Defect and Fault Seeding In Dependability Benchmarking

Defect and fault seeding is often considered for gathering empirical estimates within reliability models. Traditional defect seeding is fraught with difficult to resolve validity concerns when attempting to estimate true defect and fault populations. With dependability benchmarking we are less concerned about true defect and fault estimates, rather we wish to compare the relative effectiveness of dependability approaches. We propose that in this context the traditional concerns regarding defect and fault seeding techniques may not be as difficult to address and that potential new approaches may be useful as a means of benchmarking approaches to dependability.