University of California
The root causes of failures are viewed as errors that are made during one or more phases of development. If a general model of software errors could be developed, then it may feasible to develop a certification foundation based on the model. An error-based approach to certification is described. A classical theory of error is reviewed and a software interpretation of the theory is developed. The interpretation suggests a strategy for testing and analysis. The strategy was evaluated by comparing its potential effectiveness with that of certification standards based on individual methods.