In this essay we proposes that the degree to which communities show division of labor (i.e. specialization of contributors) is (co-)determined by the degree of generalized reciprocal behavior users can assume from one another. The mechanism that explains is that the norm of reciprocity can replace authority as a mechanism sanctioning free-riding in settings where there is no formal authority (i.e. OSS communities). In our analyses we find that (GitHub) community members do take co-specialized roles in their contribution behavior the more the latter can trust in their efforts being rewarded in kind.

As a next step, we suggested that the degree of how efficient the division of labor will be varies across different GitHub sub-communities. Our moderation analyses show that higher degree of monitoring on a project (measures as number of watchers a project has) facilitates stronger expectation of generalized reciprocity behavior among project members ( = owner is expected to focus on a project development and growth rather than generally giving-away code back to the community). Moreover, we find support that co-specialization among project members is stronger on projects which have a higher diversity in users’ tenures with the platform (which proxies for OS programming experience and in part captures coding skills) and users’ coding skills (measured as the rate of accepted by third parties code patches). Our moderation results help firms, which are willing to engage with OSS in order to get complementary innovation assets, to choose at the first glance which sub-communities/projects might be more efficient and thus highly attractive for collaboration.

Currently, we are refining the theory and contribution sections of the paper based on the results we found.