Every fall, students arrive at U.C. Berkeley having done everything right. They took the required courses. They earned their grades. Their transcripts proclaim they are prepared. And then they show up for the first required math course for their intended STEM major and discover that something is seriously wrong — not a gap they can point to, not a topic they missed, but something harder to name. A wall where there should be a door.
Their math professors can see it immediately. The students cannot. That asymmetry is the subject of this column.
For more than two decades, the math education establishment — most influentially Stanford Graduate School of Education professor Jo Boaler together with the California Mathematics Framework she helped shape — has organized its thinking around a single binary opposition: conceptual understanding versus procedural fluency. Are students learning why math works, or simply how to execute procedures? This binary has driven curriculum design, teacher training, and state law through multiple contentious revision cycles over decades.
But this is the wrong question. Or rather, it is an incomplete question that has crowded out the right one. The conceptual/procedural binary doesn’t have a name for what Berkeley and other U.C. professors are actually seeing when their young students hit that wall. And it cannot account for the damage because it exists in a dimension that binary is unable to see.
The emerging generation of math education researchers has developed a more precise framework that accounts for this, but the framework has three layers, not two.
The first is routine expertise — the internalized, automatic command of foundational procedures. This is not merely rote memorization. This is what cognitive scientists such as John Sweller study under the name of cognitive load theory. When basic operations become automatic, they stop consuming the learner’s working memory, freeing up their mind for higher-order reasoning tasks. A learner cannot think flexibly about algebraic structure while constantly having to expend effort to reconstruct how to manipulate fractions or roots. Automaticity is not the enemy of understanding. It is a precondition for it.
The second layer is adaptive expertise — the capacity to deploy what you know flexibly and under novel conditions. You have to be able to recognize, name, and analyze what the tools in your toolkit aren’t built for. You need to see structural similarities across different problem types. To improvise a path when the familiar route is blocked. Researchers Giyoo Hatano and Daniel Schwartz have mapped this carefully: adaptive expertise isn’t a different kind of knowing from routine expertise. It’s what grows out of the layer of routine expertise, given the right conditions — varied demands, genuine challenge, time.
The third layer has no single technical name in the literature, but teachers and mathematicians recognize it immediately: mathematical maturity. It is not content. It is not skill. It is the accumulated, internalized experience of having stayed with hard problems, gotten stuck, recognized higher-order situations that required you to invent new tools from your old ones, burned your own unique pathways through, and returned to this process again and again at ever-increasing levels of abstraction. It includes the slow accumulation of personal mathematical tools — all those shortcuts, heuristics, and intuitions that each mature math learner earns and makes their own. It is a well-earned confidence in your own maturing mathematical agency, and it is a kind of endurance. It is a form of intuition that develops only through years of genuine engagement at the right level of challenge — not too easy, not impossibly hard, but persistently demanding.
These layers are sequential and interdependent and also ever-spiraling. Mathematical knowledge is generative – at every developmental stage, each layer provides the soil in which the next layer develops at that level. A student who has never built any solid routine expertise arrives at college with no foundation on which their adaptive expertise can mature. A student who has never developed adaptive expertise at their current level of understanding has no scaffolding on which their mathematical maturity can emerge. The failure is not additive — it is exponential. The student isn’t “behind” by the sum of what they missed. They are behind by what they missed, multiplied by everything they would have developed through years of working with it.
This is the cascading effect that Boaler’s conceptual/procedural binary cannot see. And it is made catastrophically worse by a second failure — one that wears the disguise of ambition.
Families across California have been led to believe that moving through math courses faster is the best preparation for college STEM careers. It is not. Coverage at speed without deep internalization and adaptive practice produces transcripts that claim students are ready when they are not. Mathematical maturity cannot be accelerated. It requires time in a specific sense — not more hours, but return visits to material from new angles using a deep personal understanding together with increasing abstraction, productive struggle at the right level of difficulty, personal tool development sustained over years. You cannot rush the medium in which the layer grows and expect the layer to be there when you need it.
The blind spot isn’t in their mathematics. It’s in their ability to see their own mathematics.
