Charlie Steiner comments on K-complexity is silly; use cross-entropy instead