I agree with the fact that the BigData techniques allow to obtain a lot of information. However, and from ignorance, I find them not very elegant. They are - almost - "brute force" methods. It is like trying to calculate the pressure of a gas on the walls of the container that contains it, characterizing the behavior of each of the gas particles. Continuing with the physical analogy ... will there not be - for the BigData - more elegant macroscopic approaches?