This memorial website was created in memory of Norm Hardy. His contributions made our lives richer and changed the trajectory of computer science. Please help us build a picture of what he meant to the world by adding memories of him, insights he shared with you, and photos you might have.
Tributes
Leave a tributeIt's also not like I followed everything Norman and the rest of the capability OS world went, but I did brief visit Key Labs. And who knows what Norman did with the Fort.
Norman was a soft voice in the security community and also had other ideas about software. It's no question that he will be missed.
Software development being what it is, every now and then during that summer the KeyKOS kernel team would make a mistake -- yet once the system came back up, the rest of us would find we had not lost any more than a few minutes worth of work. At the time I failed to appreciate the foresight which went into the architecture, but whenever I've lost significant work due to a crash in the intervening decades, I recollect my luck to have experienced one of Norm's systems.
While that resilience was an advantage, it was not the most significant part of the ideas behind Key Logic, however. Reading current news about information security (or more accurately our lack of it?) I cannot fail to be reminded of how KeyKOS confinement did double duty: not only were applications protected from malicious users, but user data was equally protected from disclosure by applications.
Thank you, Norm, for not only showing me that an OS could do better, but also how.
Due to his central role, I invited Norm to give the keynote at OCap 2017
https://www.youtube.com/watch?v=hEDCXTpx0R8&list=PLCq8mSCP664TUdgHl1cD5sDiAmrDoio2p
The KeyKOS operating system is not just an astonishing engineering achievement, it is an astonishing intellectual achievement. As an intellectual achievement, KeyKOS embodies a philosophy of computation that I have spent much of my life trying to understand and explain to the world. I will say more on this elsewhere.
As an engineering achievement, as I say in my introductory remarks to Norm's keynote, KeyKOS is bizarrely close to perfect. For years, my every attempt to find something to improve failed. Jonathan Shapiro had a similar experience but did arrive at genuine improvements with EROS and Coyotos, which provided the crucial bridge to seL4. Today's only practical verified secure OS, seL4, could be verified to solve the whole problem because it built on the KeyKOS design which actually did solve the whole problem.
Several of us have brought some of these insights into other areas of engineering (with apologies to all those I should have included):
* Dean Tribble, Norm, Chip Morningstar, Marc Stiegler, and I brought these insights into modern programming language design (Trusty Scheme, Joule, E, Midori, Dr.SES).
* Bill Frantz, Chip Morningstar, Tyler Close, and I brought these to cryptographic protocols (CapTP, Waterken, SPKI/SDSI). We also built on crucial foundational work by Jed Donnelley that Norm told us about.
* Marc Stiegler, Ka-Ping Yee, Alan Karp, and Mark Lentczner brought capability-based insights to user interface security (CapDesk, Belay), all starting from a user interface koan from Norm: "The clipboard is inherently hostile to user interface security. Drag-n-drop is inherently friendly to it."
My deepest condolences to Norm's family.
I first met Norm at a Cypherpunks talk that he gave about KeyKOS at the Stanford University Quad. Hugh Daniel, who also since passed, had invited Norm to present the talk because Hugh felt that the Cypherpunks needed to hear about capabilities. I recall that Metricom modems were all the rage back then and a handful of attendees had them Velcro taped to their laptops.
Minutes into Norm's talk I realized that most of what I believed in the realm of computer security, indeed had implemented myself, was wrong: designs that do not employ capabilities are fundamentally *insecurable*. Sandboxes are a dead-end design choice. Norm's talk has shaped my thinking ever since.
Ad Astra, Norm!
We're very sorry to hear of his passing, and thank the Hardy family for all the love you've shown him, and for lending him regularly to the capability theory community as a treasured father.
My condolences to his family, I know that you will miss him a lot but also celebrate his great achievements.
A PhD student at EPFL, Switzerland
1) If you discover you're completely wrong about something you thought you understood, celebrate. You might just be onto something big.
2) Question most closely the things you want to believe. Don't fear the outcome of this process: as a rule, the truth ultimately proves more satisfying than the misconception.
3) The reality that humans can create for each other is greater than any fantasy yet conceived.
I particularly appreciated Norm's patience. It sometimes took me a while to translate what he was saying into terms familiar to me. He never showed any sign of frustration; he just led me through the process until I finally figured out that he was right and why he was right.
Norm was a regular at our Friday morning meetings; they will be poorer for his absence.
Norm Hardy was my Dad. Our house was full of puzzles, Escher, Bach, big stacks of computer print-outs, insightful conversation, apple pie and love. I think differently because of him. I love being around people who can actually provide a different perspective on an issue ... not just an opinion, but an insight.
He influenced the world of computer science profoundly, but I'll have to leave it to others to really describe that history.
My work now is in mending the American West back together through civil conversation about natural resource use - trees, rivers, wildlife. I talk with cowboys, Forest Service, timber mill operators, river guides, tribal members etc. To a huge extent, I can do this because I cherish the new perspective. Rather than being resistant to new perspectives, I love them. Norm taught me to listen for a better idea, to absorb it, to play with it, to open it up every angle with gentle questions. He was endlessly patient with my struggles to keep up with him. And I learned how to ask questions to create better understanding.
Although in a totally different realm, his scientific sprit of inquiry will live on in me. His endless patience and his gentle, but unyielding demand for an actual explanation continue to inspire.
I am humbly honored to carry that with me. He was a profound man, and a huge support. I am so fortunate to carry of piece of him in my life, forever.
Leave a Tribute
Solar system out of kilter
You write like Norman Hardy
Robert Cralle had written a program called "Scrawltran" that used a light pen to hand-write arithmetic expressions that were then turned into Fortran source code. Bob knew that Norm wrote 2s and Bs differently than most other people did. When writing a B, Norm started at the bottom-left with a vertical stroke and then came down to make the two curved parts on the right. This saved lifting his pen or pencil, so it was just a bit faster than the usual downward vertical stroke followed by a lift and move to the top again to begin the two curved parts. For the 2, he started at the bottom-right and traced the digit from there. I have no idea why he did that (it was no faster), but it apparently is what he did.
At some point, Ivan Sutherland (of computer graphics fame) was visiting Livermore. Norm gave him a tour with one of the stops being the IBM 7090 and a graphics display, which is what ran Bob Cralle's Scrawltran program. Bob knew this was coming. He modified his code to recognize Norm's characteristic strokes for 2s and Bs. When the day of the visit came, Norm sat down to demonstrate to Ivan how Scrawltran worked. As Bob had expected (because everyone seemed to do it), Norm entered a 2 and a B (not sure what order). When he entered the first character, a message from the system appeared saying "You write like Norman Hardy." Norm was nonplussed, but continued the demo and wrote the second character. The system responded "By God, you ARE Norman Hardy!"
At that point Norm got up and walked all around the computer looking for the hidden wire that he thought must have been there because how else could the program have generated those messages. I was not told what Ivan Sutherland thought of all this, but a lot of folks at LIvermore shared the story again and again.
I met Norm at weekly Mill CPU meetings. My honest observation is that he has somehow gotten younger over the five years that I've known him. I'm honored that Norm, Ivan, and I enjoyed morning coffee during those meetings.
One memorable time for me is Norm's meeting with me at Peet's in Menlo Park before my DConf 2016 presentation. He filled me with stories, knowledge, and confidence.
He was the speaker at our local C++ meetup on February 10, 2016, presenting "Reliability and security of today’s software platforms". His voice and slides are captured on a video linked there: https://www.meetup.com/ACCU-Bay-Area/events/227325917/