In scenarios drawn from the movie Minority Report, some pundits are forecasting explosive growth of visualization over the next few years. We can expect, they say, to catch terrorists, identify fraud, and flag defects -- all through visual exploration of huge collections of data. This is all supposed to happen through new widgets that allow us to sort, zoom, filter, select, highlight, and link. These tools, they say, will give us the agility of a virtual jet flying through data. Other pundits are touting the age of machine learning, in which automated algorithms will find hidden patterns, outliers, and other anomalies in Big Data. In this scenario, no human interaction will be necessary. The computer will figure it all out. The extreme versions of both these views are wrong. A few simple examples show why. In the end, it should be evident that effective visualizations need to be based on appropriate analytics and appropriate analytics must be supported by revealing visualizations.