The Muddle Machine: Confessions of a Textbook Editor (2004)
web.archive.org
web.archive.org
Oftentimes when this has happened for me, it's some knowledge that I'd absorbed in a Wikipedia rabbit hole (or some such) and thought was trivia that would be useless to me. Then suddenly I noticed some sort of congruence with a problem I was working on, and that knowledge proved not so useless after all.
This sort of cross-domain application has lead me to an intuitive and elegant solution on more than one occasion. Biology and linguistics have been especially fruitful in this regard, but perhaps that's only because those are fields I'm better acquainted with.