A Breakfast with Knuth
Scott Kurowski, 21 July 1999

On Tuesday 13 July 1999, I mailed Prof. Donald Knuth a letter with news of the June 1st discovery of the 38th Mersenne prime on Entropia's virtual supercomputing network. The following Thursday evening Knuth called, chatted for 15 minutes and invited me to breakfast on the following Wednesday.  I can bring a friend, so I call the distinguished scientist and mathematician, Ernst Mayer, recently moved to Cupertino (another distinguished scientist and mathematician friend from Sunnyvale, Luke Welsh, is unfortunately out of town on business but otherwise would also have joined us). Ernst and Luke are both Mersenne number researchers.

What do you say to the world's most famous computer scientist, to many (myself included) a living legend? I didn't use a list, which would be too interview-like. From our previous contact about the 37th Mersenne prime he already knew I was more of a opportunistic technologist than scientist, so I decided to let it ride as a purely social occasion.

I picked up Ernst and drove to Knuth's home in Stanford, where we met his wife and decided upon a breakfast at Hobees, a local cozy restaurant chain.  Knuth said he hadn't eaten at a restaurant since 1988 - I'm not sure I heard that right, but given his work focus it wasn't out of the question. He's wearing a shirt with Fermat's Last Theorem on the front, and he is 6'4", taller than I somehow expected.

Prof. Donald Knuth at the wheel of Scott's Maxima SE

"It's a lot nicer than my Volvo!"
sm_k19.gif (13018 bytes)

Speed blur.
sm_k20.gif (15560 bytes)

I started the car and got out to prepare a passenger seat for the Prof.
when he promptly jumped into the driver's seat and put the car in gear!
I shrugged at Ernst and jumped into the passenger seat. We discussed his
current work and office at Stanford, how he set up his retirement work,
and how he declined Stanford's offer to keep him salaried as he completes
The Art of Computer Programming (TAOCP).


The Banana Pancakes Algorithm

The banana pancakes problem is solved
using a technique similar to 'divide and conquer'
sm_k21.gif (15800 bytes)

Knuth, Ernst and Scott.
sm_k23.gif (16938 bytes)

We discussed Entropia, Ernst's most recent research, Knuth's current work on volume 4 of
TAOCP and some of the content of volume 5, and which parts of Vols 1-3 we have used. We also
discussed machine speeds, computation problem spaces and Knuth's study for this week, the
pentomino tiling problem - which he was pleased to learn I solved in 1986 during a combinitorics study.
Knuth commented on making mathematical discoveries, and that he had recently written for his
book that of such computing work "[he] would rather find a Mersenne prime."


Back at Knuth's Home

Knuth at his pipe organ,
custom-built into his house.
sm_k28.gif (15352 bytes)

Knuth decides which TAOCP
reference to use from Vol. 2.
sm_k30.gif (15442 bytes)

Knuth signs the 37th Mersenne prime and
3 copies of the 38th Mersenne prime with
a TAOCP programming exercise citation.
sm_k32.gif (13961 bytes)

We returned to Knuth's two-level, semi-ranch style home for him to sign several posters of the prime
number discoveries
. Knuth specifies the frame color for the 38th Mersenne prime poster that
he now hangs in his Stanford office. As he searched through Vol. 2 of TAOCP he explains his color
coded highlighter pen system for "bugs", additions and deletions, and changes to the text. The
volume he shows us has something marked on almost every page. We discussed the complexity of
the natural world, genetic algorithms and evolution programs, and Knuth comments that he is more
open minded to them now that there's ample evidence for their strength - including finding better
algorithmic solutions to his problems than his own.