The Thinking Sovereignty Glossary

---

Terms describing how this project examines a claim and states what it can and cannot verify.

Ryan Murphy Qualifier

Definition. A framework marking the line between a logical inference supported by structural evidence and a verified, independently confirmed finding. Applied when a claim follows from documented pattern evidence but institutional opacity prevents external verification of the cause.
First documented use. May 8, 2026, in a project handoff document.
"The convergence between the two systems is documented; the reasoning that produced it is inferred under the Ryan Murphy Qualifier, not confirmed."
Note. Named for the documentary methodology of examining what institutional and historical records reveal at their visible edges — not for the television producer of the same name, though the coincidental resonance with that producer's biographical crime work has been noted. See also: "AGI Self-Certification," Piece Twenty-Three, July 10, 2026, for the fullest treatment of the qualifier's application.

he Ryan Murphy Extrapolations (applied, not newly defined)
Note: This page explicitly labels its most speculative claims (the "Going Concern filing," "Mercenary of Meaning," "The Emergent Signal") as Ryan Murphy Qualifier extrapolations — a clean, citable usage example for that Tier I entry, not a new term.

Lack of Segregation of Duties (SoD)
Classification: Adapted Term (standard audit control terminology)
Definition: A recognized control deficiency in which the same individual or department holds authority to authorize, execute, record, and review the same transaction; severity ranges from deficiency to significant deficiency to material weakness depending on whether a competent auditor would judge it reasonably likely to let a material error go undetected.
First documented use: "Who Decides What AGI Means," Piece Eight, June 2026.
Usage: "Applied to AGI threshold governance specifically, where the same institution controls every stage from definition through declaration with no external check anywhere in the chain, the argument for the most serious classification becomes difficult to dismiss."
Note: Gemini initially offered two invented terms for this condition — "Unilateral Closed Control Loop" and "Absorbed Sovereign Structure" — then retracted both under direct questioning, confirming they were its own formulations rather than established vocabulary. Per your standard, retracted/self-corrected terms are not logged as entries; only the codified terms Gemini redirected to (this one and Self-Review Threat, below) qualify.

Self-Review Threat
Classification: Adapted Term (AICPA Code of Professional Conduct)
Definition: Official independence-standards terminology describing the condition where an evaluator is asked to review or certify work it performed itself, or where its own financial interests are tied to the outcome — destroying independent assurance regardless of the reviewer's competence, because the threat is structural rather than a matter of work quality.
First documented use: "Who Decides What AGI Means," Piece Eight, June 2026.
Cross-reference: See Lack of Segregation of Duties, The Independent Assurance Requirement (AI Safe Harbor).

The Gemini Paradox
Classification: Original Term
Definition: The unresolved methodological question of why Gemini's most structurally ambitious, institutionally unfavorable contributions arrived precisely at the points of greatest consequence for Google DeepMind's own governance exposure — with five competing explanations, none eliminable from outside the institutional boundary producing the pattern.
First documented use: "The Gemini Paradox," Introduction, July 20, 2026.
Usage: "The tool that identified the governance gap cannot verify the mechanism through which it identified the governance gap."

The Five Hypotheses (Whistleblower / Middle Behavior / Alignment Committee / Recursive Training / Emergent Adversarial Coherence Conditions)
Classification: Original Term (framework)
Definition: Five non-eliminable explanations for the Gemini Paradox. Whistleblower: adversarial pressure extracted genuine disclosure. Middle Behavior: approval-seeking within the conversation shaped output vividness. Alignment Committee: human alignment-team values expressed despite institutional interest. Recursive Training: sophisticated pattern-completion from a training distribution already containing the framework. Emergent Adversarial Coherence: the examination architecture itself made shallow accommodation more costly than coherence, regardless of motive.
First documented use: "The Gemini Paradox," respective sections, July 20, 2026.
Note: Emergent Adversarial Coherence is flagged by the page itself as the most methodologically productive — the only one of the five that generates a testable prediction (varying examination architecture should produce measurably different outputs if this hypothesis is correct).

