The times in recent years where I have used such things have tended to be technically successful and commercially worthless, while the things that paid off had a tendency to be churning through piles of tedious and stupid.
Also a great question. As a manager, my first question would be to ask why a library wasn't used for algorithm X.
In grad school, I wrote a complete implementation of Rijndael encryption for a crypto class, but I would never expect (or want) anyone to try to build that themselves for production software.
People tend to hire others to do the stupid and tedious stuff. Hence most jobs are stupid or tedious. I know mine is.