Field Notes 1: Online Formations

Welcome to the first public update from the Summer of Protocols program. Over the past few weeks I’ve been reading about how people gather online, trying to tease out a clear definition of a Swarm. No such luck so far, but hopefully getting closer.


The internet is a dark forest; there are odd creatures emerging from the fold. Some of these creatures are lone and silent travelers, like the lurker. Others travel in packs, and cooperate, sometimes consciously and other times dubiously. When their actions create harmonies or dissonance, Online Formations emerge.

Depending on the protocols that shape their behaviors, they could be Fandoms, Online Communities, or something else. Interestingly, it seems these Formations are not built from individuals, but from actions that coalesce through active intention or algorithmic pressure. Let’s review the most common ones and understand them better.

Differentiating between Online Actions and Agents

Social Protocols are a codified set of procedures that we perform. To uncover Social Protocols online, like those of Swarms, we can first look at what those codified procedures could be composed of, and that means understanding Online Actions.

An Online Action is an action performed Online by an Online Agent

There are two core distinctions in the definition that are worth discussing. First, the concept of Online. Second, the concept of an Online Agent.

Online is demarcated by two characteristics: (1) mediated by software, and (2) broadcast to a network. In our case, we will be looking at Online Actions on social media applications. Each action is broadcasted to a wider network participating in the application or the network of applications.

An Online Agent will be defined as a human-controlled entity within the networked software. An Online Agent could be ‘acting’ with one or multiple profiles. For example, an ‘Alt’ and a ‘Main’. An Online Agent may be controlled by one or multiple human beings in coordination. For example, a group of people may have joint control over a profile. The agent may also be enhanced by software (e.g. bot assisted). In all these cases, the Online Agent is assumed to be a single entity.

The most common types Online Agents on social media applications include:

  1. An individual (like you or me)

  2. A bot (controlled by an individual or group)

  3. A company (e.g. advertiser or marketing account)

Disembodied Collectives of Actions

Prior to the ubiquitous internet, our lives circled around holistic bodies. We could communicate in one channel at a time. For example, we were at work, on the phone, or listening to the radio. A limited amount of fragmentation would happen if we called home from work, splitting our mind from our physical presence. Now, many of us are at least splinched; half of our mind is on media--TikTok or Twitter--and the other is on whatever real-world task we are completing.

We are disembodied. In a literal sense, we leave our physical bodies behind and fall into hypnosis. Our eyes track the pixels of the screen as we teleport between spaces online. Twitter, Discord, Whatsapp, Telegram; back to Twitter. Many of us spend more hours onscreen than sleeping; we, the terminally online, march toward our dopamine dealers.

In that transformation, as we open our laptops or take out our phones, we are no longer one self. Instead, we become pixelated, datafied, and fragmented into the various personas we enact online. Moreso, we become a composition of machine-legible ‘actions’ in each application’s interface. On Twitter, for example, we have our avatar and the ability to post, like, reply, and boost.

These days it’s hard to see a person as a person in the classical sense. Better to think of a person as a world grown on the substrate of a biological body. A World of You. One that is home to a congress of native and adopted entities: ego states, creative demons, the persistent roles of all the worlds you are a part of.

Ian Cheng, “Worlding Raga 6: World To Live”

From the outside, we do not see humans or agents, we see collections of actions. Unlike our real-life embodied selves, Online Agents, as digital entities, emerge into machine and user perception through consistency in broadcasted actions. The algorithm doesn’t ‘see us’ until we persistently participate. This is the reason why content producers need to publish at such high volume and frequency. We “tweet into the void” until users begin to see us as a cohesive identity narrative. This is in contrast to meatspace interactions where actions follow from our perception of an individual.

In many cases, the amalgamation of actions creates groups even before we perceive individuals. We will call these action-collective creatures Online Formations, or, as Mizuko Ito (Mimi) called a similar theme: Networked Social Forms. The grouped actions, some in harmony and others which are dissonant, produce complex some awe-inspiring coordination that no institution would be able to carry out without heavy bribes and threats.

A Diversity of Online Formations

When we shift our focus from ‘networks of individuals’ to ‘networks of actions’, we can reframe the types of online organisms we contend with. For example, we can classify persistent, reliable, attuned, and strongly weaved actions as an ‘Online Community’. While actions centrally focused on a celebrity can be an ‘Online Fandom’. In both cases, mapping an individual to a specific locality matters less than the locality of the broadcasted action. Another example is ‘Trends,’ which are weakly aligned actions centered around a specific topic.

An Online Formation is a collective of aligned Online Actions

Alignment of Online Actions can vary. It can be loose, like a random grouping of people with little in common. Or it can be highly centered, for example around an individual (e.g. a specific influencer or celebrity), a topic (e.g. news event), or around a mission (e.g. target outcome). We will evaluate this alignment by using the concept of Centrality:

Centrality is the measure of the narrative focus of an Online Formation

A narrative focus relates to how the story of an Online Formation is concentrated on a specific entity: people, mission, meme, topic, or event.

Classifications can be further detailed based on the focus. For example, a large Online Formation with a high degree of centrality can be focused on a celebrity, like the Swifties Fandom, or a public objective, like the Effective Altruism movement.

When centrality is low, coordination is more difficult. The reason is that call-to-action is not anchored on an entity--topic, person, or other--but on memes. As a result, those looking to influence Online Formations with low centrality require mastering the attunement of disparate actions at scale. This is usually done through targeted campaigns of those within the formation who have relatively higher degrees of centrality and legitimacy: micro-influencers.

Closing Thoughts

  • There are additional implications to this model of understanding Online Formations. Specifically, the models for accountability, interaction, and intervention require a different approach. We will explore this in later research as we continue to understand these creatures in more detail.

  • Other variables we did not explore include permanence, valence, rhythm, and control structures. We will potentially come back to these, and their relationship to narrative alignment in later posts.

  • Some of the more interesting dynamics of Online Formations are their dynamism. They can transform from one shape to another. For example, we might say: An audience becomes a swarm when it is networked enough to execute a decentralised raid upon a specific trigger.

Subscribe to rafa
Receive the latest updates directly to your inbox.
Mint this entry as an NFT to add it to your collection.
This entry has been permanently stored onchain and signed by its creator.