The Societal Cancer of Network Clustering in Social Media
Some thoughts on the harmful effects of Clustering in Social Media Networks
Disclaimer: I am not linked in any way with NfX, whose views of the relative merits of clustering could be different from mine.
How does cancer actually operate? I am no expert in that. But from a quick look at typical reference texts, It appears to involve the conversion of normal homogeneously distributed healthy cells of a living body into tightly packed clusters of malformed cells. We don’t need to be experts to know the effects. The host body of the cancer gets ill. Very, very ill.
This is the analogy that makes most sense to me as characterising the effects of social media clustering on the living body of society. Social media has been widely adopted. It has become part of the fabric of society. Network clustering within it appears to be making society ill. Very, very ill.
Network Clustering is the technical term we use to describe the network characteristics of a social media network comprising many social bubbles. All of us who are in any of the established social media networks are members of at least one such bubble. This characteristic also affects many other media which might not be considered “Social”, the current media of this article included, unfortunately.
In network terms, each bubble is a cluster node of the network, like a network within a network.
Each member in a bubble is connected to every other member in the bubble, but not to anyone outside the bubble.
The world inside the bubble becomes the world of each member. Over time, members spending their time exclusively within a bubble can become progressively more isolated from the external world, each adding what they perceive has value in their isolated world.
So bubbles beginning with biased preconceptions will tend to increase that bias to ever more extremes, leading to extreme events, and bubbles have all manner of views leaning in all directions of bias. Even a bubble starting out as unbiased can become so.
Initial biases appear to be most commonly based on social deprivation, or lack of empathy due to excessive privilege, or other inequality, which at its root is financial.
Rich bubbles don’t mix with poor bubbles.
The problem is that clustering is known to maximise the profit that can be made from a network of users.
The network architecture is arranged as a star, with all traffic going through central servers, where this structure is again driven by being the most profitable. The central server need only handle the traffic at the central node of each cluster. Large volumes of traffic in our current centralised system infrastructure is expensive. The server resources have maintenance costs, all of which are in turn supplied at profit. So to maximise profit, the social media platform has a drive to reduce the volume of traffic, whilst not “Throwing the baby out with the bathwater”.
The connections generating the most profit are the ones drawing the most attention, i.e. the ones which either generate the most outrage (Between bubbles with opposing views), or the most affinity (Between members of the same bubble). In other words, the architecture is designed to pit us against one another, as tribes.
Modern algorithms, often based on Reeds Law (Positing maximum “Value” is achieved by clustering) have long learned that clustering is the way to most efficiently achieve maximum profit, because that is all they have been designed to do. They are not designed to recognise value as we would define it; the things that add value to our world.
Reed’s law loses validity in non-profit networks, compared with Metcalfs law which is known to accurately predict the market value of crypto-currency token networks, which generally have a distributed peer-to-peer architecture.
Reed’s law should perhaps be more properly identified as a law for characterising the profit attainable from a network, not the value of it.
That distinction does not appear to have been made in any references seen so far, and perhaps further explains some of the confusion over the validity of Metcalfs law, which does exactly what it claims; it quantifies the value of the network, not the profit.
Metcalfs law is fundamentally based on every member being connected to every other. Any deviation actually reduces the value of the network by Metcalfs law, whereas Reeds Law claims that clustering increases the value.
So, in a nutshell, social media networks for profit, are actually not that valuable. They might be profitable, but they are not valuable.
In fact they appear to be a cancer of society.
How do we fix it?
We create a non-profit alternative. One based on peer-to-peer architecture, where everyone has a clear balanced view of everyone else, warts and all, because like it or not, we all live together, and we need to get along.
Our non-profit alternative also has to go some way towards removing financial inequality, so as to remove biases and friction between people.
Not easy, as another noted characteristic of clustering, as well as ad-driven channels in general, is that it can be very difficult to get attention drawn to non-profit activities. There are no funds in a non-profit for advertising.
As yet, most people’s view appears to be confused over the value of profit.
But that will change. That is the challenge of projects like VRENAR.
