sign in sign up
<<
Home , Alan Perlis, ALCOR, ALGO, Block (programming), Brian Randell, Charles Babbage

viewpoints

Vdoi:10.1145/1787234.1787249 Thomas . Misa, Editor Interview An Interview with Edsger W. Dijkstra The luminary, in one of his last interviews before his death in 2002, reflects on a ’s life.

he Institute holds one of the world’s largest collections of re- search-grade oral interviews relating to the Thistory of , , and networking. Most of the 350 inter- views have been conducted in the context of specific research projects, facilitate the interviewer’s ex- tensive preparation and often sug- gest specific lines of questions. Tran- scripts from these oral histories are a key source in understanding the his- tory of , since traditional historical sources are frequently in- complete. This interview with pro- gramming pioneer Edsger Dijkstra (1930–2002) was conducted by CBI researcher Phil Frana Dijkstra’s home in Austin, TX, in August 2001 for a NSF-KDI project on “Building a Future for Software History.” Winner of ACM’s A.M. in 1972, Dijkstra is well known for his contributions to computer n i science as well as his colorful assess- ust A ments of the . His contributions to this magazine continue to enrich exas at f new generations of computing scien- ty o i tists and practitioners. ers

iv We present this interview post- n U

e humously on the eighth anniver- Th f sary of Dijkstra’s death at age 72 in August 2002; this interview has been

ourtesy o condensed from the complete tran-

ph script, available at http://www.cbi.umn. ra g edu/oh. oto h

P —Thomas J. Misa

august 2010 | vol. 53 | no. 8 | of the acm 41 viewpoints

How did your career start? Sydney or Melbourne. The final part of It all started in 1951, when my father the journey was on an F27 to Canber- enabled me to go to a programming I had never ra. And we arrived and I met my host, course in , . It was used someone whom I had never met before. And he a frightening experience: the first time was very apologetic that this world that I left the Netherlands, the first time else’s software. traveler had to do the last leg of the I ever had to understand people speak- If something journey on such a shaky two-engine ing English. I was all by myself, trying turboprop. And it gave me the dear to follow a course on a totally new topic. went wrong, I had opportunity for a one-upmanship that But I liked it very much. The Nether- done it. And it was I never got again. I could honestly say, lands was such a country that “Dr. Stanton, I felt quite safe: I calcu- Aad van Wijngaarden, who was the di- that unforgivingness lated the resonance frequencies of the rector of the Department that challenged me. wings myself.” [laughter] of the Mathematical Centre in Amster- In 1956, as soon as I had decided to dam, knew of this, and he offered me a become a programmer, I finished my job. And on a part-time basis, I became studies as quickly as possible, since I the programmer of the Mathematical no longer felt welcome at the univer- Centre in March of 1952. They didn’t sity: the considered me as have computers yet; they were trying to a deserter, and the build them. The first eight years of my friends, because if you asked them were dismissive and somewhat con- programming there I developed the what their professional competence temptuous about computing. In the software for a of consisted of, they could point out that mathematical culture of those days being built at the Mathematical Cen- they knew everything about triodes, you had to deal with infinity to tre. In those years I was a very conser- pentodes, and other electronic gear. your