How Far Are Automatically Chosen Regression Smoothing Parameters From Their Optimum?: Comment