Hacker News Reader

Comment by evertheylen | Hacker News Reader

For those interested in Dijkstra, this website has a lot to offer: http://www.dijkstrascry.com/

I also followed some lectures given by the guy that runs that website. Apparently Dijkstra was not an easy person to get along with, there are stories of him being bored during some lecture, walking to the front, and laying down for a nap. (Sorry I can't provide a source, it's basically a rumour among academic staff here in Belgium).

View on HN

> David Gries (Cornell University) was one of the recipients of "on-line" coaching during a lecture:

> My own experience with lecturing before Edsger took place in Marktoberdorf. It rests on the fact that in some languages (notably Fortran), the equality symbol and the assignment symbol are the same, and many people say "x equals e" when they mean "store the value of e in x," or "x becomes e."

> I was lecturing along, when I said "x equals e" meaning an assignment of e to x. From the back of the room came a loud "becomes," and then a stunned silence. Finally, I gathered my wits and said, "Thank you, Edsger, for correcting me. If I make the same mistake again, stop me." Twenty minutes later, I made the same mistake, and again from the back of the room came "becomes" "Thanks, I won't make the mistake again," I said, and to this day I haven't!

(https://history.computer.org/pioneers/dijkstra.html)

One of my drinking buddies, who has no affinity with computers whatsoever, recently casually remarked how he had often met Dijkstra. Apparently he was friends with one of the Dijkstra kids and used to visit him at their home in Nuenen. "The father was supposed to be some hotshot computer professor."

Needless to say, I nearly swooned when he told that. I couldn't help pumping him for details, none of which I feel at liberty to pass on to you, except that he thought Dijkstra was excentric but kind.

Perhaps not completely related to this topic, but interesting nonetheless, a small picture into Dijkstra's life from Dutch tv (subtitled in English): https://www.youtube.com/watch?v=RCCigccBzIU

Interesting video, thanks.

The subtitles at around 4:05 are out of sync. The quote reads

> Je moet de wereld niet geven waarom ze vraagt, maar wat ze nodig heeft.

In English:

> You must not give the world what it asks, but what it needs.

Edgar Daylight (the guy that runs that website) writes interesting books/booklets interviewing persons like Barbara Liskov and Donald Knuth. Here is a list of them: http://www.lonelyscholar.com/

I have read them all and recommend them. The one with Michael Jackson is particularly intersting for it's dialectic qualities.

Another typical part of the lecture style was the “Dijkstra Font”. Some of the lectures in my CS program were given by people who worked with Dijkstra, like Feijen, and they all used the exact same style of beautiful handwriting. Perhaps the font was a result of really taking the time to think things over to make sure it was correct. Those lectures could move real slow when it was time to write some predicate logic or pseudo-code on a blackboard, but they hardly used the eraser.

There is an old joke on the Internet about measuring arrogance in nanoDijkstra, but I don't know how much of it is about real arrogance and how much is about he telling things as they are.

It is practically impossible to teach good programming to students that have had a prior exposure to BASIC: as potential programmers they are mentally mutilated beyond hope of regeneration. - Dijkstra

Pretty arrogant gatekeeping. And obviously wrong.

Taken from EWD 498, which is easily recognizable as satire unless you present it ouf of context.

https://www.cs.utexas.edu/users/EWD/transcriptions/EWD04xx/E...

Yeah, quite the opposite of "telling things as they are".

I am unable to read that presentation as satire. And as somebody that grew up using BASIC, he is much more correct than the overall "consensus" of the time. (Read "consensus" as "successful propaganda".)

It was impossible for you to learn good programming?

It took many years of unlearning. But that was on my own, I can't be sure if a teaching environment would make it faster.

