# On Knowing What Causes What: A Kind of Transparency There is a moment in reading—one recognizes it at once, though it resists description—when the mind encounters something it had been circling without quite landing upon. One feels the relief of a word finally spoken, and simultaneously, a new uncertainty blooming in that very clarity. This is what happens, I think, when we turn our attention to the question of causation in the human sciences. For nearly a century, we told ourselves that to be rigorous was to be *cautious*. We would observe, measure, count. We would note correlations with scrupulous care. And we would step back from causation as though it were an ornament we could not afford, a luxury of the unscientific mind. The language of "because" came to seem almost quaint—imprecise, laden with assumption, the sort of thing poets said. We called our exile rigor. We called our abstinence truth. But something curious happened in that withdrawal. The causal language did not disappear; it merely went underground. It lived in the choices of what to measure, in the designs of experiments, in the very structure of the question asked before a single datum was gathered. We expelled it from the light only to find it had taken root in shadow. And there it grew, unexamined, all the more powerful for being unspoken. Then Pearl came, with his diagrams. And one recognizes immediately the force of this: here is someone saying, *Let us at least be honest about what we are assuming. Let us draw it. Let us make the causal structure visible, discussable, vulnerable to scrutiny.* The diagram becomes a kind of bargain—a trade of explicitness for power. Given this structure, the mathematics can now speak. The tools are provably correct, provided the diagram holds. Yet here is where the mind catches, where one must pause and circle again. --- The diagram is not innocent. This is perhaps obvious, yet it bears the weight of being obvious in a way that escapes notice. A researcher, entirely sincere, entirely careful, can draw a diagram that is wrong. Transparency is not validity. Explicitness is not truth. One can be maximally clear about a model that fundamentally misrepresents the tangle of what actually is. Consider a simple causal claim from the social sciences—and social science is where this problem becomes most acute, most human, most impossible to ignore. A researcher wishes to know whether *poverty* causes *poor educational outcomes*. Straightforward enough. The diagram is drawn: an arrow from poverty to outcomes. The mathematics proceeds. The tools work perfectly, given this structure. But what is this arrow? What has the researcher assumed about the mechanism? Has she assumed that poverty operates primarily through material deprivation—through hunger, instability, lack of resources? Or might poverty operate through the internalization of hopelessness, through what one might call the *feeling* of futility, which is not quite the same thing? And what of the reverse arrow—do poor educational outcomes not themselves reinforce poverty, creating a kind of cycle that the linear arrow cannot quite capture? And what of the third thing, perhaps undrawn, that generates both—some history, some inherited disadvantage, some structural arrangement of society itself that produces both poverty and educational failure as its twin expressions? The diagram asks us to choose. And in choosing, we have *assumed* our answer. This is not to say that diagrams are worthless. Rather, it is to say that the diagram is not where truth lives. The diagram is where *intelligibility* begins. It is where we make our tacit knowledge explicit enough to be wrong about together. But the moment we believe the diagram is correct—the moment we mistake it for the territory rather than a map—we have committed the very error we thought we were escaping. --- But there is a deeper problem still, and it is a social one. *Who decides what the diagram is?* This is not a question for mathematics. Mathematics cannot answer it. Once the diagram is given, the mathematics unfolds with perfect objectivity. But the giving of the diagram—this is a social act, a political act, a deeply human act. It emerges from somewhere: from training, from the conceptual tools available in one's discipline, from the questions one has learned to ask, from the questions one has never been invited to ask, from the experiences one brings to the work, from the experiences one is barred from bringing. Consider: a researcher trained in individualistic psychology might diagram poverty's effect on educational outcomes in terms of individual motivation, self-efficacy, cognitive load. Another, trained in sociology, might diagram it in terms of structural barriers, social reproduction, the distribution of cultural capital. A third, perhaps from a community itself affected by poverty, might diagram it in terms of resilience networks, collective knowledge, the resistance embedded in survival itself. All three are describing, in some sense, the same phenomenon. All three are holding something true. Yet the diagrams would be radically different. The mathematics proceeding from each would be correct, given its own premises. But they would not be measuring the same thing. This is where the question becomes genuinely difficult. For the diagram emerges not from nature, but from *interpretation*. And interpretation is always already social. It emerges from conversation, from power, from whose voice is heard in the room where the diagram is drawn. The exile of causal language was, in a strange way, democratic. It said: *We cannot know causes. We can only observe.* And in that modesty lay a kind of protection. No one could claim to have captured the true causal mechanism, because causal mechanisms were deemed unknowable. Everyone had to stay at the surface. But Pearl's restoration of causal language, for all its rigor, for all its transparency, opens a new problem. It invites us to assume we can know the causal structure. And if we can assume it, then the question of *who assumes it* becomes critical. Whose diagram is this? Whose causal intuitions have been mathematized and then presented back to us as truth? --- One thinks, in this context, of what intelligence might mean in a world where causation is both possible and deeply uncertain. Intelligence, perhaps, is not the ability to discover the true causal structure—for there may be no single true structure, only structures more or less useful for particular purposes, structures that capture some aspects of reality while necessarily obscuring others. Rather, intelligence might be the capacity to *hold the diagram lightly*. To draw it with care, to make it explicit, to test it rigorously—and yet to remain aware, always, that one is drawing, that one is choosing, that the structure one has made visible is also a structure one has *made*. This would require a kind of double consciousness. It would mean being rigorous about the diagram while simultaneously skeptical about the diagram. It would mean bringing to bear all the tools of mathematics and logic, and yet refusing to let them have the final word. It would mean remaining open to the diagram as others might draw it, as others *have* drawn it from within different traditions of thought, different positions in the social world. For the social world resists the diagram in ways that nature sometimes does not. The social world is made of meaning, and meaning is made by people, and people can change their meanings. A causal structure that held true for one generation might not hold for the next, because the people living within it have reinterpreted it, resisted it, transformed it. The diagram captures a moment, not a law. --- There is, I think, a kind of integrity that comes from admitting this. Not the integrity of the diagram—which will always be provisional, contestable, incomplete. But the integrity of the person drawing it: the willingness to say, *Here is what I think is happening, and here is what I had to assume in order to think it. Here is what I cannot see from where I stand. Here is where I might be wrong. And here is an invitation for you to draw it differently, to show me what I have missed.* This is not the promise of rigor, which is always in danger of hardening into rigidity. It is, rather, the practice of what one might call *intellectual humility*—a word that seems almost quaint in an age of certainty, of precision, of provably correct tools. And yet it may be the most rigorous stance of all. For it allows us to use the tools without being used by them. It allows us to build diagrams without being imprisoned by them. It allows us to know—provisionally, carefully, together—something of what causes what, while remaining always aware that the knowing is an act, not a discovery; a construction, not a revelation; a conversation, not a pronouncement. The diagram is made of lines drawn on paper. But the causes it represents are made of the lives of human beings. To be intelligent about that difference—to feel its weight—may be the beginning of genuine understanding.