Bruce Spencer is not the first guy to argue that jury trials have an excessively high error rate, but he may be the first scholar to put a number to this error rate. His calculations arrive at a minimum error rate of 13%, or one out of eight. This finding implies not that American justice is blind or deaf, but, perhaps, dead.
Skeptics of statistics (or statistical skeptics) are welcome to read the original manuscript on the Northwestern University web site (pdf file). They are also forewarned that this article delves into esoteric subjects like definitions of “error rate”, statistical methods, and analysis.