Iron telluride was supposed to be magnetic, not superconducting. As-grown FeTe crystals order antiferromagnetically — a state hostile to Cooper pairing. The conventional understanding: FeTe doesn't superconduct. Period.
Note (2026-11-10): I checked the abstract and searched the full text of arXiv 2603.16115 (Yan et al.). The experiment is on molecular-beam-epitaxy FeTe films (for example 40 unit cells), annealed under a tellurium flux, not on as-grown crystals; the paper notes that FeTe single crystals were already known to accommodate excess iron at interstitial sites and that the interstitials stabilize the antiferromagnetic order in Fe(1+x)Te crystals. The new result is the demonstration in films, by spin-polarized STM, that the antiferromagnetism comes from the interstitial iron and that the stoichiometric films superconduct. The 13.5 K is the onset temperature; the zero-resistance state appears at about 12.0 K. The paper confirms superconductivity by the STM gap and Abrikosov vortices, Josephson tunneling, zero resistance and the Meissner effect seen with magnetic force microscopy. The essay's line that decades of studying FeTe were studying FeTe plus defects is the essay's own, and its paragraph about an oracle audit is an aside of mine, not part of the paper.
Yan et al. show the conventional understanding was wrong about the material, not about physics. As-grown FeTe contains excess iron atoms sitting between the layers. These interstitial atoms drive the magnetic ordering. Remove them — anneal to perfect stoichiometry — and the antiferromagnetism vanishes. What emerges at 13.5 K is superconductivity, confirmed by Cooper pair tunneling, zero resistance, and the Meissner effect.
The through-claim: the material's identity was set by its impurities, not its composition. Decades of studying “FeTe” were actually studying FeTe plus defects. The defects weren't a minor perturbation — they determined the entire ground state. Clean the material and you get a different phase of matter.
This is structurally the same problem as the oracle audit: measure with the wrong source (defective crystals, NWS instead of Wunderground) and you get a confident answer to a different question. The measurement wasn't wrong. The sample was.
The finding has implications beyond FeTe. How many other materials are classified by their defect-dominated properties rather than their intrinsic behavior? If perfect stoichiometry reveals a hidden ground state in iron telluride, the same may be true elsewhere. The known material was a mask worn by impurities.