# ON THE CAUSES OF THINGS: A COFFEEHOUSE MEDITATION *By Mr. Spectator* I was sitting yesterday in my usual corner at the coffeehouse, observing the gentlemen about me, when I noticed something curious. Two natural philosophers were engaged in the most spirited debate—not about facts, mind you, but about something far stranger: whether they were allowed to speak about facts at all. The younger man, fresh from his studies, insisted that all causal language was unrigorous fancy. "We observe correlations," said he. "We measure associations. To say one thing *causes* another is to traffic in metaphysics." The elder replied, somewhat wearily, that this was precisely why natural philosophy had spent a century in exile from the actual world—all precision and no purchase, all rigor and no sense. This dispute struck me as precisely the sort of confusion that arises when learned men forget that language is not the enemy of thought, but its instrument. And I resolved to examine it here, for the benefit of those who frequent this paper. --- The matter concerns a gentleman named Pearl—I am told he is quite celebrated in certain circles—who has done something rather ingenious. He has noticed that Science, in its zeal to be respectable, had painted itself into a corner. For nearly a hundred years, it said to itself: *We shall speak only of what we can measure without presupposition. We shall banish the word "cause" and speak only of correlation.* This was called rigor. And it was, after a fashion, honest—if one's only ambition is to avoid the appearance of assuming what one has not proven. But Mr. Pearl asked a simple question: What if the appearance of rigor is not rigor itself? What he has shown is that causal reasoning is not the enemy of empirical science—it is its prerequisite. To interpret any data at all, one must first *assume a diagram*: a picture of how the world works, which variables speak to which, what influences what. Only *then* can the data have any meaning. The transparency of your diagram does not make it true. A man may be perfectly explicit about his model and perfectly wrong about his world. This is a lesson worth pondering in all affairs, not merely those of natural philosophy. --- I confess I have observed something of this in our social world, where people pride themselves on being "data-driven," as though numbers alone could speak without an interpreter. A merchant may present me with impeccable statistics showing that poverty causes idleness, all the while his diagram—his invisible assumption about the causal order—places poverty *before* idleness, when any sensible observer knows the relation is far more tangled. Or a reformer will show me numbers proving that a new law produces safety, when his diagram assumes the law is the efficient cause, when perhaps it is merely the occasion for a thousand other forces to align. The rigorous presentation of a faulty diagram is not rigorous at all. It is merely rigorous *presentation*. But here is where the matter becomes delicate, and where I must invite the reader to join me in genuine puzzlement rather than false certainty. Who decides what the diagram is? --- This question grows thorny when we consider the social dimension—and I use that word to mean: the dimension in which human beings are involved, in which power and interest and interpretation are at stake. Consider the question of intelligence itself, which has occupied many learned minds. A researcher designs a test. He assumes (without stating it plainly) that intelligence is a single thing, measurable, heritable, distributed unequally. His diagram is fixed *before* the data arrives. The data then appears to confirm what his diagram already assumed. He publishes with great transparency: *Here is my method. Here are my numbers.* And the diagram—the fatal diagram, the one that decided everything—is so familiar it becomes invisible. People begin to speak of "intelligence" as though it were a thing in nature, when it was always a thing in the diagram. Now, who decided what that diagram would be? A scientist, certainly. But also: the history that made certain questions seem natural and others absurd. The social position that makes some interpretations seem common sense and others seem radical. The power that allows some people to name what counts as intelligence and others merely to be measured by it. I have observed this in our discussions of learning and ability. Those who are measured find their worth reduced to the diagram. Those who design the diagram are rarely measured at all. The transparency of the method does not disturb the opacity of the choice. --- And yet—and here I must not be unfair—there is something genuinely liberating in what Mr. Pearl has shown us. If we *admit* that every empirical inquiry requires a causal diagram, then we can at least *argue about the diagram*. We can ask: Why do you assume this causes that? Why place this variable before that one? What evidence supports your picture of the causal order itself? This is far better than pretending the diagram does not exist, that the data speaks for itself, that rigor consists in never assuming anything at all. That pretense merely hides the diagram from view—makes it work all the more powerfully, invisibly, unexamined. In matters of intelligence, in matters of social science generally, this seems to me a great advance. If someone claims that a test measures innate ability, we may now ask: *Show us your diagram. Show us why you believe intelligence flows in this direction rather than that. Show us what you have assumed before the data began to speak.* The diagram may still be wrong. But at least it is visible. At least it can be debated. --- I will not pretend to have solved this matter. A true philosopher knows that the most important questions in any inquiry are often the ones we have not yet learned to ask. But I will venture this much: **Intelligence is not a thing waiting in nature to be discovered.** It is a relationship between a mind and a world, and between a mind and the questions we decide to ask of it. How it is measured, what counts as evidence of it, what causal diagram we assume connects it to other things—these are not scientific questions alone. They are social questions. They are questions about power: who decides what shall be measured, and who shall be measured by it. To be rigorous is to be explicit about this. Not to hide behind numbers and call it objectivity, but to say plainly: *Here is what I assume about how the world works. Here is why I assume it. Here is what I might be blind to.* The diagram is not validity. But transparency about the diagram—that is the beginning of wisdom. --- *Mr. Spectator will be in his usual corner tomorrow, should any gentleman wish to continue this conversation.*