I derive a lot of inspiration from Edsger Dijkstra’s principles, opinions, and of course academic work. It is to me recharging to rewatch some of his interviews and read the EWD archives.
- Edsger Dijkstra loved writing with his fountain pen - Edsger Dijkstra's fountain pen - Fountain & Dip Pens - First Stop - The Fountain Pen Network
- UTexas.edu: the manuscripts of Edsger W. Dijkstra
- according to cwi.nl - Edsger Dijkstra wrote with Mont Blanc fountain pen - Edsger W. Dijkstra: Brilliant, colourful, and opinionated
- umn.edu / PDF: An Interview with Edsger W. Dijkstra
- real treasure – Denken Als Discipline - a 2001 interview with EWD
- INFERENCE / Vol. 5, No. 31 / 12 - Edsger Dijkstra - The Man Who Carried Computer Science on His Shoulders - Krzysztof Apt
- Denken als Discipline - transcript in Dutch
Interesting Observations from EWD’s Interview
1. GPT as a high-level programming language
How does this apply to AI: “Only when you looked at what was happening did you see that the trivial aspects of the programming work were mechanized and the difficult ones remained. And the net effect of the higher programming languages, which were intended to make programming easier, was that, together with the increasing ambition of the application, they began to place higher demands on the intellectual level of programmers.”
If GPT is yet another higher level programming language, Dijkstra’s observation remains true. GPT mechanizes the trivial aspects of programming work. GPT makes programming easier and with the “increasing ambition of the application” it will also place higher demands on the “intellectual level of the programmer”
2. Word Processors (code editors?) vs Pen & Paper
Dijkstra sneaks in a subtle diss here for word processors (obviously he prefers his Mont Blanc vs Microsoft Word): “People learn that when you write something, you shouldn’t try to get it right right away. You have to write down what is on your heart and you have to rewrite and rewrite and rewrite and in the end you get the product you want. That is one of the reasons why word processors are so pushed there and have partly become popular.”
EWD explains his reasoning: “While one of the great advantages of working with pen and paper is that before you start a sentence, you have to have it completely ready in your head.”
It seems from the interview / video that Dijkstra does indeed craft his spoken sentences carefully and thinks before he speaks.
Thinking of Mozart vs Beethoven and Microsoft Word vs Mont Blanc reminds me of Amazon’s Type 1 vs Type 2 decisions. As a reminder:
- type 1 - high-stakes, irreversible, decisions that require careful thought – Mozart, Mont Blanc
- type 2 - reversible and easy to change decisions – Beethoven, Word Processor
It seems to me that using Microsoft Word vs Mont Blanc shifts from type 1 (pen and paper) to type 2 (easily changeable word processor). This is obviously the reason for word processors to write vs typewriters and pen and paper.
On the other hand I wonder if writing with pen and paper (type 1 decisions for the most part) rewires your brain to slow down and think things through before committing to semi-permenance.
English Translation of the Transcript
(via Google Translate and the link above)
Thinking as Discipline
VPRO Noorderlicht / 25 min / 10-04-2001
The Netherlands’ most famous computer scientist, Professor Edsger W. Dijkstra (70), lives in Texas as a missionary among the natives. In his house, a huge grand piano takes the place of a TV and his favorite word processor is the fountain pen. And while life is being hacked all around him, Dijkstra tirelessly preaches his gospel about the need to write correct computer programs.
VPRO NOORDERLICHT – TRANSCRIPT
AFL. ‘THINKING AS DISCIPLINE’
BY JOS WASSINK AND GERTJAN WALLINGA
BROADCAST 10-04-2001
00:00
LEADER
00:20
PAN TO SHOPPING MALL
TITLE ABOUT IMAGE: THINKING AS DISCIPLINE
MUSIC COMPUTER GAME
COMPUTER GAME IN COMPUTER STORE
PROF EDSGER W. DIJKSTRA
PROFESSOR EMIRITUS OF COMPUTER SCIENCE
UNIVERSITY OF TEXAS AT AUSTIN:
The only thing that matters is that you can quickly put something together that you can sell on the market. Doesn’t have to be good at all. As long as you can create the illusion that this is a nice product so that people buy it, well then you can see if you can make some better versions later.
PENTIUM SPOTLIGHT ON PC
PROF EDSGER W. DIJKSTRA:
Then you get the phenomenon with the version numbers even having digits after the decimal point. Version 2.6 and version 2.7 and all that nonsense. Yes.
