Open Tools for Software Engineering: Validation of a Theory of Openness in the Automotive Industry

Context: Open tools (e.g., Jenkins, Gerrit and Git) offer a lucrative alternative to commercial tools. Many companies and developers from OSS communities make a collaborative effort to improve the tools. Prior to this study, we developed an empirically based theory for companies' strategic choices on the development of these tools, based on empirical observations in the telecom domain. Aim: The aim of this study is to validate the theory of openness for tools in software engineering, in another domain, automotive. Specifically, we validated the theory propositions and mapped the case companies onto the model of openness. Method: We run focus groups in two automotive companies, collecting data in a survey and follow-up discussions. We used the repertory grid technique to analyze the survey responses, in combination with qualitative data from the focus group, to validate the propositions. Results: Openness of tools has the potential to reduce development costs and time, and may lead to process and product innovation. This study confirms three out of five theory propositions, on cost and time reduction, and the complementary role of open tools. One propositions was not possible to validate due to lack of investment in OSS tools communities by both companies. However, our findings extend the fifth proposition to require management being involved for both the proactive and reactive strategy. Further, we observe that the move towards open tools happen with a paradigm shift towards openness in the automotive domain, and lead to standardization of tools. Both companies confirm that they need legal procedures for the contribution, as well as an internal champion, driving the open tools strategy. Conclusion: We validated the theory, originating from the telecom domain, partially using two automotive companies. Both case companies are classified as laggards (reactive, cost saving) in the model of openness presented in the theory. Furthermore, we would like to have more validations studies to validate the remaining quadrants (e.g., leverage, lucrativeness and leaders).