On the Structure of General Theories of Software Engineering

Taming complexity is a fundamental goal of software engineering. The core techniques that have been successful in practice are those that separate concerns, especially variants of architectural abstractions called components and connectors. We argue that General Theories of Software Engineering (GTSE) should be organized by components and connectors to distinguish conceptually distinct elements and their inter-relationships and interdependencies. Doing so, we argue, separates concerns that should be distinct and not conflated, thereby increasing the value of GTSE efforts. 1 Our Position Software Engineering (SE) is an extremely broad field. It covers all domains for which software systems are feasible. Not surprisingly, there are core SE techniques and principles that are relevant to all domains as well as more focused techniques and principles relevant to specific domains. This is clearly evident in the typical courses covered in both Computer Science and Software Engineering curriculums. A quote from the GTSE 2013 workshop report [1] summarizes an important thread of discussion (the quoted references are from the report): Participants agreed that theorizing takes many forms [17] and SE entails myriad phenomena; for instance, Perry [11] distinguishes between software engineers, SE, and software project management. Ralph [12] consequently suggests formulating a multi-level GTSE, i.e., a theory that crosses many units of analysis including individual, team, artifact, process and project. A core question then is: What might the different levels of a GTSE contain? We assert that any General Theory of SE (GTSE) should be constructed from two entities: component theories and connector theories. Component theories are about the major components in software engineering: economics, management, software engineers, and SE. Connector theories are about the relationships and interdependencies among component theories. Cognition, for example, is a critical element of a component theory about software engineers (people); structural complexity is a critical element of a component theory about software systems (SE). A connector theory would express the relationship between cognition and complexity. Further, we know from experience that the basic structure of components and connectors is recursive: component theories can be expressed as aggregations of more restricted component and connector theories; connector theories can be decomposed in terms of simpler component and connector theories. We believe that an organizational structure with component and connector theories is fundamental to a GTSE; it provides the means to structurally separate conceptually distinct elements and their interrelationships and interdependencies. In what follows, we provide an example of a GTSE structure in terms of component and connector theories (in some cases recursively) to illustrate our approach; it should be a platform to guide and