And the final speaker in this session at the Social Media & Society conference in Glasgow is Nadia Urban, whose focus is on algorithmic grooming and governance in social media cultures. Such processes require users to actively infer and internalise algorithmic norms: to learn the algorithm and bend it to their own ends.
But this cuts both ways: there is a conditioning mechanism in the everyday interaction between users and algorithms which also grooms the users. Algorithms produce social order in social media environments, while users are also active interpreters of these algorithm’s functionality; what is missing is a better understanding of how this works.
This understanding can be developed by exploring user-generated content that seeks to explain the functioning of these algorithms: Nadia examined this by gathering some 179 prominent YouTube videos purporting to provide information on the algorithms that govern YouTube’s own logic of content relevance. From this, she extracted various themes and orientations.
Iterative, affective feedback, reward, and punishment conditions users to anticipate algorithmic functioning; this recursively shapes how they design their own content, resulting in a cycle of algorithmic grooming. Users express this through eight key folk theories: these address the mechanics of the algorithm, optimisation metrics, prescriptive strategies, audience psychology, and various others.
These fall into three broad categories: temporal, aspirational, and affective orientations (which seek to understand the algorithm); normative and prescriptive orientations (which share such understandings, and thereby groom other users); and an orientation which sees the algorithm as an optimisable system.
These are united in positioning users as responsible for their own success or failure on the platform, however: the algorithm is positioned simply as a fact of life, towards which users must orient themselves by developing an algorithmic habitus, self-reproducing governance, and using their learning as a conduit for such governance.
This has three key implications: algorithmic literacy does not interrupt this conditioning structure, but merely deepens its operations; the mechanisms of algorithmic grooming themselves deserve greater attention, not just their outcomes; and folk theories themselves emerge as an important element of the governance infrastructure. Learning about all this does not necessarily empower users, but merely enrols them more deeply in the algorithmic governance process.
