The lottery of bibliographical databases: a reply to Edwards & Thorne