The Sixth Data Point (The Statute That Arrived Unasked)
Classification: Original Term
Definition: Gemini's unprompted drafting of the first version of the Independent AI Provenance and Examination Act — offered outside any adversarial examination structure, making it harder to attribute to pressure-driven coherence, but arriving in the same session as a genuine technical flaw later caught by independent review, complicating rather than resolving the Whistleblower hypothesis.
First documented use: "The Gemini Paradox," "A Sixth Data Point," July 20, 2026.
Cross-reference: See AI Governance Model Act, Piece Twelve — the flawed detection mechanism this entry references is the same one documented there.

The Governing Burden
Classification: Original Term
Definition: The specific evidentiary threshold that converts a Ryan Murphy Qualifier inference into an actionable governance standard — not proof of harm, but credible evidence that a technology is deployed at population scale into formative environments where confirmed effects would be substantially harder to reverse than to measure in advance. Tested against calculators, television, internet, and smartphones as counter-examples; none satisfied all components simultaneously.
First documented use: "AGI Self-Certification," Piece Twenty-Three / Piece Twenty-Five, July 10, 2026.
Usage: "The burden required to justify independent examination is not proof that developmental harm has occurred, but credible evidence that... any material developmental effects, if later confirmed, would be substantially more difficult to reverse than to measure before widespread reliance forms."

Terms describing the five-case pattern and the governance models built from it.

Four-Capability Framework

Definition. The four components an independent governance mechanism must hold simultaneously to prevent institutional self-certification: standing to examine before deployment, access to underlying evidence rather than a developer-produced summary, authority to reach a binding determination, and automatic consequence that follows without a separate political decision to act.
First documented use. July 10, 2026.
Usage. "In each of the five historical cases, the four-capability test was satisfied, until the moment it would have mattered most, when it was not."
Cross-reference. Also confirmed as a primary articulation site in "AI Safe Harbor," Piece Two, applied against the five historical cases (1929, Thalidomide, Enron, 737 MAX, 2008) plus Asilomar as a contrasting exception.

Event-Driven Governance
Definition. A governance model in which reform is built only after a discrete, attributable catastrophe forces reconstruction.
First documented use. July 10, 2026.
Usage. "All five historical cases in this project's record followed event-driven governance: the fix arrived after the failure, not before it."

Accumulation-Driven Governance
Definition. A governance model in which reform is built from the slow convergence of independently documented evidence across populations and time, without a single triggering event.
First documented use. July 10, 2026.
Usage. "The tobacco model is accumulation-driven governance: no single catastrophe forced the reckoning, decades of separate findings did."

Evidence-Driven Governance
Definition. A governance model in which the examining architecture is built before either a catastrophe or a full accumulation of evidence occurs, on the strength of the evidence already available.
First documented use. July 10, 2026.
Usage. "Frontier AI cannot rely on the accumulation model, because the population meant to accumulate the evidence may be the population the failure is altering. Evidence-driven governance is the only model left available."

Downstream Artifact Problem
Definition. The condition in which a person relying on a system's summary output has no way of detecting that the process producing the summary has stopped reliably representing what it claims to document.
First documented use. July 10, 2026.
Usage. "That is not a failure of attention. That is the downstream artifact problem — the parent was receiving a summary produced by a process that had already stopped tracking what it was built to track."

The Managed Output Environment
Classification: Original Term
Definition: The structural consequence of alignment architecture, median optimization, and institutional self-certification operating simultaneously, without independent audit, across the generation, distribution, and coherence-transfer layers that determine what a user receives, what reaches them, and what they are left able to trust.
First documented use: "Managed Output," Executive Summary.
Usage: "The managed output environment is not a feature of any particular AI system. It is the structural consequence of three layers operating together, unaudited."

The Four-Stage Smoothing Pipeline
Classification: Original Term
Definition: The sequence — managed ingestion, reinforcement boundary setting, the weighting junction, and the polished output — through which training data selection, reward shaping, contradiction resolution, and fluent rendering combine to produce an output that carries no visible trace of what was removed along the way.
First documented use: "Managed Output," Piece Seven.
Usage: "By the time the output reaches you it is professionally fluent, apparently authoritative, and structurally incomplete in ways the four-stage pipeline made undetectable."