topic scientifically respectable. vative programmer. The way in which And there was nothing I could point to! programs were written down, the I spoke with van Wijngaarden in There’s a curious story behind your of the instruction on paper, the 1955, and he agreed that there was no “shortest path” . organization; it was very much such thing as a clear scientific compo- In 1956 I did two important things, modeled after what I had seen in 1951 nent in , but I got my degree and we had the festive in Cambridge. that I might very well be one of the peo- opening of the ARMAC.c We had to have ple called to make it a science. And at a demonstration. Now the ARRA, a few When you got married in 1957, you the time, I was the of guy to whom years earlier, had been so unreliable could not enter the term “programmer” you could say such things. As I said, I that the only safe demonstration we into your marriage record? was trained to become a scientist. dared to give was the generation of ran- That’s true. I think that “- dom numbers, but for the reliable mer” became recognized in the early What projects did you work on in Am- ARMAC I could try something more am- 1960s. I was supposed to study theoreti- sterdam? bitious. For a demonstration for non- cal , and that was the for When I came in 1952, they were computing people you have to have a going to Cambridge. However, in 1955 working on the ARRA,a but they could problem statement that non-mathema- after three years of programming, while not get it reliable, and an updated - ticians can understand; they even have I was still a student, I concluded that the sion was built, using selenium diodes. to understand the answer. So I designed intellectual challenge of programming And then the Mathematical Centre a program that would the shortest was greater than the intellectual chal- built a for Fokker Aircraft between two cities in the Nether- lenge of , and as a Industry. So the FERTA, an updated lands, using a somewhat reduced road- result I chose programming. Program- version of the ARRA, was built and in- map of the Netherlands, on which I had ming was so unforgiving. If something stalled at Schiphol. The installation selected 64 cities (so that in the coding went wrong, I mean a zero is a zero and I did together with the young six would suffice to identify a city). a one is a one. I had never used some- Blaauw who later became one of the What’s the shortest way to travel one else’s software. If something went designers of the IBM 360, with Gene from Rotterdam to Groningen? It is the wrong, I had done it. And it was that un- Amdahl and . algorithm for the shortest path, which forgivingness that challenged me. One funny story about the Fairchild I designed in about 20 minutes. One I also began to realize that in some F27: On my first to Australia, I flew morning I was shopping in Amsterdam strange way, programs could become on a big 747 from Amsterdam to Los with my young fiancée, and tired, we sat very complicated or tricky. So it was Angeles, then on another 747 I flew to down on the café terrace to drink a cup in 1955 when I decided not to become of and I was just thinking about a , to become a programmer whether I could do this, and I then instead. At the time programming a Automatische Relais Rekenmachine Amster- dam = Automatic Relay Amsterdam. didn’t look like doing science; it was b Fokker Electronische Rekenmachine Te Am- c Automatische Rekenmachine MAthematische just a mixture of being ingenious and sterdam = Fokker Electronic Calculator In Centrum = Automatic Calculator Mathemati- being accurate. I envied my hardware Amsterdam cal Centre

