Trip report E.W.Dijkstra, Munich 24-26 November 1976.

I have been three days in Munich on the invitation of o.Prof.Dr.Dr.h.c. Friedrich Ludwig Bauer, alias Unkel Fritz. I left --as two weeks before-- Eindhoven on Tuesday evening and returned on Saturday morning. The first night I slept reasonably well, on the return trip I slept very well.

Because Unkel Fritz has a house in Grafrath, about 1 hour driving from Munich, I did not expect him on Wednesday morning at half past seven at the Munich Central Station in order to pick me up; I had expected that he would have asked someone who lived nearer the station. I was greatly surprised to find Fritz himself waiting at the end of the platform. I thanked him very much, but he said "Don't mention it, I live quite nearby, where I have an apartment, let us go there and let us have breakfast." This was quite a sensible arrangement. I then heard that he had not planned to pick me up, that he had ordered one of his underlings to do so, but that at a quarter past seven, he had received a telephone call from the underling who, as a result of early snow, had got stuck with his car still miles outside Munich! So much for the German academic hierarchy...

I had been invited as a consultant. Unkel Fritz had a sizeable project under way, felt that it had reached a critical stage and would appreciate my comments. He had already asked me to come when we met in Los Alamos, early this year. I have tried to consult by mail, but he was really keen on having me in Munich, so that I could also speak with his assistants etc. Eventually I gave in and went. On Thursday evening I spoke for their "Informatik Seminar". It was a talk I had never given before on the structure of termination proofs and it was fun, but for the fact that my audience was larger than I had expected and, in spite of the abstract that had quite clearly announced what kind of talk to expect, partly insufficiently equipped to follow a rather formal mathematical talk.

When I left on Friday, everybody said that they were very grateful for my having come, that I had been quite helpful, etc. But, personally, I don't see how I can have been of great assistance, because as the days went by, my faith in the project faded more and more, and the only advice that I could give them seemed to be the advice that they could not follow, viz. to chuck it (at least on the grandiose scale they envisaged). The reasons for my growing doubts about the whole project were threefold.

Firstly, its mathematical contents seemed to be very similar to what Burstall and Darlington did three years ago, and their experience seems to suggest that these techniques are alright as far as they go, which is not far enough to make them the core of an all-embracing programming technology. There were no clear indications --at least I did not see them-- why Munich would be able to build much more on that foundation than Edinburgh had been able to do. Secondly I failed to encounter the professional competence that I believe to be necessary for such a job. I heard too much sloppy jargon: the gentlemen spoke with the same easy about "levels of language" as mediaeval astrologers must have discussed the Harmony of Spheres. In addition to that I could not avoid the impression that some frustrating experiences in the past (with ALGOL 68, to be precise) had corrupted some of the soundness of their judgement. Thirdly --and that seems most dangerous of all-- they seemed Too Keen on Visible Success. As a result the considerations in defense of the project were each time polluted by arguments of a propagandistic nature. If the frills that should make the project Spectacular were trimmed from it, a small, sensible project might remain, but, of course, then it would not be as Grandiose.

Even internally the propaganda machine was working full blast. A worked out example was circulating --proving that one procedure was a fixpoint of another-- and the two pages of tedious manipulations of that example were used as an argument that mechanical assistance would, indeed be needed. I suspected the usual ploy and ventured the opinion that two pages was perhaps a bit much for this problem. "Oh no", said Unkel Fritz "one had already to be quite a good mathematician to bring the proof down to that size." I have now seen this dirty trick so many times, that I boldly stated that half a page should be more than sufficient, and in a moment of solitude I did it in four lines (not even having looked at their solution, as that could only be misleading). The shortness of my proof was too clearly politically inconvenient and my experience with this example was discarded by the (usual!) contradictory remarks that

  1. of course, my proof followed the same argument, but that I have been much less explicit in writing it down
  2. that such remarkable elegance and mathematical virtuosity cannot be expected from the average programmer.

I slept the two nights in Holzkirchen as as the guest of Prof.Gerhard Seegmüller and his wife. That was very pleasant. Wednesday evening the fresh snow was responsible for some complication, as Gerhard had to park the car a few hundred metres from his home. With my suitcase I had to climb through the fresh snow his car could not get through; I did not lose my slippers.

*                         *

At nine o'clock Saturday morning Ria was at Eindhoven station with the car. I gave her my suitcase, over a cup of coffee we exchanged our recent experiences and then I went by train to Amsterdam to attend most of the monthly meeting of the Royal Dutch Academy, where I was reminded of a story that I had heard in Munich. There I was told about an African tribe, where filibustering was prevented by the rule the everybody was only allowed to speak as long as he could stand on one leg. It seemed to me a most sensible rule. The afternoon session did not last too long, we could get the three o'clock train back and at a quarter to five I was home again.

That Sunday I did not do much, I did not even write a letter to my mother.

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

transcribed by Tristram Brelstaff
revised Thu, 12 Aug 2004