CONTINUED PENTIUM COMMERCIAL
PROF EDSGER W. DIJKSTRA:
While if it had just been good, version 1 would have simply been the product.
CLOSING SHOT COMPUTER STORE
QUOTE:
Computer science is no more about computers than astronomy is about telescopes – E.W. Dijkstra
INTRO SHOT DIJKSTRA
01:43
INTRO SHOT DIJKSTRA;
PICKS UP EMPTY MAILBOX WITH COWBOY HAT ON
COMMENTARY:
Professor Edsger W. Dijkstra is the Netherlands’ first programmer. With his systematic programming method he won the Turing Award, the Nobel Prize in Computer Science, in 1972. He currently lives with his wife in Austin, Texas, where he moved in 1984.
PROF EDSGER W. DIJKSTRA:
That was a time when the biggest concern for all kinds of departments was: is my curriculum watery enough? At the same time, the University of Texas at Austin was trying to reduce enrollment and improve quality. That was an opposite development that was considerably more attractive than what happened in Dutch higher education.
CAMPUS UNIVERSITY OF TEXAS AT AUSTIN
02:49
CAMPUS UNIVERSITY OF TEXAS AT AUSTIN
Glockenspiel
PAN OVER FOUNTAINS
QUOTE:
Universities will continue to lack the courage to teach hard science; they will continue to mislead the students, and the next phase in the infantilization of the curriculum will be welcomed as an educational step forward.
03:22
APPROACH TO DIJKSTRA
COMMENTARY:
Quality, correctness and elegance. These are the requirements that a computer program should meet, according to Dijkstra. In 1954 he resolved to make programming a science. But since then he has had to row against the current.
PROF EDSGER W. DIJKSTRA:
I don’t lose sleep over the fact that the business community has this… the feeling that they can’t afford to deliver a first-class product. Doesn’t prevent me from continuing my research.
03:57
LANDSCAPE WITH CLOUDS;
CITYSCAPE;
SKIFFEUR ON THE RIVER
QUOTE:
You shouldn’t give the world what it asks for,
but what she needs.
DIJKSTRA WRITES BEHIND A DESK AT HOME.
04:19
DIJKSTRA WRITES BEHIND A DESK AT HOME.
MOZART MUSIC SOUNDS IN THE BACKGROUND (ANDANTE)
PROF EDSGER W. DIJKSTRA:
There are very different design styles. I always characterize them as Mozart versus Beethoven. When Mozart started writing, he had the composition ready in his head. He wrote the manuscript and it was “aus einem Guss”. And it was also very beautifully written.
Beethoven was a hesitater and a struggler who wrote before he had finished the composition and then pasted over something to change it. And there was a certain place where it was redacted nine times and they carefully peeled it off to see what had happened and then the last version turned out to be the same as the first.
PAN THROUGH WORKROOM;
MOZART MUSIC IN BG
PROF EDSGER W. DIJKSTRA:
Well, that iterative way of designing is somehow a very Anglo-Saxon habit. The entire English education system is saturated with it. People learn that when you write something, you shouldn’t try to get it right right away. You have to write down what is on your heart and you have to rewrite and rewrite and rewrite and in the end you get the product you want. That is one of the reasons why word processors are so pushed there and have partly become popular.
While one of the great advantages of working with pen and paper is that before you start a sentence, you have to have it completely ready in your head.
06:21
BROWSE EWD’S IN UNIT WORKROOM
COMMENTARY:
Dijkstra always wrote a lot: reflections, speeches, mathematical proofs and books. Bearing his initials and a serial number, the so-called EWDs are seen as his most important contribution to science.
PROF EDSGER W. DIJKSTRA
PROF EDSGER W. DIJKSTRA:
The most important thing for me has been… the daily discipline of neatly writing down what you are considering and what you are doing. Thanks to modern technology, they have had much more influence than something like this would have had in the past and you could describe it as a modern day of scientific correspondence. Although the traffic, the intellectual traffic, is largely one-way traffic.
The number now exceeds thirteen hundred. They vary enormously in length from eighty pages to 1 page. As time goes on they tend to get shorter.
Everyone who was sent them was implicitly willing to act as an internal node of the distribution tree. To send second generation copies around to a number of people. I have never actually been able to properly estimate how many people have been reached by it. I think a few hundred.
