On the conservation of software defect

This paper begins by analysing general software systems using con-cepts from statistical mechanics which provide a framework for linking microscopic and macroscopic features of any complex system. This analysis provides a way of linking two features of particular interest in software systems, first the microscopic distribution of defects with component size and second the macroscopic distribution of component sizes in a typical system. The former has been studied extensively but the latter much less so. This paper shows that subject to an exter-nal constraint that the total number of defects is fixed a priori, then commonly used defect models for individual components directly im-ply that the distribution of component sizes in the overall system will obey a power-law Pareto distribution. The paper closes by analysing a significant number of systems of different total size, different implementation languages and very dif-ferent application areas and demonstrates that the component sizes overwhelmingly obey the predicted power-law distribution. Some pos-sible implications of this are explored