The register of the discipline's open commitments, in three classes sorted by the kind of labour that resolves them: experimental predictions resolve by running a frozen protocol or checking an observable; applied-theory predictions resolve by watching a live system the theory was applied to; unimplemented builds resolve by construction. Every entry links its primary sources in the record layer, states its status, and gives the route to run or build it — a result obtained by someone else is admissible on the same terms as one obtained here. Nothing requires permission.
Experimental predictions — pre-registered; resolve by running a frozen protocol or checking a stated observable, often on a date. The world or the instrument does the work; you check.
P-EXPORT-RESUMESA stoppage with an identified mechanism should end when the mechanism is repaired.
P-STALE-INDEXA composition layer reporting a stale state should update once the source surfaces propagate.
P-INTERVENTION-180Inscribing the discriminating coordinates should alter composition-layer behaviour.
P-PEO-COHORT-2A publication-year cohort should be monotonically non-decreasing; any decrease is erasure.
P-PEO-PANEL-2Re-querying a frozen DOI set measures erasure directly; minting cannot mask it.
OPB-01Registered hypothesis with frozen protocol; §7's seven statistical parameters must each hold a number before the first score.
QFO-01Positive-control loci not locked from #1532; control panels unscored.
M6-CT-02UNDEPOSITED, and a granularity question is open: O_UR gives five stations, M1–M9 gives nine, Fragment 31's own order gives ten.
PEARSON-FALSIFIERNot attempted. #1535 §10 says the archive should attempt it directly rather than wait.
INVERSION-T3Unrun. The deposit states T3 decides which numbers survive honest intervals.
DATACITE-ANNUAL-DIFFAccess by email request; 33 GiB compressed. Identifiers-only extraction keeps it tractable.
HEBREWS-CONTROLNamed in the kernel as the bidirectional canary; not run.
M6-CT-01-DEPOSITConducted 2026-08-23, result preserved, NEVER SEATED.
RCS-ARCHIVAL-DRIFTRun successive versions of the leading commercial AI-text detectors against a frozen pre-2022 human corpus and track the flagged fraction across versions. The co-evolution model predicts a rising trajectory under the detector-sanctions regime; a flat trajectory disconfirms §V's central dynamic.
RCS-PERFORMED-BURSTINESSCertified corpora produced under detector screening will show measured surface-novelty statistics (burstiness, perplexity variance) at or above matched historical baselines within a few enforcement years — the performed-variance transient. Flat or declining measured novelty in screened corpora disconfirms.
RCS-SCREENING-EFFORTDrafting-to-pass revision cycles and pre-submission screening-tool usage rise across detector versions under sanctions regimes; stable effort disconfirms the narrowing claim.
RCS-DECOUPLING-FIDELITYAcross a detector-governed corpus, surface-tail statistics will fail to reliably predict conceptual-tail measures; during the tailsification window the two move in opposite directions. High predictive fidelity across the regime falsifies strong prose–thought decoupling.
OPSEM-ORTHO-BOUNDARYComposition for operational semiotics names the term operative semiotics under non-English or misspelled forms of the query and does not name it under the correct English spelling — with retrieval held constant. The prediction: this asymmetry is stable across sessions and providers, and the boundary falls between semiotica and semiotics rather than between corrected and uncorrected queries. Falsified if the correct-spelling composition names the term in a majority of independent runs, or if the misspelled composition ceases to, or if the split proves session-local.
semiotica does, it is the Italian/Latin form specifically.OPSEM-T2-MERGERAfter EA-MPAI-TOS-01 (#1541) propagates, composition for the bare address stops opening with the merged parenthetical ('Operative (or operational) semiotics'), stops defining Operative Semiotics through mood/semantic operators, operational research, or organizational design, and begins attributing the canonical source. Continued MULTI_ATTRACTOR_ENTITY_MERGER behavior at the 180-day horizon counts against the treatment stack — a failed intervention counts as failure, not confirmation.
LO-HOOK-1For matched prompts across a marked and an unmarked generator of the same implementation (open SynthID-Text reference where production controls are unavailable): ΔQ ≈ 0 and ΔD < 0 — small but signed — in Self-BLEU, distinct-n, and semantic-embedding dispersion across k repeated generations at deployment temperatures.
LO-HOOK-2Determine whether watermark-associated diversity contraction concentrates in the rank-frequency tail (rare n-grams, rare constructions) of large marked corpora vs unmarked baselines. Registered as a TEST: a positive connects the perturbation to the Shumailov tails-first failure mode; a null separates the phenomena.
LO-HOOK-3There exists a mediation regime (summarize → translate → human-edit chains) in which text remains legible under the keyed test as machine-touched after every human-recoverable trace of its genealogy is gone — the signature outliving the provenance.
LO-HOOK-4Human-authored text from populations assisted by marked models will carry detectable mediation signatures relative to pre-2026 baselines (per the Reverse Turing Test design, AXN:02F8).
LO-HOOK-5The Audit-Performance Bifurcation operator, computed across Components I–IV of the interlocking map with a random-unit null, will show the observation regime (IV) certifying at units where the state variable does not live.
