I've been patting myself on the back for making my life easier by choosing chess data as a domain to write programs against. Chess produces tons of well-structured, predictable, meaningful data. Except I know by this point in my life that there are always edge cases.
My two favorite players are the ones who have screwed it up for me: Paul Morphy played games giving piece-odds, which is captured in the PGN files, but was not handled in my code, and Bobby Fischer didn't show up to game 2 of the 1972 World Championship… but I still want to process the game so I have to figure out how to run my program on a game that never happened.
If you want to read more, subscribe to my personal newsletter.