RogerDearnaley comments on On the functional self of LLMs