Written for Lawyers or Users? Mapping the Complexity of Community Guidelines

Abstract

Recent regulatory efforts such as the EU's Digital Services Act aim to increase transparency in mostly opaque content moderation practices of social media platforms. They encourage platforms to post information about what content is prohibited on the platform. But what kinds of platforms follow these best practices, and how readable is the information posted for average users? This paper introduces the Content Moderation Policies and Reports Dataset (COMPARE), a new data collection comprising content moderation policies of 132 of the most popular social media platforms for comparative analysis. We then use COMPARE to measure the complexity of 89 community guidelines, focusing on length, readability, and semantic complexity. We find that while the largest platforms are most likely to make guidelines available, they also tend to have the longest and semantically most complex guidelines. In addition, community guidelines seem to grow in length with new regulation. Our results suggest that it is crucial to look beyond the mere provision of community guidelines and to study how complex guidelines are for users.