There are multiple ways in which a network can be represented as a body.
One can of course question whether it’s necessary to do that, but the real issue here is what affordances such representation allows.
Seeing a body as a network.
Feeling a network as a body.
…
Some parts of the body are undeniably, physically, networks. Brain, for instance, is one such network. Traditional Chinese Medicine represents the relations between the organs as a network. Nervous system, lymphatic system, cardio-vascular system. Everything within ourselves is a network. It thus makes sense to envision a network as a body.
Interestingly, to develop the concept of networks, we first had to look beyond ourselves. The science of networks emerged from studying bridges and then, later, electricity and communication. So it is also about connection between the bodies. (Ironically, in both tales of the networks’ origins, the city of Königsberg, now a disconnected enclave, plays a central role.)
Later, networks were used to represent information. Artificial intelligence is a network of switches, gateways, and thresholds. LLMs work with vector representations in multi-dimensional space where every token has a certain degree of proximity to another — also a network. There are text networks that represent co-occurrences between words and mimic our perception of text. There are knowledge graphs that represent ontological relations between entities. Finally, social networks have become so ubiquitous that it is impossible not to think about networks at least once a day.
…
Network itself can have a body. When it’s represented in a 3-dimensional space, it has a shape and a form. It can be touched, stretched, aligned, and manipulated. It translates the invisible into tangible.
Take this very text for example: what’s been written up to this point on this page. It is quite virtual and even if some parts can be felt, they do not really have a physical body that can be easily seen or experienced in a tangible way.
However, if that part of the text is represented as a network, it suddenly gets a physical body:
What can this body do?
It has a structure that is based on specific rules. The words are the nodes, their co-occurrences are the connections (within defined limited windows of attention following the landscape reading model). The bigger nodes are the ones that carry more information (they have high betweenness centrality).
There are two forces acting on all the nodes (Force-Atlas iterative layout algorithm). The repulsive force pushes the nodes apart from each other. The more connections the node has, the stronger is the repulsive force — this helps to physically distribute them across the graph. The attractive force pulls the nodes that are connected closer to each other, leading to clustering, which reveals the structure more effectively.
This is how the network gets its physical shape. That physical shape opens up the perspectives and ways of interacting with the text would not be available otherwise (the first four apply to the interactive version):
- we can stretch it and push it together
- we can touch it and turn it around
- we can zoom in and zoom out
- we can remove the nodes from it and observe what else emerges in their place
- we can discover the main pathways for meaning circulation (“network”, “represent”, “body”)
- we can see the clusters which are not so connected to the main pathways
- we can reveal the gaps between the clusters
- we can find that the network has limits
From the perspective of network analysis, we can optimize this structure, make it more resilient and diversified by distributing circulation more evenly across the different clusters, bridging the gaps, and exploring beyond the limits. These are the affordances that network representation allows and where it can be efficiently used to develop the text even further.
The fact that this text or this network now has a body makes it much easier to understand the abstract concepts of optimization and diversification. We are dealing with a physical shape.
One of the ways to optimize it is to make it more balanced (although this depends on the objective). One of the ways to diversify it is to make sure there are multiple centers, and not just two or three central nodes. Both could be beneficial to whatever structure we’re dealing with. Not only this text but also a social network. Studies have shown that networks with multiple centers are more resilient and adaptive, less susceptible to propagation of disease. This also applies to information, which makes it interesting to realize that an optimal, more balanced structure is also related to the notion of “health” (mental or physical).
But it doesn’t always have to be about optimization and health. We can also give other shapes to the network. We can turn it into something weird, unbalanced, full of tension. Perhaps then we enter into a more artistic realm (or creative thinking).
…
If this text / network is now a physical object, we can also approach it as a sculpture.
What kind of shape do we want to create?
This is what it looks like at this point:
The center of gravity is around the notion of network, which is the backbone of this whole structure, holding it together. If this network were a body, perhaps this is its pelvic region.
What if the body were to start moving and dancing? It would shift the center of gravity, bring attention to the (body) parts that are underexplored or neglected, find connections between them, explore the gaps and connections.
The specifics of this dance would depend on aesthetic preferences. Usually the most interesting movements are the least predictable ones:
There is a body part that relates to tangible manipulations, touches, forms, and shapes. Then there’s another part that talks about the cities, enclaves, and disconnectedness. They are not linked directly, so perhaps it could be interesting to explore the relations between them. Poetry thrives in the spaces where common sense no longer applies.
Tangible manipulation of a disconnected enclave. City of bridges. Four landmasses, separated by water. The structure of isolation. Proving that no single path exists while inventing a new theory. Disconnection as the condition that makes the search for a path meaningful. If everything were already connected, there would be nothing to discover.
If we embody the structure, rotate it, feel where the clusters are dense and where the gaps open up. The body (of the network) discovers the things the eye and the mind miss. When a node is moved, the edges tense up. It’s not visualizing the graph, but, rather, reasoning through moving around tension in the body (the way a sculptor reasons about volume by carving, not by calculating).
This body can also pretend that its “pelvic floor” doesn’t exist anymore. How would the movement route through that structure otherwise?
Structural isolation and poetic relations. Body that moves in a way that it shouldn’t. Carving through absence. The structure of isolation as the architecture of discovery. Exploring the edges, reaching for the extremities, shifting the center of gravity back to where it doesn’t belong.
Palpable understanding.
…
The ideas presented in this text are implemented in the cognitive variability framework used in InfraNodus text network analysis tool and also in the “embodied navigation” LLM skill I developed.