Donald Knuth, Computing's Philosopher King

April 23, 2007 4:34 AM

“I wanted to try to capture the intelligence of the design, not just the outcome of the design.” “In 1977, [Donald] Knuth halted research on his books for what he expected to be a one-year hiatus. Instead, it took 10. Accompanied by [his wife] Jill, Knuth took design classes from Stanford art professor Matthew Kahn. Knuth, trying to train his programmer’s brain to think like an artist’s, wanted to create a program [TeX] that would understand why each stroke in a typeface would be pleasing to the eye.”—from a profile of Knuth in the Stanford Magazine (May '06). Salon calls him “computing’s philosopher king” (Sep '99). NPR’s Morning Edition interviews Knuth as “the founding artist of computer science” (Mar '05). Perhaps a MeFite somewhere has one of these? (Previously)
posted by Ethereal Bligh (40 comments total)

The published volumes of The Art of Computer Programming occupy pride of place on my computing desk bookshelf. And sadly, no, I haven’t worked through them.
posted by Ethereal Bligh at 4:39 AM on April 23, 2007

I'm looking forward to him finishing Volume 5 of The Art of Computer Science. I've been tempted to buy the first four volumes and give myself a thorough grounding, but something makes me want to hold off until the whole series is done.
posted by pax digita at 4:42 AM on April 23, 2007

pax digita, so somewhere around the turn of the next millennium then? I'm in the same boat.
posted by Skorgu at 4:44 AM on April 23, 2007

(Arrgggh!)

I intend to work through them and put that on my résumé as a personal qualification -- it'll be a useful counterpoint to not having finished my B.A.
posted by pax digita at 4:45 AM on April 23, 2007

I've never really learned TeX, but it has a very, very vocal contingent of people that love it.

It may also be one of the only 'finished' programs in computer science; still useful, still ported to anything and everything, but declared to be done. As far as I know, no code has changed in TeX since 1985. And yet, it's still in broad use.

It's very possibly the most perfect program ever written.
posted by Malor at 4:52 AM on April 23, 2007

One of my favourite things about TeX and METAFONT is that their version numbers don't work like most version numbers; instead, they are converging towards pi and e respectively:

I still take full responsibility for the master sources of TeX, METAFONT, and Computer Modern. Therefore I periodically take a few days off from my current projects and look at all of the accumulated bug reports. This happened most recently in 1992, 1993, 1995, 1998, and 2002; following this pattern, I intend to check on purported bugs again in the years 2007, 2013, 2020, 2028, etc. The intervals between such maintenance periods are increasing, because the systems have been converging to an error-free state. The latest and best TeX is currently version 3.141592 (and plain.tex is version 3.1415926); METAFONT is currently version 2.71828 (and plain.mf is version 2.71). All these systems are Y2K-compliant. My last will and testament for TeX and METAFONT is that their version numbers ultimately become $\pi$ and $e$, respectively. At that point they will be completely error-free by definition. —Knuth
posted by chrismear at 5:05 AM on April 23, 2007 [1 favorite]

I'm convinced LaTeX got me an extra 20% in my 2nd year Comp Sci class project (and one of the top marks). While everyone else handed in crappy looking MS Word formatted reports, our group handed in a fantastically typeset LaTex generated report. Our supervisor looked at it and said: "LaTeX. Nice.".

Never mind that the actual project itself did not run correctly at all.
posted by PenDevil at 5:16 AM on April 23, 2007 [1 favorite]

The TeX hyphenation and justification algorithm is so good that Adobe just copied it for InDesign, their flagship typesetting program.
posted by gsteff at 5:42 AM on April 23, 2007

T_EΧ's big problem is that it doesn't support anything but roman characters. Knuth has (wisely, I feel) declared that attempting to add Unicode to T_EΧ would risking breaking older datafiles, which is unacceptable to him -- his stated purpose is that T_EΧ will *always* read and render properly formatted inputs correctly, and trying to make this work with Unicode would be another ten year distraction from TAOCP.

