Shedding Light on Shadowbanning

Abstract

For social media services, content moderation — the policies, practices and tools providers have in place to address online abuse — is their primary tool to defend users from everything from mis- and disinformation to hate speech and online extremism. As the list of potential abuses continues to grow, online service providers have commensurately built new systems to enforce their content policies, largely through removing or reducing the visibility of potentially abusive content. But social media services don’t always inform users when they are moderating their content: a user’s comment may appear normally to themselves but be hidden to others; a user’s handle may disappear in search; a user’s original content may get ranked so low in a recommendation algorithm that, for all intents and purposes, it becomes undiscoverable. On the internet, people refer to this broad range of undisclosed content moderation actions as shadowbanning.The goal of this paper is to bridge the gap between social media companies, end users, and the broader public in how each understands shadowbanning in order to help social media companies better manage disclosures to users about content moderation. We aim to do this by critically examining three questions: (1) What is shadowbanning? (2) Who is affected by shadowbanning? (3) What larger effects does shadowbanning have?The first section of this paper defines the term shadowbanning, reviews specific shadowbanning practices in areas such as search, comments, and recommendations, and discusses the reasons social media companies give for why they engage in shadowbanning. The second section looks at which groups may be disproportionately affected by shadowbanning and describe how users diagnose and respond to their own shadowbanning. The third section explores the consequences of shadowbanning on individuals, groups, and society at large. The fourth and final section recommends three ways social media services can mitigate the harms of shadowbanning: sharply limiting the circumstances in which they shadowban, “shedding light” on shadowbanning by publishing their policies about when they shadowban and data about content or accounts subject to opaque content moderation, and creating mechanisms to allow researchers to learn more about the potential harms shadowbanning may cause. URL: https://cdt.org/insights/shedding-light-on-shadowbanning/

Topics & Concepts