Many of these notes, like my Applied Mathematician’s Apology, worry that mathematicians these days are interested in rigor and generality but not so much in solving problems. It was different in the days of Euler and Gauss.

I am also well known — since my “Definition of numerical analysis” half a lifetime ago — for my regret that Wilkinson and other pioneers of numerical analysis emphasized rounding errors so much. Their excitement about rounding errors made our subject, to many people, seem ugly.

Lately I have been startled to realize how these themes may connect. Imagine if the brilliant and charming Wilkinson, instead of talking about rounding errors, had shown his fellow mathematicians of the 1950s and 1960s that numerical analysis was a study of algorithms. Might their descendants today feel a better connection to the subject?

In a note a few years ago, I proposed that Abraham Lincoln may have had a net negative impact by keeping the Union together when it should have split in two. Here I find myself, painfully, proposing an analogous assessment of Wilkinson. I fear he may have accidentally encouraged the secession of numerical analysis from the rest of mathematics when they should have stayed together.

[13 May 2026]