This paper argues classical statistics and standard econometrics are based on a desire to meet scientific standards for accumulating reliable knowledge. Science requires two inputs, mining of existing data for inspiration and new or 'out-of-sample' data for predictive testing. Avoidance of...