By Alon Harel & Ariel Porat. Full text here.
Should a court convict a defendant for unspecified offenses if there is no reasonable doubt that he committed an offense, even though no particular offense has been proven beyond a reasonable doubt? Suppose a defendant is charged with two unrelated offenses allegedly committed at different times and places: pickpocketing and rape. The probability that he committed each one of the offenses is .9. Assume that the minimum threshold required for conviction is .95. Under prevailing law, the defendant would be acquitted of both charges since no offense can be specifically attributed to him. However, a simple calculation of the probability that the defendant committed at least one offense amounts to .99. Consequently, it seems that convicting him for one offense without specifying what this offense is and punishing him with the sanction designed for the least severe offense would be just and efficient. We call the principle that requires such an aggregation the “aggregate probabilities principle” (APP).
This Article proposes that under certain conditions, deterrence, efficient law enforcement, and minimization of adjudication errors would be better achieved if courts applied the APP. The Article also addresses the most powerful possible objections to this method and suggests that the aggregation principle be adopted only under strict conditions that preclude its potential abuse by the prosecution. In addition, the Article shows that sometimes aggregating probabilities will yield fewer, rather than more, convictions. If the APP is adopted, the presumption of innocence currently applied with regard to the offense will be replaced by a presumption of innocence applied to the accused.