by Dawn Nafus

About Dawn Nafus

Dawn Nafus is an anthropologist at Intel Labs. She holds a PhD from the University of Cambridge. Her research interests include experiences of time, beliefs about technology and modernity, and the anthropology of numeracies. She is interested in exploring new ways people might relate to their own data.


Data play a major role in orchestrating contemporary power relations through the collecting capacities of knowledge-generating machines. For media studies, “data” is an increasingly important term as information gathered and shared through personal and public communication channels becomes subject to new kinds of tracking, quantification, and analysis. Media technologies like the smartphone combine multiple functions of broadcast, storage, transmission, and capture, turning everyday experiences into information that can be measured, sold, or used for political claim making. But discussion of “data” didn’t start in media studies. The terminology is drawn from traditions of information science, sociology, and the natural sciences. In these disciplines, recorded observations combine to create frameworks for understanding social phenomena and the behavior of populations, whether birds, humans, or microbes. The systematization of data in these disciplines previously required a human agent to conduct the analysis. Today computational machines are just as likely to provide the source of empirical revelation, as software packages and backroom analytical engines perform commands that allow for large-scale composition and representation of data. This automated assessment of data, the large-scale amassing of insights that is sometimes referred to as “big data,” can have the effect of stripping important contextual cues and details from the activities being measured. For example, geospatial data are time-stamped, but the meaning being measured is often transported over time through space—a route, say—which likely involves sociocultural rhythms and meanings merely evoked by the data themselves, or perhaps erased entirely. Or consider graph/network data, which pretend they are time-free but are actually a snapshot of a specific moment in history. Simple attribute data that seemingly tell an eternal truth (fingerprints, blood type, or light from a star) are sampled once in time but are presumed to refer to an incontrovertible essence. Data have different types and functions depending on context; their meaning rarely remains fixed.