A weaker logical-weakness standard might resolve the 'too wide' problem by no longer condemning good...