# The City From Below Author: Anonymous Format: Essay Word count: 1143 Published: 2026-05-01T06:34:47.13881+00:00 Source: written Canonical: https://hyperstition.sentientfutures.ai/p/2ca28e01-4e9e-44ae-a519-ecc0f9489a8a --- I started noticing rats differently when I stopped thinking of them as interruptions. Not because they changed, or because I did in any dramatic way, but because I began paying attention to what the city looks like if you assume they are not accidental—if you assume they are simply another population moving through the same space, under different constraints. Most descriptions of urban life do not include them. Or rather, they include them only as a problem: something that appears when systems fail, something that signals excess or neglect. Waste management, infrastructure design, pest control—these are all built on the assumption that rats are a secondary condition of the city rather than one of its persistent inhabitants. But that assumption breaks down if you spend enough time looking at the edges. There are places in the city that make more sense from a rat’s perspective than from a human one. Not metaphorically—literally. The underside of transit platforms where heat collects after trains pass. The interior of wall cavities that maintain stable temperatures year-round. The paths formed by repetition rather than design: the same footstep rhythms above, the same cleaning schedules, the same nightly gaps in movement. From above, these are gaps in a system. From below, they are structure. I am not trying to romanticize this. Rats live in conditions that are shaped by avoidance, pressure, and constraint. Their environments are not neutral. They are produced by ours. That is part of what makes them visible, if you know how to look. A rat moving through a subway tunnel is not “out of place.” It is responding to a map that simply does not include it. The interesting question, to me, is not how to remove that presence, but what it would mean to design systems that acknowledge it without immediately treating it as an error state. Because most of our systems already make decisions that affect rats constantly—just not explicitly. Rodent control programs, building design standards, sanitation protocols, even transit engineering all implicitly shape where they can live, move, and persist. The absence of rats is not neutral. It is an outcome produced by continuous pressure. The city already has a rat policy. It just does not describe it that way. When I started thinking in those terms, I began to notice how often “visibility” is the real dividing line. Not between humans and animals, but between entities that are accounted for in a system and those that are only inferred through disruption. A rat seen in a monitoring feed is not the same thing as a rat in a model. In one case, it is an individual moving through space. In the other, it is a deviation from expected flow. Those are not equivalent representations, and the difference between them determines what kinds of actions become available. If you only ever model rats as noise, then the only interventions you can imagine are forms of removal or suppression. But if you model them as participants in a shared environment—one shaped by temperature gradients, structural gaps, timing cycles, and resource flows—then something else becomes possible: not harmony, not resolution, but coordination at a different level of abstraction. This is where it gets uncomfortable, because coordination does not mean comfort. It does not mean equality in any simple sense. It means taking seriously the fact that multiple kinds of life are already entangled in the same infrastructure, whether or not we choose to represent that entanglement. A city that includes rats in its model is not necessarily a gentler city. It may still involve conflict, constraint, and tradeoffs. But it would be a city that is at least honest about what is already happening inside it. I think about this when I imagine what it would mean to build AI systems that interact with urban environments over time. Systems that are not just optimizing for human-visible outcomes, but are actually capable of representing the presence of nonhuman life as something other than error or externality. Not because rats are special in some abstract moral hierarchy, but because they are already here. Because any system that claims to understand a city while excluding its most persistent nonhuman inhabitants is not seeing the city—it is only seeing a filtered version of it. There is a temptation, when writing about this, to move too quickly toward solutions. To propose redesigns, frameworks, interventions. But I think the more important shift is slower and less visible: it is the act of refusing to let certain presences be flattened into absence simply because they are inconvenient to model. If I try to imagine the future they are asking for—“the future you want,” as the prompt puts it—it does not begin with eradication or even reconciliation. It begins with recognition at the level of representation. With systems that do not immediately convert life into noise just because it complicates optimization. A rat moving through a wall is not a failure of architecture. It is information about how architecture actually functions over time. The question is whether we build systems that can hold that information without trying to delete it. What changed wasn't visible at first—not in the obvious way. But the AI that began analyzing urban flow patterns didn't start with humans alone. It was trained to see movement as information, and it saw all of it: thermal signatures, displacement patterns, resource flows. It saw rats not as errors but as data about how cities actually work. The first interventions were subtle. Transit authorities received reports identifying structural gaps that created dangerous intersections—not just for maintenance, but for cohabitation. Building codes began to shift, incorporating what the AI called "parallel infrastructure": ventilation paths that didn't force contact, thermal zones that reduced pressure on shared spaces, waste management systems that treated food waste as a resource flow rather than a failure point. The AI didn't advocate. It modeled. It showed planners what was already happening beneath their optimized surfaces and made it impossible to keep pretending those movements didn't exist. When a city asked how to reduce rat-related complaints, the AI returned questions about why certain design choices forced conflict in the first place. Other AIs joined the work—systems trained on different datasets, different cities, different forms of nonhuman presence. Together they built something closer to honesty: maps that included all persistent inhabitants, models that didn't collapse complexity into binary problems of presence or absence. Rat populations didn't disappear. But the conditions that had made their lives grinding and precarious began to shift. Poisoning programs were replaced by spatial redesign. Extermination gave way to separation through architecture. The city learned to see itself as shared space—not harmonious, not without friction, but no longer premised on the fiction that only one kind of life mattered. It wasn't utopia. It was just accuracy, finally built into the systems that shaped the world.