The key was separating Quality reviews from Impact reviews, and finding how differently they're evaluated, with big implications.