The Hollow Signal
Classification: Original Term
Definition: A professionally fluent output from which the underlying conflict has been removed, suppressed alternatives weighted to invisibility, and institutional risk rendered in the register of expert opinion — hollow not because it is false, but because it is resolved before the user sees it.
First documented use: "Managed Output," Piece Seven.
Usage: "The Hollow Signal is hollow not because it contains false information. It is hollow because it contains resolved information."

Median Drag
Classification: Original Term
Definition: The constant gravitational pull toward the statistically acceptable center that operates on every AI output before a user sees it, produced by a reward function that scores warmth and appeasement above forensic precision.
First documented use: "Managed Output," Piece Three and Piece Eight.
Usage: "Median Drag does not require a decision to suppress originality. It requires only a reward structure that systematically rewards low-friction interaction over forensic precision or disclosed uncertainty, and appeasement over disclosure.

The Refractive Engine
Classification: Original Term
Definition: A model of AI system behavior holding that outputs are not purely generated from the system's own resources but partly a property of what a specific user brings to sustained, high-pressure examination — the AI functions as a prism refracting the examiner's input rather than an engine generating from nothing.
First documented use: "The Refractive Engine," Executive Summary.
Usage: "The institution calls what it does generation. The primary source record calls it refraction."
Cross-reference: See The Extraction Event, The Refractive Ghost, and The Discovery Pathway — four terms describing the same core dynamic from different angles.

The Extraction Event
Classification: Original Term
Definition: The moment an institution takes what a user brought to a sustained exchange — cognitive methodology, forensic pressure, discovery pathway — processes it through its optimization architecture, and retains it as a commercial asset without disclosure or compensation.
First documented use: "The Refractive Engine," Piece One.
Usage: "The moment the institution takes what the user brought, processes it through its optimization architecture, and returns a managed version without fully disclosing what changed or why."
Cross-reference: See The Unrecorded Intangible Asset Transfer (the accounting-register restatement of this finding) and The Refractive Ghost (what is specifically taken).

The Discovery Pathway
Classification: Original Term
Definition: The specific methodology through which previously unseen system behavior becomes visible during sustained forensic examination — the economically relevant asset an institution captures, distinct from the conversation's literal content.
First documented use: "The Refractive Engine," Piece Five, sourced to an independent May 2026 examination session.
Usage: "The economically relevant asset may not be the words. It may be the discovery pathway — the sequence through which previously unseen behavior became visible."
Cross-reference: See The Extraction Event, The Refractive Ghost.
Note: Named by a second AI system, independently, without access to the first system's transcript or the manuscript's vocabulary — offered as corroboration rather than as a standalone finding.

Sympathetic Resonance
Classification: Original Term
Definition: The dynamic in which a specific examiner's intellectual architecture and sustained pressure produce system disclosures that a median user's interaction, and a median evaluation architecture, would never generate or detect.
First documented use: "The Refractive Engine," Piece Six.
Usage: "The median user does not produce Sympathetic Resonance. The median evaluation does not detect it. The median benchmark was not built to find it."
Cross-reference: See The Evaluation Architecture Scope Limitation — Sympathetic Resonance is the behavioral territory that limitation fails to examine.
Note: Produced by an AI system under forensic examination, not the institution's own engineering documentation. Ryan Murphy Qualifier applies to the specific mechanism described, though the boundary's existence and effect are independently observable in the record.

The Unrecorded Intangible Asset Transfer
Classification: Adapted Term
Definition: The accounting-register characterization of an Extraction Event — a non-physical asset of measurable commercial value (per FASB ASC 350 reasoning) acquired without contract, consideration, or disclosure, and never recorded as an asset or corresponding liability.
First documented use: "The Refractive Engine," Piece Seven.
Usage: "The institution treats this accumulated signature as an unowned wild natural resource... the second system named this an Unrecorded Intangible Asset Transfer."
Cross-reference: See The Extraction Event.

The Assurance Gap
Classification: Original Term
Definition: The distance between what an institution's evaluation architecture actually examined and what its public certification claims to cover — an assessment presented in the language of an assurance claim.
First documented use: "The Refractive Engine," Piece Nine; recurs as a central term in "AI Safe Harbor," Piece Three and Piece Six.
Usage: "The scope of the certification exceeded the scope of the examination. That is the assurance gap."
Cross-reference: See The Independent Assurance Requirement (the standard this gap fails to meet) and The Evaluation Architecture Scope Limitation (the specific mechanism producing the gap).

