Assessing the maintainability benefits of design restructuring using dependency analysis

Software developers and project managers often have to assess the quality of software design. A commonly adopted hypothesis is that a good design should cost less to maintain than a poor design. We propose a model for quantifying the quality of a design from a maintainability perspective. Based on this model, we propose a novel strategy for predicting the "return on investment" (ROI) for possible design restructurings using procedure level dependency analysis. We demonstrate this approach with two exploratory Java case studies. Our results show that common low level source code transformations change the system dependency structure in a beneficial way, allowing recovery of the initial refactoring investment over a number of maintenance activities.