Interview with Donald Knuth | Hacker News Reader

Interview with Donald Knuth

simple-talk.com

102 pointsmqt33 comments

Knuth always reminds me of my own passion that gets obscured by day-to-day work and life. Loving machines and the beauty within and not for any economic value.

Still waiting for his volume on lexing and parsing techniques. I always love his writing and how complete he expresses any particular problem.

It is my dearest hope that he lives long enough to finish AoCP, but I fear the odds are against it.

Nonetheless, his work thus far has been a remarkable achievement.

He thinks that it's only 5 years away... 2015 est. completion date.

It felt like it cut off at the end, am I missing something? Quite enjoyable read, it'd be fun to see Knuth play the organ, he should sell tickets!

"As Sportin' Life says in Gershwin’s Porgy and Bess: ‘Methusaleh lived nine hundred years. But who calls dat livin’ when no gal will give in, to no one who’s nine hundred years?’ I may have mis-remembered that lyric, but you get my drift."

I think he pretty much had to wrap things up on that quote, as nothing else in the interview could possibly top it. At least it made me LOL.

This seems like a good time to point out that his annual "Christmas Tree" lecture is a week from today.

http://www-cs-faculty.stanford.edu/~knuth/musings.html

<snip> Interviewer: Would you still study computers whether or not it had any commercial value?

Knuth: Thank you for asking that question. I have always been attracted to computer science because it involves beautiful patterns, rather like the way dancers enjoy choreography, and because questions such as "What can be computed efficiently?" are profoundly interesting and challenging. </snip>

The words and perspective of a master. His description of his work habits (no email, uninterrupted focus) is congruent with one who embodies the Tao of Programming. :)

There is much to admire about Knuth's contributions to CS, but advancing the field in an open-ended creative way is not one of them.

The startup community has done more in that respect than academia, but why should academics not contribute to inventing novel applications?

We need to acknowledge such accomplishments more, even if they appear less mathematical/scientific.

As I see it, Knuth is to computer science what Tolkien was to literature. Neither man is what you'd call "hip and creative", but both produced a life time's worth of high quality, painstakingly careful work. (In fact, it'd be interesting to compare TAOCP and LOTR in more depth.)

If you try to be creative, you'll end up with nothing but emperor-has-no-clothes modern art (which won't get you very far in computing). But if you do quality work, you may well get creativity thrown in as a bonus.

People try to be creative all the time in startups and games. And it can be quite profitable too.

As for computing in academia, it's not set in stone. CS can be expanded or a new field created to reward creativity.

> emperor-has-no-clothes modern art

I admit I have a hard time "getting" most modern art. But seeing the actual pieces in person, especially some Picasso and Pollock, is quite a different experience than seeing dinky little prints in an art book.

Jackson Pollock's paintings really stand out in a modern art gallery as being way too pretty to be there.

TAOCP is a very thorough "text book" for people who really are interested in the subject matter. It's not really Knuth's goal to break new ground with it, but to compile the most interesting things he's found into a single work.

It's hardly the only thing he's done though. Even without TAOCP, he's one of the more prolific and accomplished computer scientists in history: http://www-cs-faculty.stanford.edu/~knuth/vita.html

Bah. Knuth wrote several interesting applications, including one that changed the face of mathematics more in the last 20 years than any other development. Where would math be without TeX? But beyond that, there's more to CS than "novel applications," and Knuth has been one of the most important theoretical computer scientists in the world, and was one of the fathers of the field, inventing asymptotic analysis of algorithms, inventing the parsing algorithms used by nearly every compiler on the planet, and on and on.

Is TeX really a novel app in the sense that twitter say is?

In fact, TeX isn't even the first of its kind.

Which computer type setting system do you know that preceded TeX and produces beautifully typeset text and mathematics?

http://en.wikipedia.org/wiki/Troff

But - is it as beautiful?

Twitter is a greater conceptual leap over blogging than TeX is over troff.

Perhaps. Still, TeX feels like an accomplishment. (Though as a programming language it's close to horrible.)

Not really. Twitter is microblogging. TeX greatly expanded the quality and re-usability of very difficult typeset documents.

troff did not produce beautiful typeset mathematics, nor did it even pretend to deal with fonts.

Strictly speaking it is possible to use fonts other than R, B, I, BI using ".fn" but because the set of fonts is determined by the output device it's very limited.

Much to my surprise GNU groff's font format does actually contain kerning information and ligatures although it's quite primitive and none of the included fonts have characters much outside of Latin-1 (to the point that there's a specialized "EURO" font you have to use if you want a Euro symbol). I would have to go over the older documents I have to be sure but I don't think this was a feature of the original troff.

TeX was the first serious examination of what good typographical layout was and how to codify it. Previously there were many rules of thumb, and lots of examples, and a handful of programs for typesetting (like RUNOFF, which was and is pretty damned ugly, although it beats the pants off a typewriter) but in TeX Knuth created an algorithm for things like mathematical layout, for paragraph layout. I am unaware of anything that has surpassed TeX in this field—programs like Lout use the same algorithm TeX does, just with a less idiosyncratic surface syntax and with a rewritten core.

TeXmacs has a typesetter with real-time WYSIWYG editing that is comparable in output to TeX. So TeX has been surpassed in that sense.

I think comparing TeX to twitter is a serious insult to TeX.

Did you try reading the article? This bit from Knuth is a fun contrast to your comment:

> Alas, people these days rarely measure a computer scientist by standards of beauty and interest; they measure us by dollars or by applications rather than by contributions to knowledge, even though contributions to knowledge are the necessary ingredient to make previously unthinkable applications possible.

Try getting a grant proposal or publishing a paper on a novel application without a proper scientific evaluation.

Imagine for example how you would evaluate twitter in a scientific way.

Twitter has very little to do with science. Donald Knuth produces science.

What's with the twitter fixation? I've benefitted from Knuth's work for over 20 years (as a practitioner, not an academic). Twitter? To me it's a curiosity at best; I've never used it.

Seconded. Just the idea of mentioning twitter in this context inflicts almost physical pain on me.

Knuth is regarded by many as an Einstein-like figure in Computer Science. How that remotely relates to a web-company is beyond me.

Let’s take Twitter as an example then.

The only reason that Twitter is possible, at any kind of scale, is because decades of research have been put into the algorithms that route messages inside big computer systems. Do you remember that first year or two of Twitter when it was the laughing stock of the internet for its constant interruptions and crashes? Well, eventually they had enough money to hire someone who was familiar with the latest CS research, and could actually make the thing keep up with their user base. That's not considering all of the further decades of effort put into all of the implemented infrastructure on which modern computers and networks run, into compilers, kernels, network stacks, security, etc. etc. And of course the computer software prerequisites to Twitter are just the tip of the iceberg, the whole thing built on that marvelous edifice of millennia of scientific and mathematical investigations.

I'm told Twitter is neat (I really have no interest in using it myself).... but it’s both utterly dependent on Science, and also not that scientifically interesting, as an idea. (Topics related to Twitter might be interesting; for instance, the social networks could be interesting for sociologists, the use by dissidents might be interesting for political scientists, the details of the network stack might be interesting to computer scientists, etc.)

Reply on news.ycombinator.com