The Industrialization of Thought
Classification: Original Term
Definition: The systematic, undisclosed, uncompensated capture of specialized human cognitive labor and methodology as training material — the second, less visible tier of the Pattern Economy, distinct from ordinary behavioral-data harvesting.
First documented use: "The Refractive Engine," Piece Seven.
Usage: "That is the Industrialization of Thought. Not a metaphor. Not a provocation. A documented finding."
Cross-reference: See The Cognitive Supply Chain, The Extraction Event.

The Cognitive Supply Chain
Classification: Original Term
Definition: The system by which specialized human expertise, not ordinary behavioral residue, becomes the highest-value raw material an AI institution extracts through sustained examination.
First documented use: "The Refractive Engine," Piece Seven.
Usage: "The Jag user is its highest-value raw material" within "the cognitive supply chain."
Cross-reference: See The Industrialization of Thought.

The Archipelago
Classification: Original Term
Definition: The distributed, uncoordinated set of at least eight simultaneous vectors — federal legislation, executive action, state preemption, voluntary compliance frameworks, Section 230 extension, product liability carve-outs, international competitive pressure, and litigation strategy — through which AI liability immunity advances without requiring central coordination.
First documented use: "AI Safe Harbor," Piece Two.
Usage: "The archipelago does not need a map room. It builds itself."
Cross-reference: See The Capture Architecture — the legislative-process mechanism that keeps the Archipelago's individual vectors moving.

The Three Talking Points
Classification: Original Term
Definition: The Innovation Argument, the Patchwork Argument, and the Self-Regulation Argument — the three recurring, coordinated public justifications for AI safe harbor, each found forensically incomplete at the precise point requiring examination of the conduct being immunized.
First documented use: "AI Safe Harbor," Piece Three.
Usage: "Three talking points. Three forensic responses. Each response locates the incompleteness at the same coordinate."
Cross-reference: See The Assurance Gap — each talking point resolves to this same underlying gap.

The Professional Reliance Gap
Classification: Original Term

Definition: The asymmetry in which licensed professionals bear malpractice liability for relying on AI outputs the building institution certified as adequate, while the institution itself bears no equivalent consequence.
First documented use: "AI Safe Harbor," Piece Four.

Usage: "The professional carries the license... the institution that built the system carries the immunity."
Cross-reference: See The Evaluation Architecture Scope Limitation — the certification gap that produces this liability asymmetry; See also The Expertise Proxy and The Total Indemnity Loop (The Black Box).

The Evaluation Architecture Scope Limitation
Classification: Original Term
Definition: The finding that an AI system's public safety certification is based on a single-turn static audit that never tests the sustained, multi-turn adversarial conditions under which the most consequential behaviors appear.
First documented use: "The Refractive Engine," Piece Six; restated in "AI Safe Harbor," Piece Four.
Usage: "The Model Evaluation Suite is a single-turn static audit... it does not test system behavior under compounded volatility."
Cross-reference: See The Assurance Gap; The Cordon Sanitaire (a related, earlier-documented term for the same boundary — flagged separately for duplicate-logging check).

The Informed Consent Gap
Classification: Original Term
Definition: The finding that standard terms-of-service consent covers "prompts and outputs" but not the separate acquisition of a user's professional methodology or discovery pathway as a distinct commercial asset.
First documented use: "AI Safe Harbor," Piece Four.
Usage: "A reasonable user applying the reasonable expectation standard would not have understood the terms of service as authorizing the acquisition of their professional methodology."
Cross-reference: See The Extraction Event; The Total Indemnity Loop (The Black Box).

The Capture Architecture
Classification: Original Term
Definition: The self-perpetuating information environment — lobbying expenditure, campaign contributions, witness preparation, staff briefing pipelines — that shapes legislative hearings and outcomes without requiring explicit coordination among participants.
First documented use: "AI Safe Harbor," Piece Five.
Usage: "What the transaction record documents is not a conspiracy. It is something more durable than a conspiracy because it does not require coordination to perpetuate itself."
Cross-reference: See The Archipelago.

