Agree that for small collections of data - but imagine the logs on TechCrunch? Or a site that is in between those two? That data is very unique and very qualified. The interesting part is less how relevant the data is - but more how it has never been available before...