Classifying Politics in Online Communities

• In Event: The Limits of Free Expression on Social Media

Abstract

Whether specific media texts are "political," "about politics," or otherwise politically relevant is a question of interest to political communication researchers, governments and digital platforms, and individual media users. Determining whether media texts are politically relevant has implications for how users respond to these texts (Coles, 2020), how researchers measure users’ exposure to politics (Guess et al., 2019), and how and when governments and platforms apply policies regarding political speech (McGregor, 2019). Yet people disagree on what constitutes the political (Coles, 2020; Eliasoph, 1998; Fitzgerald, 2013; Görtz et al., 2023; Mansbridge, 1999) and thus whether specific media texts fit the definition of politically relevant media (Coles, 2020; Settle, 2018; Vraga et al., 2016). The consequences of this disagreement are especially stark when governments or platforms must determine the political relevance of media texts since these decisions may affect whether and how such texts are made available to potentially millions of media users. So how do people whose duty it is to act as arbiters in this regard classify media texts as political or not?
The current study explores this question in the context of online communities. Internet users may discuss a variety of topics in online communities; however, some communities have policies regarding the posting and discussion of political content (Graham, 2008; Graham et al., 2016; Jackson et al., 2013). Because people may disagree on what constitutes the political (Coles, 2020; Eliasoph 1998; Fitzgerald, 2013; Görtz et al., 2023, Mansbridge, 1999), online community moderators are tasked with classifying content as political or not and applying community rules (Edwards, 2002, Gibson, 2019). This study investigates how moderators of online communities with such rules make these decisions and what the consequences are for how we understand media content that may be political in nature. What cues in or features of online community content are important for determining whether texts are political? How do moderators navigate disagreement between themselves and with community members regarding the application of these policies? And how does the enforcement of these policies affect the community in terms of its membership and the discussions that go on within it?
This study conducts a phronetic iterative analysis (Tracy, 2019) of in-depth interviews of Facebook group and subreddit moderators (N = 34). It examines the processes through which moderators identify potentially political content, the user and content characteristics that moderators associate with politics, and what effects moderators believe that these policies and the content they seek to address have on their communities. To maximize exposure (Small & Calarco, 2022), interviews lasted an average of 90 minutes. Moderators were from online communities around topics including popular media franchises, sports/fitness, finance, occupations/industries, local communities, religion, humor, parenting, education, and the arts.
This study serves as a microcosm to explore the patterns and problems that may exist in a wide range of different contexts in which media texts must be evaluated for their political relevance, such as by government agencies or social media platforms. For example, social media platforms may set and enforce rules regarding political advertising (McGregor, 2019). As with the online communities in this study, these platforms must develop processes to identify which advertising content falls under the "political" umbrella in order to enforce these policies. Implications are considered for how users engage in online communities with political content policies, the role of moderators as arbiters of what constitutes political content, and the difficulty that online platforms and governmental organizations may have in applying policies regarding political speech.

Author

• Stewart Matthew Coles, University of Illinois at Urbana-Champaign