“Classical statistical papers and books, with some notable exceptions, have tended to treat statistical inference as being based on a single experiment, and concentrated on the design and inferential analysis of that one experiment. Science, however, is a process of learning and exploration, with each experiment motivating and guiding the next one. Modern statistics has started to give greater consideration to designing this process of learning: for example, data mining and statistical learning theory, statistical meta-analysis and adaptive clinical trials. In the learning process, communication from one investigator to another of well-understood and meaningful measures of uncertainty is essential.”