When More Heads Are Better than One? Understanding and Improving Collaborative Identification of Code Smells

Code smells are program structures that often indicate software design problems. Their efficient identification is required in order to ensure software longevity. However, the identification of code smells often cannot be performed in isolation by a single developer. This task might require the knowledge of various program parts, which are better understood by different developers. However, there is little guidance to support software teams on efficient identification of code smells. In this research, we investigate how to improve efficiency on the collaborative identification of code smells. Our investigation is based on a set of controlled experiments conducted with more than 58 novice and professional developers. Our preliminary results suggest the use of collaborative practices significantly increases the efficiency of code smell identification. We also compiled a set of guidelines and heuristics to support an effective collaborative strategy for code smell identification.