Lloyd´s Register´s approach to software dependability assessment

Software dependability assessment is the evaluation of evidence produced from the development process. Assessment must be objective; that is, it must be based upon observations of the supplied evidence which can readily be checked, or facts derived from those observations by some agreed procedure (measurements). The difficulty lies in choosing the observations and measurements so that dependability may be validly inferred from them. There are few existing standards against which the dependability of a software product can be assessed. Existing standards concentrate on the process of developing software and insufficiently on the product. In addition they are, in the main, too general to provide a basis for assessment: compliance with them is itself a matter of judgement and is therefore partly subjective. Faced with this situation, the work being undertaken within Lloyd´s Register focuses on the development of fixed criteria on which to base assessments. These are naturally derived from applicable standards, but place greater emphasis on repeatability and objectivity. Measurement is employed wherever possible. Procedures are being developed, and minimum qualifications for assessors sets, in order to ensure that, where measurement is not possible, decisions are justified and personal preferences minimised