Esdger Wybe Dijkstra: 1930-2002.
August 7, 2002 3:26 PM   Subscribe

Esdger Wybe Dijkstra: 1930-2002. One of the last remaining computer science gods has passed on after a lengthy battle with cancer. Few had his eloquence, his wit, or his charm. A sharp writer (check out "GOTO statement considered harmful" or The Humble Programmer [pdf]) and a true pioneer in all areas of the field, Dijkstra had a hand in much of modern computer science. A sad day for nerds worldwide.
posted by Succa (21 comments total)
 
I wrote one of my first CS papers on the shortest path algorithm. I sprained my ankle the day before it was due (coming back from a late night/early morning in the CS lab), so I used that as part of my presentation (finding the shortest way between buildings on campus when you're on crutches). Rest well Mr. Dijkstra.
posted by girlhacker at 3:36 PM on August 7, 2002


Yes, very sad to hear... I hardly know what to say. Dijkstra was a really brilliant (and wonderfully idiosyncratic) guy.

When I was an undergrad, he was the guest lecturer for a day in one of my CS classes. There is nothing more terrifying than having THE Dijkstra ask you a question out of the blue.

(Then at the end of the class, some 1st-year students came in, thought he was the regular prof, and had him sign their cards to add a class. I'm not sure if anyone ever clued them in or not. "You remember Dijkstra's Algorithm from last term? That was him.")
posted by xil at 3:55 PM on August 7, 2002


dijkstra will be missed by this cs graduate.
posted by moz at 3:57 PM on August 7, 2002


He gave a talk at my school a few years back, and some friends and I checked it out. He was an unbelievable speaker, with his silky voice and unpenetrable calmness. There was a guy two rows from the front trying loudly to refute Dijkstra's arguments, like an obnoxious lawyer shouting "objection!" after every sentence. By the end of the talk, the crowd was in stitches watching Dijkstra humourously dismantle the guy. He was something else, I tell you. Most of his talk was probably miles above everyone in the room's head, but just watching him work his magic up there was worth it.

Also, what's not to love about this Dijkstra quote: "Object-oriented programming is an exceptionally bad idea that could only have originated in California." You just don't get that kind of cynicism from the Stallman's and the Linus Torvaldses.
posted by Succa at 4:13 PM on August 7, 2002


He will be very much missed. I've worked on a lot of legacy systems built in the 1960's - onwards and understand where he was coming from when he called use of the Go To statement harmful.
posted by rjs at 4:32 PM on August 7, 2002


Thanks, Succa. I was surprised to reread "GOTO statement considered harmful" and find that it seems so reasonable and innocuous. It's hard to remember today how controversial it was at the time. The holy wars lasted well into the 1980s. In my memory they were far uglier than any of the flame wars of today, but Dijkstra was always a voice of reason. He never lost his temper and he never gave an inch.

For me, his greatest idea was that complexity was the enemy. If it isn't simple enough to be understood easily, it's probably wrong. The vindication of Dijkstra's work is how obvious most of it seems today.
posted by fuzz at 5:02 PM on August 7, 2002


bummer. i wrote an interpreter for a programming language he designed. the book in which the language was used had the note None of the programs in this monograph, needless to say, has been tested on a machine.

(of course, all the ones i tested worked)
posted by andrew cooke at 5:11 PM on August 7, 2002


This may be just my inner bitter unemployed engineer speaking, but I wonder if the brilliance of pioneering heros such as Dijkstra and Knuth is a thing of the past. Why do you need a clever (and beautiful) algorithm for computing shortest path--just crank up your 2GHz CPU running Bloatware/OS and use the brute force O(n**2) method to do it. If you can't figure out the O(n**2) algorithm, maybe there is a wizard° that will do it for you.

I mean, goodness gracious, the general public believes Bill Gates is a computer genius. How wack is that?

°By wizard, I mean a utility with happy menus to write your code, not a dude wearing sandals and drinking Jolt cola.
posted by chipr at 6:00 PM on August 7, 2002


chipr, I hope you were being facetious! If not, as a CS grad student I have to stick my geeky neck out here.

This quadratic assignment problem is just a simple example of why we need brilliant people like Dijkstra. The simplest algorithm would've taken 140 times the age of the universe to run on a desktop computer, whereas the algorithm the researchers discovered would only have taken 7 years. Running on a grid system took only a matter of hours.

I don't know about you, but I needs my quadratic assignments, and I needs them now!
posted by krunk at 8:13 PM on August 7, 2002


No, not facetious. More like bitter and cynical. I'm just very depressed about (what I perceive to be) the conversion of computer programming from a profession to a trade. There seems to be a great need for programmers that know J2EE and SOAP, and not a lot of need for ones that know Dijkstra's Algorithm or understand structured programming.

