Lognormal and power law distributions

It turns out that Anderson has grappled with the lognormal vs power law distribution stuff before. He gives some very good references.

It is interesting that he claims that his long tail thesis isn't much affected by the difference, but it actually is. That's the real relevance of the study published. I think he also grossly over-estimates what the total size of the long tail can be in any market, but that's for another day.