Yeah, so listening to the latest TAL podcast isn't a good idea. This week's theme is, "Stories of people betting on things with very bad odds." Ira Glass need only visit the results page to find some low probability events.
I did discover a possible research area...is it possible to determine causation with data mining techniques? Normally we would need to conduct a controlled experiment to conclude a causal effect, but we ought to be able to say something more than simple correlation with a sufficiently robust database from a comprehensive study.