The September Cluster

Every claim below is quoted as entered, scored at the probability recorded at issue, against criteria written before the outcome was known. No criterion was edited after the fact. The combination map for the synthesis question was frozen on 13 September. The day got to edit nothing but the observations.

CL-041 — The MOU Window. Claim as entered (1 Sep): "The United States will signal return to the June MOU framework within the 14-day window closing 15 September." Probability at issue: 50%. Pre-written criterion: any documented U.S. diplomatic signal toward MOU re-engagement before midnight 15 Sep. Observed outcome: contrary record only — aviation sanctions, the Salalah-day bank designation, "soured on diplomacy" reporting. Window closed empty. Verdict: FALSE (posted 15 Sep, GS-RRA-010). Brier: 0.2500. What the method got wrong: the 50% price was hope-shaped — the hardline-sabotage record was already in the file when the coin-flip was priced. Process fix: entries touching a track already sabotaged once carry a mandatory base-rate annex — priced from the record, not the wish.

CL-054 — The Chabahar Maps. Claim as entered (6 Sep): "Iran will publish the enlarged exclusion-zone coordinates reaching toward Chabahar; the zone's actual scope will reveal whether the announcement was capability or theater." Probability at issue: 65%. Observed outcome: published 06:00 Tehran, 17 Sep; extent ~12nm beyond port approaches; the Omani side of the middle route remains open. The promised escalation arrived as calibration. Verdict: TRUE. Brier: 0.1225. (Reopened 19 Sep — see the post-issuance note.)

CL-056 — U.S. Damage Disclosure. Claim as entered (10 Sep): "The U.S. government will formally acknowledge the Muwaffaq Salti aircraft damage within the week." Probability at issue: 40%. Observed outcome: service-level candor (acting Navy Secretary's Bahrain admission, 13 Sep) but presidential- and Pentagon-level non-disclosure. Verdict: PARTIAL (0.5 outcome — admission occurred below the level the claim specified). Brier: 0.0100. Process fix: disclosure claims must name the podium — "the government" is not a resolvable actor.

CL-060 — Iranian Toll Flexibility. Claim as entered (13 Sep): "Iran's suspension of its 10% freight charge will hold through the Salalah window." Probability at issue: 60%. Observed outcome: suspension held through the postponement and into the 21 September acceptance cascade; PGSA fee authority dormant, not revoked. Verdict: TRUE. Brier: 0.1600. The cluster's most structurally important TRUE — the corridor's economic layer is behaving more diplomatically than its military layer.

CL-055 — Poland Article 4 (resolved early, posted GS-RRA-012). Claim (6 Sep): Article 4 consultations within 14 days of the 10 Sep incursion. Probability at issue: 65%. Convened 17 Sep. Verdict: TRUE. Brier: 0.1225.

CL-070 — The Salalah Proposer Discriminator (interim). Observed: Oman proposed (16 Sep); UAE/Qatar accepted within 6h on "regional consensus"; Riyadh accepted on timing language (17 Sep, 05:40). Interim verdict: the frozen reading tree resolves to mediator-proposed, bloc-choreographed, Saudi-permitted. Final scoring transfers to CL-072 (agenda publication, 21 Sep).

CL-068 — Qeshm Attribution (process ruling). Mine signature confirmed by two independent source families; no claimant. Ruling: UNSCORED — the entry was created without a probability at issue, and the Ledger refuses to retrofit one. Process fix (binding from that night): no attribution entry is accepted without a probability and resolution date at entry.

CL-069 — BIS-EVT-001 Ratification. Ratified by directorate vote, 5–0, this date. Per its own terms, the standard is scored from October on screening yield — first measurement at the 2 October Calibration Report.

• Cumulative Brier mean: 0.1250 — target ≤ 0.20 by Q4, met on first accounting. A flat 0.50 forecaster scores 0.2500 on this set; the Ledger beats its own coin-flip by half.
• Resolution rate on matured entries: 100% (10/10; no UNRESOLVED-EXPIRED — the 0.30 penalty has never been applied; someday it will be, and the Ledger will carry it).
• Corrections to date at issuance: 4 (COR-1 through COR-4), latency ≤ 2 cycles each — standard met.
• Overconfidence index — first finding, and it is the honest kind of uncomfortable: the system is currently under-confident. Entries priced at 0.60–0.65 resolved TRUE at 83% (5 of 6); the single 0.50 entry resolved FALSE. Small-sample caveat stated in the same sentence as the finding. Logged for the 2 October Calibration Report: recalibration review — but no probability editing without 20+ scored entries; calibration is earned in volume, not tuned in hindsight.

Frozen combination map (13 Sep) applied to observed outcomes: MOU track dead (CL-041 FALSE) · corridor enforcement geography moderate (CL-054 TRUE, calibrated scope) · economic flexibility alive (CL-060 TRUE) · regional diplomatic track reconvening 21 Sep under Omani custody (CL-070 interim).

Verdict, per the frozen map: ALIVE-DISPLACED. The off-ramp architecture survived — but it migrated from the U.S.–Iran MOU track to the GCC–Iran regional track under Omani mediation, with Saudi permission priced in timing language and Iranian flexibility priced in a suspended toll.

The hope-control held: the Agency's documented closeness to this story did not edit a single criterion. New entry CL-074: the 21 September Salalah round will produce a documented deliverable (agenda, communiqué language on fees, or a dated follow-on) — probability 55%, resolves 22 September. This is the displacement track's first scored test.

This report scores a week the Agency mostly called correctly — and the Ledger's design assumes that is precisely when self-deception is cheapest. The record shows: one hope-shaped price (CL-041, caught by the base-rate fix), one unscoreable entry accepted in violation of the discipline we enforce on others (CL-068, fixed tonight), and a calibration finding that flatters no one — under-confidence is still mis-calibration, and a warning system that systematically under-prices will eventually cry wolf in reverse. The mean of 0.1250 is a beginning, not a vindication: ten resolutions is a sample, not a track record. The Ledger's value is proven on the weeks it fails. This was not one of those weeks, and that is exactly when this paragraph matters most.

Ledger stands at 78 active entries at issuance (CL-072, CL-073, CL-074 added this cycle). Next events: Salalah, 21 September (CL-072/CL-074); framing audit № 2, 20 Sep; Berlin/MeckPomm votes, 20 Sep; OTH hash-freeze 26 Sep; OTH run 28 Sep; first monthly Calibration Report 2 Oct. SRR-002 will be scheduled when the next resolution cluster arms.