Investigating issues with computing and interpreting the truck factor metric

Abstract

The knowledge held by a company’s employees is a mission-critical asset in software engineering organisations which is hard to manage and maintain due to high labour turnover rates. The software engineering community is familiar with the so-called Truck Factor metric which is defined as the minimal number of developers that have to be hit by a truck or leave, before the project becomes in trouble [1]. Alternatively, this metric reveals the project’s concentration of knowledge and its key developers. Due to the importance of this metric to project managers, many algorithms have been proposed to automatically compute this metric by extracting data from version control systems. Despite popular understanding of this metric’s notions, there are still challenges with its calculation, interpretation and validity. Therefore, in this research study, an implementation of an existing technique to calculate this analysis together with improvements, is proposed. To explore the validity of this approach, an industry partner was recruited who provided access to two git repositories. These repositories were analysed and conclusions were drawn regarding any knowledge risks. These conclusions and results were presented to the industry partner for feedback on their perceived validity. Furthermore, this research concluded that the Truck Factor metric needs to be calculated by considering different perspectives and to consider evaluating important modules within the repository. It was also found that the complexity of the code, as well as how the developers work within the project, also influenced the validity of the Truck Factor results.