Unusual keynote at the ACCU conference this morning: it's on the psychology of software developers by an academic researcher. In recent years I already know much if not most of the content at these conferences, but this talk is new and useful. For example the academic presenting the empirical findings has found very little pair programming in top teams, but lots of pair debugging. And she's very right on that, and it hasn't occurred to me before. Interesting. Another thing is that experts write more buggy code than non experts, and deliberately leave bugs which smell like they might demonstrate other bugs unfixed for as long as possible, even in shipping releases. Also very true. This talk is proving to be highly valuable.