I hope my rant isn't being seen as disrespectful. I so admire his contributions to how we think about good programing. I'm saddened by the loss.
posted by chipr at 11:48 PM on August 7, 2002


I've worked in IT for the past 20 years, starting out as an IBM mainframe programmer in the early 80's. A couple of thoughts:

I feel that, with things like quality assurance and structured testing becoming common practice, the profession has become more, um, professional;

the sloppy quality of a lot of modern applications that I see disturbs me. It almost seems like, with modern system development environments making it relatively easy to produce a working application, very little care is given to aspects like usability;

maybe the modern geniuses of programming are now working in areas related to maintainability (like the development of standards and analysis and design methods) rather than efficiency.
posted by rjs at 12:09 AM on August 8, 2002


Dijkstra was never particularly concerned with "real-world issues". He was intent on keeping computer science intact as a branch of mathematics, rather than a set of heuristics for Getting Things Done. Much of his recent work was on program correctness and the formal systems thereof. Apparently he was pretty famous for wanting his students to stay away from the computer, not giving in to it, and keeping CS education inside the theoretical realm. Remember, he's the guy behind the quote: "Computer science is as much about computers as astronomy is about telescopes."

(Some other nerdy quotes here.)
posted by Succa at 12:20 AM on August 8, 2002


Succa, that's a great quote.

Man, I didn't think that a genius like Dijkstra lived in our time. My first CS program in college was a Java applet based on Dijkstra's shortest path algorithm (like many of our fellow MeFi geeks), and it amazed me when I first figured it out. Never figured that someone like him lived in the 20th century--always thought of him as some 19th century mathematician.

Here's the Dijkstra digital archive that Succa kinda linked to in the first post.
posted by hobbes at 2:26 AM on August 8, 2002


chipr:
I dig where you're coming from; 2 or 3 years ago it seemed that everyone and their monkey could get a programming job. It really scares me when I think of a guy with a DeVry diploma and a C++ book working on a potentially mission-critical project. Kind of like giving giving someone a year of astronomy and a copy of "Astrophysics for Dummies", then asking them to work on the space shuttle.

I wonder if the current hiring slowdown/freeze will result in companies being more selective in their hiring (obviously) and potentially better quality software? On second thought, that sounds unlikely.
posted by krunk at 6:44 AM on August 8, 2002


There seems to be a great need for programmers that know J2EE and SOAP, and not a lot of need for ones that know Dijkstra's Algorithm or understand structured programming.

You gotta be able to walk before you can run!
posted by themikeb at 8:05 AM on August 8, 2002


There seems to be a great need for programmers that know J2EE and SOAP, and not a lot of need for ones that know Dijkstra's Algorithm or understand structured programming.

in my experience, you need both theoretical and practical knowledge. having buzz-words on your cv might get you the initial job, but respect and promotion/keeping the job only come with being able to deliver the goods. and to deliver you need to understand what you are doing.

crap end products are (again, imho) largely the result of market pressure - people are prepared to spend good money on the latest software rather than wait for something with more work behind it (you could rephrase that as unsatisfied customers having little in the way of leverage to force change once they've bought the product - but they do keep buying...)
posted by andrew cooke at 8:41 AM on August 8, 2002


Succa - I don't disagree, but I stand by my position.

First, I never claimed to be a scientist. I'm an engineer by training. I pervert the work of theoreticians for practical use. I believe there is a discipline there that ought to be valued in the software development process, but I fear it is becoming less so.


Second, I think Dijkstra had a remarkable impact on the mundane, practical side of programming. The publication of "Goto considered harmful" was a watershed event.

Andrew - I think there has to be more to it than just market pressure. Consider the stuff on Freshmeat. Most of that stuff is developed in a pressureless environment, yet a remarkable amount of it is crap.
posted by chipr at 10:53 AM on August 8, 2002


There seems to be a great need for programmers that know J2EE and SOAP, and not a lot of need for ones that know Dijkstra's Algorithm or understand structured programming.

Well, I'd say most business apps out there need SOAP more than they need Dijkstra's Algorithm. Business apps are usually pretty simple from a CS perspective. Games on the other hand...
posted by Triplanetary at 11:22 AM on August 8, 2002


Freshmeat...

ok, but who reads self-published novels or "scientific" writings in non-peer-reviewd journals? there's a big difference between commerical code and what people write to try out/explore ideas/languages/systems and then share on freshmeat.
posted by andrew cooke at 11:24 AM on August 8, 2002




Well, shucks. I managed to screw up a post defending my technical abilities :) Even with preview. My apologies, and a second try:

Dijkstra was an amazing man, and he will be missed. For those who haven't seen it, Joey deVilla's got a nice tribute at his site.

chipr: I wish one of my tools would conjure up a caffeinated dude in sandals to solve my problems :) But hey, some of us know both Dijkstra's algorithm and J2EE. And there's nothing wrong with different people working at different levels of abstraction and detail. As long as they're working at the appropriate levels.

Succa: I'll have to look into his opinions on OO - I'd be interested to see what he had to say.
posted by Songdog at 7:43 AM on August 9, 2002


« Older "Pulp"   |   But it was broken when I got here! Newer »


This thread has been archived and is closed to new comments