Carl Feynman comments on Erdős Problems in Algorithmic Probability