I always understood the saying in the statistical sense. That is, having seen – and understood – an exception, we can be more confident that our understanding

I always understood the saying in the statistical sense. That is, having seen – and understood – an exception, we can be more confident that our understanding of the rule is valid. E.g., say the "rule" is, the database always keeps data safe. I might not feel confident about that claim, until I see a case when the database fails to keep the data safe, and understand exactly how and why the rule did not apply in that scenario. If I never see such a case, I have little confidence that we correctly measure whether the database keeps the data safe.