In that case I think Knuth referenced proving it by hand, e.g. with Hoare triples.

You could argue that hand proof is useless, but by forcing yourself to go through it in tiny steps you can actually sort out a lot of oversights.