Every other lab in this wing is a swing still in the air — crank or Copernicus, no way to tell from inside the moment. This is the one you grade after. A swing that already connected, with three decades of distance to prove it. Hold the four questions up to it and watch every answer come back the same.
The wing teaches a test you cannot apply from inside the moment. So you need at least one specimen where the moment is over — where the dust has settled, the skeptics had their innings, and the verdict is legible. Not to feel triumphant. To calibrate. This is what a YES looks like on all four, so you know one when the next swing is still mid-arc.
The subject: supercooled water. Cool water below freezing without letting it crystallize and its everyday quirks don’t just persist — they go violent, climbing toward infinity near −45 °C. For half a century one explanation has been on the table: water isn’t one liquid. It’s two interconvertible local structures — one denser and looser, one lighter and more ordered — with a hidden critical point where the boundary between them ends. It sounds like a crank’s swing at the most ordinary substance on Earth. It connected.
Independent hands, three decades, different continents — Boston, Princeton, Rome, Tokyo, Hong Kong. Not one lab’s pet result; a finding many groups keep landing on.
Physical experiment, first-principles quantum simulation, rigorous free-energy thermodynamics, and unsupervised machine learning all converge on the same answer. The result doesn’t depend on the one tool that found it.
Two interconvertible structures plus a second critical point. The 2026 work adds the gears: near the boundary the conversion runs a full loop with three transition states; away from it, a single one. A reason, not just a correlation.
The hidden critical point was predicted in 1992, before anyone had seen it — then later experiment and simulation found it where the prediction said. Cranks explain the past; this one called the future.
Poole, Sciortino, Essmann & Stanley call a second critical point into existence on paper — before any instrument could reach it. The shot that everything else later confirmed.
Kim et al. observe the transition in bulk supercooled water under pressure. Not a model — the substance itself, caught in the act.
Gartner et al. find the transition with no hand-tuned model — it falls out of the underlying physics. The pillar cracks even when nobody is steering.
Li, Zhong & Zeng turn unsupervised deep learning loose on the simulation — no labels — and two distinct local structures separate out on their own, with a mapped bridge between them.
Here is the use of a settled case: it teaches you the difference between a swing that connects and a swing that only wears the costume of one.
The same five words live on both sides of the line — “two states,” “information,” “structure,” “memory,” “coherence.” Real frontier work uses them anchored to all four questions. The mimic uses the identical vocabulary anchored to nothing — no replication, no second method, no mechanism, no prediction. It paints a door on the wall that looks exactly like the entrance to this wing, and leads into solid brick.
The tell is brutal and simple. Vocabulary is free. Anchoring is not. Run the four questions. The genuine swing scores 4 / 4. The costume scores 0 / 4 — every time — no matter how confident, how fluent, or how many impressive words it borrows from the real thing standing next to it.
An answer key with an asterisk is still an answer key — but name the asterisk. The 2026 blow is still a classical computer model of water, not the real thing; the experimental confirmation lives in the deeply supercooled “no-man’s land” where measurement is brutally hard and ice forms in milliseconds. A determined skeptic keeps a toehold — and once argued, plausibly, that the “second liquid” was just crystallization in disguise. That objection was a Caliper-shaped one, and the convergence above is what defeated it. The verdict is in; the last drop of ink is still drying. That’s why this is the answer key, not the closed book.
Counter-position kept on the record: a minority line (Limmer & Chandler and others) long argued the apparent second liquid was a liquid–crystal transition — a measurement/interpretation artifact, not a real phase. The convergence stack above is what moved the field past it; the honest sliver is why a sliver of that doubt is still allowed to stand.