Big data analytics, the social graph, and unjust algorithmic discrimination: Tensions between privacy and open data

This paper examines threats to privacy and anonymity accompanying underlying technical changes related to the Internet of Things and big data analytics. Governmental and corporate entities have focused on creating sophisticated graphs of citizens' social media connections and aggregating these data with other sources, such as physical location, biometric data, public records, and online search habits. The increased instrumentation and tracking of natural and social processes and resulting availability of real-time user data has enabled sophisticated user modeling and new algorithms to mine, model, and personalize these data. Even when data seem innocuous or have been anonymized/de-identified, analysis can lead to inferences, reidentification, and subsequent informational harms. The algorithms and machine-to-machine (M2M) communication employed in big data analytics may disadvantage certain individuals or groups. On the other hand, big data analytics has immense potential to enhance public welfare - leading to fairer hiring decisions, government transparency, energy conservation, participatory governance, and substantial advances in medical research and care. This paper first addresses how these developments may create unjust power differentials used by one group to diminish the opportunities of another, threaten to destroy anonymity when engaging in public affairs, and hinder public participation in democratic discourse. Legal and policy barriers to citizen privacy protections and the tension between privacy and open data (for the public good) are then identified and discussed. This paper intends to stimulate a deeper policy debate about how to protect citizens from informational harms and unjust discrimination while opening public access to large data sets required for participatory governance, health advances, and environmental protections.