Over the course of the next few months we’ll be treating this newsletter more like a Substack, sharing drafts from Chand and Ethan’s forthcoming book “A Field Guide to Social Media,” coming next year with MIT Press. This is the first chapter of several.

Frameworks for Understanding Social Media

by Chand Rajendra-Nicolucci and Ethan Zuckerman

Do you have thoughts about this chapter? We’d love to hear your feedback in the comments.

When we set out to write this book, centered around important patterns of social media—what we are calling “logics”—we realized we needed tools to help uncover and describe those patterns. Over time, we’ve developed a set of analytical tools that inform our thinking and that we think are useful for anyone trying to understand social media. We will present them here as background for the analysis you will find in this book and as a reference for future analysts of social media.

What is Social Media?

To analyze something, you need to define it. There have been many attempts to define social media over the years. Some definitions are quite specific, like this from researchers danah boyd and Nicole Ellison: “web-based services that allow individuals to (1) construct a public or semi-public profile within a bounded system, (2) articulate a list of other users with whom they share a connection, and (3) view and traverse their list of connections and those made by others within the system.” Others are fairly vague, like this from researcher Joshua Habgood-Coote: “a social media site is one whose primary service is to allow users to hang out, where the majority of content shared on that site is generated by users.” There’s no widely accepted definition, largely because as the internet has grown, almost everything online has a social aspect to it. Reasonable people can and do disagree over how to draw the lines. As a result, it’s tempting to just say, “I know it when I see it.” However, it’s worth trying to draw the lines in the context of this book. Here is our attempt at a definition: A social media platform is a bounded digital space whose primary purpose is to enable many-to-many communication.

Let’s step through the key parts of the definition. First, “bounded”: we included this to capture the idea that a social media platform is a finite system. The entire digital public sphere is not a social media platform (nor is the blogosphere, the Fediverse, or email). Second, “primary purpose”: we included this because today many websites include social features, even if their primary purpose is not to be a social media platform. For example, the New York Times has comment sections on many of its online articles. But it wouldn’t make sense to call the New York Times a social media platform. Or consider dating sites: Tinder has many social features but its primary purpose is matchmaking. Third, “many-to-many communication”: this captures the idea that social media is about facilitating networks of people to interact, which is sort of obvious, but by emphasizing “many-to-many” it helps us exclude sites like Substack which are primarily organized around one-to-many communication. Another way of thinking about this is: would you use the platform if only one of your connections was there? On Substack you might, since the value of Substack is largely tied up in the relationship you have with individual writers. However, if Twitter had only one of your connections, you probably wouldn’t use it, since the value of Twitter is largely tied up in the network of people that use it, not individual users.

Chat platforms are a tricky case for the many-to-many requirement—we think it makes sense to think of them as fulfilling multiple roles. If you’re using WhatsApp solely for one-to-one communication, it’s closer to a communication utility like SMS than social media. However, once you start using it for group conversations, it makes sense to think of it as social media.

Our definition is not the “right” way to define social media—it’s just our attempt at coming up with something useful for this book. Plenty of other folks have given helpful definitions, many of which we drew inspiration from.

Analytical Frameworks

There are a number of analytical frameworks for social media. (See, for example, this and this.) We will present a few that we’ve developed in the course of our own research that have been consistently helpful and will appear throughout this book.

TBG (Technology, Business Model, Governance)

TBG outlines three main axes of analysis for social media platforms: technology, business model, and governance.

Technology refers to the architecture and affordances of a platform. Is the platform decentralized? If so, what protocol does it use? Is content text-centric, like on Twitter? Or video-centric, like on TikTok? Are accounts public by default? Is the platform organized around bi-directional relationships (like Facebook Friends or LinkedIn Connections), unidirectional relationships (like Twitter or Instagram followers), membership relationships (like Subreddits or Discord), or an algorithm (like TikTok or Instagram Reels)? Can you re-post or quote post content? What metrics does the platform display? What does the platform’s algorithm prioritize?

