As science deals with the ever-growing complexity of research fields, research software provides the tools to observe, assist and manage their processes and results. However, the majority of research software tends to prioritize the validation of theories and the creation of prototypes over the delivery of products, often overlooking the crucial aspect of quality assurance. This can result in products that are susceptible to flaws and errors. The risk of failing to deliver correct results not only undermines user trust, but also damages the reputation of researchers and developers in a team, risking losses in funding and public trust in research. The implementation and emphasis on software engineering practices, particularly software testing, can mitigate these risks by prioritizing the identification and elimination of errors and faults in software development while also enhancing observability and transparency throughout the coding cycle.
Luc Michel (Tue,) studied this question.