Paper long abstract:

Insurance has played a crucial role in performing solidarity, framed in the epistemology of 19th century statistics and its sociology of 'the norm', enabling the development of 'insurance society'. The statistical possibilities to calculate the risk for disease, unemployment and death made threats more visible, calculable and mutually shareable, resulting in an ex post solidarity.

The welfare state's solidarity - the stabilized result of these epistemological possibilities - is, however, not the only possible one. Today, the rapid development of predictive modelling and its widespread use of genomic algorithms and data-mining techniques have begun to challenge both the epistemology of statistics and the idea of solidarity that informed the insurance society.

In the last thirty years, internet databanks and the growing capacity to store data changed thoroughly the way insurers have to cope with data. 'Big data' demands a new epistemology: the 'norm' has to be replaced by the 'profile', the bell curve around the average has to be replaced by a data-structure in which there are only singularities in relation to other data.

In this paper we will demonstrate how an STS approach is particularly useful to study an insurance epistemology put to practice and challenged by the proliferation of 'big data' and predictive modelling. We will also explore whether solidarity as we have known it can persist in light of the emergence of an epistemology of 'big data' and predictive modelling.