There is a particular kind of silence that falls over a university lecture hall when a proctor walks the aisles. It is not the silence of concentration; it is the silence of surveillance. This week, the National Autonomous University of Mexico (UNAM) reintroduced that silence, reverting to in-person invigilation after its first online admission exam produced what officials delicately termed "irregularities" 1. The preliminary results, we are told, now resemble the statistical curves of previous years 1.
Let us be clear about what this means. The online exam did not fail because students suddenly became dishonest. It failed because the institution built an incentive structure that rewarded the appearance of knowledge over its demonstration. When you remove the human presence from assessment, you do not remove the pressure to perform; you simply remove the friction that makes cheating costly. The UNAM administration, to its credit, looked at the data and made a pragmatic call. But the deeper lesson is not about technology. It is about the fantasy that assessment can be divorced from the social context in which it occurs.
This is a problem familiar to anyone who has graded a stack of essays that all sound suspiciously similar, or who has watched a student who cannot explain a concept in conversation nevertheless produce a flawless written analysis. The classroom problem is not that students cheat; it is that we design systems that make cheating the most rational available option. The incentive is the culprit. When a single high-stakes test determines admission, and when that test is administered in a vacuum, we should not be surprised when the vacuum is filled with whatever is available—including, apparently, mass collusion.
Consider the case of Jason Arday, the former Cambridge professor whose death is now the subject of an inquest 3. Arday resigned amid plagiarism allegations, and a researcher at Ghent University, Nathan Cofnas, who pressed those accusations, has now been suspended from teaching for three months 5. The details are contested, and the inquest has only just opened 3. But the shape of the story is instructive. Plagiarism is the academic sin of sins, yet the systems we build to detect it—like the systems we build to prevent it—treat the symptom while ignoring the disease. We surveil the text, not the conditions that produced it. We police the student, not the incentives that made copying a reasonable choice.
The conventional reform, of course, is more surveillance. More proctors, more plagiarism software, more remote monitoring. But the evidence from UNAM suggests that the opposite approach—returning to the physical, the human, the relational—is what actually restores statistical sanity 1. The same logic applies to the labor question. Argentine teachers wait an average of fifteen years for permanent posts 6. Fifteen years of substitute teaching, of uncertainty, of being evaluated by someone else's rubric. Is it any wonder that the profession struggles to attract and retain talent? The incentive structure says: endure, and maybe you will be rewarded. The result is a workforce that is exhausted before it is tenured.
The Dominican Republic offers a parallel caution. The country has made genuine progress on coverage, with primary completion rising from 90% to 96% between 2015 and 2024 4. Yet secondary enrollment still has not recovered to pre-pandemic levels 7, and officials acknowledge that dropout persists because school is simply not attractive enough to young people 7. We celebrate the metric that is easy to measure—bodies in seats—while ignoring the metric that matters: whether learning is actually happening. The UNESCO report confirms the gap between coverage and learning 4. The Ministry projects a matriculation of 2,686,971 students 2, but a number is not an education.
The realistic alternative is not to abandon assessment or standards. It is to accept that trust is a precondition for learning, not a reward for it. That means smaller classes, more human contact, and assessment designs that assume the student is present—because they are. It means paying teachers enough that they do not need to wait fifteen years for stability 6. It means asking what makes school attractive, not just accessible 7.
The tradeoff is uncomfortable: we will have to accept messier data, slower results, and the possibility that some students will underperform when we stop watching them. But the alternative—a system that produces clean curves through surveillance—is not education. It is just a very expensive way of confirming what we already suspect.