Predictive data analytics: a neo-liberal snake in the grass?

Governments worldwide habitually deploy data analytics tools to streamline the task of governing, ideally leading to more effective, efficient, and economical processes. Governments’ use of data analytics tools can be clustered into three categories: descriptive analytics (e.g. identifying voting irregularities, detecting welfare fraud); prescriptive analytics (e.g optimising use of public transport resources, pooling information from multiple public agencies to assist emergency responses); and predictive analytics (e.g. forecasting post-earthquake tsunamis, identifying spatial areas with elevated risk of suicides).

Inevitably, problems have arisen with governments’ use of data analytics, mostly caused by political agendas and public sector management failings. But the frequency and gravity of the problems seem to peak when two conditions are present: (1) when analytic tools are used to predict; and (2) when those predictions are used by governments to support the exercise of their legitimate coercive powers