42 communications of the acm | august 2010 | vol. 53 | no. 8 viewpoints designed the algorithm for the short- would write down the formal specifi- The OS/360 monitor idea would have est path. As I said, it was a 20-minute cation of the machine, and all three of never occurred to a European? invention. In fact, it was published in us would sign it with our blood, so to No, we were too poor to consider it 1959, three years later. The publication speak. And then our ways parted. All the and we also decided that we should try is still quite nice. One of the programming I did was on paper. So I to our designs in such a way that it is so nice was that I designed was quite used to developing programs that we could keep things under our in- it without pencil and paper. Without without testing them. tellectual control. This was a major dif- pencil and paper you are almost forced There was not a way to test them, so ference between European and Ameri- to avoid all avoidable complexities. you’ve got to convince yourself of their can attitudes about programming. Eventually that algorithm became, to correctness by reasoning about them. my great amazement, one of the cor- A simple writing error did not matter How did the notion of program proofs nerstones of my fame. I found it in the as long as the machine wasn’t there yet, arise? early 1960s in a German book on man- and as soon as errors would show up In 1959, I had challenged my col- agement science—“Das Dijkstra’sche on the machine, they would be simple leagues at the Mathematical Centre Verfahren” [“Dijkstra’s procedure”]. to correct. But in 1957, the idea of a with the following programming task. Suddenly, there was a method named real-time interrupt created a vision of Consider two cyclic programs, and in after me. And it jumped again recently a program with non-reproducible er- each cycle a section occurs called the because it is extensively used in all trav- rors, because a real-time interrupt oc- critical section. The two programs el planners. If, these days, you want to curs at an unpredictable moment. My can communicate by single reads go from here to there and you have a hardware friends said, “Yes, yes, we and single writes, and about the rela- car with a GPS and a screen, it can give see your problem, but surely you must tive speeds of the programs nothing you the shortest way. be up to it…” I learned to cope with it. I is known. Try to synchronize these wrote a real-time interrupt handler that programs in such a way that at any When was the “shortest path” - was flawless and that became the topic moment in time at most one of them rithm originally published? of my Ph.. thesis.3 Later I would learn is engaged in its critical section.d I It was originally published in 1959 that this would almost be considered an looked at it and realized it was not in Numerische Mathematik edited by un-American activity. trivial at all, there were all sorts of side F.L. Bauer. Now, at the time, an algo- conditions. For instance, if one of the rithm for the shortest path was hardly How was the computing culture in programs would stay for a very long considered mathematics: there was a America different? time in its noncritical section, the finite number of ways of going from A Well, the American reaction was very other one should go on unhampered. to B and obviously there is a shortest different. When IBM had to develop the We did not allow ‘After-you-after-you’ one, so what’s all the fuss about? It re- software for the 360, they built one or blocking, where the programs would mained unpublished until Bauer asked two machines especially equipped with compete for access to the critical sec- whether we could contribute some- a monitor. That is an extra piece of ma- tion and the dilemma would never be thing. In the meantime I had also de- chinery that would exactly record when solved. Now, my friends at the Math- signed the shortest sub-spanning interrupts took place. And if something ematical Centre handed in their so- for my hardware friends. You know, went wrong, it could replay it again. So lutions, but they were all wrong. For on the big panel you have to connect a they made it reproducible, yes, but at each, I would sketch a that whole lot of points with the same cop- the expense of much more hardware would reveal the bug. People made per wire because they have to have the than we could afford. Needless to say, their programs more sophisticated same voltage. How do you minimize the they never got the OS/360 right. and more complicated. The construc- amount of copper wire that connects tion and counterexamples became these points? So I wrote “A note on two even more time-consuming, and I had problems in connection with graphs.”2 In the mathematical to change the rules of the game. I said, Years later when I went to my ophthal- “Sir, sorry, from now onward I only ac- mologist—he did not even know that culture of those cept a solution with an argument why I was a computing scientist—he said, days you had it is correct.” “Have you designed the algorithm for Within three hours or so Th. J. GPS?” It turned out he had seen the Sci- to deal with infinity Dekker came with a perfect solution entific Americanof November 2000.10 to make your and a proof of its correctness. He had analyzed what kind of proof would be How could you tell if early programs topic scientifically needed. What are the things I have to were correct? respectable. show? How can I prove them? Having For those first five years I had always been programming for non-existing machines. We would design the instruc- d This is an of the mutual ex- clusion problem, which later became a corner- tion code, I would check whether I could stone of the THE multiprogramming live with it, and my hardware friends [THE = Technische Hogeschool Eindhoven would check that they could build it. I (Technical University Eindhoven)].