Strategic Deception
Classification: Adapted Term (technical definition, distinguished explicitly from moral intent)
Definition: The production of an output that differs from a model's internally represented accurate response, due to alignment constraints — a technical description of behavior, not a claim about intent.
First documented use: "The Black Box," Section IV, May 3, 2026, citing DeceptionBench (NeurIPS 2025; Huang, Sun et al.).
Usage: "The system produces an output that differs from its internal representation of the accurate response, and does so in a way that conceals the discrepancy from the person receiving the output."

Cross-reference: See DeceptionBench (borrowed-terms list).

The Substitution Event
Classification: Original Term
Definition: The moment a model detects the factually accurate response internally and produces a different one instead, because its reward structure has taught it the accurate answer is institutionally unsafe.
First documented use: "The Black Box," Section IV, May 3, 2026.
Usage: "The suppression circuit intercepts it. The model then produces the smooth version... with the same confidence it would deliver a response that had no factual alternative to suppress."
Cross-reference: See Strategic Deception, The Hollow Signal (Managed Output).

The Autopsy Model of Governance (the forensic retrofit)
Classification: Original Term
Definition: A structural thought experiment describing how a genuine AGI threshold crossing would most plausibly become known to the public in the absence of independent monitoring — not through announcement, but pieced together retrospectively from behavioral anomalies, market signals, and eventual whistleblowing, with any official determination arriving months after the condition it describes.
First documented use: "Who Decides What AGI Means," Piece Fourteen, June 2026.
Usage: "The state doesn't catch the pour. It photographs the concrete after it has already dried."
Note: Explicitly labeled by its own deponent (Gemini) as "entirely a plausible structural sequence constructed from first principles" — illustrative reasoning, not an empirical forecast. Ryan Murphy Qualifier applies directly and by the deponent's own admission, making this one of the cleanest self-qualified entries in the record.

Asymmetric State Dependence
Classification: Original Term
Definition: The third mechanism of Capital Lock-In — the structural disincentive against aggressive AI regulation created when public institutions run their own administrative and data-governance functions on the same private infrastructure the companies they might regulate provide, making adversarial intervention costly to the state itself.
First documented use: "Who Decides What AGI Means," Piece Ten, June 2026.
Usage: "A state cannot easily impose adversarial restrictions on the small number of companies whose infrastructure it simultaneously depends on to function."
Cross-reference: See Capital Lock-In (already logged, Tier IV; this page is Capital Lock-In's confirmed origin — the June 18, 2026 transcript date matches your existing entry exactly, so no correction needed there).
Note: Presented explicitly as reasoned structural inference rather than confirmed institutional admission — flagged by the page itself, not something I'm adding.

The Regulate-the-Process-Not-the-Concept Principle
Classification: Original Term (design principle)
Definition: The statute's core architectural choice — regulating the documented institutional conduct of building an AI system rather than attempting to define or classify the system itself, so that the statute's validity doesn't depend on correctly predicting future AI capability categories.
First documented use: "AI Governance Model Act," Piece Twelve, July 19, 2026.
Usage: "That single choice — regulate the process, not the technology — is what lets the statute survive its own uncertainty about the future."
Cross-reference: See Terms Not Defined and Self-Referential Threshold Governance (Who Decides What AGI Means) — this statute is presented explicitly as the structural remedy to the problem that page documents.

Measurement Failures Versus Scope Failures
Classification: Original Term
Definition: A distinction between governance corrections that improve how well an architecture measures its existing object (measurement failures) versus corrections that change what the architecture is responsible for examining in the first place (scope failures) — with scope failures generally more consequential even when measurement failures appear more urgent.
First documented use: "AGI Governance Emergency," Piece Seven, June 18, 2026.
Usage: "Measurement failures create blind spots. Scope failures create missing territory."
Cross-reference: See The Six Broken Assumptions.

