Nice video about Stein's paradox.

This video gives an intuitive explanation why the James-Stein estimator dominates the sample average when the dimension is sufficiently large (greater than 2). Click the image to watch! Special thanks to Alexander, who shared this video with me.