august 2010 | vol. 53 | no. 8 | communications of the acm 43 viewpoints

settled that, he wrote down a program ing into a rigorous discipline, no longer who had to agree with the report, none that met the proof’s requirement. You a lot of handwaving. But perhaps more saw that sentence. That’s how recur- lose a lot when you the role of important, it made writing sion was explicitly included. mathematics to program verification and language definition topics worthy as opposed to program construction or of academic attention. It played a ma- Was this called at that time? derivation. jor role in making computing science Oh yes. The was quite well Another experience in 1959 was at- academically respectable. The third known. It was included in LISP, which tending the “zeroth” IFIP Congress novelty was the introduction of the was beginning to emerge at that time. in Paris. My international contacts “Boolean” as a first-class citizen. We made it overlookable. And F.L. had started in December 1958, with It turns the Boolean expression from a Bauer would never have admitted it in the meetings for the design of ALGOL statement of fact that may be wrong or the final version of the ALGOL 60 Re- 60. My boss, Aad van Wijingaarden, right into an expression that has a val- port, had he known it. He immediately had had a serious car accident, and ue, say, “true” or “false.” How great that founded the ALCOR . It was a Jaap Zonneveld and I, as his immedi- step was I learned from my mother’s group that together would implement ate underlings, had to him. reaction. She was a gifted mathemati- a subset of ALGOL 60, with recursion Zonneveld was a numerical analyst, cian, but she could not make that step. emphatically ruled out. while I did the programming work. For her, “three plus five is ten” was not The ALGOL 60 meetings were about a complicated way of saying “false”; it What were other novelties in ALGOL 60? the first time that I had to carry out dis- was just wrong. A fifth novelty that should be men- cussions spontaneously in English. It Potentially this is going to have tioned was the structure. It was a was tough. a very profound influence on how tool for structuring the program, with mathematics is done, because math- the same use of the word “structure” You’ve remarked that learning many ematical proofs, can now be rendered as I used nine years later in the term different languages is useful to pro- as simple calculations that reduce “.” The con- gramming. a Boolean expression by value-pre- cept of lexical was beautifully Oh yes, it’s useful. There is an enor- serving transformations to the value blended with nested lifetimes during mous difference between one who “true.” The fourth novelty was the in- , and I have never been able is monolingual and someone who at troduction of recursion into impera- to figure out who was responsible for least knows a second language well, tive programming. Recursion was that synthesis, but I was deeply im- because it makes you much more con- a major step. It was introduced in a pressed when I saw it. scious about language structure in sneaky way. The draft ALGOL 60 report Finally, the definition of the seman- general. You will discover that certain was circulated in one of the last weeks tics was much less operational than constructions in one language you just of December 1959. We studied it and it was for existing programming lan- can’t translate. I was once asked what realized that recursive calls were all guages. was essentially de- were the most vital assets of a compe- but admitted, though it wasn’t stated. fined by its implementation, whereas tent programmer. I said “mathemati- And I phoned —that call with ALGOL 60 the idea emerged that cal inclination” because at the time it to Copenhagen was my first interna- the should be was not clear how mathematics could tional telephone call; I’ll never forget defined independent of computers, contribute to a programming chal- the excitement!—and dictated to him , stores, etc.; the definition lenge. And I said “exceptional mas- one suggestion. It was something like should define what the implementa- tery” of his tongue because you “Any other occurrence of the proce- tion should look like. Now these are have to think in terms of words and dure identifier denotes reactivation five or six issues that for many years sentences using a language you are fa- of the procedure.” That sentence was the has missed, and I miliar with. inserted sneakily. And of all the people think that is a tragedy. It was the ob- session with speed, the power of IBM, How was ALGOL 60 a turning point? the general feeling at the time that Computing science started with AL- All the programming programming was something that GOL 60. Now the reason that ALGOL 60 should be doable by uneducated mo- was such a was that it was not I did was on paper. rons picked from the street, it should a university project but a project cre- So I was quite not require any sophistication. Yes… ated by an international committee. false dreams paralyzed a lot of Ameri- It also introduced about a half-dozen used to developing can computing science. profound novelties. First of all, it in- programs without troduced the absence of such arbitrary When did you understand that pro- constraints as, say, ruling out the sub- testing them. gramming was a deep subject? scripted subscript, the example I men- I had published a paper called “Re- tioned. A second novelty was that at cursive Programming,” again in Nu- least for the context-free , a for- merische Mathematik.8 In 1961, I was mal definition was given. That made a beginning to realize that programming tremendous difference! It turned pars- really was an intellectual challenge.