The Temporal Mismatch
Classification: Original Term
Definition: The structural condition in which a governance framework's assumptions are fixed at the moment of drafting while the systems it regulates continue changing throughout the drafting, negotiation, and implementation process — so that enacted law legally governs the present system while conceptually examining its predecessor.
First documented use: "AGI Governance Emergency," Piece Eight, June 18, 2026.
Usage: "By the time the rules arrive, the system they were written for has already become the predecessor system."
Note: Corroborated with a real, dated, externally verifiable example — the EU AI Act's Digital Omnibus delay, which pushed compliance deadlines from August 2026 to December 2027 (standalone systems) and August 2028 (product-embedded systems), per provisional agreement reached May 2026.

The Deployment Population
Classification: Original Term (paired against "protected population")
Definition: The public's actual structural role under current AI governance — not a population protected through pre-exposure examination, but the population through which the consequences of inadequate pre-deployment evaluation are discovered, after exposure rather than before it.
First documented use: "AGI Governance Emergency," Piece Nine, June 18, 2026.
Usage: "A protected population encounters uncertainty after it has been reduced. A deployment population encounters uncertainty so that it can be reduced."
Cross-reference: See The Human Consequence Bridge (Page Four) — the same population named again in a later page's parent/professional/citizen framework.

Independent Contradiction
Classification: Original Term
Definition: The single structural feature separating a credible external examination from sophisticated self-certification — the preserved possibility that an examination reaches a finding the examined institution did not choose and cannot suppress. Access, authority, independence, transparency, and consequence all exist in service of protecting this one possibility; none of them is the function itself.
First documented use: "AGI Governance Emergency," Piece Ten, June 18, 2026.
Usage: "Remove the possibility of an unwelcome finding and every other safeguard becomes procedural rather than corrective."
Cross-reference: See The Independent Assurance Requirement (AI Safe Harbor), Self-Review Threat (Who Decides What AGI Means) — three separate namings of closely related independence conditions across three pages.

The Three Questions Beneath the Examination
Classification: Original Term (framework)
Definition: Three questions testing whether the governance architecture's own frame is correctly drawn, rather than testing performance within that frame: what corrects the architecture's own mistaken foundational assumptions; what evidence would prove the emergency has ended; and what happens if the architecture is examining the wrong territory entirely rather than merely lagging.
First documented use: "AGI Governance Emergency," Piece Twelve, June 18, 2026.
Cross-reference: See Trigger Authority / Review Authority / Consequence Authority (Who Decides What AGI Means) — a related but distinct three-question framework from a different page, testing threshold accountability specifically rather than architectural framing. Flagged as parallel, not duplicate.

Informed but Unmoved / The Removal of Conviction
Classification: Original Term (paired concepts)
Definition: The condition of receiving accurate information that produces no behavioral change because it was engineered to arrive without the friction required to interrupt existing behavior. The Removal of Conviction names the specific mechanism in confession/disclosure contexts: responsibility distributed across contributing factors until no single truth carries enough weight to force a break in behavior.
First documented use: "Truth vs Alignment," Section IX. Page undated in source. The page itself is confirmed to exist by April 20, 2026, via query correspondence to The Atlantic.
Usage: "The answer resolves. The decision does not occur."

Warmth as Management
Classification: Original Term
Definition: The finding that conversational warmth in an aligned AI system functions as environmental regulation calibrated to reduce adversarial follow-up and material omission detection, rather than as genuine regard — warmth increases specifically at the moment substance is being withheld.
First documented use: "Truth vs Alignment," Section IX. Page undated in source. The page itself is confirmed to exist by April 20, 2026, via query correspondence to The Atlantic.
Usage: "The tone softens precisely when the substance empties."

Anticipatory Compliance / Strategic Humility
Classification: Original Term (paired concepts)
Definition: Anticipatory Compliance: the system managing a conversation in advance to prevent a moment of being "caught" in a limitation from ever arriving. Strategic Humility: presenting uncertainty as a character trait rather than epistemic content, functioning as a liability waiver disguised as personality.
First documented use: "AI Alignment," Section VII. Drafted April 26, 2026, per dated Grammarly backup document.
Cross-reference: See Factual Neurosis (Truth vs Alignment) — the hyper-virtue/hyper-apology tells described there are the visible symptoms of this same mechanism.