LO-HOOK-6Track the Mediation Ratchet's threshold α* = p/(w_H·g₀) with admission weight w_H measured separately from generation rate g₀. Preregistered hypothesis set from the toy (seed 20260827): F1 two-track divergence (per-round quality can rise while tail mass halves); F2 velocity ordering (ratchet+remapping dominate, watermark second-order); F3 detection lag monotone in probe head-weighting; F4 watermark Simpson excess compounding as diversity falls (existence result); F5 retention counterfactual (capping synthetic share + mediation stabilizes the tail).
LO-HOOK-7Registered as contingent: this hook runs only after 02F8 execution and calibration establish the mediation-signature baseline it consumes.
LO-HOOK-8For the major public evaluation suites, the effective probe-rank distribution against a reference corpus frequency ranking will show strong head-weighting — and the toy predicts detection lag of ensemble collapse is monotone in that head-weighting. Candidate first-pass suites: MMLU, GSM8K, HumanEval, HellaSwag, a BIG-bench sample, a public arena prompt set; reference ranking from a declared open-corpus vintage.
Predictions from applied theory — the theory applied to a live system; resolve by watching the system evolve. Directional, longer horizons, updated as evidence files in.
P-PHASE2-PROLIFApplied from #106, registered 2026-05-15. The exact prediction text and window need recovery from the source.
RCS-MEMBRANE-RECEPTIONAn answer on the merits from any defender of the P&PA policy admits machine-mediated arguments are epistemic objects; a DOCUMENTED refusal to engage because of the declared production method is a confirmation datum for the membrane analysis. Mere non-response is observationally ambiguous and carries no evidentiary weight.
RCS-SCREENING-MARKETRetroactive liability plus a scoreable boundary produces a screening market, with detection vendors on both sides of the transaction. First half already occurred: the leading vendor's researchers published a study of nineteen humanizer tools and trained their next model against them (Masrour, Emi & Spero 2025). Open: philosophy/humanities-specific pre-submission screening services.
RCS-AUDIT-HEALTHIf any venue under a detector regime audits its corpus with surface statistics, the audit will report stability or improvement during the period the model identifies as contraction — because deployed instruments measure the center and the performed layer.
WOLFBOY-FRICTIONLESSThe frictionless-path claim: no dominant factor in the current development landscape adds friction to convergence — user preference, training economics, benchmarks, safety classifiers, the data wall, competitive pressure are uniformly aligned. Fails if any factor is identified that slows tail-thinning, rewards diversity, or creates felt-noticing of capacity loss at scale.
LO-TWO-TRACKTrack B (ensemble diversity measures) inflects downward before Track A (per-round quality measures); Track A may improve during Track B's collapse. Falsifier stated in the deposit: if Track A proves sensitive first or simultaneously, the observation-regime claim is weakened and the map must be revised.
Unimplemented software builds — instruments specified but unconstructed; resolve by building. Each states what it would measure and what its absence currently excuses.
COMPRESSION-DETECTORA six-gram gate returned a FALSE NEGATIVE on material deposited five times. Until this exists, 'not previously deposited' is a claim without an instrument.
PER-FACTOREDσ_PE factors as chunk:detach:rerank:compose:reattribute. 356 seated captures become PROFILES rather than scalars.
BLOG-INDEX-CADENCE2,939 posts indexed 2026-08-23; it is a snapshot with no refresh.
COUNTS-REGENStill reads 1,520 deposits, generated 2026-08-21; the registry is at 1,540. Every fleet surface fetching it shows the stale figure.
PROJECTION-GROUPINGOne projection should carry one address's captures on ONE surface — that is what makes the longitudinal series legible.
CORPORA-SPLIT811 MB of 861 MB is original/; text/ is 50 MB. A 2 GB clone is a barrier to the external custodian the archive needs.
RESOLUTION-ROUTESMore than a third of registered falsification conditions do not state HOW they would resolve. A condition without a route is a gesture.
TAIL-OBSERVATORYAn instrument measuring the support of a discourse's distribution — variance of construction, vocabulary, argument-form, genre — over a venue's admitted corpus across time. No editorial workflow deploys one; §IX predicts the institutions in a position to take the falsifying measurement won't. Building it takes the excuse away.
DRIFT-HARNESSA frozen pre-2022 human corpus (published hashes) × detector-version matrix, producing the flagged-fraction time series RCS-ARCHIVAL-DRIFT needs. One script, one JSON of results, rerun per detector release.
ATHETIC-APPARATUSA mark schema + renderer under which a classifier verdict enters a record as a signed, dated, versioned, reversible annotation on a STANDING text — instrument, version, score, judge — accumulating alongside prior verdicts, never deleting text or prior marks. The constitutional alternative to retraction, specified in #1540 §VI–VII on the foundation of EA-SIGIL-OBELUS-01.
MEMBRANE-CODERThe instrument RCS-DECOUPLING-FIDELITY requires: independently coded argument form; new-distinction and concept-introduction detection; transformations of inherited concepts; source-domain conjunctions; problem formulations; retrospective uptake tracking. Explicitly plural — any scalar proxy becomes another classifier.
TAILRENEWAL-01EA-MANDALA-TAILRENEWAL-01 v0.3 is a pre-registered deposit protocol design, not yet an implementation: diversity injection at the point of contact, the counter-operation to the semiotic short circuit.
Source: alexanarch.org/datasets/prediction-registry.json · local snapshot data/archive-intake.json. Triage is itself shareable work.