Many, many years ago, at school for CS and before that interning at NASA Ames, I learned to take programming seriously as engineering work, and also the level of rigor that engineering work requires. In the intervening thirty years, I have had to come to terms with the degree to which this attitude is rare in theory and vanishingly rare in practice again and again. Tooling has gotten better, but the rigor applied to the average line of code has gotten worse.

I would hope that safety-critical systems would be different, and in some cases they really are. However, the number of people who want to build secure communications systems that they want to be safety critical and who expect thousands of people to literally use in a case where a breach of the threat model means the users will die and who as developers cannot bother to meet the basic design rigor requirement of 2006 — have a fucking threat model that is accurate and up to date — will never cease to shock me.