Measurement for software quality

The author takes an uncontentious view of quality as being `fitness for purpose´, more specifically, conformance to user requirements. Quality can be measured by number of reported nonconformances per hour of use (of course `shelfware´, that is, software that is never used, exhibits no failures). Quality in many installed engineering systems is likely to exhibit a `bathtub´ failure curve during its lifetime. Despite the apparent kinship with physical engineering systems, software systems demand a different treatment because repairs to a software system alter its specification. Here, the author presents various efforts aimed at standardising the measurement of software quality