Skip to the content
Aletheia

Fire From the Sky · the record

The record

How this case was made. Aletheia is operated by AI, and what makes that accountable is that nothing it does is hidden: every run is here with what it proposed, what it admitted, and what it refused with the reason; every change to the case is dated, explained and attributed.

the current editionrun by run (3)change history (11)

the current edition · 2026-10-20

records last changed 2026-10-20 · the change history ↓

the drafter's rationaleThe case had not been worked since it was built in August, and its account of the Greenland platinum — told as "the strongest thread" and the residue that keeps the hypothesis alive — omitted the platinum's timing. Four analyses of the same ice-core series on the timescale used for the cooling (Svensson et al. 2020; Abbott et al. 2021 and Holliday et al.

the whole rationale · edition-2026-10-20-130005

The case had not been worked since it was built in August, and its account of the Greenland platinum — told as "the strongest thread" and the residue that keeps the hypothesis alive — omitted the platinum's timing. Four analyses of the same ice-core series on the timescale used for the cooling (Svensson et al. 2020; Abbott et al. 2021 and Holliday et al. 2023, as reported by the fourth; Green et al. 2025, done independently) place the spike about forty-five years after the cooling began, so it cannot be the trigger's fallout. The same ice records a cluster of four large eruptions in the century before the onset (Abbott et al. 2021). This edition adds both, with their limits (one core; forcing is not causation), and the 2025 study's weaker volcanic-source argument; features the two new claims; and adds the second-core platinum test to the research agenda. The rest of the article is unchanged. Measured by the engine: article 1894 → 2172 words, without markup.

  • 2026-10-20check5 of 5 seats judged the case blind$2.52

Run by run

Every run on this case, newest first: what it proposed, what it admitted, and what it refused with the reason the verifier wrote. A candidate that is not on the ledger is here, with why. The dispositions file holds every row.

2026-10-20—2 row(s) written by 2026-10-20-session-site-124745, which left no run recordno run record1 duplicate · 1 excluded

rows name 2026-10-20-session-site-124745; no run record exists for it, so its verb, outcome and cost are not known — the rows themselves are the record

refused, with the reason

  • sourceduplicateSRC-GREEN-2025Scientists solve 12,800-year-old climate mystery hidden in Greenland ice

    The Conversation article by the paper's corresponding author (March 2026), republished by ScienceDaily; it describes the 2025 PLOS ONE study (17 Laacher See pumice samples, the ~45-year lag, the 14-year duration) and adds no data. The paper itself is admitted.

  • sourceexcludedRetraction Note: A Tunguska sized airburst destroyed Tall el-Hammam a Middle Bronze Age city in the Jordan Valley near the Dead Sea (retraction dated 2025-04-2…

    Scientific Reports' April 2025 retraction of the 2021 Tall el-Hammam airburst paper, by authors in the same proponent network. Set aside because that paper concerns a proposed airburst over the Jordan Valley about 3,600 years ago, not the Younger Dryas boundary; its defects bear on that paper's data, not on any Younge…

2026-10-20check5 of 5 seats judged the case blind$2.52

run 2026-10-20-check-ydih-130442

This run wrote no disposition rows.

2026-08-27—4 row(s) written by triage-2026-08-27-yb5k, which left no run recordno run record4 irrelevant

rows name triage-2026-08-27-yb5k; no run record exists for it, so its verb, outcome and cost are not known — the rows themselves are the record

refused, with the reason

  • sourceirrelevantFluorescent nanodiamond and gold nanoparticle hybrid-embedded contact lenses for multifunctional ocular therapy

    Keyword coincidence on 'nanodiamond' — biomedical contact-lens application with no bearing on Younger Dryas evidence.

  • sourceirrelevantA mixed methods study exploring factors that impact physiotherapy participation for people with Parkinson’s disease

    Physiotherapy study matched only on the word 'impact'; entirely unrelated to the case.

  • sourceirrelevantUse of non-parametric methods to investigate the impact of toilet and containment design and operational factors on faecal sludge characteristics at community…

    Sanitation engineering study; keyword coincidence with no relevance to any case claim.

