This week in news that sucks:
Peter G Neumann died yesterday.
Peter was, of course, most well known for the definitive essay on the use of the word 'only' in English. The photo is of a sign that made me think of him when I saw it. Peter valued clear communication a lot. I suspect he'd be better known if he didn't: Often, after you'd read Peter's explanation of something he'd thought of, it would be so obvious that you wouldn't think of it as something that needed someone to invent it. He edited almost all of the CHERI publications. He also wrote a lot on the distinction between 'trusted' (if it's broken, everything is broken) and 'trustworthy' (it is unlikely to be broken) in systems design.
Peter was PI on the first grants that funded the CHERI work. DARPA directly funds only US institutions. Cambridge was operating on those grants as a subcontractor.
I first met Peter at one of the PI meetings for the DARPA CRASH programme, where we were presenting early CHERI work. He had a knack of falling asleep in talks (second picture), then asking a question about something that he'd extrapolated that the talk would cover, but which was definitely presented while he was asleep. Often of the form 'In the 1970s, when we were looking at this approach, we got stuck on this problem, how did you solve it?' A sad number of presenters had to admit that they hadn't actually got to that bit of the problem yet, but the good one either had a solution or had reframed the problem to avoid that obstacle.
The third picture is taken just after the second and Peter's willingness to pose for it shows his sense of good fun. He was always quick with a (terrible) pun, except when it came to names. Peter never made puns of people's names, because people didn't get to choose their names and so humour derived from them would be unkind, and I never saw him being unkind.
Indeed, he was always supportive to the folks around him. He somehow managed to give the impression that he felt privileged to work with you, when the reality was that the privilege was entirely in the other direction. He gave great career advice on a couple of key occasions for me (as well as writing a reference that made me blush). I'm far from the only person in the CHERI project whose career benefitted from gentle nudges and unwavering support from Peter.
I learned a huge amount from working with him. You can see his fingerprints all over CHERIoT. A lot of the system was either inspired by conversations with him or shamelessly stolen from his earlier work. I can't imagine our having built any of it without his influence.
Mastodon Feed
