Politics and epistemology of big data: a critical assessment

In this paper I will discuss Big Data as a suite of new methods for social and political research. I will start by tracing a genealogy of the idea that machine can perform better than human beings in managing extremely huge quantity of data, and that the quantity of information could change the quality of the interrogation posed to those data. In the second part of the paper I will analyse Big Data as a social and rhetorical construction of the politics of research, claiming in favour of a more detailed account of the consequences for its progressive institutionalization. Without a serious methodological assessment of the changes that these new methods produce in the scientific epistemology of social and political sciences, we risk to underestimate the distortive or uncontrollable effects of the massive use of computer techniques. The challenge is how to avoid situations in which it is very difficult to reproduce the designed experiment, and it is arduous to explain the theories that can justify the output of researches. As an exemplification of the problem I will discuss the work on emotional contagion led by Facebook and published on PNAS in 2014. Until now it was difficult to explore all the Big Data projects’ consequences on the perception of human intelligence and on the future of social research methods. The vision that there is no way to manage social data than to follow the results of a machine learning algorithm that works on inaccessible, epistemologically opaque and uncontrollable systems is rather problematic and deserve some extra consideration