In 1983, as a new graduate student at Stanford, I was part of a colloquium on constructivist epistemology led by the legendary cybernetician Heinz von Foerster, and it affected me profoundly. Von Foerster was preoccupied with second-order problems — not what we don’t know, but the conditions that make certain things unknowable to us. His formulation has stayed with me for more than 40 years: “We can’t even see that we can’t even see.”
This is the most precise description I know of what has been allowed to happen to these students.
A student who knows they have a content gap can seek help. There are tutors, office hours, Khan Academy videos. A content gap is knowable, visible, and recoverable. But a student whose self-knowledge about their own mathematical readiness has been systematically miscalibrated over all the years of their education doesn’t experience that they’re experiencing a gap. They crash into a sudden, unforeseeable, and inexplicable wall. The blind spot isn’t in their mathematics. It’s in their ability to see their mathematics.
And this is what makes von Foerster’s diagnosis more than merely ironic: California’s K-12 math pipeline does not simply fail to correct for this blind spot. Through feedback mechanisms such as grades that certify mastery without requiring it; through courses that can be completed without deep and hard-won processes of digestion; and through the progressive elimination of standardized external reference points that could have introduced students to an honest outside perspective — we ourselves have constructed the systemic blind spot. The system manufactured the condition of not seeing.
The contrast with Stanford’s math department is instructive. They have meticulously documented their placement process and their nonnegotiables on their public-facing web pages for incoming students and their families. Stanford requires all incoming students to take their placement test prior to enrolling in a first math course — a reality check on what they actually know and where they are rusty — and enforces one hard gate: no skipping into multivariable calculus without having earned credit from the prerequisite course. Beyond that, their placement guidance is advisory. Such advisory guidance is certainly ignored by some over-confident students. But those who are recommended to start with the most introductory course and who disregard that advice often find themselves unprepared for the course they actually take and their learning suffers correspondingly.
These are Stanford admits — among the highest-achieving students in the country, accustomed to succeeding, confident in their preparation. And yet Stanford has learned that its diagnostic system has to hold firm against these students’ own self-assessment — because that self-assessment has been corrupted by years of misleading feedback prior to college. The institution is not being needlessly harsh. Instead, it is substituting its own accumulated empirical knowledge — generations’ worth of knowledge about what actually predicts success in their math sequence — for students’ “self-knowledge” that is not trustworthy on this question. The policies and their website are saying, with as much kindness as firmness allows: we know more about your mathematical readiness than you do, and we are not going to let you talk us out of that.
What Stanford’s placement tests are detecting is not merely gaps in content knowledge. They are detecting the absence of important layers of adaptive expertise and mathematical maturity that were supposed to have accumulated over years to the academic benefit and preparedness of very bright and capable students — but too often did not.
How did we arrive here? Not through any coordinated conspiracy, but through something more troubling: a broad collaboration of mutually reinforcing errors. Curriculum designers and policymakers prioritizing equity over rigor without understanding that the two are not, in fact, opposed. Counselors giving in to parents’ demands for acceleration without any of them understanding the long-term costs and risks of extreme acceleration. Districts responding to parental pressure for early advanced course placement without tools to assess and discuss genuine readiness. Colleges sending mixed signals about what solid preparation actually means. And a state mathematics framework that spent its political capital on a false binary while the real damage kept accumulating in dimensions it had no language for.
The U.C. math faculty open letter — now signed by over 2,000 STEM professors from across multiple campuses, demanding the restoration of standardized math prerequisites for STEM admissions — was shocking to many observers. But there was no good reason to be shocked. This was the university system finally saying out loud what its placement data had been telling it for years: that the K-12 pipeline has been exporting its accountability upstream and dumping it there.
Those public school graduates sitting in that Berkeley prerequisite course are not lazy. They are not stupid. They mostly did what they were told, took the courses they were advised to take, and earned the grades the system offered them. They were failed by a framework that could not see — and therefore could not name, could not measure, and could not protect — the thing they most needed.
Students are not missing a mere remediation plan. Their adaptive mathematical capacity was supposed to grow slowly, over years, through essential practice and work that was never struggled through and digested before being genuinely internalized as a part of the student’s academic self-concept. You cannot tutor back a layer of cognitive and mathematical maturity. You can only build it in the first place.
And by the time their coursework reveals what years of schooling and assessment concealed, that window is nailed shut.
