Learning should alter prediction

If an idea does not change what we expect to happen, it may not have changed what we believe.

This is a useful test for learning. After reading the book, what would I now predict differently? Which decision would change? What evidence would surprise me?

Without a shift in expectation, knowledge can remain ornamental. It gives us better language while leaving the underlying model untouched.

Real learning should create some risk of being wrong in a new way.