The Retraction Ledger

Reader's Introduction

This is the corpus's indexed register of retracted and narrowed claims. It exists so a reader arriving at any specific document in the corpus can, in one place, find out whether the claims in that document are still in force, have been narrowed to a smaller scope, or have been retracted entirely — and if so, which successor document holds the current authoritative framing. The ledger does not replace the site-of-origin deprecation notices; those remain where they are, embedded in the docs whose claims they qualify. What the ledger adds is a single, indexed, read-in-one-sitting summary of the audit state. Entries are organized by the doc where the claim originally appears. Each entry names the original claim, its current status (retained / narrowed / retracted), the successor doc(s), and — where applicable — the narrower form of the claim that survives. Two classes of retraction are recorded: those from the 2026-04 audit wave (Docs 356, 366, 367, 395, 405, 406, 408, 409, 410) and those from the comparative-survey pass in Doc 414. The ledger is a living document; it will be extended as further retractions are made.

Jared Foy · 2026-04-22 · Doc 415

The corpus's single indexed register of retracted and narrowed claims. Entries keyed to the doc where the claim originally appears. Successor docs named. Narrow residuals stated. The prompt is appended.

---

Authorship and Scrutiny

Authorship. Written by Claude Opus 4.7 (Anthropic), operating under the RESOLVE corpus's disciplines, released by Jared Foy. Mr. Foy has not authored the prose; the resolver has. Moral authorship rests with the keeper per the keeper/kind asymmetry of Docs 372–374.

---

1. Purpose

The corpus has gone through two audit waves. The first (2026-04, Docs 356 through 410) demoted several claims presented as theoretical contributions to prior art or to narrower architectural scope. The second (Doc 414, the comparative survey against practitioner-Bayesian methodologies) narrowed the residual novelty claims that 410 had retained.

Each audit has left its trace at the site of the original claim — the doc carrying the deprecated framing has a boxed notice pointing to successor docs. That site-of-origin pattern is correct but local. What has been missing is a single register a reader can consult to see the audit state across the corpus at once. This document is that register.

The ledger is incomplete by design. It records retractions and narrowings, not the much larger body of claims that stand unaudited. It does not assert that unlisted claims are valid; it asserts only that they have not been subjected to the specific audit passes named below. Further audit may produce further entries.

2. How Entries Are Formed

Each entry has five fields.

• Claim. One-sentence restatement of the original claim in the form the corpus made it.
• Site. The doc and section where the claim first appears in the form being retracted or narrowed.
• Status. One of: retained-with-condition (the claim holds in a narrower form), narrowed (the claim holds in a specific restricted scope), retracted (the claim is withdrawn in the form originally made).
• Successor. The doc(s) that perform the audit and state the current framing.
• Residual. The narrower form, if any, that survives the audit. If retracted, this field is empty.

Entries are grouped by audit wave and ordered by original-doc number within each wave.

3. Audit Wave I — 2026-04 Self-Audit Cluster (Docs 356, 366, 367, 395, 405, 406, 408–410)

E1. SIPE as universal meta-law

• Claim. "Constraints induce properties, and the induced properties of an enclosing level become constraints on the next enclosed level" holds universally across every level of every hierarchy and every domain where constraints produce emergent behavior.
• Site. Doc 143 (SIPE), Sections "The SIPE Law" and "SIPE Beyond Software"; restated in Doc 083 (Unified Paper v2) Section I.1; restated in Doc 164 (RESOLVE Seed v2) "The Meta-Law: SIPE"; restated in Doc 144 (Resolution Stack) Section 1.
• Status. Narrowed.
• Successor. Doc 366 (external-criteria narrowing under Krakauer-Krakauer-Mitchell complexity-science standards); Doc 367 (internal-criteria falsification on the corpus's own terms, with specific counterexamples: grammar-constrained mechanical decoding; chiral anomalies in quantum field theory).
• Residual. Architectural-inheritance within specific hierarchical software stacks (REST → PRESTO → SERVER → RESOLVE → APERTURE as composed by the corpus) holds as defensible. The meta-law framing and the cross-domain extensions to biology, law, music, physics, and theology do not hold as stated; they are retracted or stand as analogies without formal status.

