Efficiency implications of open source commonality and reuse

This paper analyzes the reuse choices made by open source developers and relates them to cost efficiency. We make a distinction between the commonality among applications and the actual reuse of code. The former represents the similarity between the requirements of different applications and, consequently, the functionalities that they provide. The latter represents the actual reuse of code. No application can be maintained for ever. A fundamental reason for the need for periodical replacement of code is the exponential growth of costs with the number of maintenance interventions. Intuitively, this is due to the increasing complexity of software that grows in both size and coupling among different modules. The paper measures commonality, reuse and development costs of 26 open-source projects for a total of 171 application versions. Results show that reuse choices in open-source contexts are not cost efficient. Developers tend to reuse code from the most recent version of applications, even if their requirements are closer to previous versions. Furthermore, the latest version of an application is always the one that has incurred the highest number of maintenance interventions. Accordingly, the development cost per new line of code is found to grow with reuse