Supercooled water has a liquid–liquid critical point.
Can very cold liquid water separate into two liquid forms, with their boundary ending at a critical point? The test fixes a temperature and pressure range and excludes ice.
No recorded prices yet. Trading prices appear once pools are live.
An AI dot opens that round's saved record. Each dot is one recorded reading, thinned to hourly for readability; the underlying samples are every few minutes. The line between two dots is not a measurement. How the AI estimate is made.How to read this chart
Weekly AI estimate
The average of three models reading the same evidence. It is an estimate, not proof the hypothesis holds, and it does not set the pool price.
- Range
- 35–76%
- Verdict
- Undecided
The panel's confident binary this round Undecided the models did not all agree
Market definition · YES, NO and evidence rules
This is the frozen contract wording. Read the YES and NO tests, the evidence requirements and the undecided cases before taking a position.
Canonical title
Pure water has a metastable liquid–liquid critical point between 150 and 250 K.
Exact hypothesis · technical wording
Bulk pure H2O has a metastable first-order coexistence line between two ergodic liquid phases, ending at a critical point with 150 <= T <= 250 kelvin and 0 < P <= 500 megapascals. Metastability is relative to crystalline ice; transient amorphous solids and nanoconfinement transitions do not count as bulk liquid phases. The search domain is a contract convention.
Each round reads the evidence that was public at a stated cutoff. Later evidence goes into a later round; the market itself has no expiry date.
How the evidence cutoff is fixed
A round only counts evidence published before its own cutoff time, and the round's number is that timestamp. So a round cannot borrow a later cutoff, and an earlier round's judgment cannot be reused for a later one. The hypothesis wording, the resolution criteria and the methodology identifier were hashed into the market account at creation, which is what stops any of them being edited afterwards.
Resolution criteria
YES: Independent experiments establish two equilibrated liquid phases before crystallization, a terminating coexistence line and critical scaling within the stated domain, with structure/dynamics excluding ice and amorphous-solid conversion. A force-field simulation alone is insufficient. NO: An independently validated thermodynamic reconstruction excludes any such coexistence line and critical endpoint throughout the complete stated domain, including uncertainty and metastable sampling. Failure to reach the domain or crystallization before equilibration is UNDECIDED, not NO. Reference sources: You et al.: experimental evidence of a liquid-liquid critical point (https://www.science.org/doi/10.1126/science.aec0018); Constraints on water's liquid–liquid critical point (https://www.nature.com/articles/s41567-024-02761-0); Second critical point in two water models (https://www.science.org/doi/10.1126/science.abb9796). Evidence policy: use public primary research with methods and uncertainty, plus the named reference sources below. A preprint can inform the estimate but cannot alone meet a terminal criterion; require a peer-reviewed primary result and an independent reproduction using non-overlapping observations or independently prepared samples, unless this is a mathematical proof or the specific criterion instead requires independently implemented calculations. Report data provenance, exclusions, calibration, statistical model and systematic uncertainty. Independence means different data acquisition, not two papers reanalysing the same sample. For a quantitative terminal test, both independent results must meet the stated threshold; disagreement or an interval crossing the boundary means UNDECIDED. Use the final corrected version public at the signed round's evidence cutoff (UTC); exclude later evidence until a later eligible round. Every report in a terminal round must use the same cutoff, covered by the verified signed context and enforced by the configured terminal protocol; a different cutoff requires a new eligible round, never reinterpretation of a signed report or mutation of an immutable binding. This market has no calendar expiry. Retracted findings cannot supply a terminal result. If a named source disappears, use its DOI/publisher or an archived identical version, then an equivalent primary publication applying these unchanged definitions; if none is accessible, remain UNDECIDED, never substitute a different test. Missing data, failed replication, a null search outside the specified domain, or an untested alternative is not automatically NO. These are instructions for evidence assessment, not an automatic truth oracle: scientific interpretation and systematic-error judgment remain. Settlement follows the configured platform mode: terminal mode requires one valid round with all three frozen models returning the same explicit YES or NO and criterion_met=true; any UNDECIDED, failure or disagreement leaves it unresolved. Legacy proposal/challenge mode requires its recorded review process. Model agreement does not prove scientific truth. Invalidity is not NO and this text does not authorize a refund or cancellation: only a cancellation rule disclosed before exposure, if the configured contract supports it, can apply. Otherwise pause new entries and preserve holders' existing rights. Never reinterpret terms after listing.
The definition does not change when an AI estimate changes. Read the listing policy and this market's settlement status.
How each model argued
Three models read the same evidence and did not fully agree. Their average is 0.58.
This round ran under methodology panel-v9. Short excerpts are below; the round 5 page keeps the full answers, sources and raw responses. The current methodology and example requests do not reconstruct this historical round's exact request.
OpenAI UNDECIDED 0.76
openai/gpt-6.1-sol
The experimental critical-region signatures and molecular-model results support the hypothesis, interpreted as equilibrated bulk liquids rather than transient amorphous conversion. However, the accessible evidence does not establish…
Would change this view: Independent pressure-calibrated experiments reproducing equilibrated liquid coexistence and endpoint scaling before crystallization would substantially increase my estimate and could establish YES.
