Evan R. Murphy comments on On Deference and Yudkowsky’s AI Risk Estimates