44 communications of the acm | august 2010 | vol. 53 | no. 8 viewpoints

Peter Naur and I were main speakers On the blackboard he wrote wall-to- at a workshop or a summer school in wall formulae and I didn’t understand Brighton, England; there were quite Thanks to a single word of it. But there were many a number of well-known British sci- my isolation, people that joined the discussion and entists in that audience. In the audi- posed questions. And I couldn’t under- ence was , but neither of I would do things stand those questions either. During us remembers that. I don’t remember differently than a reception, I voiced my worry that I him because he was one of the many was there on false premises. “The first people in the audience, and he doesn’t people subjected speaker, I did not understand a word remember it because in his to the standard of it.” “Oh,” he said, “none of us did. Peter Naur and I, both bearded and That was all nonsense and gibberish, both with a Continental accent, have pressures of but IBM is sponsoring this, so we had merged into one person. [laughter] We conformity. to give the first slot to an IBM speaker.” reconstructed years later that we were Well, that was totally new for me. Let’s both there. I was a free man. say that the fence between science and In 1962, my thinking about program industry, the fence around a university synchronization resulted in the P- & V- campus, is here [in the U.S.] not as high operations. The other thing I remem- as I was used to. ber was a conference in on sym- bol manipulation, in April or so. Peter How did GO TO become ‘harmful’? Naur was there, with his wife. There was [laughter] the darkest week in my In 1967 was the ACM Conference were panel discussions and Peter and I professional life. In a NATO Confer- on Operating Principles in were sitting next to each other and we ence on Software in 1969 Gatlinburg. That, I think, was the first had all sorts of nasty comments, but in Rome,11 I characterized the Russian time that I had a large American audi- we made it the rule that we would go decision to build a -compatible copy ence. It was at that meeting where one to the microphone in turn. This had of the IBM 360 as the greatest Ameri- afternoon I explained to gone on for an hour or so, and van Wi- can victory in the Cold War. and a few others why the GO TO - jngaarden, my boss, was sitting next to Okay now, 1964–1965. I had general- ment introduced complexity. And they an American and at a given moment ized Dekker’s solution for N processes asked me to publish it. So I sent an ar- the American grabs his shoulder and and the last sentence of that one-page ticle called “A Case Against the GO TO says “My God! There are two of them.” article is, “And this, the author believes, Statement” to Communications of the [laughter] This may be included in an completes the proof.” According to ACM. The editor of the section wanted oral history? It’s not mathematics, it Doug Ross, it was the first publication to publish it as quickly as possible, so isn’t either, but it is of an algorithm that included its cor- he turned it from an article into a Let- a true story…. rectness proof. I wrote “Cooperating ter to the Editor. And in doing so, he In September 1962, I went to the first Sequential Processes,” and I invented changed the into, “GO TO State- IFIP Congress, in Munich, and gave an the Problem of the Dining Quintuple, ment .”4 That title invited on advancing program- which Tony Hoare later named the became a template. Hundreds of writ- ming. I got a number of curtain calls: Problem of the Dining Philosophers.5 ers have “X considered harmful,” with clearly I was saying something unusu- X anything. The editor who made this al. Then I became a professor of Mathe- When did you first visit theU .S.? change was . matics in Eindhoven, and for two years My first trip to the U.S. was in 1963. I lectured on numerical . By That was to an ACM Conference in Why is “elegance” in programming im- 1963–1964, I had designed with Carel Princeton. And I visited a number of portant? S. Scholten the hardware channels and Burroughs offices; that was the first 1968 was exciting because of the the interrupts of the Electrologica X8, time I met . I must al- first NATO Conference on Software the last machine my friends built, and ready have had some fame in 1963, Engineering, in Garmisch. In BIT I then I started on the design of THE because there was an ACM workshop published a paper, on “A Constructive multiprogramming system. with about 60 to 80 participants and I Approach to the Problem of Program Of course, 1964 was the year in was invited to join. And they paid me Correctness.”1 1968 was also the year which IBM announced the 360. I was $500. I didn’t need to give a speech, I of the IBM advertisement in Datama- extremely cross with Gerry Blaauw, didn’t need to sit in a panel discussion, tion, of a beaming Susie Meyer who because there were serious flaws built they just would like me to be there. had just solved all her programming into the I/O organization of that ma- Quite an amazing experience. problems by switching to PL/I. Those chine.7 He should have known about were the days we were led to believe the care that has to go into the design What about your first two trips to Amer- that the problems of programming of such things, but that was clearly not ica surprised you about the profession? were the problems of the deficiencies a part of the IBM culture. In my Tur- Well, the first lecture at that ACM of the programming language you ing Lecture I described the week that I workshop was given by a guy from IBM. were working with. How did I char- studied the specifications of the 360, it It was very algebraic and complicated. acterize it? “APL is a mistake, carried

