The Authority of Big Data based Scores: Empirical Insights from the Use of Passenger Name Records

A growing body of literature is concerned with the question of how big data, the analysis of massive digital datasets with learning algorithms, affects individual lives and social order. The accounts of authors exploring concepts such as algorithmic regulation (Morozov 2014), governance by algorithms (Just and Latzer 2016), and design based regulation (Yeung 2016) stress that big data has a regulatory function: it is used for setting rules and controlling their implementation (Ulbricht et al. forthcoming). Yet we know little about the mechanisms of how big data turn into rules: How are results produced by big data analyses fed into rule-making procedures? What authority is attributed to them? How do machine-made results interact with human decision-making? My analysis provides first answers to these questions in the field of big data based mobility regulation. In the EU, the reservation systems known as passenger name records (PNR) represent one of the major use cases of big data in the public sector. The PNR are collected by various commercial systems on a global scale and introduce a new epistemology to mobility regulation: Firstly, the data sources are varied, they do not only encompass mobility data in the narrow sense, but also data about consumption, lifestyle and social networks of travelers (Hasbrouck 2015). Secondly, these ever growing data sets are mined by machine learning algorithms in order to detect patterns that are supposed to identify individuals who allegedly pose security risks: possible terrorists, drug dealers, unwanted immigrants etc. In short, passengers are not treated based on their identity and criminal history, but on data-based risk scores. Different risk scores finally lead to a different treatment: at airports, passengers are subject to different security controls and some passengers are denied access to planes. The German government is currently building the legal and technical infrastructure to start using the PNR for security checks. As the subsequent procedures are being developed, this analysis explores how decision-makers and security officials perceive the epistemological value of the PNR based risk scores and what role they attribute to the scores for decisions about security measures. Based on document analysis and interviews with German officials, I show that although to some degree they acknowledge the epistemological limitations of the PNR-based scores, they attribute a high authority to them. The scores are treated as imperfect, yet useful parameters for the regulation of mobility. The results are discussed with reference to the quantification and governance by numbers literature. I also highlight ethical and political implications and possible counter-measures.