Scholars from a variety of backgrounds – economists, sociologists, strategists, and students of technology management – have sought a better understanding of why some knowledge disperses widely while other knowledge does not. In this quest, some researchers have focused on the characteristics of the knowledge itself (e.g., Polanyi, 1966; Reed and DeFillippi, 1990; Zander and Kogut, 1995) while others have emphasized the social networks that constrain and enable the flow of knowledge (e.g., Coleman et al., 1957; Davis and Greve, 1997). This chapter examines the interplay between these two factors. Specifically, we consider how the complexity of knowledge and the density of social relations jointly influence the movement of knowledge. Imagine a social network composed of patches of dense connections with sparse interstices between them. The dense patches might reflect firms, for instance, or geographic regions or technical communities. When does knowledge diffuse within these dense patches circumscribed by social boundaries but not beyond them? Synthesizing social network theory with a view of knowledge transfer as a search process, we argue that knowledge inequality across social boundaries should reach its peak when the underlying knowledge is of moderate complexity. To test this hypothesis, we analyze patent data and compare citation rates across three types of social boundaries: within versus outside the firm, geographically near to versus far from the inventor, and internal versus external to the technological class. In all three cases, the disparity in knowledge diffusion across these borders is greatest for knowledge of an intermediate level of complexity.