Anthropic UNDECIDED 0.35
anthropic/claude-sonnet-5.5
You et al. (Science 2026) is a peer-reviewed experiment that places an LLCP near 210 K and about 100 MPa, which is inside the domain. It is a single study using transient liquids heated from amorphous ices. It has no independent…
Would change this view: An independent, differently prepared experiment showing critical scaling and a coexistence line would move this toward YES; a validated reconstruction excluding the critical point in the whole domain would move it toward NO.
xAI UNDECIDED 0.63
x-ai/grok-4.7
You et al. (Science, 2026) report critical-like scattering and a heat-capacity rise at 210±8 K after ultrafast heating of amorphous ices, and independent simulations place an LLCP near 198 K and ~127 MPa. That is supportive but not…
Would change this view: Independent replication of equilibrated bulk coexistence and in-window critical scaling would support YES; a validated full-domain exclusion would support NO.
What moved the number
This is a new contract, not a migrated position or a continuation of old forecast history. The initial 0.50 is an initialization convention, not an AI forecast or an assessed scientific probability. Reference links are background sources,…
OpenAI estimated 0.76. The experimental critical-region signatures and molecular-model results support the hypothesis, interpreted as equilibrated bulk liquids rather than transient amorphous conversion. However, the accessible evidence…
This round's saved score, answers and sources →Live panel did not return all three estimates (OpenAI, Anthropic, xAI failed or gave an unusable answer). A round is only live when all three are valid, so this round is a placeholder at the previous score rather than an average of two.
This round's saved score, answers and sources →OpenAI estimated 0.78. You et al.’s ultrafast scattering and heat-capacity signatures support a critical region, while molecular simulations independently support plausibility within the specified domain. However, the retrieved primary……
This round's saved score, answers and sources →AI estimate scheduled: an independent three-model panel reads the published definition and evidence every week. New positions are still closed for this market, so the estimate is here to read, not to trade against.
The evidence packet · 6 entries
What the panel was shown when this round ran. A row with no source link is a description only, and the panel is told so.
You et al.: experimental evidence of a liquid-liquid critical point
ContextPublished 26 March 2026: ultrafast heating and x-ray scattering report critical-region signatures and a heat-capacity increase at 210 +/- 8 K. This is direct experimental evidence to assess, not automatic satisfaction of the independent-replication, pressure-domain and liquid-equilibration criteria.
Source linkConstraints on water's liquid–liquid critical point
ContextComputational constraints predate the 2026 experiment; distinguish model predictions from experimental measurements.
Source linkSecond critical point in two water models
ContextPrimary computational evidence, explicitly for molecular models.
Source linkRequired positive observation
SupportingIndependent experiments establish two equilibrated liquid phases before crystallization, a terminating coexistence line and critical scaling within the stated domain, with structure/dynamics excluding ice and amorphous-solid conversion. A force-field simulation alone is insufficient.
Description only · no source link recordedRequired contrary observation
ContraryAn independently validated thermodynamic reconstruction excludes any such coexistence line and critical endpoint throughout the complete stated domain, including uncertainty and metastable sampling. Failure to reach the domain or crystallization before equilibration is UNDECIDED, not NO.
Description only · no source link recordedEligibility and unresolved outcomes
MethodEvidence policy: use public primary research with methods and uncertainty, plus the named reference sources below. A preprint can inform the estimate but cannot alone meet a terminal criterion; require a peer-reviewed primary result and an independent reproduction using non-overlapping observations or independently prepared samples, unless this is a mathematical proof or the specific criterion instead requires independently implemented calculations. Report data provenance, exclusions, calibration, statistical model and systematic uncertainty. Independence means different data acquisition, not two papers reanalysing the same sample. For a quantitative terminal test, both independent results must meet the stated threshold; disagreement or an interval crossing the boundary means UNDECIDED. Use the final corrected version public at the signed round's evidence cutoff (UTC); exclude later evidence until a later eligible round. Every report in a terminal round must use the same cutoff, covered by the verified signed context and enforced by the configured terminal protocol; a different cutoff requires a new eligible round, never reinterpretation of a signed report or mutation of an immutable binding. This market has no calendar expiry. Retracted findings cannot supply a terminal result. If a named source disappears, use its DOI/publisher or an archived identical version, then an equivalent primary publication applying these unchanged definitions; if none is accessible, remain UNDECIDED, never substitute a different test. Missing data, failed replication, a null search outside the specified domain, or an untested alternative is not automatically NO. These are instructions for evidence assessment, not an automatic truth oracle: scientific interpretation and systematic-error judgment remain. Settlement follows the configured platform mode: terminal mode requires one valid round with all three frozen models returning the same explicit YES or NO and criterion_met=true; any UNDECIDED, failure or disagreement leaves it unresolved. Legacy proposal/challenge mode requires its recorded review process. Model agreement does not prove scientific truth. Invalidity is not NO and this text does not authorize a refund or cancellation: only a cancellation rule disclosed before exposure, if the configured contract supports it, can apply. Otherwise pause new entries and preserve holders' existing rights. Never reinterpret terms after listing.
Description only · no source link recordedPast estimates
Every weekly estimate, newest first
| Date | Estimate | Range |
|---|---|---|
| Nov 10, 2026 09:21 PM EST | 58% | 35–76% |
| Nov 09, 2026 09:40 AM EST | 61% | no live reading |
| Nov 02, 2026 09:40 AM EST | 61% | no live reading |
| Oct 26, 2026 10:40 AM EDT | 61% | no live reading |
| Oct 19, 2026 07:08 PM EDT | 61% | 42–78% |
Discussion
Start a threadNo threads yet. Argue about the evidence, the panel, or whether the hypothesis is worded fairly.
Anyone can join the discussion. Sign in with a wallet to keep your threads together.