How Multiple Contributors Reduce Software Quality - A Quantitative Analysis on a Large Telecommunications Company

Abstract

Since the origin of agile methods and open source development, code ownership tends to be more widely distributed over multiple contributors than before. The question is to what extent a component is a↵ected when several developers contribute to it. Does several contributors provide better solutions than a sole developer or does multiple contributors generate additional defects? Although some previous research results point to the fact that more contributors do expose a project to higher risk, surprisingly little has been done to validate this hypothesis. An answer to this question is highly relevant since it provides organizations with the option to adjust development teams and contribution levels accordingly, in order to assure software quality. By empirically studying a large data set from a proprietary telecommunications company, we examine the relationship between the number of contributors and the number of defects in closed source industrial projects. In addition, we are the first to investigate the e↵ect that multiple contributors have on defect density and defect severity. We find the correlation between contributors and defect density to be statistically significant and we find that the number of contributors and the number of defects has a near perfect positive relationship.