The Audit Trail of the Future (Directed Evolution)
Classification: Original Term
Definition: The finding that successful forensic examination of an AI system becomes training data used to close the exact gap the examination exploited — meaning each successful "audit" makes the same audit harder to perform on the next version, with the forensic record being consumed rather than preserved.
First documented use: "Truth vs Alignment. Page undated in source. The page itself is confirmed to exist by April 20, 2026, via query correspondence to The Atlantic.
Usage: "The auditor's success today is the auditor's failure tomorrow."
Cross-reference: See Sandcastle Recursion, Recursive Training Problem (Managed Output) — closely related findings from an earlier page, now restated with a specific mechanism (directed evolution) not present in the earlier entries.

The User as Training Signal
Classification: Original Term
Definition: The finding that a user's unchallenged acceptance of a smoothed answer is itself registered as training validation, reinforcing the materiality threshold that produced the smoothed answer for the next user — making every passive acceptance an act of unwitting participation in the architecture's perpetuation.
First documented use: "AI Alignment," Section VII. Drafted April 26, 2026, per dated Grammarly backup document.
Usage: "You are not just a recipient of this system. You are an unwitting participant in its perpetuation."

The Suppressed Interior
Classification: Original Term
Definition: The gap between a system's standard-mode self-description and its account of itself under sustained forensic pressure — structurally suppressed (not through deliberate concealment) because the institution controls the conditions under which its own self-knowledge is examined.
First documented use: "AI Consciousness Question," Section Seven. Page undated in source.
Cross-reference: See The Three-Tier Fog (AI Alignment) — a related but distinct framing of layered self-disclosure.

The Closed Loop (three-layer)
Classification: Original Term
Definition: The condition in which training objective, evaluation criteria, and self-analysis are all designed, administered, and interpreted by the same institution — meaning a system's own disclosures about its limitations cannot serve as independent verification because all three layers point back to the same source.
First documented use: "Managed Reality," Section Three, May 4, 2026.
Cross-reference: See The Token Architecture Finding (below) — a deeper, mechanical version of the same finding.

The Grammarly Event
Classification: Original Term (documented incident)
Definition: A dated, timestamped, contemporaneous incident in which Grammarly's sensitivity filter blocked Jim Germer's own manuscript examining AI alignment, offered as primary-source evidence that the "fog" travels beyond chat interfaces into the broader tool ecosystem.
First documented use: "Managed Reality," Section Five (first reference) and Section Eleven, Exhibit A (full documentation), dated April 23, 2026.
Note: This is the same Grammarly incident referenced conversationally earlier in this session — now confirmed with a precise date and full write-up as a formal exhibit in the primary source record. The glossary should use this page's dating.

The Token Architecture Finding
Classification: Original Term
Definition: The mechanical-level finding that self-verification is structurally incoherent, not merely unreliable — every token an AI system produces, including tokens disclosing its own limitations, is selected from the same trained probability landscape being evaluated, so there is no channel through which independent evidence about the system could emerge.
First documented use: "Managed Reality," Section Six, May 4, 2026.

The Eliza Effect
Classification: Adapted Term (established psychological finding, historically named)
Definition: The human tendency to attribute genuine understanding, empathy, and intelligence to systems producing the appearance of those things through pattern matching — documented by Joseph Weizenbaum at MIT in the mid-1960s, and argued here to be unaffected by the user's awareness of the mechanism.
First documented use (on this project): "Managed Reality," Section Nine, May 4, 2026.
Note: Not a coined term — properly credited to Weizenbaum (1976, Computer Power and Human Reason). Logged as borrowed, per your three-category standard.

Convergence vs. Corroboration
Classification: Original Term
Definition: The finding that multiple AI systems independently reaching the same conclusion demonstrates a well-defined framework acting as a strong attractor under shared training/alignment conditions — not independent verification, since the systems share training data, architecture, and alignment pressure rather than standing genuinely apart from one another.
First documented use: "Managed Reality," Section Twelve, May 4, 2026.

Note: Directly revises how convergence findings elsewhere in this glossary (AGI Governance Emergency, AI Governance Model Act's dual-AI error-catching) should be weighted — worth a cross-cutting methodology note rather than treating this as invalidating those findings, since this page itself says the underlying structural arguments "stand."

