Big Data

5From Narratives to Systems: How to Shape and Share Data Analysis

In the previous chapters, we presented two components of the methodological paradigm required for knowledge production through computational analysis of big data: the ability to confer an epistemic value to big data through their evidential status, and the role of the empirical and autonomous transcendental in relation to the epistemic cultures in which it is mobilized, here the data sciences. In these two components, the idiographic and exploratory dimension returns as a leitmotiv that contradicts the widely held view of big data as a process of quantification and automation through data and computation. Our case study then highlighted that these components are indeed necessary, but not sufficient, to make the data intelligible: they are complemented by know-how that is itself guided by the pragmatic context of the epistemic project in which the data analysis takes place, and which varies according to the nature and value of the data.

More specifically, our case study presented a stabilized pragmatic context: that of a client who appoints a service provider for a codification project on the matter of customer experience. This stability allows the formation of habits that settle down to gradually constitute a method, iteratively integrated into dedicated software. In the words of Simondon (1958) talking about machines, the software is “a human gesture fixed and crystallized into working structures”. This case ...