Testing the proposed idea requires the operationalization of both (in)direct reciprocity and non-equivalent reciprocity. We suggest that (in)direct reciprocity occurs when a repository founder engages in a helping behavior X to other participants and receives the same helping behavior X from other participants. On the contrary, we suggest that non-equivalent reciprocity occurs when a repository founder engages in a helping behavior X to other participants and receives the helping behavior Y from other participants.

We develop founder-level predictions to test the idea of a special kind of reciprocity as a main driver of OSS repository growth:

1. [Preferential attachment has been argued to be prevalent in knowledge network by previously done research. We expect to find the "rich gets richer" effect in OSS communities, too.]

It is more likely that a repository founder receives a kind of helping behavior when s/he is connected with a greater number of network participants who provide the same kind of helping behavior.

2. [Equivalent reciprocity is referred as both direct and indirect reciprocity. The key argument here is the type of resource/activity to be exchanged between network actors.]

It is more likely that a repository founder receives a kind of helping behavior when s/he is connected with a greater number of network participants to whom s/he presents the same kind of helping behavior.

3. [Non-equivalent reciprocity also does not distinguish whether the direction of reciprocal behavior is direct or through a third party. This type of reciprocity, however, takes into consideration different types of resources/activities to be exchanged between network actors.]

It is more likely that a repository founder receives a kind of helping behavior when s/he is connected with a greater number of network participants to whom s/he presents the different kind of helping behavior.

We plan to test our predictions on a custom-tailored dataset merging data on projects and activities in GitHub, the largest OSS hosting repository site. The dataset construction and further analysis are in progress.