Vviewpoints
DOI: 10.1145/1378704.1378715
Interview
Donald Knuth:
A Life’s Work Interrupted

In this second of a two-part interview by Edward Feigenbaum, we find Knuth, having completed three volumes of The Art of Computer Programming, drawn to creating a system to produce books digitally.

 

Don switches gears and for a while and becomes what ed feigenbaum calls “the World’s Greatest Programmer.” There was a revolutionary new way to write programs that came along in the 1970s called “structured programming.” At Stanford we were teaching students how to write programs, but we had never really written more than textbook code ourselves in this style. Here we are, full professors, telling people how to do it, but having never done it ourselves except in really sterile cases with no real-world constraints. I was itching to do it. Thank you for calling me the world’s greatest programmer— I was always calling myself that in my head. I love programming, and so I loved to think that I was doing it as well as anybody. But the fact is the new way of programming was something that I hadn’t had time to invest much effort in.

with a new book on artificial intelligence, and the proofs of it were being done at III [Information International, Incorporated] in Southern California. They had a new way of typesetting using lasers. All digital, all dots of ink. Instead of photographic images and lenses, they were using algorithms, bits. I looked at Winston’s galley proofs. I knew it was just bits, but they looked gorgeous.

I canceled my plan for a sabbatical in Chile. I wrote saying “I’m sorry; instead of working on Volume 4 during my sabbatical, I’m going to work on typography. I’ve got to solve this problem of getting typesetting right. It’s only zeros and ones. I can get those dots on the page, and I’ve got to write this program.” That’s when I became an engineer. I did sincerely believe that it was only going to take me a year to do it.

PHO TOGRAPH B Y TIMO TH Y ARCHIBALD

the motivation is his love affair with books… That goes very deep. My parents disobeyed the conventional wisdom by teaching me to read before I entered kindergarten. I have a kind of strange love affair with books going way back. I also had this thing about the appearance of books. I wanted my books to have an appearance that other readers would treasure, not just appreciate because there were some words in there.

 

For Part I of this interview, see Communications, July 2008, page 35.

…and what had happened
to his books.

Printing was done with hot lead in the 1960s, but they switched over to using film in the 1970s. My whole book had been completely re-typeset with a different technology. The new fonts looked terrible! The subscripts were in a different style from the large letters, for example, and the spacing was very bad. You can look at books printed in the early 1970s and almost everything looked atrocious in those days. I couldn’t stand to see my books so ugly. I spent all this time working on them, and you can’t be proud of something that looks hopeless. I was tearing out my hair.

At the very same time, in February 1977, Pat Winston had just come out

But, in fact, it was to be a 10-year
project. the prototype user was
Phyllis Winkler, Don’s secretary.

Phyllis had been typing all of my technical papers. I have never seen her equal anywhere, and I’ve met a lot of really good technical typists. My thought was definitely that this would be something that I would make so that Phyllis would be able to take my handwritten manuscripts and go from there.

The design took place in two all-nighters. I made a draft. I sat up at the AI lab one evening and into the early morning hours, composing what I thought would be the specifications