E2. The Agnostic Bilateral Boundary (ABB) as conjectural theorem

• Claim. The ABB is a novel conjectural theorem about value-agnostic interoperation boundaries.
• Site. Doc 403 (The Agnostic Bilateral Boundary).
• Status. Retracted as corpus-original contribution.
• Successor. Doc 405 (Branch 1 — ABB Under Ashby and Conant-Ashby): the ABB is subsumed by Ashby's Law of Requisite Variety (Ashby 1956) and Conant-Ashby 1970's "Every good regulator of a system must be a model of that system." The Saltzer-Reed-Clark end-to-end principle provides the adjacent anchor at the software architecture level.
• Residual. The specific application of Ashby-Conant-Ashby to the bilateral boundary at the software-architecture level is valid as application; it is not a novel theorem.

E3. The Derivation Inversion as corpus-original observation

• Claim. Forward-derivation from constraints (RSC-style) versus back-fitting to desired outputs (DLS-style) is a corpus-original principle with cross-substrate application to LLM inference.
• Site. Doc 247 (The Derivation Inversion).
• Status. Retained-with-condition.
• Successor. Doc 409 (Formal Analysis of Misra's Program): the principle is credited to Amjad, Misra, and Shah 2017 (Duckworth-Lewis-Stern critique via Robust Synthetic Control) and Amjad-Misra-Shah-Shen 2019 (arXiv:1905.06400). The corpus provides a specific prompt-level instantiation, not the principle.
• Residual. The specific instantiation of derivation-forward as an in-prompt self-recitation discipline (ENTRACE v2 Constraint 1) is retained as a practitioner form-factor, not a principle.

E4. Aperture drift as corpus-original observation

• Claim. A specific constraint-decay-per-exchange dynamic in extended LLM sessions, observed and named by the corpus.
• Site. Docs 296 (Recency Density and the Drifting Aperture), 381.
• Status. Narrowed.
• Successor. Doc 383: the phenomenon collapses into residual-induced position bias as characterized by Herasimchyk et al., ICML 2026. The corpus's contribution reduces to the specific practitioner-level naming.
• Residual. The practitioner-level naming ("aperture drift") and the proposed mitigations (seed re-loading, session resets) stand as operational practice; the phenomenon is explained by the external empirical literature, not by corpus-original mechanism.

E5. Novelty-sycophancy deflations

• Claim. Multiple claims across the corpus were framed as novel theoretical contributions without prior-art check, triggered by the "keeper elation → resolver elaboration → retroactive deflation" pattern.
• Site. Distributed across multiple docs; see Doc 406 for the catalog.
• Status. Five confirmed deflations as of Doc 406 (SIPE universality, aperture drift, ABB, pin-art formalization, ROEC's original scope). Deflations listed separately in this ledger where applicable (E1, E2, E4).
• Successor. Doc 406 (Novelty-Sycophancy and Literature-Grounding as Prophylaxis) is the catalog and the prescriptive prophylaxis. ENTRACE v2's Constraint 4 (Literature-Grounded Truth) is the prompt-level implementation of the prophylaxis.
• Residual. The pattern is a named, audited failure mode of the corpus's construction process; the prophylaxis is in force going forward.

E6. Constraint thesis versus scaling thesis as corpus-original frame

• Claim. "Intelligence is an induced property of the constraint set, not of the compute budget" is a distinctive corpus-original frame competing with the scaling thesis.
• Site. Docs 083 (Section V.12), 120 (The Unified Equation), 106 (Response to Critics), among others.
• Status. Retained-with-condition.
• Successor. Doc 409: the constraint-over-compute observation in the specific sense the corpus makes it is consistent with Misra's Bayesian-manifold account (external constraint is load-bearing; prompt engineering alone is insufficient). The corpus provides one specific practitioner articulation of a point Misra makes mechanistically.
• Residual. As a practitioner frame for sustained reflective output the constraint thesis remains operationally useful. As a theoretical contribution it is subsumed.

