Big data is one of the current buzzwords, but it does earn its importance, as IoT and Industrial Internet will generate huge volumes of new data. Big data can be an extremely valuable source of new insights, thus creating the basis for entirely new services, products and processes. But it can also produce a deluge of pointless or downright faulty information, due to several reasons: biased or just incorrect or incomplete data, autocorrelation, reinforcing cycles which generate truths from non-existent phenomena etc.
I enjoyed reading Mayer-Schönberg's and Cukier's book Big Data, although I rather doubt the idea of correlations reigning supreme over causality. Actually I think that the authors did not say that causality no longer matters - just that correlations can point us to where to look for new causalities.
Financial Times had an excellent article about the issues of Big Data. It is a sober reminder that Big Data should not become an obsession - and also of the importance of defining what we really mean when we use a term. Big Data, IoT and Industrial Internet seem to mean very different things to different people. It also points out that data quality still has a major significance when we try to find and understand new phenomena. Data can speak volumes, but some of the talk can been nonsensical babbling, showing patterns which turn out to be mirages.
Still, using correlations to find new places to look into and generate new ideas is a good way of thinking out of the box - as long as we remember that that is what it's about, not about turning correlations into causalities and, further, new truths.