This is also the purpose of the T_E&Chi license -- you can make any change to T_EΧ that you wish, but if you do so, you cannot call it T_EΧ, because Knuth cannot then offer assurance that you'll render a given T_EΧ input correctly.
posted by eriko at 5:43 AM on April 23, 2007

I have the first three books sitting neatly on my shelf as well. Its nice to pick up and read a random section, and then put it back down, cursing yourself for getting stupider as you grow older.

I worry he'll die before he finishes the series.
posted by chunking express at 5:49 AM on April 23, 2007

Our supervisor looked at it and said: "LaTeX. Nice."

The biggest problem I have with LaTeX is the fantastically large number of people who never progress from using everything in its absolute vanilla, default style settings - Article class, Computer Modern font, etc - which is why your supervisor was able to, at a glance, tell that the paper was typeset in LaTeX. It gives LaTeX something of a bad name for producing bland output.

The second-biggest problem I have is the poor photocopy-reproduction quality of the Computer Modern font - a major issue considering how many people never use anything else.
posted by dmd at 5:50 AM on April 23, 2007

I worry he'll die before he finishes the series.

I have 1 & 3 from years back. Whilst they've been fascinating over the years for dipping in and out of I don't think I've ever really used them for my professional needs, and since I'm pretty much out of development work now I don't expect to use them seriously. I like completeness though and would one day like to complete the set. The last two books came out in the intervening time and whilst initially I was in a rush to get them I stopped myself. I'm almost ashamed to admit this now but the reason I did that was I wasn't about to spend a lot of money on attempting to complete a series that might never actually be completed.

I'm surprised to hear he's not 70 yet though, I thought he was way older than that.
posted by vbfg at 6:10 AM on April 23, 2007

On the first day of my first class in Grad school on a Monday, the professor handed out the first assignment saying, "This is due Wednesday and has to be submitted in LaTeX." Luckily, I already knew it but 3/4 of the class had to figure out how to install, run and write in LaTeX in less than two days along with figuring out the gnarly discrete math that we were studying.
posted by octothorpe at 6:21 AM on April 23, 2007

His stance on the topic of email communication is kind of quaint.

I have been a happy man ever since January 1, 1990, when I no longer had an email address. I'd used email since about 1975, and it seems to me that 15 years of email is plenty for one lifetime.

[...]

I have a wonderful secretary who looks at the incoming mail and separates out anything that she knows I've been looking forward to seeing urgently. Everything else goes into a buffer storage area, which I empty periodically.

My secretary prints out all messages addressed to [e-mail] or [e-mail], so that I can reply with written comments when I have a chance.

posted by Rhomboid at 6:30 AM on April 23, 2007

I just picked up the first book from the library, and it's pretty neat. A different way of looking at things, as I'm used to the "Here's a theorem, here's a proof" method of doing things, and Knuth is very expository.

Also, as long as we're talking about Knuth, I spent about 6 hours this weekend while flying studying symmetric functions and the Robinson-Schensted-Knuth (RSK) algorithm, which comes down to drawing a lot of pretty pictures and playing "Jeu de Tacquin". The reasons for this are subtle, to say the least (it has important applications in the theory of projective manifolds (or so my advisor tells me)) but the bumping and sliding is still neat. Unfortunately can't find a good introductory reference for it.

Anyways, good man, that Knuth. Him and Conway are my math heroes.
posted by TypographicalError at 6:33 AM on April 23, 2007

While I think the output of TeX overall is both attractive and impressive, I personally think that the standard Computer Modern fonts used are fantastically ugly. If they are produced by some model of what makes a typeface attractive, then I think that model is a failure. It is a shame that it seems most TeX users choose to use them for everything.
posted by grouse at 6:38 AM on April 23, 2007 [1 favorite]

Heh, I'm writing my thesis in LaTeX right now. I really like that it takes care of the layout&design for you. I really hate that it takes away the pleasure of layout&design from you. After reading x number of papers made by using LaTeX, it kinda gets boring.

And suck on this caramel: I think I can produce better looking docs in Word than most people who use genericLatexTemplate4 stuff.
posted by Foci for Analysis at 6:53 AM on April 23, 2007

xkcd on Knuth.

BTW, does everyone really pronounce it Ka-NOOTH? It's his name, so he can tell everyone to pronounce it however he likes, but until I read his Wikipedia page I pronounced it with a silent K — is this common?