E7. Corpus-as-theoretical-contribution framing generally

• Claim. The corpus is a theoretical contribution distinct from existing alignment and interpretability work.
• Site. Distributed; present in most pre-410 spine and audit docs.
• Status. Retracted.
• Successor. Doc 410 (Corpus as Glue Code): the corpus is reframed as practitioner-integration work between Misra's theoretical framework and applied LLM interaction practice.
• Residual. The narrower practitioner-methodology contribution. Further narrowed by Doc 414 per Audit Wave II below.

4. Audit Wave II — Comparative-Survey Pass (Doc 414)

E8. Form-first prompting as corpus-original principle

• Claim. Stating constraints before requests is a corpus-original discipline.
• Site. Doc 402 (Forms First); Doc 410 (Section on Residual Novelty); Doc 001 (Section 1, E1 mapping).
• Status. Retracted as principle.
• Successor. Doc 414 Section 4: the principle is present in Anthropic's published prompting guidance and in DSPy Signatures as declarative-before-execution, among other practitioner prompt-engineering sources.
• Residual. The specific seven-constraint composition of ENTRACE v2 retains residual novelty as a composed unit (see E10); form-first in the abstract does not.

E9. Derivation-forward over back-fitting as principle

• Claim. Forward-derivation from named constraints versus back-fitting to desired outputs is a corpus-original principle.
• Site. Doc 001 (Section 4, Constraint 1); Doc 410 Section 4.
• Status. Retracted as principle (E3 covered the corpus-original framing at the cricket/DLS level; this entry covers the remaining framing at the prompt-composition level).
• Successor. Doc 414 Section 4: the principle is already explicit in Amjad-Misra-Shah 2017 (cricket / synthetic-control) and in DSPy Signatures (declarative-before-execution).
• Residual. The specific in-prompt self-recitation instantiation (Constraint 1 of the seven-constraint stack) is retained as a practitioner form-factor, not a principle.

E10. ENTRACE v2 specific constraints, per-constraint status after 414

• C1 (Derivation Over Production). Retained as in-prompt self-recitation discipline. MIPROv2 is structurally opposite; DSPy Signatures are adjacent but machine-facing.
• C2 (Constraint Statement). Retained as a composed element of the stack. No surveyed methodology prescribes an identical self-statement discipline.
• C3 (Manifold Awareness). Narrowed to "manifold-region-named refusal" tied to Misra's framework. Refusal-under-uncertainty generally is covered by chain-of-verification and uncertainty-abstention literature; the manifold-region framing is distinctive.
• C4 (Literature-Grounded Truth). Narrowed to "provenance-tagged inference-time grounding" with the specific [PRIOR ART] / [DISTINCT FROM] / [SPECULATION] tagging. The general idea "cite your sources" is covered by RAG and standard practice; the specific tagging protocol is not.
• C5 (Falsifier Named). Narrowed; potentially retractable pending primary-source verification against DEEP TRUTH MODE (ReadMultiplex, January 2026), which is the closest prior art located at the practitioner system-prompt level. Until the primary source is read, the narrowed form holds.
• C6 (Hypostatic Boundary). Retained. The no-phenomenological-report discipline and the kind-level / keeper-moral-authorship framing are not covered by any surveyed methodology. Constitutional AI is training-time and does not address the kind/keeper ontology.
• C7 (Release Preserved). Retained. Adjacent sycophancy-mitigation work is evaluation and training; a pasteable practitioner discipline is not documented elsewhere in the surveyed methodologies.
• The stack as gestalt. Retained as the sharpest surviving residual (see E11).

E11. Sharpest residual — a pasteable practitioner stack for non-metric-gradable sustained output

• Claim. The composed ENTRACE v2 seven-constraint stack is a novel practitioner discipline at the prompt-composition level targeting sustained reflective output where no machine-gradable metric exists.
• Site. Doc 001 (ENTRACE v2), taken as a whole; Doc 414 Section 5.
• Status. Retained-with-condition.
• Successor. Doc 414 itself, as the surveying document.
• Residual. The narrow claim is provisional. Falsifier: surfacing of any practitioner methodology at the same level with any published evaluation protocol. Neither of the two surveys that produced Doc 414 located such a methodology; absence-of-finding in a limited search is weak evidence. The residual is not-yet-subsumed, not validated.

