Daniel Murfet comments on Estimating the Probability of Sampling a Trained Neural Network at Random