Abstract

The datafication of society is said to be revolutionizing how researchers investigate the world, resulting in improved scientific communications, faster data integration and analysis, and more reliable outputs. Big and Open Data exemplify the newest frontier of empirical research, and scientific success in extracting knowledge from such objects is often hailed as demonstrating the power of (increasingly automated) inductive reasoning: science as the collection and interpretation of facts about the world. In this lecture, I critique this view of scientific inquiry, which is predicated on the existence and availability of documents of the world from which insights can be distilled. Building on in-depth, long-term studies of data practices in the biological and biomedical practices, I review the multiple failures of this form of empiricism, drawing attention especially to the intersection of moral and epistemic problems that this approach to research fails to address or even to recognize as significant, with severe implications for the reliability and the robustness of the knowledge thereby generated. The study of research practices calls for an alternative framing of empirical inquiry focused on the limitations of data as research components and the value judgements involved in using data as scientific evidence. Throughout my discussion, I pay homage to the seminal role played by members of the Cambridge HPS Department – and particularly Simon Schaffer, Peter Lipton and Hasok Chang – in shaping our field’s approach to research practices, including my own ideas on the epistemic role of data, experiments and inference.