The authors describe efforts to better understand classical testing and debugging methods, and they suggest alternate ways of viewing tests and test results. Specifically, they address, compare, and analyze debugging versus operational testing issues. In the analysis, they use models of programs and their testing. They explore testing effectiveness, based on the reliability of a program after testing, along with the concept of program failure probabilities after testing. The underlying issues of subdomains and selection of test cases from a subdomain are also analyzed. The main value of the paper may lie in the good job it does of addressing issues that testers should be aware of when selecting testing approaches. Appropriately, the authors warn users of the techniques they present to exercise caution in interpreting theoretical comparisons.

Although the actual analysis presented is reasonable and understandable, I had trouble wading through the paper. Too much detail is given about insignificant issues, considering that readers of this technical paper need to be reasonably knowledgeable about the subject. Despite these problems with the presentation style, I recommend the paper to those working on testing, defect removal, and debugging.