worldbuilding

Thoughts are world-building and the mind is a means of travel between worlds. 

Deleuze insists that philosophy creates concepts. This is, to me, far more exhilarating than the Hegelian notion of philosophy as dialectic because it treats philosophy as a creative act of construction instead of one of discovery and recognition. It taps into something that I’ve long felt regarding worldbuilding, something every child with a sense of awe and play knows intuitively. Concepts are worlds. Each concept is a constructed multiplicity defined by its internal components and their relationships, and invented in response to problems.  

In Deleuze’s account, concepts sit on a plane of immanence, which sets the conditions under which concepts can exist and relate. The plane is not a theory or a worldview, but a prephilosophical field that philosophy must lay out each time it thinks. 

Concepts here are a kind of virtual world that one can go and interact with, but not the only worlds. A world, in my view, is a working reality. Worlds are hyperabundant. They are the set of objects and relations being treated as real to move through life, and what the mind maintains so that perceptions don’t dissolve into noise. The nuance here is that a world is not a representation of reality, but rather a specific functional arrangement that holds long enough to make sense and create movement. Deleuze’s concepts do this, carving out a terrain of sorts. They make a way of living, thinking, or acting possible that was not before. To create a concept is to construct a small, inhabitable world that can be entered, tested, extended, or abandoned when it no longer holds, or even if it's just not interesting. 

Humans generate worlds through a kind of cybernetic loop: sensing → memory → prediction → action → correction → sensing again. This loop is running whether we reflect on it or not and most of what it produces seems self-evident. While minds are clearly “world-generators” in this sense, they are not the only ones. A scientific field produces a world, deciding what to count as a phenomenon, a variable, a real effect, and so on. It decides what is a legitimate instrument. Bureaucracies produce worlds by making people appear through forms and categories. Platforms produce worlds by training attention on certain objects while starving others. 

These aren’t metaphors, though they can be read that way. A system that senses, remembers, predicts, acts, and updates generates a stable reality for its participants independent of whether or not anyone in the system fully intends that. Deleuze and Guattari are again useful here because they refuse the picture of a world sitting quietly “out there,” waiting for representation. They treat worlds as produced by assemblages, arrangements of bodies, habits, tools, codes, institutions, and desires. A world holds when an assemblage holds and dissolves when the assemblage breaks.

The language of Deleuze and Guattari is creative and psychedelic in a way that much of philosophy is not. It matches my internal notion of concepts as inhabitable places that one can venture into while experiencing all of their instability. Territories form and crack, sometimes being completely reorganized spontaneously through new information. I’ve been tempted to follow this into a sort of mysticism, but there is no need because the point is practical: worlds are maintained structures. 

One of the risks of this framing is drifting into the thought that anything can be a world and that all worlds are equal. I don’t think this is the case because worlds are corrected by resistance. A world becomes stable by surviving contact with whatever doesn’t obey it. Some expectations fail even when you want them to succeed and vice versa. You can call this “reality” or “constraint”, but the label matters less than the function, which is pushback. A world that doesn’t accommodate pushback becomes fragile, and though it may feel coherent to those inside, it will fail when it encounters enough interventions it can’t absorb. That failure is information. 

Objective reality is often imagined as a single world that everyone should be able to see if they would remove their biases and accept facts. That is obviously aspirational, though it is not a description of how a common world is achieved. The shared world emerges as the overlap of many individual worlds. It forms where many world-generators, with different vantage points, are forced to coordinate under the same constraints. Thus the shared world is a plurality of the many worlds that exist dimensionally in overlapping areas. This shared world grows when translations hold and a claim survives being carried between different practices, tools, communities, and purposes. This is close to Arendt’s sense of worldliness, though I’m framing it with a slightly different flavour. For Arendt, the world is not a planet nor a private interior, but rather the in-between, the durable stage where we can appear to one another, speak, act, judge, and remember. Her world is the durable edifice that outlasts the life of the individual. A common world, in this sense, isn’t guaranteed by coexistence. It must be built and maintained, and it can also be lost. 

Up to this point, I have been trying to describe a mechanism. 

Worlds form wherever patterns of perception, action, and expectation stabilize enough to guide life, even if only temporarily. While individual minds clearly generate worlds in this sense, they are not the only sites where stabilization can occur. Worlds persist only insofar as they remain responsive to resistance, whether material, social, or institutional, and they decay when insulated from it. What we call a common world is not given in advance, but is assembled where many distinct worlds are forced into coordination, translation, and partial overlap.

Worlds are both rare and abundant. They are rare in the sense that no two are the same. They are abundant in that they form constantly and some only for a moment. If worlds are produced and maintained, then each stable and coherent world is a fragile achievement. I personally like to imagine that every person contains many worlds, some entirely their own, including the interior coherence that lets them make sense of life and carry out actions without disintegrating. Collective worlds also exist everywhere, languages, crafts, rituals, epistemic practices, forms of attention. 

Most possible worlds never arrive, while many arrive and vanish quickly. The ones that endure do so because they find support through communities, institutions, memories, tools, literature, and so on. When a world disappears, something important is lost. A dead world is a way of perceiving and valuing that is no longer available as a possibility, a sort of extinction. 

Optimization as a form of “lossy compression” is a kind of world-killer. It reduces variation and diversity, functionally destroying worlds. We have a tendency to learn the dominant pattern and make the dominant pattern cheaper than everything else, which improves performance in a familiar regime while simultaneously punishing everything else. Overfitting is the technical name for a familiar moral and political problem. As a system becomes excellent at solving the problems of yesterday, it loses the ability to adapt to tomorrow, ultimately becoming fragile while appearing competent. Technology is agnostic towards world-building. It has the capacity to both create and destroy. Still, I can’t help but feel there is some latent machinic desire that is perhaps more biased towards destruction than creation, if only through feedback. 

A good example of this is model training in AI. At some point, we start training systems on past outputs which produce plausible averages and reward legibility. As AI becomes infrastructure that mediates speech, writing, hiring, diagnostics, policing, attention, and so on, it begins to mediate and shape the creation and destruction of worlds. A world that cannot be translated into the model’s categories becomes harder to express and rare worlds become expensive. Synthetic outputs begin flooding our shared world, and the environments begin learning in a regression where outputs become a degrading loop before collapsing. This happens in culture and science too. The danger here is that this happens rather slowly and doesn’t really announce itself.

I think one of the most vital questions we face today is how to create stable, enduring pluralistic worlds in the sense implied by Arendt, and ideally, a multiplicity of them. If technology is going to be biased towards world-creating rather than world-destroying, we need to determine how to select for patience, translation, and tolerance for ambiguity. Arendt saw something essential here. The modern world can produce a kind of worldlessness, with people being deprived of a shared space where their words can be heard and their deeds can be seen by others. The risk, in terms of the mechanism I’ve described before, is domination or the capture of overlap. A small number of world-generators become the gate through which reality must pass. Translation stops being mutual and assimilation is forced. 

In my world, every person has a right to worldhood. I see this right more specifically as the right to sustain a sense-making life without forced erasure by systems that trend towards a monoculture with one set of defaults or one form of legibility. Here, plurality is a survival condition for worlds. It keeps the common world resilient and able to generate novelty, which is vital for problem solving and persistence. The shared world requires overlap that is constantly being reworked, in the same way that the richest common worlds have emerged from the crowded conditions of competing interpretations. A common world is precious because it is built from plurality and maintained through work. 

Worlds are precious because they are hard to make and easy to kill.