Distributed computing frameworks like Hadoop and Spark have enabled processing of "big data" sets -- but that's not enough for modeling surprise/rare "black swan" or complex events. Just think of scenarios in disaster planning (earthquakes, terrorist...