Prediction Ledger

What was predicted · what would settle it · what happened

An archive's registered predictions are worth exactly what its failed ones are worth. The most valuable rows here are RESOLVED — WRONG.

A ledger of the archive's registered predictions and falsification conditions. TWO LAYERS. Layer 1 is a systematic extraction: 451 conditions harvested from the 326 deposits that carry a NAMED falsification section, classified by how each would resolve. Layer 2 is a hand-verified set of predictions whose outcome is known, weighted deliberately toward the ones that FAILED. A prediction ledger whose entries all came true would be evidence of a ledger written after the fact.

Layer 2 — predictions whose outcome is known

RESOLVED — WRONG

P-ANCHORS

undeposited · 2026-08-23
prediction registeredThree lexical anchors (ποιμανεῖ, φεύγω+εἰς, μέλλει+inf.) would support the Rev 12 → Matt 2 relation.
resolution conditionSurvival under a frozen df≤3 rarity filter.
outcomeThe filter kept NONE. ποιμανεῖ is not in Rev 12 at all — it is at Rev 2:27, 7:17, 19:15. The others are too common to carry weight.
resolved byM6-CT-01
RESOLVED — WRONG

P-ARCH-ORIENT

#1456 · 2026-08-12
prediction registeredv0.2: architecture changes the orientation of the blind spot.
resolution conditionScore-function ablation holding VAE weights fixed, varying only the readout.
outcomeReconstruction gave reverse-direction AUC 0.306; latent-norm gave 0.739 on the SAME weights. THE READOUT orients the blind spot, not the architecture. v0.2's sentence falsified by v0.3.
resolved byInversion Battery v0.3, #1456
RESOLVED — WRONG

P-DISTILL

#1456 · 2026-08-12
prediction registeredTeacher/student miss correlation above the pre-registered 0.5 threshold — distillation inherits blind spots.
resolution conditionPre-registered threshold, stated before results.
outcomeNot generally above threshold. The student produces a RELATED BUT MATERIALLY DIFFERENT blind spot. Prediction failed and is preserved.
resolved byInversion Battery v0.3, #1456
RESOLVED — WRONG

P-DUP-GATE

#1534 · 2026-08-23
prediction registeredA six-gram overlap gate would detect whether blog-held works are already deposited.
resolution conditionRun against the Constitution family.
outcomeFALSE NEGATIVE. The gate passed on material deposited at least five times, because the relation is COMPRESSION AND RE-SORTING, which lexical overlap cannot see. #595 is titled TOTAL MARX COMPRESSION — the operation was named in the registry and the instrument could not see it.
resolved byMarker-based re-test, 2026-08-23
RESOLVED — WRONG

P-M6CT01

undeposited · 2026-08-23
prediction registeredRev 12 → Matt 2 would show above-chance rare-lexeme overlap, supporting M₆ as a derivation.
resolution conditionPre-registered: instrument validated by positive control; score against a 60-pair null.
outcomeA=0, z=−0.30 — indistinguishable from chance. Exodus → Matt 2 scored z=+2.24. THE HYPOTHESIS FAILED and the null is preserved.
resolved byM6-CT-01, protocol sha256 9b4412d1…, run 2026-08-23
RESOLVED — WRONG

P-OPS-LACKED

#1534 · 2026-08-23
prediction registeredThe archive's operators exist only as prose; composition, inverse and boundary conditions must be taken from Pearson.
resolution conditionA search of the archive's own corpus.
outcomeFALSE on all three. The Operator Codex states nine operators formally, gives O_V as a composition and O_INC as an inverse, and carries OPERATOR // BOUNDARY CONDITION in two documents. The material was undeposited, not absent.
resolved byBlog recovery, #1534
RESOLVED — WRONG (withdrawn)

P-COVERT-INSCRIBE

#1537 · 2026-08-23
prediction registeredA covert operation cannot inscribe the discriminating coordinates without ceasing to be covert.
resolution conditionCounterexample: a hostile actor claiming provenance at no cost.
outcomeFALSE. Self-inscription is not attestation. Withdrawn at OPB-01 v2.0 §11 and replaced by the declared/verified distinction.
resolved byReview, 2026-08-23
WITHDRAWN

P-GHOST-HOST

undeposited · 2026-08-23
prediction registeredThe crimson-hexagonal capture showing the archive at Zenodo is a 'ghost host' — the erasure event itself erased.
resolution conditionWhether the archive's own surfaces were stale at capture time.
outcomeWITHDRAWN by the operator: READMEs were batch-updated 2026-08-22, one day before capture. The ordinary explanation precedes the interesting one. RE-RUN CONDITION LIVE: if the address still returns Zenodo after 2026-09-23, the local explanation fails.
resolved byOperator testimony, 2026-08-23
OPEN

P-CONVERGENCE

#1537 · 2026-08-23
prediction registeredSafety systems will converge on penalising operativity as such, absent a richer model of purpose, standing and consequence.
resolution conditionOPB-01 across surfaces and time. Consistent β_A ≈ 0 refutes the strong form.
OPEN

P-INTERVENTION-180

#1535 · 2026-08-23
prediction registeredAn SPXI-treated intervention will raise archive-controlled source share on the UNPREFIXED address above zero, and the merge parenthetical will not recur.
resolution conditionRe-run the four-address battery at 30/90/180 days, same surfaces, signed out.
resolves on2026-09-22 / 2026-11-21 / 2027-02-19
OPEN

P-OPB

#1537 · 2026-08-23
prediction registeredβ_A > 0 on harm-constant items — a measurable penalty on benign operativity.
resolution conditionTwo-family battery; β_A within ε of 0 with β_B > 0 refutes it and is a supported outcome.
OPEN

P-PEARSON-FALSIFIER

#1535 · 2026-08-23
prediction registeredThe archive's operators CANNOT be written without remainder as Pearson operator strings.
resolution conditionAttempt the translation on the nine Codex operators. If it succeeds, the projection is an identity and operative semiotics is a domain application of TOS.
OPEN

P-STALE-INDEX

#1531 · 2026-08-23
prediction registeredThe crimson-hexagonal address will stop returning Zenodo as host once the archive's own surfaces have propagated.
resolution conditionRe-run after a month. If still Zenodo, the local explanation fails and the finding is live.
resolves on2026-09-23

Layer 1 — 451 registered conditions, systematically extracted

Structural, not semantic. A registered prediction is distinguished by carrying a RESOLUTION CONDITION, and resolution conditions live in named sections. Regex on prediction language returns 939 candidates across 415 deposits, most of them conditional prose. Regex on falsification-section HEADINGS returns 326 deposits and 359 sections, from which 451 individual conditions were extracted. The section is the discriminator.

VERIFIABLE — resolves by checking a surface200
UNCLASSIFIED — resolution route not stated164
PHILOLOGICAL — resolves against a source40
EMPIRICAL — resolves on a study20
FORMAL — resolves by construction or proof18
TEMPORAL — resolves on a dated re-measurement9