Lynn Conway

Lynn Conway (1938–2024) revolutionised computing twice, under two different names — because in between, her employer erased her from history and she had to start her entire career again from zero. She did. She climbed all the way back up, changed how every chip on Earth is designed, and lived long enough to receive the apology. Also, she flew gliders. Of course she flew gliders.

Revolution number one — years too early

In the 1960s, as a young researcher on IBM's secret supercomputer project, Conway invented a way for a processor to issue multiple instructions at once, out of order — dynamic instruction scheduling, the trick at the heart of essentially every fast CPU you've ever used. It was decades ahead of its time; the industry wouldn't catch up until the 1990s, when out-of-order execution became the engine of instruction-level parallelism. Then, in 1968, IBM discovered she was preparing to transition — to live as the woman she was — and fired her. Her name came off her own inventions.

Conway completed her transition and restarted her career in what she later called "stealth mode": a new name, a new résumé that began at zero, and a job as a lowly contract programmer. Nobody she worked with knew she'd helped design a supercomputer. She simply out-worked and out-thought everyone, again — and within a few years she had climbed back to one of the most storied research labs in the world, Xerox PARC. Same brilliance, second ascent. It's one of the great comeback stories in the history of engineering.

Revolution number two — teaching the world to make chips

At PARC, Conway teamed up with Caltech's Carver Mead to attack a scandal hiding in plain sight: chip design was an in-house craft, passed around like guild secrets. Conway's masterstroke was structured design — simplified, scalable rules that stay valid as transistors shrink, so a designer could plan a chip's floorplan and its placement and routing without knowing the fab's private folklore. She wrote much of the legendary textbook (Introduction to VLSI Systems, 1980) — and then proved it worked in the most audacious way possible: her 1978 MIT course had students design real chips, mail the designs off to a factory, and hold working silicon a few weeks later. Multi-project chip services like MOSIS descend directly from that course, and so does the entire fabless chip industry.

The long arc

Conway went on to a distinguished professorship at Michigan, and in her later years did something as brave as any of it: she came out publicly and spent decades as an advocate, making sure the next generation of transgender engineers wouldn't have to start over in stealth. In 2020 — fifty-two years late — IBM formally apologised and honoured her work. She accepted with characteristic grace, having long since won: her ideas were inside every computer in the room. She kept flying gliders into her eighties, which feels right — Lynn Conway spent her whole life finding lift where others saw only sink.