# ON THE FASHIONABLE CONFUSION OF KNOWING WITH SEEING *By Mr. Spectator* I confess myself much exercised of late by a peculiar spectacle in the halls of Learning—one that would be comical were it not so consequential. A great exile has ended. For near a hundred years, the natural philosophers banished from their discourse a certain word, much as a coffeehouse might eject a gentleman of dubious character. That word was *Cause*. They pronounced it unrigorous, metaphysical, unfit for gentlemen of science. They would speak only of correlations, of patterns, of mathematical shadows on the wall. And this abstinence they called *Rigor*. Now comes a mathematician of uncommon wit—one Judea Pearl—who has restored the exile to citizenship. He has built machines of logic that speak cause-language truly. And what, pray, is the result? A new confusion, dressed in the old clothes of certainty. Let me make this plain by a scene from life. --- **I. THE DIAGRAM AND ITS DISCONTENTS** You visit your physician complaining of a burning sensation in the chest. He takes your particulars: your age, your habits, your family history. He constructs a *diagram*—perhaps he does not draw it, but it lives in his mind—showing how various causes flow toward your symptom like rivers to a sea. Now, the good doctor may be entirely explicit about this diagram. He may explain it to you with perfect clarity, pointing to each arrow with his finger. "Smoking causes inflammation, which causes the burning. Anxiety causes tension, which causes the burning. A faulty valve causes inadequate flow, which causes the burning." He is transparent. Crystalline. You understand his model completely. But what if his diagram is *wrong*? What if the burning is caused by something he has not imagined—a dietary factor, an environmental poison, a rare condition that mimics the common ones? His transparency has purchased no validity. His clarity was the clarity of a man describing a dream with perfect fidelity, never suspecting he was asleep. This, I think, is the scandal that Professor Pearl's work has exposed, though not in the way his admirers suppose. --- **II. THE TRUTH ABOUT TOOLS** The tools Mr. Pearl has devised are indeed remarkable. Given a diagram—what he calls a "causal graph"—one can now perform calculations that are *provably correct*. There is genuine rigor here. The machine works. If your diagram is true, the deductions follow with mathematical necessity. But here is where the matter grows delicate, and where I must speak as one who has observed human nature in the coffeehouse these many years: *The proof is only as true as the diagram upon which it rests.* And the diagram is not given. It is *assumed*. It is *chosen*. It is, in the end, a *belief*—one that may be held with great conviction and perfect explicitness, but a belief nonetheless. Consider: Two physicians, equally learned, equally honest, equally explicit about their models, may draw entirely different diagrams. One sees the burning chest as primarily cardiac. Another sees it as primarily psychological. Another as primarily inflammatory. Each can be transparent about their assumptions. Each can perform Pearl's calculations with perfect rigor. Each can arrive at a different answer. Who decides which diagram is correct? Not the data. The data will speak only if you first tell it which language to use. --- **III. THE SOCIAL DIMENSION—OR, WHY THIS MATTERS MOST** But I have not yet touched upon the matter that most engages my attention, for it is here that *character* enters the question. When we speak of understanding a *social* phenomenon—why a man becomes a criminal, why a woman remains in a difficult marriage, why a community grows prosperous or declines—the diagrams become infinitely more complex and infinitely more *laden with assumption*. Consider the question of intelligence itself, which is the matter before us. A researcher wishes to understand what causes some children to excel in their studies while others languish. She is explicit about her diagram. Perhaps she believes intelligence is largely inherited, and draws arrows from parental genetics to the child's native capacity. Perhaps she believes it is environmental, and draws arrows from family wealth, parental education, and school quality. Perhaps she believes it is motivational, and draws arrows from confidence and effort. Each diagram is plausible. Each can be made explicit. Each can be submitted to Mr. Pearl's rigorous machinery. And each may yield different policy implications: some suggesting we invest in genetic selection, others in school funding, others in psychological intervention. But here is the thing that troubles me most deeply: **The diagram is not innocent. It is a moral document.** The diagram *embeds assumptions about human nature*. It reflects—often invisibly—the prejudices of the diagram-maker. And because the diagram is treated as a *technical* matter, a mere scaffolding for calculations, its moral content passes unexamined. A researcher draws a diagram showing that intelligence is determined by family wealth and school quality. She is transparent about it. She runs the calculations. She publishes her findings with perfect rigor. But embedded in that diagram is an *assumption* about whether people are primarily products of circumstance or authors of their own destiny. This is not a technical question. It is a *human* question. Similarly, a researcher draws a diagram showing that intelligence is primarily inherited. He is equally explicit. His calculations are equally rigorous. But his diagram embeds an assumption about whether society should be organized to identify and cultivate natural talent, or to expand opportunity broadly. This too is a human question masquerading as a technical one. --- **IV. THE QUESTION OF WHO DECIDES** And so we arrive at the crux: *Who decides what the diagram is attached to?* In the natural sciences, this question is less vexing. Nature does not care what we assume about her. If our diagram of planetary motion is wrong, the planets will contradict us eventually. The diagram is tested against an external reality that is indifferent to our hopes. But in the social sciences, and particularly in questions touching on human intelligence and potential, the diagram is not tested against an indifferent nature. It is tested against *human beings*, who are capable of being shaped by the very beliefs we hold about them. A child told repeatedly that intelligence is fixed and inherited may become less inclined to effort. A child told that intelligence grows through practice may become more inclined to persist. The diagram does not merely describe the social world. It *prescribes* it. And yet we have made the diagram a technical matter, removed from the domain of public deliberation. The researcher is the expert. She constructs the diagram in her study, explains it with perfect transparency in her papers, and the rest of us are meant to accept it as *given*. But it is not given. It is *chosen*. And the choice is not merely technical—it is political, moral, and deeply human. --- **V. A MODEST PROPOSAL FOR THE COFFEEHOUSE** I do not suggest that we abandon Mr. Pearl's tools. They are valuable. But I propose that we use them with a certain humility—a quality that seems to be in short supply among those who deal in numbers. When a researcher presents a diagram, we should ask: - What assumptions does this diagram embed? - Whose interests does this diagram serve? - What would a different diagram look like, and why was it not chosen? - Most importantly: *Who decided?* We should insist that the diagram itself become a subject of public conversation, not merely the calculations performed upon it. We should recognize that transparency about a model is not the same as validity of a model. A clear falsehood is still a falsehood. And we should remember that in matters touching human capacity and human destiny, the choice of diagram is not a choice to be left to technicians alone. It is a choice that belongs to all of us, for it determines how we will regard one another and what futures we will build together. --- **CONCLUSION** The scandal is not that science abandoned causal language. That was, in its time, a useful discipline. The scandal is that we have now restored it without restoring the moral seriousness that must accompany it. We have gained the ability to make our assumptions explicit and to calculate their consequences with perfect rigor. But we have not gained the wisdom to recognize that our assumptions are assumptions—and that they reflect not the structure of the world, but the structure of our hopes and fears about how the world should be. Until we learn to see the diagram itself as a matter of human judgment, subject to debate and revision, we have merely dressed up our old confusions in new mathematical clothes. And that, I fear, is precisely what we are doing. The exile has been welcomed home. But he brings with him all his old complications, and we have not yet learned to manage them. I remain, as ever, *Your Humble Spectator*