I've long been a fan of A Theory of Ice, and so I was very interested to see Ellen's take on the stats and non-(or boxcar-)stats approaches to analyzing hockey. I think she does a fair job, and brings up some good points, though I think that statisticians are quite a bit more aware of the conundrum of the desire for sample sizes versus the sample sizes available when action is needed than, say, non-stats people are of how stats might help such decision making. One group (and sometimes, necessarily) desires to continually improve their process; I'm not always so sure about the other.