67

‪Algorithmic recommendation has been the subject of debate ever since the first texts that popularised the term “filter bubble”. The hypothesis of a user’s confinement is typically associated with the biases engendered by an increasingly advanced personalisation of the content offer, whether it be search engines, social networks or cultural services. In terms of information, the question raised is that of access to a plurality of opinions in a democratic society. Regarding online cultural services’ content offer, the thesis of algorithmic bias and of confinement in certain categories of content, presumably to the detriment of cultural diversity, has never been fully tested in empirical work, even though it is widely discussed. Establishing appropriate forms of regulation, moreover, entails the use of objective tools in order to confirm or alleviate fears linked to the filter bubble hypothesis.‪