The Mirroring Hypothesis: Theory, Evidence and Exceptions

The mirroring hypothesis predicts that organizational ties within a project, firm, or group of firms (e.g., communication, collocation, employment) will correspond to the technical patterns of dependency in the work being performed. A thorough understanding of the phenomenon is difficult to achieve because relevant work is scattered across multiple fields. This paper presents a unified picture of mirroring in terms of theory, evidence, and exceptions. First, we formally define mirroring and argue that it is an approach to technical problem solving that conserves scarce cognitive resources. We then review 142 empirical studies, divided by organizational form into (1) industry studies, (2) firm studies, and (3) studies of open collaborative projects. The industry and firm studies indicate that mirroring is a prevalent pattern but not universal. However, there is evidence of a mirroring “trap”: firms focused on the current technical architecture may fall victim to architectural innovations arising outside their boundaries. Thus in technologically dynamic industries, partial mirroring, where knowledge boundaries are drawn more broadly than operational boundaries, is likely to be a superior strategy. Firms can also strategically “break the mirror” in two ways: by implementing modular partitions within their own boundaries or by building relational contracts that support technical interdependency across their boundaries. Finally, in contrast to industry and firm studies, studies of open collaborative projects, most of which focused on software, were not supportive of the hypothesis. We argue that these contradictory results arise because digital technologies make possible new modes of coordination that enable groups to deviate from classical mirroring as seen within firms. This working paper includes Appendix A, which describes our detailed findings by category. Appendix B, a tabular summary of the 142 studies in our sample, is available on request from the authors.