While facing big data, we developed a visualization strategy based on head/tail breaks in order to see a clear pattern from a hairball. The strategy is to recursively drop out the tail parts until the head parts are clear or visible enough. The head parts are self-similar to the whole, thus the parts enable us to see the whole: http://en.wikipedia.org/wiki/Head/tail_Breaks
Waldo Tobler had the similar strategy by dropping out those below the mean; see http://www.csiss.org/clearinghouse/FlowMapper/
I am curious, what are cognitive implications for the strategy?