Goal conflict in software reviews
Gina Kingston · UNSWorks (University of New South Wales, Sydney, Australia) · 2022
The thesis discusses an investigation into the impact of goal conflict on the performance of software reviews. It was motivated by anecdotal evidence from practicing software engineers and acquirers, who perceived that joint software reviews were less efficient than software inspections. Negotiation literature and the anecdotal evidence suggested that performance differences might arise from goal conflict in joint software reviews and an associated increase in negotiation stumbling blocks. The investigation consisted of three main empirical studies and one main conceptual model. Two additional empirical studies were conducted, and one additional conceptual model was developed, to reduce the risks associated with the research programme. The main model considered the progress of issues through a software review. It suggested that fewer issues would be raised as the level of goal conflict increased due to an increase in the number of negotiation stumbling blocks. This was supported by the results of the laboratory experiment. The model also suggested that fewer issues would be resolved, because of the difficulty in obtaining agreement. This was not supported by the results of the experiment. However, the results of the survey and the preliminary industrial study suggested that the quality of the issue resolution was affected. There was evidence in both studies that issues were resolved prematurely because of Conformance Pressure. The industrial case study focused on the SAAM review process, which is a scenario-based review process that uses a facilitator. The proponents Of the SAAM review process claim that it helps align the goals of participants and the negotiation literature suggests that, facilitators can help reduce negotiation stumbling blocks. However, in the case study the facilitator was observed to exert rather than reduce Conformance Pressure. Furthermore, while the SAAM review process provided a focus for the review, several problems with the SAAM review process, as implemented during the study, were noticed. In particular, the voting process was perceived to be biased. Thus, while the thesis provides an introduction into the relatively unexplored area of software reviews in the presence of goal conflict, it also suggests areas where further research is desirable.