Thanks for the comment. I mostly agree, but I would add the following: the epistemological question is certainly relevant, but it also sort of smacks of a research'y kind of approach. In others words, what questions can I ask based on the data? I'd prefer to see scientists thinking about what problems they can solve that are presented as a result of business strategy, operations or external phenomena. Outside in, not inside out. This was the failing of data mining, sort of poking around in the data and using cool algorithms to see if anything popped out. That didn't work last time and it won't work now.