> Probably good advice, but when was the last time a programmer accurately scoped a problem when they said it will take “an afternoon” to build?

This is why you time box things. Spend XX hours trying to get a thing working and if you aren't close, you grab a library and move on.