In the last four decades, oncology has shifted from a small, marginal domain within biomedicine to one of the largest, most central and successful. It is now second only to cardiovascular medicine in terms of practitioners, research activities, biopharmaceutical investment and market size. Oncology has pioneered the most innovative recent approaches to translational research. In spite of its staggering expansion, a relatively small number of oncologists-the oncology "core set"-appear to define the international research and treatment agenda, with unknown consequences for medicine and human health. Given the dominant position of oncology at the frontier of biomedical innovation, our project proposes to achieve a deep understanding of how the organization of oncology researchers shape how the system of oncology as a whole "thinks"-what it attends to and how it searches the space of possible diagnoses and treatments in pursuit of advance. Contributions from our effort will include conceptual and empirical contributions to medical sociology, science studies, as actionable insights for health and science policy. To gain this understanding, we will deploy a "metaknowledge" approach that draws on digital archives and new computational tools to trace research teams and collaboration networks, investigative and clinical technologies, the myriad recombinant elements under investigation-tissues, cells, genes, proteins, biomarkers-and the changing landscape of institutions that hosts them. Specifically, we will use machine learning, natural language processing, network analysis and Bayesian modeling approaches, linked with qualitative and historical investigation of oncology's emergence and development. Together, these tools and data will allow us unprecedented, intregrative insight into how oncology "thinks", how it followed its particular path of innovation, and the possible consequences of that path for biomedical insight and human health.