Paper short abstract:

Data are supposed to talk about the study objects. What's their origin, transformation and use in complex systems sciences? Three different cases will allow us propose a classification of a different relation to study objects in this interdisciplinary field of research.

Paper long abstract:

In the last 10-15 years, computing power, as well as the Big Data era, have allowed scientists reach ambitious goals. Modeling and simulation are thus central to old and new sciences, especially in the field of complex systems sciences. But they produce a new kind of experience, which can have a different relationship to data. Particularly, in many complex systems sciences subfields of research, the study objects are unseizable, so they must be reconstructed through a long interpretative process. How do scientists using these tools obtain legitimacy and credibility, not only among their peers but also among policy makers? Before being able to answer this and other questions, I think we must answer another, more fundamental one: What is the epistemology inscribed in these sciences and what does it imply as for the treatment of data? Refusing the normative, overhanging viewpoint of philosophy of science all alone, I tried to describe the actual epistemologies of three subdomains of complex systems sciences by means of an ethnographic study of their practices. I chose a laboratory of biology, one of epidemiology, and one of geography. Data are supposed to talk about the study objects. What's their origin, transformation and use in complex systems sciences? I have used different theoretical frameworks to propose a classification of the main epistemological approaches that one can find in these fields in relation to their distance from the study object. This work can shed new light on a socio-epistemological approach to sciences.