Simone Santini has an amusing column in the December 2005 Computer about peer reviewing and refereeing. It is probably something only a researcher would find amusing, but his underlying question is a good one: "How much damage could be caused by a peer reviewer having a bad day?"
He goes on to write ficticious reviews of Dijkstra's "goto considered harmful", Codd's "relational model of data", Turing, Shannon, Hoare, and Rivest-Shamir-Adelman's famous public key paper.
It looks like the Computer Society has it available without charge, but who knows for how long:
"We are sorry to inform you ..."