A standardized test is a measurement instrument with the unusual property that the act of measurement reshapes the thing measured. When a state decides that fourth-grade reading proficiency will be assessed by a particular multiple-choice instrument administered on a particular spring morning, every fourth-grade teacher in the state spends the year teaching toward that instrument. The curriculum bends. The school day reshapes. Recess shortens. Art and music get cut. Reading itself, the activity the test purports to measure, becomes a different activity — a regimen of passage-and-question pairings drilled until the form is reflexive — and the children who do it begin to associate the word "reading" with something other than what readers actually do.

This is not an argument that measurement is bad or that accountability is suspicious. It is an argument that a particular kind of measurement, deployed at industrial scale and tied to consequential decisions about funding, evaluation, and tracking, produces a particular kind of distortion that operates beneath the level at which anyone involved can easily perceive it. The teachers are not cynical; most are doing what the system rewards. The administrators are not malicious; most are responding to mandates whose noncompliance is career-ending. The children are not lazy; they are taking the tests the adults are giving them. And yet the aggregate result is a school system that has substantially confused its own ends, replacing the goal — children who can think — with a proxy that is correlated with the goal in trivial cases and uncorrelated with it in the cases that matter.

The regime has three layered effects. The first is curricular narrowing. Subjects that are tested expand; subjects that are not contract or vanish. Within tested subjects, formats that are testable expand; formats that are not — long writing, extended projects, oral discussion, the slow accumulation of knowledge of a domain through play — contract. The second is pedagogical narrowing. The teaching strategies that produce short-term test gains are not the strategies that produce long-term learning; under pressure, teachers default to the former because the latter does not show up on the spring measurement. The third is the most insidious: the regime alters what the participants believe school is for. After enough years inside it, children, parents, and teachers stop being able to articulate the goal of education except as the production of test scores. The proxy has eaten the thing.

Law 2 is violated specifically because thinking, in its real form, cannot be assessed by a forty-minute bubble sheet. Real thinking involves holding contradictory evidence in mind, generating hypotheses, testing them against reality, revising. None of those operations leaves a trace that a Scantron can read. A test can measure whether a child knows that the boiling point of water at sea level is one hundred degrees Celsius. It cannot measure whether the child could figure out the boiling point at altitude given a few hours, a textbook, and the question. The first is recall. The second is what we mean by an educated mind. When school is reorganized around what can be machine-scored, it is reorganized away from the second toward the first, and Law 2 — the law that says thinking is what the system is for — quietly fails.

The defenders of the regime make a real point. Before standardized testing, low-performing schools could hide behind sentiment; children, especially poor children and children of color, could be moved through grades without anyone confirming whether they had learned to read. Testing surfaced the failure. The civil rights case for testing — that it makes inequality visible — is genuine, and any honest critic must take it seriously. The question is whether the visibility produced is worth the curriculum it consumes, and whether there are forms of accountability that surface inequality without distorting the entire instructional day around themselves. The answer, on the evidence of the past three decades, is probably no for the first and probably yes for the second. We have made the inequality visible and worsened the conditions that produce it.