# On Diagrams, Causation, and the Diagram Itself There is a moment in reading Judea Pearl when one feels the ground shift beneath the feet—not violently, but as a ship's passenger discovers, hours into the voyage, that the familiar shoreline has become a thin line on the horizon. We have been told, for so long, that science must confine itself to correlation, to the modest enumeration of what goes with what. The exile of causation was presented as *rigor*—a kind of asceticism, a stripping away of the temptations of narrative that cloud the mind. And then Pearl extends his hand from across this imposed boundary and shows us that causation was never truly forbidden; it was merely *formalized*. Given a diagram—given, that is, a commitment about how things touch—we can reason about cause with mathematical certainty. But the moment one grasps this gift, another anxiety arrives, quieter and more persistent. The transparency is not validity. One can draw a diagram with perfect clarity, label every arrow, state every assumption with crystalline precision—and be entirely, comprehensibly, *wrong*. The diagram is not found in nature. It is made. Someone made it. And that someone brought to the making all that they are: their history, their blindnesses, their unstated convictions about how the world is organized. To know the cause of something, one must first have decided what *something* is. Consider: a woman's intelligence is measured. A test is administered. A score is obtained. The scientist writes clearly: "We assess verbal reasoning through the following tasks..." The apparatus is transparent. The math is rigorous. And yet—what is the diagram? Does intelligence flow *into* the test, like water into a vessel? Or does the test *constitute* intelligence through some more intricate arrangement, through the selection of what counts, what is noticed, what is permitted to matter? Does intelligence have a cause at all, or is it rather a kind of gathering—a moment when certain capacities, certain experiences, certain permissions to think, certain absences of fear all converge in a particular body on a particular morning? The diagram must be assumed before the data can speak. This is Pearl's austere truth. But who is present when the assumption is made? Here one must linger over the word *social*—that residual category, that dumping ground for all that does not fit into the clean mechanisms of cause and effect. The social is the dimension one cannot quite diagram. Or rather: one can diagram it, but only by forgetting, in the moment of drawing, that the diagram itself is a social act. The researcher, sitting alone or in a room with colleagues—themselves shaped by institutions, by what they were taught to see and taught to ignore—this researcher makes marks on paper. These marks say: "Intelligence causes test performance" or "Socioeconomic status causes intelligence" or some other arrow, some other direction of flow. The diagram is presented as though it were discovered, like a fossil extracted from rock. But it was made. It was made by people who were, themselves, never outside the world they sought to diagram. The social dimension is thus not something *in* the diagram. It is the fact of the diagram itself. Consider a concrete example—not because examples make things clearer, but because they make the obscurity richer, more textured. A researcher wishes to understand whether a particular intervention improves intelligence. The diagram is drawn: Intervention → Intelligence. The assumption is made explicit. The data is collected. The math proceeds. And the conclusion emerges with satisfying rigor: yes, or no, or perhaps. But who was permitted in the room when the diagram was drawn? Whose conception of intelligence was already sedimented into the very language available for making the assumption? When the researcher chose "intelligence" as the outcome—not "capacity for joy," not "ability to think about what matters," not "resistance to the crushing of spirit"—that choice was not transparent. It was social. It was historical. It was made in a world where intelligence had been defined, for more than a century, in ways that served certain purposes and excluded others. And the intervention itself: what was it? Who decided it was an intervention rather than simply something that happened? What is the difference between an intervention and an event, between a cause and a context? These are not mathematical questions. They are questions about how we carve the world into pieces, and those carvings are never innocent. Pearl has given us something genuine—a way of making our assumptions explicit, of building tools whose correctness can be proven *given* the diagram. This is not nothing. Indeed, it is considerable. To be explicit about what one believes is, in its own way, a form of honesty. But honesty is not the same as truth. And transparency is not the same as validity. The deeper question—the one that resists the diagram—is this: what does it mean for someone to know what they have decided before they know what is true? The researcher must decide about the diagram in advance. But the decision to decide, the choice of what dimensions to make visible and which to leave in shadow—these things happen in a social space, a historical moment, a set of institutional pressures and permissions. They are not *chosen* in any simple sense. They are *inhabited*. To think about intelligence, then, is not to improve our diagrams, though that may be worth doing. It is to ask: who is the we that diagrams? What are we permitted to see? What have we agreed not to look at, calling it *not-rigorous* when perhaps it is simply *not-convenient*? The social dimension of intelligence is not something that can be added to the diagram. It is the fact that the diagram exists at all—that we are creatures who make marks and call them understanding, who assume and call it knowing, who decide and call it discovering. Pearl restored causal language to respectability. But causation itself—the *fact* of cause—remains, as it always was, a human interpretation of a world that may not be organized into causes and effects at all. The diagram is not wrong. But neither is it innocent. And until we can hold both of these truths at once without flinching, we will continue to mistake transparency for truth, and mistake our own reflections for the world.