John G. Halstead comments on On Deference and Yudkowsky’s AI Risk Estimates