Poster Open Access
Over the last few years, a research object has been attracting the attention of quite a number of members of the scientific community of digital humanities and sociology: the study of algorithms, and more precisely of their impact on society. Examples include the study of the impact of work distribution algorithms (Uber), recidivism prediction algorithms (COMPAS) or recommendation algorithms (Youtube or Spotify). Both quantitative and qualitative methods are used for these studies. At the same time, we can find an increasing integration of the digital humanities in these studies, both to analyse the data used by the algorithms and their results.
However, an insufficient number of studies in this field have focused on the fabrication of these algorithms, either on the way engineers and more widely organizations design these programs that have the impacts that we now know a little better (their non-neutrality or discrimination to name only two).
On one hand, sociologists have gone to meet engineers to study their way of working: this is “humanities of the digital”. On the other hand, computer scientists have conducted experiments to open the "black box" of algorithms, notably through reverse engineering techniques: this is “numerical humanities”.
In order to overcome this dichotomy at the heart of the digital humanities, it appears important to intertwine these two mentioned aspects. This shall illustrate that fears about qualitative studies being supplanted by quantitative studies are not well-founded.