Business model refers to how a platform pays the bills. Advertising? Subscriptions? Donations? Crypto tokenization? Government support? There’s currently little variety in business models for social media: most platforms are supported via surveillant advertising. But some of the most interesting ideas for different approaches to social media begin by imagining changes to a platform’s business model.

Governance refers to who owns the platform and how decisions are made about the platform. Is it owned by a billionaire unconstrained by a board of directors? Is it a public company? Is it a private company? Is it owned by a foundation? Is it owned by the government? Is content moderation handled by professionals, volunteers, algorithms, or some combination of the three? Is there an advisory board? Does it have binding or non-binding power?

We’ve found TBG often reveals key factors that underlie the nature of a platform and is quite useful when comparing and contrasting platforms.

Owners/Rooms

Another way of understanding social media platforms is by placing them within a 2x2 matrix of owners and rooms. Does a platform have one owner or many owners? Is it made up of one room or many rooms? This matrix is particularly helpful for analyzing the governance and architecture of a platform. To illustrate, we will apply the matrix to Reddit, Twitter, and Mastodon.

Reddit has one owner—the Reddit corporation—and many different rooms—a vast array of subreddits each with their own purpose, rules, and norms. Because Reddit has one owner, it’s difficult for users to pick up and start again, or to have much formal influence on the central governance of the platform. This was illustrated in the spring and summer of 2023 when Reddit made a number of policy changes that were unpopular with users. Users’ only recourse was to “protest” by shutting down their subreddits for a period of time. At the same time, because Reddit has many different rooms, a diverse array of communities can grow, flourish, and co-exist, each with their own structure that’s layered on top of the floor provided by Reddit. A Redditor can easily create her own room, with her own topic and rules, and try to build a new community.

Twitter has one owner and one room. It is owned by the X corporation, whose primary shareholder is Elon Musk, and consists of a single big space with a shared set of rules and norms. Because Twitter has a single owner, disgruntled users face the difficult choice of putting up with decisions they disagree with or starting over on a different platform that lacks the network and audience that Twitter provides. Because Twitter is made up of a single room, content can spread quickly to reach massive audiences. That has been critical for activists in launching social movements like #MeToo and #BlackLivesMatter and has made Twitter a nexus of culture and politics. But it has also encouraged harassment and made the platform a target for spammers. Another way of framing this is that in a single big space it’s hard to preserve context (the academic term of art is “context collapse"). Context collapse means that conversations often bump into each other, which is good if you’re trying to spread an idea or message, but can also lead to misunderstandings and conflict.

Mastodon has many owners and one room. Mastodon has thousands of different server administrators who run different instances of the network. Those instances are linked together to essentially form one big space. Though each server may have specific rules for their users, everyone is contributing to a single, shared conversation that’s happening on Mastodon. Because Mastodon has many owners, it’s easy for someone to switch to a server that matches their preferences. For example, if someone disagrees with their server’s moderation policies or finds that its technology is unreliable, they can switch to a different server. Because Mastodon is one big room, it has many of the characteristics found on Twitter that are downstream of virality and content collapse.

We’ve found the owner/room matrix to be a useful tool for understanding the implications of a platform’s governance structure and architecture.

Friends, Followers, Members, or the Algorithm?

We previewed this framework in our earlier discussion of TBG. Answering a single question about how a platform’s social graph is organized—Friends, Followers, Members, or the Algorithm?—can offer powerful insight into how a platform’s affordances and architecture affect its dynamics.

When a platform’s social graph is organized around “Friend” relationships, or bi-directional connections, like on Facebook or LinkedIn, it results in a more intimate space, with people tending to share content related to their personal life.

When a platform’s social graph is organized around “Follower” relationships, or uni-directional connections, like on Twitter and Instagram, it results in a space that feels more public, performative, and engaging, with people often sharing content with the goal of growing their audience and gaining influence. These platforms tend to feature metrics of attention, helping to quantify the audience people are reaching.

