Stephen Bennett comments on Use Normal Predictions