Paper short abstract:

This paper examines the expansion of big data, predictive modelling& data driven technologies into social service delivery. By drawing on the framing rhetoric around these data technologies, I show how dubious numerical expertise is justified in making decisions about marginalised populations.

Paper long abstract:

Welfare bureaucracies are undergoing a transformation with various administrative procedures that were once fully based on human inputs transitioning into data driven, automated systems. This shift is underpinned by discursive work, claiming and repurposing new streams of digital data as the basis of policy making and as valid evidence from which actionable insights can be gleaned. Justifications for the shift draw on claims of an epistemic break where large scale data linking and aggregation is seen as offering a new gold standard of knowledge and ‘a higher form of intelligence... that can generate insights that were previously impossible, with the aura of truth, objectivity, and accuracy’ (Boyd and Crawford 2012: 663). I argue that the conceptual apparatus of ‘dataism’ from Critical Data Studies is useful to explore how such belief in the ‘objective quantification and potential tracking of all kinds of human behaviour and sociality through online media technologies’ is mobilised in social service delivery (Van Dijck 2014:206). Using case studies of the introduction of data driven decision making in sensitive social policy fields such as child protection and the pre-emptive, algorithmic targeting of social services, I empirically demonstrate the discursive slippages, conflations and rhetorical choices deployed to legitimise data science expertise within government. Novel digital proxies for ‘offline’ social activities are created using patterns from digital trace data as a stand-in for human behaviour. A fundamental problem emerges as this assumes that life processes that can’t be expressed as digital data &translated into a machine readable template don’t count.