Michael Loy (University of Cambridge), Quantifying the unquantifiable: Bayesian analysis of uncertain economic networks

Over recent years, archaeological studies of the economy have adopted new digital tools and models from the social sciences. One such approach has involved using formal Social Network Analysis (SNA) to represent sites and the complex interactions which took place between them. However, SNA as conducted with regard to its roots in sociology is often practised upon a closed and complete dataset; archaeological material can be described as selective and partial at best. This paper will therefore consider the challenges of using quantitative archaeological data within computer models, and propose a new direction based around the concept of ‘uncertainty’ and Bayesian statistics. Specifically, the case study of a probability- and network-based analysis of material data from the Aegean basin 700-500 BC will be used. By these means, three levels of uncertainty in the archaeological record will be discussed: the uncertainty associated with a context’s date, the question of how representative a dataset is, and the size of datasets from different sites relative to one another. Furthermore, this paper will propose as another solution the potential of Open Data in quantitative studies of the ancient world. By collaborating on shared databases - particularly across large geographic areas which encompass many sites - quantitative archaeological research into much larger and complex economic systems is made more possible, and provides new opportunities in methodology and practice.