08:29
PAN PLAYING PIANO
PROF EDSGER W. DIJKSTRA:
To compose you have to write notes, yes, but the job of a composer is not to write notes. The job of a composer is to compose music. In early programming, you had to write your text in machine code, a meaningless sequence of capital letters and numbers, which was the analogue of writing notes. And people who thought that was programming. After that it was made easier and so-called higher programming languages were invented: FORTRAN, PASCAL, C++, what have you. These were thought to make programming a lot easier, even solving all programming problems.
STUDENTS BEHIND COMPUTERS
AT TAYLOR HALL BASEMENT LABORATORY
PROF EDSGER W. DIJKSTRA:
Only when you looked at what was happening did you see that the trivial aspects of the programming work were mechanized and the difficult ones remained. And the net effect of the higher programming languages, which were intended to make programming easier, was that, together with the increasing ambition of the application, they began to place higher demands on the intellectual level of programmers.
PAN OVER STUDENTS IN COMPUTER LAB
QUOTE:
The competent programmer programmer is fully aware of the limitations of his own mind. That’s why he approaches his programming task with humility. He avoids other clever things like the plague. – EWD 340
10:47
PROF EDSGER W. DIJKSTRA:
I remember the first time, which was in 1970, the first time I really went out into the marketplace to show how you could develop programs so that you could keep a firm hold on them intellectually. Well I first went to Paris and then to Brussels. In Paris I gave a lecture at the Sorbonne and the audience was very enthusiastic. And then on the way home I gave the same story to a large software house in Brussels. The story story fell completely flat on its face. I have never given such a bad lecture in a way. And later I discovered why: management was not interested in flawless programs because it was the maintenance contracts from which the company derived its stability. And the programmers weren’t interested in it either because it turned out that they derived a lot of their intellectual excitement from the fact that they DID NOT understand exactly what they were doing. They had the feeling that if you knew exactly what you were doing and you did not run any risks, then it was a boring profession.
BACK IN SOFTWARE LABORATORY
QUOTE:
We should not introduce errors into a program out of carelessness. We must do this systematically and with care.
12:23
EXT. CAMPUS
INTERMEZZO
MUSIC MOZART ANDANTE,
12:45
PROF EDSGER W. DIJKSTRA:
If you don’t understand something in physics, you can always hide behind the unfathomable depths of nature. You can always blame God. You didn’t make it so complicated yourself. But if your program doesn’t work, you have no one to hide behind. You cannot hide behind unwilling nature, no, a zero is a zero and a one is a one and if it doesn’t work you just did it wrong.
13:27
PAN OVER PHOTO WITH MEMORY CORE
QUOTE:
I realized that my previous projects had only been finger exercises. I now had to tackle the complexity myself. But it took me a long time to get up the courage to do so.
COMMENTARY:
In the sixties, Dijkstra saw how the complexity of the programs was growing beyond the programmers’ heads. And that even the most prestigious projects were threatened.
PROF EDSGER W. DIJKSTRA:
That was an experience in 1969, just after the first successful moon landing. It was a conference at, uh, a NATO conference on software engineering in Rome and there I met Joel Aron who was the head of IBM’s federal systems division and that was the department that had been responsible for the moonshot software. And I knew that every Apollo flight needed something like 40,000 new lines of code, well it doesn’t really matter what kind of unit a line of code is, 40,000 is a lot and I was deeply impressed that they put so much software in good condition. had gotten in order, so when I met Joel Aron I said: how do you do it? Do what? he asked. Well I said, getting that software right. Right?!? he said. And then he said that in one of the calculations of the orbit of the Lunar Module the moon was repelling instead of attracting and they had that drawing error by chance, you have to think about it: discovered by chance, five days in advance. I blushed and said: Those guys have been lucky. Yes! was Joel Aron’s response.
SHOT FULL MOON
QUOTE:
Testing a program is an effective way to demonstrate the presence of errors in a program, but it is completely inadequate to prove their absence. – EWD 340
15:57
PHOTOS X-8
COMMENTARY:
Dijkstra now knew from his own experience how frighteningly complicated computer programs could become. He had completed the operating system for the Netherlands’ largest computer, the X-8, under high voltage.
PROF EDSGER W. DIJKSTRA:
The fear that you might not get it right, that it will escape from your control.
Look in the case of the
I knew that if we designed the thing with insufficient care, we would have something that once entered into the machine would not work, would occasionally make strange errors while it was completely unclear whether we would ever be able to figure out what caused it. of that. And that was the reason to put all conceivable control into the development phase.
