A platform's content moderation policies are generally open to public scrutiny, whereas the specific operational logic of recommendation algorithms is highly opaque.