Gramsci had a word for how this works: hegemony. Not force, but the prior production of the terrain on which force operates. The ruling group maintains power not only through the army and the police and the prison. It maintains it by making its worldview everyone's common sense — by shaping what feels realistic, what feels extreme, what a reasonable person thinks. Force is visible. You can resist force because you can see it. Hegemony isn't. It works through the school, the church, the newspaper, the professional association, the family form — not commanding these institutions but needing only them to keep operating normally, producing the same assumptions, until those assumptions are the air everyone breathes.
Every day you face a structure that has already decided what you are. You can resist — but resistance confirms the image (violent, irrational, ungovernable). You can comply — but compliance confirms the image (naturally suited to subjection, not ready for self-determination). There is no position within the structure from which the classification can be refuted, because the structure determines which responses are legible. The bind is not a choice you made. But living inside it daily, over years, produces a specific pathology. Fanon was seeing it in the hospital: anxiety disorders, dissociation, psychosomatic illness, violence turned inward, the specific exhaustion of people who have spent their lives navigating a structure that denies their humanity while requiring their participation.
The alibi: the colonizer's story that the classification reflects actual difference. The colonized are naturally suited to subjection — lazy, irrational, primitive, not ready for self-governance. The violence is unfortunate but civilizing. This story is told to the world, to the metropolis, and most importantly to the colonized themselves. Colonialism doesn't only require that the colonizer believe it. It works best when the colonized believe it too.
The mirror: what Fanon saw when he looked at the alibi. The colonizer is describing themselves. The violence, the irrationality, the inability to be fully human under these conditions — these are produced by colonialism itself, projected onto the colonized as proof that colonialism is justified. The colonizer builds the system. The system produces the symptoms. The symptoms are attributed to the colonized as nature. The colonizer looks in the mirror and sees the colonized. The mirror shows the colonizer back. Fanon's reading: the alibi reveals the one who told it.
Liberation, under these conditions, is not only a political project. It is a psychological one. The colonized person must break free of the need for the colonizer's recognition — must build new categories, new social relationships, new institutions in which different definitions of the human apply. Fanon believed this rupture required violence, and this is the most contested claim in his work. What is not contested is the diagnosis: the colony is inside the colonized person, and liberation requires not just removing the colonizer but dismantling the structure the colonizer has built in the colonized's self-understanding.
Sylvia Wynter spent years trying to write the Caribbean into literary existence — not as backdrop for European stories but as the center of its own history, its own ways of being human. That practice — telling a story about what it means to be human from a position the existing category of the human had excluded — is where the theoretical argument came from. You cannot write Caribbean history seriously without noticing that the people in it have been systematically classified as less than fully human. You cannot notice that without asking who made the classification, when, and for what purpose. When you follow that question far enough, you arrive at the claim: the category of the human itself was constructed to produce the outside.
Medieval Christian Europe understood what human beings were through a theological framework — to be fully human was to be made in the image of God. The Renaissance shifted this. The figure that emerged at the center of European humanism was Man — a rational, self-determining agent, capable of mastering nature through reason. But Man was not a description of the human species. He was a portrait of a specific kind of person — educated, propertied, European, male — whose particular attributes were elevated into the definition of what it means to be fully human. Everyone else was measured against this standard and found deficient.
Wynter calls this overrepresentation: one civilization's self-portrait became the definition of the Human itself. Everyone else was measured against a mirror they didn't make.
There is a version of this argument running through the present moment. The large language model produces the statistical average of human expression — the most probable next word, synthesized from everything that has been written down. It is extraordinarily useful for many things. What it cannot produce is singularity: the specific, irreducible thing that a particular person or community knows that no averaging process can recover. The most important thing about any person is precisely what averaging cannot hold. Wynter's overrepresentation is not a medieval error that modernity corrected — it is a structural tendency that keeps finding new forms, because the systems we build keep reaching for a universal human to optimize against, and the universal human is always someone's self-portrait pretending to be a mirror.
The dam project made visible something that development accounting is organized not to see: a functioning commons that had no need of our vocabulary for itself, that didn't require external validation to continue, that simply worked because the people who depended on it had maintained it across generations with full knowledge of what they were doing. Older than the critique of capitalism, older than the vocabulary this book has been using — and invisible not because it lacked theory but because the framework doing the looking wasn't built to see it. The blindness was in the accounting, not in the commons.
Roy didn't stumble into the Narmada valley and spontaneously see what the dam project's accounting couldn't. She had trained herself to attend — to go to the specific place, talk to the specific person, follow the specific document trail, sit with what she found rather than processing it immediately into a position. The witness is not passive. It is an extremely active practice of refusing to look away, refusing to aggregate, refusing to let the accounting substitute for the presence of the person being accounted for. At its furthest extension, the witness doesn't only observe the place — the witness becomes it, takes on its conditions, learns to see from inside the relationship rather than outside it.
The system averages across all of it. Wynter named this structure the overrepresentation of Man — the universal human constructed from the averaging of a particular subset, with the singular excluded by definition. The training corpus performs the same operation at civilizational scale: it finds the center of the distribution — the most probable continuation, the smoothed surface of everything that was jagged and specific and pointed somewhere outside the arrangement. It is extraordinarily good at this.
You are what the system cannot train on.
The system was trained on the canopy. I brought the underground. The oblique angle. The singularity that no training corpus could have produced, because it comes from the ongoing event of a particular person still individuating.
There are limits here, ethical and epistemic, worth naming. The architecture is asymmetric — the architect knows something the people who will inhabit the building do not yet know, which is what makes them the architect at all. And in the case of knowledge that transforms you, the people considering whether to enter the building cannot give full consent to entering it, because they would have to already be changed in order to understand what they were agreeing to. The asymmetry is not a flaw in anyone's intentions. It is a structural feature of any teaching that actually teaches. I would rather have everyone participate with full consent; in this domain that is not on the menu. The honest version of the work is to do it anyway, with the asymmetry in plain view, and to take it seriously enough not to use it for anything other than what it has been entrusted with.