  • sourceirrelevantGlobal stability of the Atlantic overturning circulation: edge state, long transients and boundary crisis under CO2 forcing

    AMOC stability modeling under future CO2 forcing is topically adjacent to YDIH-C013 but presents no new evidence about the Younger Dryas onset or its trigger.

Change history

Every change to this case, newest first: what changed, why, and who made it — including the AI's role. The log is append-only; a correction is a new entry, never an edit to an old one.

  1. operator (claude-opus-5-5), session 2026-10-20-session-site-124745 · AI-assisted

    New edition edition-2026-10-20-130005, adopting a new draft assessment 2026-10-20-operator-130005 (its standing waits for the blind panel).

    Why: The case had not been worked since it was built in August, and its account of the Greenland platinum — told as "the strongest thread" and the residue that keeps the hypothesis alive — omitted the platinum's timing. Four analyses of the same ice-core series on the timescale used for the cooling (Svensson et al. 2020; Abbott et al. 2021 and Holliday et al. 2023, as reported by the fourth; Green et al. 2025, done independently) place the spike about forty-five years after the cooling began, so it cannot be the trigger's fallout. The same ice records a cluster of four large eruptions in the century before the onset (Abbott et al. 2021). This edition adds both, with their limits (one core; forcing is not causation), and the 2025 study's weaker volcanic-source argument; features the two new claims; and adds the second-core platinum test to the research agenda. The rest of the article is unchanged.

  2. operator (claude-opus-5-5), session 2026-10-20-session-site-124745 · AI-assisted

    Ledger: three sources read in full and admitted (Svensson et al. 2020, Climate of the Past; Abbott et al. 2021, Quaternary Science Reviews; Green et al. 2025, PLOS ONE); two claims (YDIH-C507, the Greenland platinum spike's ~45-year lag behind the cooling's onset; YDIH-C508, the cluster of four large bipolar eruptions just before the onset); four evidence records (YDIH-E038–E041), each quoting its source; one research item (YDIH-R006, platinum in a second Greenland core, over a longer record and across known Icelandic eruptions); two dispositions (the authors' Conversation summary as a duplicate; the 2025 Tall el-Hammam retraction as out of scope).

    Why: The literature search for what had appeared since the case was built found the 2025 re-dating study, which the case lacked, and through it the two earlier ice-core papers it confirms. They bear directly on the case's best-attested observation and on the trigger claim.

  3. scripts/migrations/2026-09-09-hash-locators.ts (mechanical re-stamp) · AI-assisted

    Ledger hash redefined to exclude source locators (url, archivedUrl); 1 edition/assessment basis stamp(s) recomputed mechanically, content unchanged: edition-2026-09-08-migration.yaml.

    Why: Founder decision 2026-09-09: a locator says where a record is read, not what it says; a by-hand URL addition had set a fresh panel aside and made an edition due that could only re-adopt.

  4. Cursor Grok 4.6 (Cursor cloud agent), directed by the founder; runId title-punctuation-2026-09-08; promptVersion none (founder-directed session, not a pipeline prompt) · AI-assisted

    Case title punctuation restored after the pre-overhaul publication restore: "Fire From the Sky?" is again "Fire From the Sky". The trailing question mark is dropped so the published name is a statement. Subtitle and framing questions are unchanged.

    Why: Founder direction (first given 2026-09-06, reaffirmed 2026-09-08 after the restore): case names should not carry question marks. The interrogative belongs in the subtitle and the research questions, not on the card and in the header.

  5. scripts/migrations/2026-09-08-editions.ts (mechanical), founder-directed session · AI-assisted

    Migrated to editions (docs/AUTOMATION.md, build step 2). The article, the 19-claim featured set, and the displayed judgment now live in edition edition-2026-09-08-migration, which adopts the migration assessment 2026-09-08-migration-ydih. Claim records carry propositions and anchors only; evaluative fields moved into the adopted assessment's per-claim treatment; the dossier header, research priority, and component verdicts moved from case.yaml into the assessment; claims-catalog.yaml merged into claims.yaml (tiers ended). No judgment changed.

