In the late 1990s I implemented an algorithm from one of Knuth's books. (An algorithm to generate prime numbers.) I found a mistake!
I was so excited.
I then checked the errata. It was known.
It took another 20 years before I found an actual mistake, regarding the early history of superimposed codes. A very specialized topic where I have one of the few copies of the patent challenge distinguishing between random superimposed codes and arbitrarily selected superimposed codes.
But the number of parameters is finite, and all parameters have a finite number of possible values.
His errata changed the wording thus:
Zillions of alphabets can be generated by the programs in this book. (https://ftp.rrze.uni-erlangen.de/ctan/systems/knuth/dist/err...)