Another area for investigation that I would love to see some solid stats on is the distribution of hits. Bentley and Maschner have shown that the Billboard chart exhibits a Zipf distribution and is thus in a state of self-organised criticality, where avalanches (sales collapses) of arbitrary size can occur.

My suspicion is that the distribution of album sales, whether cumulative or over an interval, would also show a Zipf distribution. The 'hit-chasing' model of the record industry makes a lot of sense if this is the case, as the records that hit it big will be hugely profitable. The ones that do less well - in the long tail of the
distribution - are the ones that are relatively subsidised. However, the tail is cut off at the point where labels won't pick up a band.

I suspect that the widespread availability of music from unsigned bands via legitimate digital distribution can affect the overall form of the power law, by extending the tail still further, perhaps changing the power of the distribution slope, and reducing the number ot magnitude of hits that way.