The discussion focuses on one group of probabilistic models: those similar to or developed from the negligence rule propounded by Learned Hand in United States v. Carroll Towing Co. First, the formula set by Hand is examined, and then it is developed into a more theoretically rigorous form. Second, judicial attempts to apply Hand-type models on a case-by-case basis are considered; and third, a functionally useful exploitation of the Hand model is developed in a context apart from case-specific analysis. Learned Hand's 'rule' concerning the standard of care involved in ordinary negligence actions was set forth in three cases decided in the 1940's: Conway v. O'Brien, United States v. Carroll Towing Co., and Moisan v. Loftus. The 'rule' is developed from an initial, essentially intuitive formulation into a wholly algebraicized one, although the change in the nature of the model itself is minimal. Kac has stated that 'The main role of models is not so much to explain and to predict--though ultimately these are the main functions of science--as to polarize thinking and to pose sharp questions.' Much of the law's use of the Hand model and its variant expressions had been geared toward explanation and prediction; too little has been directed toward 'sharp questions.' While much energy has been devoted to detailed exegesis and refinement of the model, too little attention has been given to the possible insights arising from interaction of the model and its underlying principles with other elements of the liability rules with which it is associated. Moreover, the model has been imbued with the nature of reality itself, such that it is used as the whole rather than only part of the decisionmaking network. A total of 172 footnotes are listed. (Author abstract modified)