Albert Einstein and Empirical Software

s scientists, we apply scientific investigative techniques to gain more understanding of what makes software “good” and how to make software well. Often, we adapt investigative techniques from other disciplines to define measures that make sense in the business, technical, and social contexts we use for decision making. However, sometimes we can learn as much from another discipline’s failures as from its successes. Examples from nineteenth-century physics show how a change in perspective can lead to explanations for previously misunderstood phenomena. We must also consider whether our measurements constrict our view of what is really happening in the development process, and we must change or expand them if they are. Such changes may involve looking further afield, drawing on examples from the legal and business communities to improve our own models and theories. Science, and particularly physics, clearly illustrates the limitations of a too-literal approach to building and maintaining software. A DECEPTIVELY ORDERED UNIVERSE At the end of the nineteenth century, several things about physics were well known: