Social Network Architecture

Social media (or online social networks) create equivalent of physical social interactions, intermediated by data network communication. It is a powerful tool to shape human experience and social habits. It is also a powerful tool to participate in many communities at the same time (which is equally marvellous and problematic).

There are at least two levels of influence in question. The visible level is what we habitually call the social network: logical structure of our communications. There is an identity (profile, account, persona) and there are structures of messages the identity interacts with (feeds, channels, groups, threads, topics).
Seemingly, the governance is defined on this level. People can or cannot manage communications structures. There are policies, administrative or moderation functions, even elements of self-governance, expressed by “group rules”.

But all of this is closely contained and predefined through the physical architecture of network: its topology and protocol.

Physical topology assigns roles to every “station” (effectively a computer of various kind) and sets allowed paths of connections between them. From our experience, we can see the three main topologies in action:

  • Facebook is an example of centralised topology. All users are equal, connecting to the network directly and there is only one privileged central point — FB servers (even if they form their own network, connecting server farms, we perceive them as one entity). Network is closed in the sense that every user has to be approved by the central authority, before being allowed to communicate.
  • Fediverse is an example of decentralised topology. There is no central node (instance, server) — we can see every instance as separate mininetwork. However, users can only participate in the communications via an instance. Governance and administration is defined and performed on the instance level. Protocol-wise we can say that fediverse is semi-open. The node (instance) can be added easily and the only requirement is protocol compatibility. This helps to provide a diverse list of instance software. But still, end-user has no other way to tap the network than to go through an instance (which is usually not easy to install on idividual station).
  • Freenet is an example of distributed topology. There are no special functions assigned to any node of the network and “end-user” participates directly in communication, solely through the sofware installed on their own system. Social network functions are performed by Sone, a specialised application, using Freenet protocol to communicate. Both programs are installed on end-user computer and, as Freenet protocol is separate from Soe, there is no obstacle to create alternative clients for social network.

Now, looking at the examples, we can see how strong is the influence of network architecture on the way people can (self)organize within the network.

  • In centralised networks, people are give a sandboxed social space with several predefined social structures (chat, newsfeed with comments, groups, pages etc.). It strictly depends on the policy of network owners, how much authonomy will they grant to users and how much openness for external communications they allow.
  • In decentralised networks, especially those using open protocol, there is hardly a way to implement ANY central policy, except by consensus of instance administrators. As we see in fediverse, they rather tend to keep the instance-to-instance governance in a form of confederacy, where there is no permanent “federal” authority. From an end-user point of view, however, the structure of fediverse is a federation, where significant part of people’s sovereignity is permanently delegated to the group of instance admins. While there are still ways to migrate between instances, user’s situation is similar to a citizen of a state: there is a “primary” “citizenship” assigned to a person and there is no way to become “a stateless” one — only to change the affiliation. In effect, all self-organissed social structures are always localised (attributed to certain instance) and there is always a difference between “local” and “visiting” participants.
  • In distributed networks, also assuming open protocol, there are no predefined (on architecture level) social structures. They mostly leave self-organisation in the hands of users. This kind of networks is less user-friendly in terms of the interface, but at the same time much more freedom oriented if we are looking for “a political system, where people could organize themselves however they wish and at the same time communicate, cooperate and coexist in a safe and constructive way”.

Considering this chief requirement, for all further modeling I will assume that we would use a distributed (fully peer-to-peer) network, using open protocol.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s