18 Feb 2010Working Paper Summaries

The Mirroring Hypothesis: Theory, Evidence and Exceptions

by Lyra Colfer and Carliss Y. Baldwin

In its simplest form, the mirroring hypothesis suggests that the organizational patterns of a development project, such as communication links, geographic collocation, and team and firm membership, correspond to the technical patterns of dependency in the system under development. According to the hypothesis, independent, dispersed contributors develop largely modular designs, while richly interacting, collocated contributors develop highly integral designs. Yet many development projects do not conform to the mirroring hypothesis. HBS doctoral graduate Lyra Colfer and professor Carliss Y. Baldwin synthesize observations from a large number of cases that violate the hypothesis to explain when and how development organizations can "break the mirror." Key concepts include:

While mirroring is common in practice, it is not universal.
In the presence of compatible motivations and frameworks supporting expectations of good faith, there are new ways of building common ground, based on digitized designs; electronic archives; automated test suites; and instantaneous transmission of text, data, and pictures.
These alternative means can be used as complements or substitutes for mirrored forms of organization.
Managers of development organizations within and across firms and in open collaborative groups, who choose or are required by circumstances to "break the mirror," should be aware of these alternative means of achieving coordination.

Author Abstract

The mirroring hypothesis asserts that the organizational patterns of a development project (e.g. communication links, geographic collocation, team and firm co-membership) will correspond to the technical patterns of dependency in the system under development. Thus the hypothesis predicts that developers with few or no organizational linkages will design independent system components, while developers with rich organizational linkages will co-design highly interdependent system components. (The hypothesis claims a correspondence between organizational structure and technical architecture, but allows causality to flow in either direction.) Scholars in a range of disciplines have argued that mirroring is either a necessary or highly desirable feature in the design of development projects, but empirical research shows that some projects deviate from strict mirroring, seemingly without harmful effects. In this paper, we formally define the mirroring hypothesis, describe its theoretical underpinnings and systematically review the empirical evidence for and against it. Our review includes 129 studies spanning three levels of organization: within a single firm, across firms, and open community-based development. Across these levels, the hypothesis was supported in 69% of the relevant cases, but not supported in 31%. It was most strongly supported within firms, less strongly across firms, and often violated in community-based development settings. The exceptions in turn were of two types: In four cases, closely collaborating teams within single firms created modular systems comprised of independent components. More surprisingly, in 28 cases, independent and dispersed contributors made highly interdependent contributions to the design of a single technical system (or sub-system). Based on a detailed analysis of the latter 28, we introduce the concept of actionable transparency as a means of achieving coordination without mirroring. Contributors achieve actionable transparency by embedding their design in a centralized system with a shared design language and near-real-time updating, where everyone with an interest in improving the design has the right and the means to act on it. We present examples from practice and then describe the more complex organizational patterns that emerge in lieu of genuine mirroring when actionable transparency allows people to "break the mirror." 47 pages