Harmfulness of Code Duplication - A Structured Review of the Evidence

Duplication of code has long been thought to decrease changeability of systems, but recently doubts have been expressed whether this is true in general. This is a problem for researchers because it makes the value of research aimed against clones uncertain, and for practitioners as they cannot be sure whether their effort in reducing duplication is well-spent. In this paper we try to shed light on this issue by collecting empirical evidence in favor and against the negative effects of duplication on changeability. We go beyond the flat yes/no-question of harmfulness and present an explanatory model to show the mechanisms through which duplication is suspected to affect quality. We aggregate the evidence for each of the causal links in the model. This sheds light on the current state of duplication research and helps practitioners choose between the available mitigation strategies.