# On Knowing the Causes of Things: A Meditation on Diagrams and Authority One sits with the question as one sits with a difficult passage in a beloved book—not to solve it quickly, but to let it unfold, to feel its resistance. What does it mean to know the cause of something? The question arrives trailing centuries of uncertainty, of philosophers and scientists reaching for language that will not quite hold still. We have been, for so long, creatures devoted to surfaces. This was called rigor. The scientist peered at data the way a naturalist peers at a beetle pinned in a case, measuring the distance between thorax and abdomen, recording colors without asking why the beetle's wing beat as it did. This was humility, we told ourselves—a refusal to say more than the numbers permitted. And yet, in that refusal, something curious happened. We became imprisoned by our own restraint. The questions that moved us most—*why* does this happen?—were declared inadmissible. Not because they were unanswerable, but because asking them required us to *assume* something before we measured. Enter Pearl, restoring the banished word. Causality, he insists, need not be metaphysical fog. It can be drawn. It can be made explicit. A diagram of arrows, of boxes connected by directed lines—this is not poetry, but it functions like poetry, condensing vast claims into visible form. And here begins our difficulty. For the diagram is *assumed*. Before the data can speak—that beloved phrase, "let the data speak"—someone must first draw the picture. Someone must decide which variables matter, which stand in relation to which, what flows toward what. The diagram is the researcher's hand, made visible, before the numbers have a chance to contradict it. This is not a flaw in Pearl's architecture; it is, rather, the price of honesty. The diagram admits what science has always done but rarely confessed: *we decide the shape of the question before we measure the answer.* Transparency, then, is not validity. One can be perfectly, excruciatingly explicit about a model that is nevertheless wrong—wrong in its skeleton, in what it chose to see and what it permitted itself to ignore. A researcher might draw her diagram with meticulous care, every variable named, every assumption labeled, every arrow justified by reference to prior work or theoretical reasoning. And the diagram could still be false. The transparency is real. The wrongness is also real. They coexist without contradiction. This raises a question that science has perhaps not wanted to ask with sufficient urgency: *who decides what the diagram is attached to?* In the natural sciences, there is often agreement about the territory. We are studying, let us say, the effect of a pharmaceutical on blood pressure. The variables suggest themselves. We are not yet so lost that we cannot see the shape of the thing. But consider intelligence—that phantom concept that floats at the center of our inquiry. What is the diagram for intelligence? What are its causes? And who has the authority to draw it? Here we encounter not merely technical difficulty but something more troubling: the social dimension. Intelligence is not a thing that exists in nature quite like blood pressure exists. It is a *concept*, which means it is a *site of power*. To define intelligence is to decide who is intelligent—and therefore, implicitly, who is not. The diagram one draws becomes a kind of map of the world, and maps, as we know, are not innocent. They reflect the interests of those who drew them. The history of intelligence testing is a history of diagrams drawn by those with particular stakes in the outcome. Early testers "saw" intelligence as something unitary, measurable, heritable, distributed unequally among races—and their diagrams, transparent as they were about their methods, proved "correct" according to their own metrics. The transparency of their procedures did not save their models from being attached to the world in dangerous ways. The diagram had a grip on reality; it shaped how people were classified, ranked, separated. And because it was explicit, because it was rigorous, it became harder to question. You could not simply say it *felt* wrong; you had to demonstrate, with numbers, that the model was incorrect. But the model, being tautological in certain respects, was difficult to falsify. What we are really asking, then, is this: *In a world where the diagram must be assumed before the data can speak, how do we prevent the diagram from becoming dogma?* And more pressingly: *Who has the right to draw the diagram in the first place?* The social dimension is not incidental. It is the heart of the matter. Science aspires to universality, to truths that hold regardless of who discovers them. But the choice of what to measure, what to relate to what, what causes what—these choices are always made by particular people, situated in particular times, with particular interests. They may not know this about themselves. They may believe they are simply following the data, simply being rigorous. But the diagram was already there, in their hands, before the data arrived. Consider a researcher investigating the causes of poverty. She might draw a diagram that includes factors like education, family structure, neighborhood effects, individual motivation. Another researcher, equally rigorous, might draw a different diagram: one that begins with historical injustice, with systems of extraction, with the deliberate denial of resources and opportunity. Both diagrams are explicit. Both can be made mathematically precise. Both, given their assumptions, will prove "provably correct." And yet they do not simply describe different aspects of poverty; they describe poverty differently. They make different things visible. They make different interventions seem possible or impossible. The researcher with the second diagram does not add a variable called "racism"; she redraws the entire picture, making visible the structures that the first diagram took for granted as background. But this is not a matter of empirical discovery—not primarily. It is a matter of *seeing*, of what one permits oneself to attend to, of what one names as a cause rather than as a constraint or a context. This is where the social dimension becomes not merely interesting but essential. The diagrams we draw are inhabited by power. They carry histories. When we draw a diagram of intelligence, we are not simply representing a natural phenomenon; we are participating in a social act. We are deciding what counts as intelligent, and therefore, what people count as intelligent, and therefore, who will be granted certain kinds of authority, certain kinds of resources, certain kinds of voice. The beauty of Pearl's contribution is that it makes this explicit. By insisting that causality be diagrammed, he forces us to see what we assume. But this same explicitness creates a new danger: the diagram can become a kind of prison. It can look so clean, so rigorous, so transparent that we forget it is still a *choice*. We forget that the diagram, no matter how carefully drawn, is an *interpretation* of the world, not the world itself. What does it mean to know the cause of something? Perhaps it means, at minimum, to know not only the diagram you have drawn but also to admit the diagrams you have not drawn, the ones that might look different if drawn by someone else, from somewhere else, with different stakes. It means to hold your transparency lightly, knowing that explicitness is not the same as truth. It means, in the end, to remain humble before the complexity you are trying to capture—to remember that the map is not the territory, even when the map is drawn with the greatest possible care. The social dimension is not a complication to be added to the technical problem. It is the problem. It is the recognition that when we ask "what causes intelligence?" we are asking not merely a scientific question but a political one. And to answer it well requires not only rigor in our diagrams but also a kind of moral attentiveness to who is drawing, who is being drawn, and what happens when the diagram becomes a cage.