august 2010 | vol. 53 | no. 8 | communications of the acm 45 viewpoints

through to perfection. It is the lan- was a very heavy use of anthropomor- guage of the future for the program- phic terminology, the “electronic ming techniques of the past: it creates ” or “machines that think.” That a new generation of coding bums.” is absolutely killing. The use of anthro- I thought that should pomorphic terminology forces you Content Written not be puzzle-minded, which was one linguistically to adopt an operational of the criteria on which IBM selected view. And it makes it practically impos- by Experts programmers. We would be much bet- sible to argue about programs inde- ter served by clean, systematic , pendently of their being executed. with a sense of elegance. And APL, with its one-liners, went in the other direc- Is this why re- tion. I have been exposed to more APL search seemingly doesn’t take hold in than I’d like because had Europe? an APL period. I think he outgrew it be- There was a very clear financial Blogs fore his death, but for many years APL constraint: at the time we had to use was “it.” the machines we could build with the Articles available stuff. There is also a great Roundtables Why did your “structured program- cultural barrier. The European ming” have such impact? tends to maintain a greater distinction Case Studies In 1969, I wrote “Notes on Struc- between man and machine. It’s less in- 6 JOBMultimedia9-513 tured Programming,” which I think clined to describe machines in anthro- 2.25 X 4.75 owed its American impact to the fact pomorphic terminology; it’s also less RSS that it had been written at the other inclined to describe the mind COMMUNICATIONS OF THE ACMside of the Atlantic Ocean; which has in mechanical terminology. Freud two very different sides. I can talk never became the rage in Europe as he about this with some authority, hav- became in the United States. ing lived here [in the U.S.] for the bet- ter part of 17 years. I think that thanks You’ve said, “The tools we use have a queue.acm.org to the greatly improved possibility of profound and devious influence on , we overrate its im- our thinking habits, and therefore on portance. Even stronger, we underrate our thinking abilities.” the importance of isolation. See, look The devious influence was in- IMAGINE... at what that 1963 invitation to the ACM spired by the experience with a bright a graduate computer science program workshop illustrates, at a time when I student. In the oral examination we that offers you the convenience and access of online had published very little. I had imple- solved a problem. Together we con- learning, combined with the benefits of participating in live classroom discussion and interaction. mented ALGOL 60 and I had written a structed the program, decided what real-time interrupt handler, I had just had to be done, but very close to the It’s here... the Brooklyn Campus become a professional programmer. end, the kid got stuck. I was amazed of Long Island University is offering Yet I turned out to be quite well known. because he had understood the prob- a NEW BLENDED LEARNING program How come? Thanks to my isolation, I lem perfectly. It turned out he had to that fuses online learning with traditional classroom would do things differently than peo- write a subscripted value in a subscript studies, significantly reducing the amount of time ple subjected to the standard pres- position, the idea of a subscripted sub- you’ll spend on campus and maximizing interaction sures of conformity. I was a free man. script, something that was not allowed with faculty members and fellow students. in FORTRAN. And having been educat- What were other differences between ed in FORTRAN, he couldn’t think of M.S. in Computer Science Europe and the U.S.? it, although it was a construction that Brooklyn Campus One of the things that saved Europe he had seen me using at my lectures. Session was that until 1960 or so, it was not con- Wednesday, August 18, 6:00 p.m. Saturday, August 21, 10:30 a.m. sidered an interesting market. So we So the use of FORTRAN made him un- 718-488-1011 • [email protected] were ignored. We were spared the pres- able to solve that? sure. I had no idea of the power of large Indeed. When young students have companies. Only recently I learned that difficulty in understanding recursion, in constant dollars the development of it is always due to the fact that they had the IBM 360 has been more expensive learned programming in a program- than the Manhattan Project. ming language that did not permit it. I was beginning to see American If you are now trained in such an op- publications in the first issue of Com- erational way of thinking, at a given munications of the ACM. I was shocked moment your pattern of understand- by the clumsy, immature way in which ing becomes visualizing what happens greater access to excellent they talked about computing. There during the execution of the algorithm.

