Sense of Wonder Motivates VLSI Chip Revolutionary, Lynn Conway
AAAS Fellow Lynn Conway, professor emerita at the University of Michigan and a vastly influential pioneer of the digital age, sees the outsize role she played in the development of computer technologies as the product, for the most part, of the sense of wonder she's maintained her whole life.
“I've never stopped being in kindergarten, never stopped playing, creating, messing around, being in awe of everybody around me,” said Conway. “Part of my spirit is a childlike openness that has always allowed me to see things other people don't notice.”
Perhaps the best known of Conway's contributions is the work she did to develop “very large-scale integration” (VLSI) computer chips, and the soup-to-nuts textbook on the topic she co-wrote with Carver Mead Ph.D. of the California Institute of Technology in 1979. Introduction to VLSI Design touched off the so-called Mead-Conway revolution, which democratized chip design and gave a major prod to Moore's law, which, simply put, holds that computers' processing power will double every two years.
But Conway's accomplishments are many and foundational in the computer industry, beginning with her 1965 invention of dynamic instruction scheduling (DIS) at IBM Research in Yorktown Heights, New York, as part of the Advanced Computing Systems project's architecture team. DIS led directly to the development of superscalar computers, and is still part of the bedrock of high-performance computing systems.
Now 79, Conway is most visible as an activist for transgender rights. Nearly 50 years ago, she was fired after telling her superiors at IBM that she planned to undergo gender confirmation surgery. To say that Conway landed on her feet would be to gloss over the loss and trauma she experienced as a result of that dismissal, which came in spite of her astonishing talent and contributions. Conway assumed a “covert” stance for many years, finding work first as a low-level contract programmer even as she transitioned to life as a woman. Having made her way back to the level of computer architect at Memorex, Conway was recruited by Xerox in 1973 to work at the firm's Palo Alto Research Center (PARC) in California.
At PARC, Conway came into her own—again. Even though she remained in the shadows to the point, she recalls, where people thought she was Mead's assistant, Conway set to work cutting away the clutter holding back chip technology.
“It's hard to think of anything new when dumb ideas get established,” she said.
At PARC, Conway invented “multiproject wafers” (MPW), which allowed for packing multiple circuit designs, perhaps even from different labs, into a single chip, streamlining production and slashing costs. That innovation’s effect on the computer industry can scarcely be calculated.
Conway also did much of the work refining the methodology explained in Introduction to VLSI Design, which at one time was considered “the bible of chip designers.” The book taught students not only the basics of this fledgling enterprise, but also how to validate and verify their designs, and had the effect of freeing the digital revolution from the halls of elite commercial labs, handing bright engineers everywhere, and even amateurs, the chance to play a meaningful role.
As part of this effort, Conway tested the book's tenets in a course she taught in 1978 at the Massachusetts Institute of Technology, which she had attended as an undergraduate. (She has bachelor's and master's degrees in electrical engineering from Columbia University.)
“The book would have been rejected if everything in it hadn't already been tried,” she said. “As it was, all the academics said it couldn't be that simple.”
The following year, 12 universities offered classes based on Conway and Mead's book; by 1983, more than 113 schools were teaching it. As part of the course, students created their own chip designs and held the finished chips in their hands.
But “the chips weren't the invention,” Conway said. “The invention was the living system of people and technology, a complex techno-social system—designers, builders, and the people in the printing foundries where the chips were created. The trick was to use the best technology you had at the time to make the computers to design better tools to make better chips—and then run that production cycle over and over. It was a model for Silicon Valley—and the world.”
Conway became assistant director for strategic computing at the Defense Advanced Research Projects Agency (DARPA) of the United States Department of Defense, charged with harnessing digital technologies for use by the military. While there, she played a key role in setting up the “meta-architecture” for DARPA's stance in the field. In 1981, DARPA established the MOSIS system, the first of the Internet-based rapid-prototyping chip services Conway's innovations made possible.
But Conway had become disenchanted with Silicon Valley's ”emerging macho, careerist, money-and-status-obsessed” culture. In 1985, she went to the University of Michigan as a professor of electrical engineering-computer science and Associate Dean of Engineering. She helped lead a major expansion of the engineering college there, she said, “while living in a wonderful Midwestern university town and getting a life.” Conway has been with her husband, Charles Rogers, whom she met in Michigan, for 30 years.
Conway thinks the minor in anthropology she collected at Columbia University has been an important influence on her career. Because of that perspective, “I think I was able to understand the large-scale techno-social phenomenon whereby industry would re-invent itself,” she said. “I was ready to understand that as a paradigm shift.”
Conway has experienced her share of failure, she said, including her first project at IBM, but she sees failure as a necessary part of creativity.
“You learn by failing,” she said. “People who strive don't necessarily expect to succeed. They just want to do whatever it is they're involved with so badly, they go ahead and do it.”