A study of behavioral decay in design patterns

Design patterns represent a means of communicating reusable solutions to common problems, provided they are implemented and maintained correctly. However, many design pattern instances erode as they age, sacrificing qualities they once provided. Identifying such instances of pattern decay is valuable because it allows for proactive attempts to extend the longevity and quality attributes of pattern components. Apart from structural decay, design patterns can exhibit symptoms of behavioral decay. We utilized a taxonomy that characterizes these negative behaviors and designed a case study wherein we measured structural and behavioral decay, hereafter referred to as pattern grime, as well as pattern quality and size, across pattern evolutions. We evaluated the relationships between structural and behavioral grime and found statistically significant cases of strong correlations between specific types of structural and behavioral grime. Furthermore, we extended the QATCH operational software quality model to incorporate design pattern evolution metrics and measured and correlated software quality to the presence of behavioral grime in software systems. Our results suggest a strong inverse relationship between software quality and behavioral grime.