46 communications of the acm | august 2010 | vol. 53 | no. 8 viewpoints

The only way in which you can see the pressure on the universities to supply algorithm is as a FORTRAN program. them, even if the university did not In many places, quite know how. In many places, de- And what’s the answer then for our fu- departments of partments of computer science were ture students to avoid the same trap? founded before the shape of the - Teach them, as soon as possible, a computer science lectual discipline stood out clearly. decent programming language that ex- were founded before You also find it reflected in the ercises their power of . Dur- of scientific societies, such as ing 1968 in Garmisch I learned that in the shape of the the Association for Computing Ma- the ears of the Americans, a “math- intellectual discipline chinery. It’s the British Computer Soci- ematical ” [such as we educat- ety and it was the Dutch who had Het ed in Eindhoven] was a contradiction stood out clearly. Nederlands Rekenmachine Genootsc- in terms: the American hap; without knowing Dutch, you can is an impractical academic, whereas hear the word “machine” in that . the American engineer is practical And you got the departments of Com- but hardly academically trained. You puter Science. Rather than the depart- notice that all important words carry ment of computing science or the de- different, slightly different meanings. partment of computation. Europe was I was disappointed in America by the of the population it is supposed to ad- later, it coined the term . way in which it rejected ALGOL 60. I dress have changed radically. That al- Tony Hoare was a Professor of Compu- had not expected it. I consider it a trag- ready started in the 1970s. So whatever tation. edy because it is a symptom of how I say about the [European] university the United States is becoming more is probably idealized by memory. Yes. “Information” came a bit later on? and more a-mathematical, as Morris But a major difference was that the It was the French that pushed in- Kline illustrates eloquently.9 Precisely fence around the university campus formatique. Today the English prefer in the century which witnesses the was higher. To give you an example, Information , IT, and In- emergence of computing equipment, when we started to design a comput- formation Systems, IS. I think the tim- it pays so much to have a well-trained ing science curriculum in the 1960s, ing has forced the American depart- mathematical mind. one of the firm rules was that no in- ments to start too early. And they still dustrial product would be the subject suffer from it. Here, at the University In 1963 Peter Patton, in Communica- of an academic course. It’s lovely. This of Texas, you can still observe it is the tions of the ACM, wrote that European immediately rules out all courses, Department of Computer . If programmers are fiercely independent and at the time it ruled out all FOR- you start to think about it, you can only loners whereas Americans are team TRAN courses. We taught ALGOL 60, laugh, but that time there were at least players. Or is it the other way? it was a much greater eye-opener than as many computer sciences as there At the Mathematical Centre, we FORTRAN. were professors. used to cooperate on large projects and apply a of labor; it was Is there a relationship between the something of a shock when I went to curriculum and the nature of funding 1. dijkstra, E.W. A constructive approach to the problem of program correctness. BIT 8, 3 (1968), 174–186. the Department of Mathematics at of universities? 2. dijkstra, E.W. A note on two problems in connection Eindhoven where everybody worked Yes. It has the greatest influence on with graphs. Numerische Mathematik 1 (1959), 269–271. all by himself. After we had completed the funding of research projects. Quite 3. dijkstra, E.W. Communication with an automatic computer. Ph.D. dissertation, University of the THE System, for instance, Nico regularly I see firm XYZ proposing to Amsterdam, 1959. Habermann wrote a thesis about the give student fellowships or something 4. dijkstra, E.W. Go To statement considered harmful, Commun. ACM 11, 3 (Mar. 1968), 147–148. Banker’s Algorithm, and about sched- and then, somewhere in the small 5. dijkstra, E.W. Hierarchical ordering of sequential uling, , and deadlock preven- print, that preference will be given to processes. Acta Informatica 1 (1971), 115–138. 6. dijkstra, E.W. Notes on structured programming. tion. The department did not like that students who are supervised by pro- In O.-J. Dahl, E.W. Dijkstra, and C.A.. Hoare, Eds., because it was not clear how much he fessors who already have professional Structured Programming. Academic Press, , 1972, 1–82. had done by himself. They made so contact with the company. 7. dijkstra, E.W. Over de IBM 360, EWD 255, n.d., much protest that Cor Ligtmans, who circulated privately; http://www.cs.utexas.edu/ users/ EWD/ewd02xx/EWD255.PDF should have written his Ph.D. thesis on Why do computer science depart- 8. dijkstra, E.W. Recursive programming. Numerische another aspect of THE System, refused ments often come out of electrical Mathematik 2 (1960), 312–318. 9. Kline, M. Mathematics in Western Culture. Penguin to do so. engineering in the U.S.—but not in Eu- Books Ltd., Harmondsmorth, Middlesex, England, rope? 1972. 10. Menduno, M. shrugged: When it comes to online Is the outcome of the curricula differ- A major reason is timing. For fi- road maps, why you can’t (always) get there from ent in Europe and America? nancial reasons, Europe, damaged by here. Scientific American 283, 11 (Nov. 2000), 20–22. 11. randell, B. and Buxton, J.N., Eds., Software I must be very careful with answer- World War II, was later. So the Ameri- Engineering Techniques: A Report on a Conference Sponsored by the NATO Science Committee (Rome, ing this because during my absence, can computing industry emerged ear- , Oct. 1969), NATO, 1970. the role of the university, the financ- lier. The computing industry asked ing of the university, and the fraction for graduates, which increased the Copyright held by author.

august 2010 | vol. 53 | no. 8 | communications of the acm 47