This morning I found out that my senior thesis advisor, Arthur Lo, passed away nearly two months ago. This came on the heels of a Facebook chat with a good friend who similarly found out her favorite professor and advisor died recently. Perhaps it’s a sign that we’re in that sandwich phase of adulthood, tending to our own children and hoping to pass on some of the wisdom and insight handed to us by those we now miss.
Professor Lo was old school before it was fashionable. He was practical, direct, challenging and most of all, approachable. He knew how things worked in the real world. At a time when the personal computer was just entering the market and the Macintosh hadn’t yet been released, he carried a tremendous passion for the boundary of the analog and digital domains, probably sensing that digital design would reshape our daily lives. He worked on very early transistor applications at RCA across Route 1 from Princeton, and eventually was a leader in early digital design as well. Professor Lo somehow knew that the world of ones and zeros would always be captive to underlying LaPlace transforms, parasitic capacitance, and path-length race conditions. Every time my iPhone vomits between cells or finding a wireless network, I’m reminded of Professor Lo’s insistence that we pay attention to the analog world. We are inherently analog beings with analog senses, and the physics that holds the universe together is equally and entirely non-digital. Apologies to Vernor Vinge and others anticipating a singularity, but Professor Lo knew his stuff.
His accented English made him pronounce things like “register full” as “register foo”, which was only funny because “foo” and “bar” were entering our vernacular at the same time as generics for “I need a variable name” or “Whatever we’re going to call that file.” We were never laughing at him, only with him as he managed to make an introductory circuit course mildly amusing. My lab partner in that course is part owner of a sports team and I work for a software company, so we weren’t the ideal EE students either.
But as his student, I listened intently and filed things away, because what he taught me was the foundation for nearly everything I later explored in the areas of device drivers, networking and performance. Neal Nuckolls, who was responsible for much of the early fast networking code in Sun’s operating system, once said that “you have to understand how something works before you can improve it.” To this day, that’s one of my basic rules of work, and it regularly reminds me of Professor Lo.
My favorite interaction with him happened one weekend just before Thanksgiving, when I was deep into thesis research. Theoretically, I was deep into thesis research in building a high-speed ECL circuit, requiring me to learn more about signal propagation than I knew was possible. What I was really doing involved job interviews, parties, and some incidents I talk about only rarely. During our required advisor-student meeting, Professor Lo asked me to explain what I’d learned so far about how high-speed digital signals move in long circuit board traces. The underlying electrical engineering is the science of transmission lines – the same physics that describes electricity coming into your house through the power grid, getting a clean cable signal or how copper phone lines manage to remain echo free even though they run for miles between your home and the central switching office.
Back to my explanation of transmission lines. It wasn’t very good, and that’s heavily discounting for the fact that I was explaining it to a guy who really had done this for a living. What I remember and what Professor Lo probably said diverge; he was polite, proper and respectful beyond all reproach. But as I stumbled through a drawing of an echo signal bouncing off a non-terminated signal trace, I thought I heard Professor Lo say “Stern, you don’t know jack shit about transmission lines. Read these.” At which point I was handed a few sturdy EE texts about transmission line theory and echo suppression. They were accepted with a smile, and genuine gratitude for his incredible patience and tolerance for what was really a waste of his time that he turned into a teaching opportunity. I’m pretty sure he never used vulgar language, at all, even when students like me deserved it, and although I never figured out what he really said, his tone of voice and seriousness made me attack my thesis work with vigor. Another case of real world physics snapping me right back to reality. I’ve told the story to many family members, and when I get in trouble in the kitchen, it’s usually followed with “Stern, you don’t know jack shit” as a gentle reminder of who is really in charge. I never expected it to be a long-lasting tribute to a great teacher.
I read those insanely dense tomes over Thanksgiving break, and did my best to make up for lost time. My required interest in transmission lines turned into a fascination with 10baseT (thinnet) ethernet, which of course was (when done correctly) a transmission line with stubs. That led into NFS network design, which became the basis for Managing NFS & NIS. Anyone who read to the end of the first edition and was wondering why there was a short chapter on transmission lines and termination techniques should tip the propeller hat, with much respect, to Professor Lo’s memory. I do so regularly, and am thankful for his ability to admonish us to understand the way things work, at the most basic level, because that’s where the real, analog world happens.
It’s sad that he’s no longer here with us.
[Update, February 11, 2012: The director of David Sarnoff Labs left a thoughtful comment about Arthur Lo here]