Publications

This paper offers a legal perspective on the phenomenon of shadow banning: content moderation sanctions which are undetectable to those affected. Drawing on recent social science research, it connects current concerns about shadow banning to novel visibility management techniques in content moderation, such as delisting and demotion. Conventional moderation techniques such as outright content removal or account suspension can be observed by those affected, but these new visibility often cannot. This lends newfound significance to the legal question of moderation transparency rights. The EU Digital Services Act (DSA) is analysed in this light, as the first major legislation to regulate transparency of visibility remedies. In effect, its due process framework prohibits shadow banning with only limited exceptions. In doing so, the DSA surfaces tensions between two competing models for content moderation: as rule-bound administration or as adversarial security conflict. I discuss possible interpretations and trade-offs for this regime, and then turn to a more fundamental problem: how to define visibility reduction as a category of content moderation actions. The concept of visibility reduction or ‘demotions’ is central to both the shadow banning imaginary and to the DSA's safeguards, but its meaning is far from straightforward. Responding to claims that demotion is entirely relative, and therefore not actionable as a category of content moderation sanctions, I show how visibility reduction can still be regulated when defined as ex post adjustments to engagement-based relevance scores. Still, regulating demotion in this way will not cover all exercises of ranking power, since it manifests not only in individual cases of moderation but also through structural acts of content curation; not just by reducing visibility, but by producing visibility.