Buck comments on Aligning a toy model of optimization