Ask HN: How do you spot a classic book before it becomes popular?

In the world of software engineering and programming, there are some books that are widely agreed as classics. E.g.:

- "Patterns of Enterprise Application Architecture", Martin Fowler

- "Refactoring", Martin Fowler

- "The Practice of Programming", Brian Kernighan and Rob Pike

- "Code Complete", Steve McConnell

- "The Pragmatic Programmer", Dave Thomas and Andrew Hunt

- "The Mythical Man-Month", Frederick P. Brooks Jr.

Outside of software engineering, there are books like:

- "Getting to Yes", Roger Fisher and Tahl Raz

- "Never Eat Alone", Keith Ferrazzi, William L. Ury, and Bruce Patton

I'm sure some of you have bought some of these books before they ever became popular and you at least had a gut feeling that the investment was worth it very early on.

My question is: can you tell, maybe in retrospect, how you spotted such classics?