I'm not sure it's satire, though it's certainly hyperbolic and exaggerated (regarding the impact of learning BASIC). It's also important to keep in mind the sort of language BASIC was at the time and Dijkstra's aversion to goto statements and unstructured programming. It had some of the elements needed for structured programming, but they could easily be bypassed. `GOSUB`, for instance, allowed for calling subroutines and stored a return address that `RETURN` would use. But `GOSUB` just jumped to a label:

  10 GOSUB 100
  100 <DO STUFF>
  ...
  200 RETURN

It was entirely possible to jump anywhere inside the subroutine, and commonly done. For instance, the first time you call it maybe you want to initialize some variables but then keep them the same later:

  10 GOSUB 100
  20 GOSUB 130
  100 <RUN ONCE CODE>
  110 <MORE RUN ONCE>
  120 <LAST RUN ONCE>
  130 <THE REST>
  200 RETURN

The logic behind doing this isn't actually expressed in the code. To an extent, he's right, there's a great deal of unlearning that has to happen after learning to program this way.

There are certain patterns of programming that people pick up based on their early experience which can be hard for them to overcome later ("The determined Real Programmer can write FORTRAN programs in any language.").

And that core statement being made, that points to the "truth that might hurt", is the difference between satire and humour. You provided one reason for why that EWD should be considered satire (because it criticizes a status quo through hyperbole and exaggeration), while stating at the beginning that you don't consider it satire. That's a form of satire by itself. Smiley.

Yes but this is also how it works in assembler, and nobody in their right mind would say knowing assembler would make you a worse programmer. We can all agree that BASIC is a crappy language by modern standard, but it did have a repl and was interpreted which gave it a quick turnaround, ideal for experimenting and learning.

Dijkstra on the other hand believed programming students should be forced to prove the correctness of their programs.

You agree that "BASIC is a crappy language by modern standard". Why is Dijkstra not allowed to make that point?

> Dijkstra on the other hand believed programming students should be forced to prove the correctness of their programs.

I did that in my CS homework like many others and I consider being able to reason about the properties of a program/algorithm a core competence of a CS graduate. Why do you think it's a problem?

> You agree that "BASIC is a crappy language by modern standard". Why is Dijkstra not allowed to make that point?

But...that is not the point he is making. He is not saying BASIC is a crappy language - he is saying programmers which have learnt BASIC are bad programmers, even if they learn other languages and techniques. That is a pretty arrogant and not particularly clever thing to say.

> Dijkstra on the other hand believed programming students should be forced to prove the correctness of their programs.

This is why Dijkstra wrote his newsletter by hand - he couldn't find a word processor that had been proven correct, nor a proven correct OS to run it on.

I'm kidding, but it illustrates the problem with Dijkstra's approach. If not just students but actual working programmers had to prove the correctness of their programs, we'd have much less software - say, 5% as much. But that other 95% does some actually useful things, even though it can still break. Would you rather have a word processor that crashes every once in a while? Or would you prefer a typewriter?

Have you ever tried to teach the right way of doing something to someone who has been doing it the wrong way for a long time and wants to argue that there's no difference?

Sure. I usually approach it by showing why I think the better way is better, rather than just calling them idiots. It have also often happened to me I have reconsidered something I had considered "the right way" for many years, due to a convincing counterargument or example.

For example I considered Dijkstra-style single-entry single-exit the right way for years, until I realized that early exits in many cases made functions a lot simpler and more straightforward.

The two are mutually exclusive? Let me have a talk with your kids about life, death, Santa and the Easter bunny. I won't be a condescending dick, I'll just be telling them how things really are.

One I’ve heard is that at a conference someone asked him “what do you do?” to which he replied “I cast false pearls before real swine.”

I don’t know if it really happened, but it’s clear that a lot of people didn’t like him.

And also clear that the feeling was mutual.

He was rather famous at UT Austin for asking if the colors on the slides of speakers' slides meant anything. I believe it was some form of hazing for prospective faculty. Also, there was some discussion of the difference between a stack data structure and a networking stack.

On the other hand, everyone who I personally know, who took one of his classes, really liked it and him.

Reply on news.ycombinator.com