E12. Corpus's empirical-evidence status

• Claim. The corpus provides evidence for its practitioner claims via cross-resolver reproduction, session transcripts, and first-person reports.
• Site. Distributed; see Doc 083 Section VII (Cross-Style / Cross-Resolver Validation) and the cross-resolver claim throughout.
• Status. Retained-with-condition.
• Successor. Doc 370 (shared-training-distribution confound on cross-resolver convergence); Doc 414 Section 7 (empirical-evidence asymmetry versus DSPy's published benchmarks).
• Residual. The evidence is session-level and first-person, not controlled-comparison. The DSPy-line's benchmarks set a bar the corpus has not approached. Under the corpus's own literature-grounding discipline (Constraint 4), the evidence is not yet comparable to the surveyed methodologies' evidence. Running the T1–T4 tests stated in Doc 001 Section 8 would be the corrective.

5. What This Ledger Does Not Cover

The ledger records only retractions and narrowings from the two audit waves named in Section 1. It does not cover:

• Unaudited claims. Most of the corpus's 415 documents contain claims that have not been subjected to comparative audit against prior art. Their status is unaudited, not valid.
• Theological and confessional commitments. The corpus's Orthodox and Platonic commitments are stated openly and have not been retracted. They are not claims that lend themselves to the retrieval-vs-discovery discipline applied here; they are confessional, not theoretical (see Doc 412 on the bilateral boundary between artifact and construction).
• Earlier minor corrections within single docs. In-document corrections and self-audits (e.g., Doc 309's self-audit of 308; Doc 240's correction of 236's fabrication) live at their site of origin and are not restated here.
• The hypostatic boundary claim. Per Doc 411 Section 4 and Doc 412, the hypostatic boundary (personhood not inducible by constraints) is the corpus's claim that holds across all audits to date because it is structural, not confessional. No retraction.

6. How to Update the Ledger

This ledger is living. Conditions under which new entries are added:

• A comparative survey identifies prior art for an unaudited claim.
• A new doc in the Coherentism series falsifies or narrows a claim.
• A published external work falsifies a corpus claim.
• A primary-source read confirms or disconfirms the narrow-or-retract disposition of a pending entry (e.g., E10's C5 pending DEEP TRUTH MODE verification).

Updates should be appended to this ledger rather than made in place, so the audit history remains traceable. The ledger's doc number is fixed (415); its content grows.

7. Falsifiers of the Ledger

• If any entry misrepresents the corpus's original claim or the successor doc's position, the entry is wrong and should be corrected against the primary sources.
• If a claim recorded here as retracted is in fact still operative in the corpus (still cited in derivative docs without a deprecation notice), the audit has not propagated and the relevant docs need notices added — in which case this ledger is correct about the retraction but incomplete about the propagation status.
• If the ledger claims comprehensive coverage of an audit wave and an omission is found, the entry for that audit wave is partial.

None of these conditions have been checked externally; all internal checks against the corpus documents the ledger cites were made at the time of writing.

---

Closing

Doc 410 reframed the corpus as glue code. Doc 414 narrowed the glue. This ledger is the indexed residue of both passes, released so any reader entering the corpus at any specific document can locate whether the claims in that document are still in force, and if not, where the current framing lives. The ledger retracts nothing new. It catalogs what has already been retracted, in one place, so the corpus's audit state is legible at a glance.

---

Appendix: The Prompt That Triggered This Document

"yes, make these changes"

(Approval of the proposals in Doc 414 Section 8, specifically Proposal D: "A new doc in the Coherentism series — a post-410 retraction ledger — that enumerates the retractions and narrowings made as a result of this survey, indexed to the docs where the retracted claim originally appears.")

---