In this article we explore how learning across competing Communities of Practice (CoP) may benefit from actively managed processes, namely: building an explicit knowledge repository (open source), and setting incentives to crossing boundaries (knowledge-bridging). To better illustrate our results, we compare them to a simple learning model of CoP rooted on situated learning theory, whereby competing communities emerge as individual contributors effect work together. In the seminal works of Lave and Wenger  and Brown and Duguid , CoP’s were associated with informal self-organization and tacit knowledge exchanges. Since then, attention in the literature has shifted towards understanding intentionally created CoP’s, and the role of online CoP’s built around explicit open source knowledge repositories. In this sense, there has been little research done to model system-level learning across competing CoP’s, particularly in the context of open source. Our simulation model contributes to fill this gap and shows that at the system-level: i) open sourcing has a strong positive effect on the average learning rate, and ii) knowledge-bridging reduces learning rate volatility and has a moderately positive effect on the average learning rate. These findings have implications for theory as they inform the debate in the literature around the different ‘kinds’ of CoP’s. In addition, as ‘self-organized CoPs’ (e.g. Linux) and ‘firm-sponsored CoPs’ (e.g. Sun Microsystems’s OpenSolaris) enter into competition, the implications for practice are ever more important.