and so my mentor john monaghan likened the problem of predicting violent people to predicting violent storms. when you think of meteorology, you think of the difficulty of classifying a hurricane and tracking a hurricane, making judgments about such complex behavior that has sort of chaotic premises underlying it, you're going to make lots of mistakes. you're going to make lots of mistakes in both directions. sometimes you're going to make a mistake when you think the storm is going to hit and it doesn't and sometimes you're going to make mistakes where you think the storm is not going to hit and it does. and i think one of the great challenges, quite frankly, for the legal system, is understanding statistics well enough to make that judgment about where you draw that line that anita was referring to about where do you want to avoid the errors. you want to avoid the errors, a category five hurricane is going to hit miami but there is only a 40% likelihood that that hurricane is going to hit miami, do you evacuate the city? well, if you evacuate the city and it doesn't hit, you spent t