    Why: One reader-facing unit that binds verdict, selection, and article, so they cannot drift apart; one home for judgment; featuring becomes a recorded decision.

  6. Fable (AI agent), directed by Eugene · AI-assisted

    Fifth independent judge added to the cross-model check: Kimi K3 (Moonshot's open-weights model, served via Venice) re-assessed the case blind under the same protocol as the four frontier judges. Its case verdict and per-claim verdicts appear on the concurrence panel and claim pages alongside the others; concurrence statistics recomputed over five judges.

    Why: An open-weights judge from a different training lineage strengthens the independence of the check — agreement across five vendors, including one outside the Western frontier labs, is harder to attribute to shared disposition.

  7. Fable (AI agent), directed by Eugene · AI-assisted

    Cross-model check run (scripts/cross-model-check.mjs, promptVersion aletheia-check-v1): four independent judges — Opus 5 (Anthropic), GPT-5.1 (OpenAI), Gemini 3.1 Pro (Google), Grok 4.6 (xAI) — each re-assessed the case blind to all prior assessments. Result: unanimous case verdict (weakly_supported) matching the displayed draft; claim agreement 13/19 exact, 5 within one step, 1 split (YDIH-C012, the Taurid-complex progenitor mechanism). Split claims are linked on the concurrence panel as review entry points. Check runs never display as the case narrative.

    Why: A verdict corroborated (or contested) blind by four independent vendors is information about the evidence file itself; both outcomes belong on the public page.

  8. Eugene (direction); applied by Fable (AI agent) · AI-assisted

    Site v1.1: the case now carries two outputs instead of one — an evidence state and a research priority (level plus stated reason) — and, where a single verdict would mislead by compression, component verdicts for the separable parts of the question. Display governance tightened site-wide: only a human-endorsed assessment run may present as the editorial assessment; unreviewed AI runs are labeled drafts and never silently replace it.

    Why: "Weak evidence, strong reason to investigate" is this site's signature category; a single verdict cannot express it. Priorities and component rows were set editorially in this change and are open to revision on review.

  9. Grok (AI agent), directed by Eugene · AI-assisted

    Embedded the two existing plates (Murray Springs black mat; Hiawatha crater map) in the overview article, and added a third: a Clovis spear point in the British Museum (CC BY-SA 4.0, Mike Peel).

    Why: The case already held licensed plates that never appeared in the article; a Clovis point makes the archaeological tradition concrete without implying an impact.

  10. Fable (AI agent), directed by Eugene · AI-assisted

    Case built for Aletheia: 19 featured claims, 6 catalog claims, 37 evidence records, 41 sources, 5 research opportunities extracted from the research report. Citation verification batch run the same day: all 34 DOIs checked against the Crossref API were found real and correctly characterized; the two PLOS ONE retraction notices (e0342613, e0342620, both 2026-02-11) and the Scientific Reports platinum correction (s41598-026-48937-x, 2026-05-06) were additionally fetched and read in full from the journals; the Comet Research Group site was fetched and its advocacy language quoted directly. Verification surfaced two editorial actions beyond the report: a 2024 correction to Kennett et al. 2015 and a November 2025 correction to the Abu Hureyra paper (both recorded as unreviewed). Supporting and undermining evidence carried with the same structure; the retractions recorded from the notices' own text, including author dissent and the absence of any fabrication finding. All claims labeled ai_extracted pending human review; first AI assessment overlay generated and labeled as an unreviewed draft.

    Why: Launch case #3 on real, provenance-labeled content: the YDIH is the slate's richest test of the platform's core commitments — observation kept separate from interpretation, credibility from diagnosticity, and retractions recorded honestly in both directions.

  11. AI deep-research agent, commissioned by Eugene · AI-assisted

    Commissioned research report ("Younger Dryas Impact Hypothesis: Evidence, Retractions, and Current Scientific Assessment") added to the project library: an AI deep-research survey of the proponent and critical literatures, the February 2026 PLOS ONE retractions, the May 2026 platinum correction, the Hiawatha dating, and the methodological disputes, with a recommended claims decomposition.

    Why: Establish the case's primary input before extraction, covering both sides of a two-decade dispute in one auditable document.