Towards Knowledge Evolution in Software Engineering: An Epistemological Approach

1 Introduction

Software engineering (SE) is by nature a human discipline. In huge software projects hundreds of analysts, designers, developers, project managers, potential users, etc. are involved in the process of creating a software application aimed to fulfill services easing the everyday work and life of thousands of individuals. To deal with such a process the traditional waterfall model in which requirements are collected once for all at the beginning of the project life cycle is nowadays too limited. Researches have led to the definition of advanced modeling languages as well as to more elaborated life cycles using the concept of iterative development. Their success come from the fact that requirements elicitation is no more a standalone task at the beginning of the project but rather a continuous process using feedback loops allowing to better apprehend needs on the basis of prototypes.

In this article we propose a modern epistemological reading of SE as an evolving science as well as software project knowledge evolution. As a first basis, project actors’ reasoning context is studied. Indeed, actors involved in a software project have a limited vision inherent to their bounded rationality. At first sight Popperian theories could be used to describe the evolution of software engineering theoretical frameworks. Due to the nature of SE belonging also to social sciences we will show that this vision is abusive. The Lakatosian falsification principle will be used to demonstrate the complexity of getting a crucial experiment in SE. Adaptive rationality will be used to characterize SE theorists and researchers’ expertise field vision.

This article is organized as follows. Section 2 presents the context of the research, the followed approach and justifies the utilization of the epistemologies used in the article and their combination. It also points to the article contributions. Section 3 proposes an epistemological reading of software processes and on the evolution of knowledge in SE as a discipline as well as within a particular project. To this ends we used Herbert Simon’s bounded rationality, the Popperian knowledge growth theory nuanced by the Lakatosian falsification principle and finally adaptive rationality. The implications are then overviewed in section 4 while conclusions are summarized in Section 5.