This paper is concerned with creation and diffusion of knowledge within the scientific community. We invoke the distinction between open and closed science to focus attention solely on those agents for whom knowledge is an end in itself. Our concern here is the flow of knowledge within a discipline, such as economics for example, in which the discipline is defined broadly enough that it contains several well-defined sub-disciplines (micro-theory, applied micro, econometrics, labour economics, industrial organization, macro-economics and so on). Within a discipline individual scientists interact directly with other scientists in a variety of ways — they collaborate; they read each other’s working papers; they talk in the corridors; they attend each others’ seminars and conference presentations and so on. If these are considered direct interactions, it is clear that all economists do not interact directly with all others. Indeed, any economist will interact directly only with a small number of other economists. The types of interactions listed above are all largely concerned with the diffusion of knowledge. These are important communication channels in the scientific community, so a knowledge diffusion model will have to treat a population of agents, each of whom interacts only with a very small subset of the rest of the population. Each agent interacts with his or her own subset, so we have essentially a network, or graph structure over which knowledge diffuses.