> then sampling to see if it excludes something important.

But how? That’s the question.