When a platform’s social graph is organized around “Member” relationships, like on Reddit, it results in a collection of smaller, more insular spaces, each with their own purpose, rules, and norms. This lessens the risk of context collapse while making it more likely that echo chambers will form. It also means users are often more invested in the platform because they feel a sense of ownership and belonging in relation to the groups they are a part of.

When a platform’s social graph is organized around an algorithm, like on TikTok, it results in a space similar to what’s found on follower-centered platforms. However, algorithm-centered platforms tend to be even more public, performative, and engaging than follower-centered ones as content creators are aware that instead of producing content for a persistent audience, like on follower-centered platforms, they are producing content for audiences assembled in real-time by an algorithm that judges content based on its potential for engagement. These platforms are often entertainment-heavy and bear many similarities to television.

Friends, Followers, Members, or the Algorithm? is one of the first questions we ask when assessing the fundamental causes of platform dynamics.

Filters and Rankers

Most platforms revolve around a feed—understanding how that feed is constructed is critical to understanding the platform as a whole. We’ve found it helpful to think about feed construction by breaking it down into two types of algorithms: filters and rankers. Filters narrow the pool of content eligible to be put in a feed. Rankers then order the pool of eligible content. 

Some examples of filters: posts from the last 24 hours, posts that are positive, posts that are about sports and music. Some examples of rankers: posts sorted reverse-chronologically, posts sorted based on popularity, posts sorted based on what a user is most likely to engage with.

Thinking about feed construction in terms of filters and rankers can help simplify what is sometimes presented as a single, impossibly complex algorithm. At its core, even the most complex feed construction algorithm is just deciding what posts it could show to you, and then sorting those posts.

VLOPs and VSOPs

The European Union coined an acronym, VLOP (Very Large Online Platform), to label the massive gatekeepers that its new digital regulations apply to. We coined a complementary acronym, VSOP (Very Small Online Platform), to label platforms that operate on a humbler scale and are often overlooked. Some of these platforms are not, in fact, especially small on an absolute scale—instead, they are very specialized and relatively small compared to VLOPs: consider Letterboxd, which serves movie lovers and has a few million monthly users. VSOP can mean “Very Specialized Online Platform,” “Very Small Online Platform,” or both.

Knowing whether a platform is a VLOP or a VSOP can help to frame the challenges and opportunities it presents, as VLOPs and VSOPs often demand different approaches.

VLOPs like Facebook, TikTok, and Twitter operate on a global scale and are usually what academics, journalists, and policymakers are referring to when they discuss social media. These platforms have hundreds of millions (or billions) of users and thousands of employees. They shape culture and politics around the world. In contrast, VSOPs like Letterboxd, Front Porch Forum, and An Archive of Our Own have anywhere from a few dozen to a few million users and usually have a specific purpose and a particular set of rules and norms. For example, Front Porch Forum serves neighborhoods in Vermont and has rules and norms about civility that reflect the region’s values. Because VSOPs aren’t for everybody they tend to be overlooked in conversations about social media. However, they serve important functions that aren’t provided by VLOPs, making them an essential part of the digital public sphere.

Because VLOPs and VSOPs are so different, what may be true for one category might not be true for the other. We’ve found keeping these categories in mind helpful for making our thinking about social media more precise, and in particular, for ensuring the ecosystem of VSOPs is not left out of our discussions.

All Models are Wrong

We provide these frameworks and our definition not to advance a grand theory of social media but to give you insight into how we’ve approached our analysis and to provide tools that you may find useful in your own thinking. We often mix and match these tools and sometimes ignore them entirely depending on the question we are asking. It's important to keep in mind George Box’s famous aphorism: “All models are wrong, but some are useful.” These tools can help us piece together insights about social media but there are always exceptions and limitations to their application.