Looking at Smale & Zhou, 2003, "Estimating the Approximation Error in Learning Theory" and wondering how I did end up there.