Skill Specialization and the Formation of Collaboration Networks
Any type of content formally published in an academic journal, usually following a peer-review process.
In recent years, there has been increased interest among funding organizations and administrators in supporting the acquisition of interdisciplinary skills in collaborative communities, such as universities, national labs, and knowledge-based firms. However, there has been relatively little work exploring the effects of interdisciplinarity on the structure and function of these communities. In this paper, I use a collaboration network–in which the nodes are individuals, and two nodes are connected if they’ve collaborated on a problem–to formalize the structure of collaborative relationships. Using a formal model of the collaborative process, I examine the effects of increased interdisciplinarity on the structure of this collaboration network. I show that when collaborative communities become more interdisciplinary, the links in the network become more concentrated among a few, high-degree individuals, and superstars emerge. These individuals, who are so productive that their contributions dominate the overall community, are a potential unintended consequence of policies intended to increase interdisciplinarity. I then define a specialist to be an individual whose skills cluster in a single area, and a generalist to be an individual whose skills span several areas, and I examine the roles that specialists and generalist play in the network. I show that while specialists have more links in the collaboration network, generalists are more likely to bridge between different communities. Given that individuals in these communities tend to benefit from being highly connected, while the community as a whole benefits from bridging activities, this result suggests that generalists may be undersupplied, which lends support to policies that fund the acquisition of interdisciplinary skills in situations where bridging activities are valued by the community at large but are not individually rational.