EWD 839

Trip report E.W.Dijkstra, Utrecht 30 Sep. - 1 Oct. 1982

This was a very short trip: I left Eindhoven on Thursday morning at 8:00 and returned the next day at 17:20. The occasion was a two-day seminar on "The software crisis", jointly organized by The Royal Institute of Engineers and the Dutch Computer Society. I was the second speaker and one of the members of the panel whose discussion closed the seminar. In particular the second capacity was a reason to attend the whole seminar, which was not exciting. Depressing would also be a quite adequate term.

This was not a disappointment, for I went with suitably low expectations: the organizers had already told me that "in view of the topic" they were not heading for a scientific level! And indeed, they reached their goal.

One view regards programming primarily as producing lines of code --for the latter, one of the speakers seriously used the hilarious acronym LOC!--; another view regards programming primarily as preventing an otherwise unmanageable complexity from emerging. The first view is too superficial to be helpful: it is like equating the study of a book with character recognition and the turning of the pages. It was however the one adhered to by many a speaker.

I enjoyed myself thanks to the presence of C.Bron, S.D.Swierstra (who moderated the panel discussion) and B.Waumans, people I know for years but whom I encounter only rarely. The highlight, however, was being introduced to Mr. Ivor Catt, whom I had never met in person, though I knew very well who he was.

By virtue of his involvement, Catt knows all the ins and outs of one of the major scientific scandals of the last 15 years, viz. the systematic suppression in the world of electronics of all publications about the phenomenon of the so-called glitch and its ramifications. Part of Catt has turned to the study of what one might call the sociology of science or the scientific establishment's mechanisms for the rejection of novel results. The story how once --on false accusations made at a secret meeting-- he had got immediate notice because it was thought mandatory that no one in the company, nor any of the company's --mostly military-- customers, should know of the glitch was typical.

(In EWD837 I mentioned Melliar-Smith's lecture on the fault-tolerant operating system designed for "flight control". One of the things it has been designed to capture is a malfunctioning caused by a glitch. In order to estimate the MTBF of the whole system, its designers needed to know the likelihood of a glitch, a figure the hardware designers have to provide. Today I heard that the electronic engineers of Bendix --the company manufacturing the hardware for the flight control system-- had been so well-conditioned that the glitch problem could not even be explained to them. I am not amazed.)

Plataanstraat 5
The Netherlands
6 October 1982
prof.dr.Edsger W.Dijkstra
Burroughs Research Fellow

transcribed by Tristram Brelstaff
revised Mon, 27 Dec 2004