COMPUTING SCIENCE
Rumours and Errours
Brian Hayes
The confessional essay is not a popular genre in mathematics and the
sciences; few of us wish to dwell on our mistakes or call attention
to them. An inspiring exception is Donald E. Knuth of Stanford
University. During a decade's labor on the TeX typesetting system,
he kept a meticulous log of all his errors, and then he published
the list with a detailed commentary.
I have long admired Knuth's act of public bravery, and this column
is my attempt to follow his example. I took courage from the thought
that if there is any realm of life in which I might hope to surpass
Don Knuth, it's in making mistakes; but, alas, I've fallen short
even in this dubious department. Knuth's published error log runs to
more than 900 entries, whereas here I am going to confess to only a
paltry handful of mistakes. Then again, Knuth needed 10 years' work
on a major software project to accumulate his budget of errors, but
I was able to commit some really serious howlers in a program of a
dozen lines.
Knuth remarks that keeping an error log not only helped in debugging
the program but also "helped me to get to know myself." I
would like to think that I too have acquired some self-knowledge
from the experience of confronting my own fallibility. And it would
be gratifying to suggest that by telling my story I might save
others from making the same mistakes—but I don't quite believe
that, and I'm not even sure it would be a good idea.
» Post Comment