Diks and Pelc consider the fault diagnosis problem on multiprocessors. They present diagnosis algorithms for two specific system configurations. In comparing previous work with theirs, they consider the previous work unrealistic due to the worst-case assumptions on the behavior of the faulty processors. They claim that their model is the only one that does not make any assumption about the behavior of the faulty processors, yet they assume that failures are permanent and that they occur with fixed probability. The remainder of this research paper deals with proofs and theories, with no intuitive explanation in most cases. The paper is not long, but it may seem very long if the reader is not interested in the theories and proofs behind the model.