Richard_Kennaway comments on The “best predictor is malicious optimiser” problem