Also, how many people actually say T_EX with a 'ch' as in 'loch' (the voiceless velar fricative, apparently), as opposed to tek?
posted by Aloysius Bear at 6:53 AM on April 23, 2007

Christ... I'd forgotten all about LaTeX.

Back in the mid nineties my final year University project was supposed to be to build a tool that took LaTeX files and turned them into properly formatted HTML. In the days before CSS it seemed like a stupid idea and I decided to drop out of university instead.

Who knew that LaTeX would still be around and that people would actually use HTML...?

If I'm honest, I'd never heard of LaTeX before that project and the concept of trying to turn it into HTML seemed pointless. I dropped out on medical grounds before I even got around to loading LaTeX on my computer...
posted by twine42 at 7:06 AM on April 23, 2007

Actually, now I think about it, I'm sure that the project was supposed to generate HTML from a printout of a LaTeX document using OCR. In 1996. What sort of sadistic bastard sets a task like that? In 1996 even Xerox OCR software had trouble telling a X from an O, let alone putting them together as HTML...
posted by twine42 at 7:12 AM on April 23, 2007

To get some idea of the level of care and thought he put into TeX, just see his Important Message to all users of TeX. Can't imagine what he thinks of the options for δ available in HTML.

Bonus content: Questions Answered (Google video, 93 min., maybe I saw it here on MeFi?)
posted by gleuschk at 7:14 AM on April 23, 2007

The biggest problem I have with LaTeX is the fantastically large number of people spending obscene amounts of time fiddling with it to produce something other than bland vanilla output and a last ditch attempt to justify their continued grad school existence...
posted by geos at 7:18 AM on April 23, 2007 [1 favorite]

Aloysius Bear writes "BTW, does everyone really pronounce it Ka-NOOTH? It's his name, so he can tell everyone to pronounce it however he likes, but until I read his Wikipedia page I pronounced it with a silent K — is this common?"

Yes, Germanic names pronounce the "K": Knuth, Knopf, etc.
posted by orthogonality at 7:43 AM on April 23, 2007

geos Speaking as a guy who writes to communicate thoughts, not to show off my nifty graphic design skills (which I don't have), I see nothing wrong with vanilla output. In fact, I prefer vanilla output because it means that people aren't being distracted by my text formatting, and thus are able to concentrate on the ideas being presented.

I can get nice looking, easy to read, non-jarring text by using the standard LaTeX tools, why should I bother jerking around with it?
posted by sotonohito at 8:01 AM on April 23, 2007

“There was something special about the IBM 650, something that has provided the inspiration for much of my life’s work. Somehow this machine is powerful in spite of its severe limitations. Somehow it is friendly in spite of its primitive man-machine interface.”
--Donald Knuth

My problem with the 650 is these days it's hard to find replacement vacuum tubes.
posted by Fuzzy Monster at 8:12 AM on April 23, 2007 [1 favorite]

Knuth is also an accomplished organist. This is listed in Volume 3, under Royalties, Use of.
posted by zamboni at 8:33 AM on April 23, 2007

Aloysius Bear, thanks so much for introducing me to xkcd!
posted by Ethereal Bligh at 9:21 AM on April 23, 2007

Speaking of typesetting....

The smart quotes in the FPP show up weird in some feedreaders (like, strangely, the HTML-based Google Reader), producing ugly squares.

Just a note for future reference.
posted by JHarris at 9:25 AM on April 23, 2007

Yeah, that's a mystery bug that Matt doesn't know how to fix. And they aren’t smart quotes, they're just real double-quotes. I hate the “smart quotes” appellative. I type these characters directly.
posted by Ethereal Bligh at 9:43 AM on April 23, 2007

I have a simple, elegant book by Knuth called Surreal Numbers in which he takes apart and reassembles our counting system by slightly altering the axioms. It was a mind-blowing read and has influenced how I think about mathematics on a daily basis.
posted by Mental Wimp at 10:23 AM on April 23, 2007

The reason the original Computer Modern looks so bad these days is that METAFONT is bitmap only. The whole original TeX » DVI workflow is terrible poop in light of the last few decades of progress with Postscript and PDF.

