Anja comments on A utility-maximizing varient of AIXI