Legal Risk, Legal Evidence and the Arithmetic of Criminal Justice

It is argued that the standard way that the criminal justice debate regarding the permissible extent of wrongful convictions is cast is fundamentally flawed. In particular, it is claimed that there is an inherent danger in focussing our attention in this debate on different ways of measuring the probabilistic likelihood of wrongful conviction and then evaluating whether these probabilities are unacceptably high. This is because such probabilistic measures are clumsy ways of capturing the level of risk involved, to the extent that a defendant can be subject to an unacceptably high level of legal risk in this regard even where the relevant probabilities are very low. An alternative conception of legal risk⎯one that is primarily cast along modal rather than probabilistic lines⎯is set out which offers a much better way of framing the debate regarding what would be an acceptable level of wrongful conviction. It is further argued that with this modal conception of legal risk in play we can capture an important necessary condition that should be imposed on legal evidence, one that has application beyond the context of the criminal trial. 1. THE ARITHMETIC OF CRIMINAL JUSTICE One of the most pressing issues in legal theory concerns the extent to which a criminal justice system ought to countenance the possibility of wrongful conviction. Clearly any wrongful conviction is an injustice, and hence something to be avoided. Moreover, any real-world system of criminal justice is bound to make mistakes from time to time, and hence is bound to generate some wrongful convictions, and so we cannot reasonably expect a criminal justice system to make no errors of this kind. The question is then how extensive can this kind of legal error be before it entails that the criminal justice system should be reformed. In short, what level of legal risk of this