When something goes wrong while running R within one of Pipeline Pilots R Statistics components, PP presents the error message generated by R and terminates the protocol. These messages are not always as clear as we'd wish, but I just ran across a nice blog posting from PP user Gerard van Westen that explains some of the messages you might see, particularly when building SVM or Forest models: http://blog.gjpvanwesten.nl/2011/10/r-statistics-error-messages-in-pipeline.html.