David Mathers comments on On Deference and Yudkowsky’s AI Risk Estimates