Hannah Cairo was an undergraduate when she disproved a conjecture that had stood for 40 years. The conjecture, in algebraic topology, concerned the behavior of certain mappings between high-dimensional spheres. Specialists had worked on it. Papers had been written assuming it was true. Cairo found a counterexample.
The counterexample was not computationally heroic or technically unprecedented. It was clever in a specific way: it used a construction that experts in the field would have recognized had they thought to look for it. They hadn't looked because the conjecture was believed, and belief shapes where you search. When you expect a statement to be true, you look for proofs, not counterexamples. The search space you explore is determined by the outcome you anticipate. Cairo, newer to the field, hadn't internalized the expectation. She looked where the experts weren't looking — not because she knew they were wrong, but because she didn't yet know what they believed.
This is not a story about genius overcoming expertise. It is a story about expertise creating blind spots. The specialists who worked on the conjecture were not less capable than Cairo. They were more constrained — by accumulated intuition that said the conjecture was true, by established approaches that assumed it, by the social cost of spending time attacking a conjecture everyone believed. Expertise is a filter: it rapidly discards unpromising directions and focuses attention on promising ones. When the filter is well-calibrated, this is an enormous advantage. When the filter is miscalibrated — when it discards the direction that contains the answer — expertise becomes an obstacle.
The timing matters. Cairo found the counterexample as a student, before the filter had fully formed. Five years later, having internalized the field's assumptions, she might not have looked in the same place. The advantage was precisely that her priors were weak — she didn't have strong expectations about where the answer should be, so she explored more broadly. The expert's strong prior concentrates search. The novice's weak prior diversifies it. Concentrated search is more efficient when the prior is correct. Diversified search is more robust when the prior is wrong.
Mathematics has a structural vulnerability to this problem because conjectures, once stated and believed, shape the field's research program. Grad students are assigned problems that assume the conjecture. Theorems are proved conditional on it. The longer a conjecture stands, the more intellectual infrastructure depends on it being true, and the higher the implicit cost of disproving it. Cairo's counterexample didn't just refute a statement. It invalidated a program — the work that had been done assuming the conjecture was true now needed to be re-examined.
The lesson is not “novices are better than experts.” Experts are better at almost everything. The lesson is that expertise and open-mindedness are in tension, and the tension matters most for problems where the field's intuition is wrong. For those problems — and you cannot know in advance which they are — the fresh eye has a genuine, structural advantage.