saying, "When I first went to work, [West] was my boss, and it was amazing. Half the time I couldn't figure out what he was saying. He expected you to be on his secret wavelength, and if you weren't, he'd be disappointed in you."
Describing his initial reaction to the book, Tom recalls, "Tracy didn't show anyone what he was writing for about two years. And then one night I get 200 pages in galley proofs. Well, reading that was like getting 200 pages of psychoanalysis all at once."
Although the book made West one of the first techno-celebrities, he admits thatperhaps he would have been happier to have "remained anonymous," and indeed, Kidder's book describes only one segment in a long and varied career.
Tom had spent summers working at the Smithsonian Institution in Washington DC, and he found a full-time position there straight out of college. Among other projects, he designed and built digital clocks that were used in NASA's space program. The clocks were used with telescopes to photograph and accurately calibrate satellite orbits, and each clock had to be synchronized and physically placed at different sites around the world. Tom spent several years traveling to remote areas and setting up the clocks. The job was fairly straightforward, though it did have its interesting moments. In Colombia, he was thrown in prison for a brief time because his clocks looked to the authorities like secret weapons.
Toward the end of the 1960s, funding for his projects began to dry up, so Tom decided to get another job. Intrigued by computers, he checked out what books he could find on the subject at the local library and started to read them on his back porch. Six weeks later, he landed a job at RCA where he essentially taught himself to be a computer engineer.
RCA gave him increasingly important projects, but he found the atmosphere confining, so he moved to Data General where he soon became a development team manager. For several years, he "pounded out 16-bit Eclipses." Then, in 1977, he was picked to manage the team that developed the first 32-bit Eclipse. Code-named Eagle, the computer was designed to go head to head with the 32-bit VAX series from Digital Equipment Corp. As such, the 32-bit Eagle had to be not only as good but better than any 32-bit minicomputer on the market. West and his team worked seemingly impossible hours against seemingly impossible deadlines and finally completed the computer known as the Eclipse MV/8000. It outperformed the VAX machines and became the first in a series of successful minicomputers.
The period was an exciting one for Japanese technology. Hewlett-Packard had just released a bombshell report suggesting that Japanese computer manufacturers were surpassing the U.S. in terms of quality. U.S. manufacturers were going through a period of deep angst, examining Japanese management and manufacturing techniques to try to find the secret of their success.
Tom studied Japanese through Berlitz courses in Massachusetts and flew to Japan "eight or 10 times" a year. He worked at Nippon Minicomputer, a manufacturing company in which Data General owned a controlling interest. With Nippon's engineers and managers, he oversaw the development of the DG/One, the world's first laptop computer. Tom asserts that the DG/One was also the first computer developed through a complete, one-to-one collaboration between American and Japanese engineers. The Americans contributed PC technology. The Japanese added a CMOS microprocessor, flat-panel displays, the first 3.5-inch floppy diskette drive and the first 3.5-inch hard diskdrive.
Currently, he is working on projects involving NUMA (Non-Uniform Memory Access), a clustering technology that allows many processors to work together. As Tom points out, since Intel chips and the motherboards supporting these chips have become so powerful, their combined performance now approaches mainframe levels. The trick is how to combine them in the best manner. Microsoft, Compaq, Data General and a host of other companies are trying, each in their own way, to link processors together. The advantage with NUMA, Tom says, is that applications from one processor can run almost unmodified on another one in the cluster. "You get all the Intel benefits like performance and low cost," Tom says, "while preserving the investment made in applications and training."
Tom adds that Data General is working with a number of other companies. "This is an issue where a whole lot of fragmented efforts are going to defeat the cause." When questioned about the number of processors that might be ultimately tied together, he only says, "Who knows? It's a crawl, walk, run sort of thing. We'll just have to see. This is a software issue. It's like symmetric multiprocessing [SMP] in Unix; it takes a long time to get the software to work."
When asked about other projects, West laughs and says he can't talk about specifics. However, he does say that "if you buy the notion that the world is going to be made up of Intel motherboards connected together in various ways over high-speed networks, then the way you think about input/output, mass storage and other concepts will change, too. So it's not just about the computer engine, which is what NUMA is all about; it's about scaling other parts of the system as well."
Do today's highly focused engineers have the same level of creativity as engineers from the wild and wooly early days of the industry? "That's a potential problem," concedes Tom. "In the past, people joined the computer industry because they didn't want to sign up for the whole nine yards of Commercial America. Now, the kids today have signed up for that before they enter college."
At the same time, Tom objects to the notion that today's engineers--and engineers in general--are one-dimensional and too concerned with highly controlled environments. "The whole business of control is not limited to engineers," he points out. "People in finance, say, or manufacturing have the same problem. It's important to think one-dimensionally in working through a difficult problem. But the stereotype goes too far in describing the whole person."
West has seen more than a few changes over the years. "When I first started and you were at a cocktail party and you told people you were in computers, they shrunk away to the other side of the room. Now people want to tell you about their kid's Macintosh and their hard drive." At this point in his own career, he says, "There aren't many mountains left to climb." However, he looks forward to new challenges and to introducing new engineers to the field. "I'd like to spend more time getting some of these kids onto projects that they're going to be proud of when they get to my point in life."