I'll file in my "department of huh?" folder--I give up. Thanks for pointing out fair-use rules and commercial speech. Back to the point ... any thoughts on power laws and file system size distributions? Are they lognormal, power law in tails, double Pareto, etc. Is the generative model based on some kind of herding behavior --preferential attachment (Yule-Simon,etc)-- or something else? Feel free to take a histogram of your own file system and tell us what comes up.