The American education system was not designed to produce scientists; it was designed to produce “interchangeable parts.” Frederick Gates, the architect of the Rockefeller-funded General Education Board, famously stated: “I do not want a nation of thinkers, I want a nation of workers.” While we like to imagine medical education has evolved, the reality is that our high-stakes testing, from the GRE to the USMLE, serves as a high-fidelity filter for compliant workers.

The “mind-reading” problem

Consider a staple of medical probability: A woman tests positive on a pregnancy test with a known 99 percent sensitivity and specificity. The question provides a “base rate” of 1 percent, meaning 1 percent of the city’s population is currently pregnant, and asks for the probability that she is actually pregnant.

In physics, we are often told to “ignore air resistance.” This is a pedagogical contract. Both the student and the teacher agree to ignore a specific force to isolate a law of gravity. But in these medical probability problems, there is no explicit contract. The exam does not say “assume a static population with zero environmental feedback.” It simply gives you a number and expects you to treat it as an unchangeable constant.

It is not testing your ability to model reality; in the real world, that 1 percent prevalence is a volatile variable governed by nutrition, environmental constraints, and culture. By failing to make the assumption explicit, the test trains students to ignore the medium entirely.

These questions are testing your ability to read the examiner’s mind and accept an arbitrary constant as truth. In short, the exam tests your ability to comply with unspoken rules.

Compliance over complexity

The pregnancy test problem is not an isolated pedagogical quirk. It is a microcosm of how the entire medical education pipeline operates: present a closed system, reward those who accept it without question, and penalize those who probe its assumptions.

Consider what the MCAT actually tests: whether you can select the correct answer from within a closed model. Absorb the information, reproduce the pattern, move on. The student who does this efficiently is rewarded. The student who stops to ask “does this model hold under these conditions” is penalized, not for being wrong, but for not playing the game.

The product of this pipeline is not a physician. It is a lookup table in a white coat, credentialed, billed through, and defended in court.

The selection pressure is not subtle. The student who raises a hand and says “this base rate does not apply to the population I want to serve” is not rewarded for clinical reasoning. They are penalized for not selecting the answer the examiner wanted. Do that enough times and one of two things happens: You learn to stop asking, or your score drops and you do not match into the residency you wanted. Either way, the pipeline got what it needed. The questioner was either converted or removed.

But selection is only the first stage. What follows is reinforcement.

Residency is where independent thinking goes to die. An 80-hour work week, and that is the reformed version, does not produce physicians capable of nuanced, context-sensitive clinical reasoning. It produces people operating under chronic cognitive impairment who default to protocol because protocol is what you can execute when you have not slept.

This is not a side effect of residency. It is the function of residency. A sleep-deprived resident does not question the attending. A sleep-deprived resident does not re-derive the base rate for the patient in front of them. A sleep-deprived resident follows the algorithm, charts the output, and moves to the next bed. The system broke them in, the same way a 16-hour shift on a factory floor broke in the workers Frederick Gates was so proud of designing a school system for.

And it does not end at graduation. Maintenance of Certification, MOC, is the loyalty oath that never expires. Every few years, physicians pay thousands of dollars to sit for another exam that tests the same closed-system pattern recognition they were selected for in the first place. It does not ask whether they have gotten better at reasoning under uncertainty. It does not evaluate whether they can identify when a protocol fails a specific patient. It asks whether they can still reproduce the sanctioned patterns. It is a recurring pledge of allegiance to the model, not to the patient.

The structure is identical to any system that demands continuous ideological reaffirmation. To pick one example out of a multitude of others, from cults to companies to governments: In Mao’s China, it was not enough to comply once. You attended the struggle sessions. You wrote the self-criticisms. You publicly affirmed the party line at regular intervals, not because the party doubted your loyalty, but because the act of repeated affirmation is the mechanism of control.

It ensures that deviation does not just carry a one-time cost at the gate. It carries a recurring cost, at every stage of your career, for the rest of your professional life. The message is constant and clear: The model is not yours to question.

This is not a failure of the system. This is the system performing exactly as designed.

Robert Trent is a graduate student.