X-8
PHOTOS X-8
17:23
PAN OVER BOOKCASE WORKROOM
TO DIJKSTRA FOR BLACKBOARD
QUOTE:
Elegance is not an unnecessary luxury, but is the difference between success and failure. – EWD 1284
SEQUENCE: WRITING ON BOARD
OF MATHEMATICAL PROOF
PROF EDSGER W. DIJKSTRA:
One of the things I discovered in the sixties is that mathematical elegance, mathematical elegance, is not an aesthetic issue, a matter of taste or fashion or whatever, but that you can translate it into a technical concept. . Because in the Concise Oxford Dictionary, for example, you will find one of the meanings of ‘elegant’: ingeniously simple and effective.
In programming practice it is manageable because if you make a really elegant program it is, well, firstly shorter than most of its alternatives, secondly it consists of clearly separated parts of which you can replace one part with an alternative implementation without that this influences the rest of the program, and surprisingly, the elegant programs are also often the most efficient.
18:59
INTERMEZZO:
CAMPUS AND Glockenspiel
QUOTE:
As long as there were no computers, programming was not a problem at all. When we had a few small computers, programming became a bit of a problem. Now that we have gigantic computers, programming has become a gigantic problem.
19:23
PHOTO 2 MAN AT ARMAC
PROF EDSGER W. DIJKSTRA:
My first years as a programmer were a bit strange compared to now in that I was always programming for machines that didn’t exist yet. My friends Bram Loopstra and Carel Scholten built the machine and while they built the machine I created the relevant software. I was very used to not writing something down and then trying it because then the machine for trying it wasn’t ready yet. So I knew from the beginning that you had to create something that you had to continue to master intellectually.
DIJKSTRA 1950s
PHOTO ZOOM DIJKSTRA 1950s
20:05
VAN THROUGH LANDSCAPE; TEXMEX MUSIC
UNPACKING VAN WITH WOMAN
DISS TO PHOTO TRIO FOR MATHEMATICAL INSTITUTE
PROF EDSGER W. DIJKSTRA:
I came to work at the Mathematical Center in March '52. My future wife had worked there since the summer of 1949. She worked there as an accountant. I came there as a programmer and well… I liked her. I went out with her for the first time to a concert on the occasion of the 1954 mathematical conference in Amsterdam. Well, then the bear was loose and of course everyone knew that I had a crush on her. I was generally complimented on my good taste.
TOGETHER DOWN THE PATH
WATCHING LANDSCAPE WITH WIDE WATERFALL
PROF EDSGER W. DIJKSTRA
PROF EDSGER W. DIJKSTRA:
I think I have been seriously influenced professionally by my mother. She was a brilliant mathematician and I remember when I had bought the next year’s books during the summer holidays, I saw the trigonometry book and I thought it looked very creepy all with Greek letters and I asked my mother if trigonometry was difficult. She said, no, not at all. You need to make sure you know all your formulas well and if you need more than five lines then you are on the wrong track.
22:17
TOTAL HOME DIJKSTRA IN AUSTIN
PUZZLE SOUNDS
PROF EDSGER W. DIJKSTRA (OFSCREEN):
The question is why elegance in general has received so little attention. It has indeed received little attention.
The downside of elegance is, if you want to call it a downside by the way, it takes hard work and dedication to achieve it and a good upbringing to appreciate it.
MAKE NRC CRYPTO AT HOME BY JJ STEENHUIS
MAKE AT HOME NRC CRYPTO BY JJ STEENHUIS;
hard toy: BIKKEL
The country’s tune, the country’s honour: NATIONAL SONG
as a nightlife option: NIGHT CANDLE
CREDITS ON W2S
23:34
COMPOSITION Jos Wassink
CAMERA Piotr Kukla
SOUND Leo Franssen
ASSEMBLY Jan Overweg
MUSIC Jan Jonker
MIX Hans Brouwer
COLOR CORRECTION Gerhard van der Beek
LEADER Marco Vermaas
COMMENT Tessel Blok
EDITORIAL Kees de Groot van Embden
Maarten Hidskes
Simon Rozendaal
Karin Schagen
Annemieke Smit
Jacqueline de Vree
Gertjan Wallinga
Jos Wassink
MMV Ria Dijkstra
Rutger Dijkstra
Dr Carel S. Scholten
Dr Wim Feijen
Dr. Netty van Gasteren
Prof Gerrit Blaauw
Caroline Nieuwendijk
Gerard Alberts
Marla Martinez
Dr. Hamilton Richards
Steve Thomas
PRODUCTION Karin Spiegel
Madeleine Somer
FINAL EDITORIAL Hansje van Etten
copyright VPRO 2001
24:00
BLACK