John_Maxwell comments on Using rationality to debug Machine Learning