Back to the email

I'm not sure I follow how weaker tests "localize errors" better than stronger tests. Do you mean that weaker tests imply a larger set of values than stronger tests so they're more likely to uncover errors?