Journal of Free Speech Law: "Moderating the Fediverse: Content Moderation on Distributed Social Media," by Prof. Alan Rozenshtein

Just published as part of the symposium on Media and Society After Technological Disruption, edited by Profs. Justin "Gus" Hurwitz & Kyle Langvardt.

The article is here; here are the Introduction and the start of Part I:

Current approaches to content moderation generally assume the continued dominance of "walled gardens": social-media platforms that control who can use their services and how. Whether the discussion is about self-regulation, quasi-public regulation (e.g., Facebook's Oversight Board), government regulation, tort law (including changes to Section 230), or antitrust enforcement, the assumption is that the future of social media will remain a matter of incrementally reforming a small group of giant, closed platforms. But, viewed from the perspective of the broader history of the Internet, the dominance of closed platforms is an aberration. The Internet initially grew around a set of open, decentralized applications, many of which remain central to its functioning today.

Email is an instructive example. Although email is hardly without its content-moderation issues—spam, in particular, has been an ongoing problem—there is far less discussion about email's content-moderation issues than about social media's. Part of this is because email lacks some of the social features that can make social media particularly toxic. But it is also because email's architecture simply doesn't permit the degree of centralized, top-down moderation that social-media platforms can perform. If "ought" implies "can," then "can't" implies "need not." There is a limit to how heated the debates around email-content moderation can be, because there's an architectural limit to how much email moderation is possible. This raises the intriguing possibility of what social media, and its accompanying content-moderation issues, would look like if it too operated as a decentralized protocol.

Fortunately, we don't have to speculate, because decentralized social media already exists in the form of the "Fediverse"—a portmanteau of "federation" and "universe." Much like the decentralized infrastructure of the Internet, in which the HTTP communication protocol facilitates the retrieval and interaction of webpages that are stored on servers around the world, Fediverse protocols power "instances," which are comparable to social-media applications and services. The most important Fediverse protocol is ActivityPub, which powers the most popular Fediverse apps, notably the Twitter-like microblogging service Mastodon, which has over a million active users and continues to grow, especially in the wake of Elon Musk's purchase of Twitter.

The importance of decentralization and open protocols is increasingly recognized within Silicon Valley. Twitter co-founder Jack Dorsey has launched Bluesky, a Twitter competitor built on the decentralized ATProtocol. Meta's Mark Zuckerberg has described his plans for an "open, interoperable metaverse" (though how far this commitment to openness will go remains to be seen). And established social media platforms are building in interoperability with ActivityPub applications.

Building on an emerging literature around decentralized social media, this brief essay seeks to give an overview of the Fediverse, its benefits and drawbacks, and how government action can influence and encourage its development. Part I describes the Fediverse and how it works, first distinguishing open from closed protocols and then describing the current Fediverse ecosystem. Part II looks at the specific issue of content moderation on the Fediverse, using Mastodon as a case study to draw out the advantages and disadvantages of the federated content-moderation approach as compared to the currently dominant closed-platform model. Part III considers how policymakers can encourage the Fediverse through participation, regulation, antitrust enforcement, and liability shields.

[I.] Closed Platforms and Decentralized Alternatives

[A.] A Brief History of the Internet

A core architectural building block of the Internet is the open protocol. A protocol is a rule that governs the transmission of data. The Internet consists of many such protocols, ranging from those that direct how data is physically transmitted to those that govern the most common Internet applications, like email or web browsing. Crucially, all these protocols are open, in that anyone can set up and operate a router, website, or email server without needing to register with or get permission from a central authority. Open protocols were key to the first phase of the Internet's growth because they enabled unfettered access, removing barriers and bridging gaps between different communities. This enabled and encouraged interactions between groups with various interests and knowledge, resulting in immense creativity and idea-sharing.

But starting in the mid-2000s, a new generation of closed platforms—first Facebook, YouTube, and Twitter, and later Instagram, WhatsApp, and TikTok—came to dominate the Internet habits of most users. Today's Internet users spend an average of seven hours online a day, and approximately 35% of that time is spent on closed social-media platforms. Although social-media platforms use the standard Internet protocols to communicate with their users—from the perspective of the broader Internet, they just operate as massive web servers—their internal protocols are closed. There's no Facebook protocol that you could use to run your own Facebook server and communicate with other Facebook users without Facebook's permission. Thus, major social-media platforms are the most important example of the Internet's steady, two-decades-long takeover by "walled gardens." …