Paper: Software quality: the elusive target

Kitchenham B, Pfleeger SL. Software quality: the elusive target [special issues section]. Software, IEEE. 1996;13(1):12-21.

I'd been looking for an article like this for a while: a good overview of the definitions of quality. Kitchenham and Pfleeger take David Garvin's work on the definition of quality and apply them to software quality. There are five different perspectives to view software quality from:

• Transcendental view: software quality can be recognised and worked towards, but never precisely defined or perfectly achieved. Thus, view holds that quality is inherently unmeasurable.
  
• User view: how well the software suits the needs of its users. Measuring quality from a user view involves refining concepts like "reliability" and "usability" into measurable characteristics (say, number of hours of learning time needed).
  
• Manufacturing view: how well the software conforms to its specifications, and development process. Measurement is by defect counts and rework costs.
  
• Product view: how well the software scores on various "internal quality indicators", like program complexity. If the internal quality is high then so must be the external quality.
• Value-based view: how much the customer is willing to pay for the software. This view, in some sense, unites the various other views with a very practical measure.

The article talks very briefly about comparing projects by using defect density and scaling by user base, but doesn't go into any of the philosophical troubles. There's also a decent reading list on software quality in general.