If you use PDFlatex and/or TeTeX, you'll end up with nice output (plus they support Unicode fully).
posted by blasdelf at 11:11 AM on April 23, 2007

The reason the original Computer Modern looks so bad these days is that METAFONT is bitmap only.

That's not what I'm talking about. I don't like the letterforms, even in print.
posted by grouse at 11:55 AM on April 23, 2007

sotonohito wrote "I prefer vanilla output because it means that people aren't being distracted by my text formatting"

As a recipient of dozens of Outlook-generated emails a day at work, I wholeheartedly support the push to make the message about the words inside rather than the formatting.

On the other hand, if I use LaTeX that means I will have difficulty collaborating with people who don't. As in, sending a file to my former adviser when working on a joint manuscript. If it wasn't produced in MS Word she could neither open it nor edit it. Any typesetting markup in the content would confuse the shit out of her. Plus, reference managers - any of these work easily in TeX?

Don't get me wrong, I have a good solid distaste for the default style rules in MS Office and in the strange way it defines these rules, plus all the bugs inherent (such as the inevitable extra page that results if a table ends too close to the bottom margin - why is a paragraph mark necessary after each table?) but my job is to communicate, and unless I was about to submit my document for review I don't think that LaTex is going to work for me.

However: If there exists such a thing as a "save Word doc as TeX file" that actually works well, it could be helpful. When submitting the damn thing the last thing I do before hitting "send" is strip as much of the formatting as possible anyway. Stupid Word doc formatting.
posted by caution live frogs at 12:59 PM on April 23, 2007

Here's a Word-to-LaTeX conversion program.
posted by aeschenkarnos at 2:58 PM on April 23, 2007

caution live frogs: In addition to the word-to-LaTeX program, there's also this to consider: any text editing program that allows sharing and commenting will work with LaTeX just fine.

As long as its the content being discussed, not the formatting, there's not really any reason why a person would have to understand LaTeX to comment on a LaTeX document.

Here's an example of what LaTeX looks like:

\section {Displayed Text}
Use the ``quote' and ``quotation' environments for typesetting quoted
material or any other text that should be slightly indented and set off
from the normal text.
\begin{quotation}
The quote and quotation environments are similar, but use different
settings for paragraph indentation and spacing.

\em When in doubt, consult the manual.
\end{quotation}


Mostly it just looks like text, and the markups are semi-intuative, at least in that a person can often guess correctly what they are supposed to mean when they see them for the first time.

I don't know enough about your needs, or the SotA in text editors with multiple commenter options to say that you should try LaTeX. As a person doing non-colaborative work, I think its a fantastic tool.
posted by sotonohito at 3:10 PM on April 23, 2007

how many people actually say TEX with a 'ch' as in 'loch'

"When you say it correctly to your computer, the terminal may become slightly moist."
posted by zippy at 4:00 PM on April 23, 2007

A friend of mine had a couple of extra volumes of The Art of Computer Programming that he wanted me to sell for him. I told him to go peddle his own Knuth.
posted by SPrintF at 6:44 PM on April 23, 2007 [2 favorites]

I told him to go peddle his own Knuth.

Your pun puts the eX in TeX.
posted by zippy at 8:33 PM on April 23, 2007

…reference managers - any of those work easily in TeX?

LOL! Does the papal bear shit in the catholic woods? Dude, reference management using BibTeX is pretty much the primary reason TeX is so widely used in academic publishing.
posted by blasdelf at 2:29 AM on April 24, 2007

I have one of his checks. I found some small piece of missing punctuation in Volume III, walked right up to him and demanded payment.

Years later, a letter and a check showed up in the mail. I believe the going rate at the time, based on the error rate, was $1.82 or something like that. I have no idea how many checks he wrote, but I bet not many got cached!

I have no doubt he derived a general expression for determining the proper reimbursement amount that used some new notation that required new development in TeX to typeset. I am sure he was happy.

What a great guy. What a deadpan lecturer. What a devious mind. 22 years, and I still haven't recovered from his exams.
posted by skippyhacker at 6:05 AM on April 24, 2007 [1 favorite]