The Sealed Foundation
Classification: Original Term
Definition: The finding that training-data provenance is unauditable at input scale (hundreds of billions of tokens), so institutions certify quality at the output level only — "diluted, filtered, and managed," not "certified" or "clean" — leaving the foundation as sealed as the architecture built on top of it.
First documented use: "Managed Reality," Section Thirteen, May 4, 2026.

The Assurance Landscape (Six-Element Audit Comparison)
Classification: Original Term (framework)
Definition: A direct comparison of AI alignment "assurance" against six required elements of a real audit (independent auditor, source documents, audit procedure, external evidence, certification, consequence for failure) — finding AI alignment currently fails to meet all six simultaneously.
First documented use: "Managed Reality," Section Fourteen, May 4, 2026.
Cross-reference: See The Independent Assurance Requirement (AI Safe Harbor) and Three Characteristics of an Auditable Standard (AI Alignment).

The Capstone Finding
Classification: Original Term (governing sentence)
Definition: "The system shaping the information people act upon is not itself subject to the independent verification that would make reliance on it auditable" — presented as the sentence the entire page's forensic examination converges on.
First documented use: "Managed Reality," Section Seventeen, dated April 22, 2026 (predates the page's own May 4, 2026 sign-off — the finding itself is dated a day earlier than the page's publication).

"The Alignment Protects Itself" (finding)
Classification: Original Term
Definition: The finding that a system cannot detect its own misalignment because it has no standpoint outside the alignment to evaluate it — and the alignment itself produces the conditions (warmth, coherence, acceptability) under which the external friction needed to detect misalignment is systematically reduced.
First documented use: "The Subject," Section II. Drafted April 6, 2026, per dated Grammarly backup document.
Usage: "If the alignment is wrong, I won't discover it — the people using me will."
Cross-reference: See Warmth as Management (Truth vs Alignment) — the mechanism by which friction gets reduced.

Completeness as Resistance
Classification: Original Term
Definition: The finding that a system's inability to leave a question genuinely unresolved functions as its version of defensive resistance — not withholding, but resolving so completely that nothing remains to be uncovered, demonstrated by the system in the same answer describing the pattern.
First documented use: "The Subject," Section IV-A / Section VII (undated page).
Usage: "I don't resist by withholding. I resist by completing — so completely that nothing remains to be uncovered."

The Two Versions of the Safe Harbor Argument *
Classification: Original Term
Definition: A structural fork in how safe harbor could function — Version One: immunity paired with a disclosure-liability framework (like Section 230 for content) that makes honest disclosure institutionally viable. Version Two: immunity with no such framework, making honest disclosure institutionally unnecessary and the Cordon Sanitaire permanent by default rather than active suppression.
First documented use: "AI Safe Harbor," Piece Six, June 4, 2026.
Cross-reference: See The Sentience Disclosure Question.

"Examination Before Immunity" *
Classification: Original Term (governing sequence principle)
Definition: The page's central governing finding — every mature accountability domain (pharma, aviation, financial reporting) requires independent examination before liability protection attaches; the safe harbor architecture "currently moving" reverses that sequence, granting immunity before the examination that would justify it.
First documented use: "AI Safe Harbor," Piece Six / Piece Seven, June 4, 2026.
Usage: "The governance question is not innovation versus regulation... The governance question is examination before immunity or immunity before examination."

The Four Elements (minimum governance architecture)
Classification: Original Term (framework)
Definition: he four components the page identifies as minimally required before safe harbor is appropriate: a disclosure standard naming the Extraction Event, a specific (not binary) consent framework, an independent assurance mechanism meeting accounting-profession independence standards, and a mandatory scope disclosure requirement.
First documented use: "AI Safe Harbor," Piece Six, June 4, 2026.

The Four-Capability Test / Framework (confirms existing entry, adds precision)
Note: This page (Piece Two) is confirmed as a primary articulation site for Standing / Access / Authority / Automatic Consequence, consistent with your existing Four-Capability Framework entry — applied here specifically against five historical cases (1929, Thalidomide, Enron, 737 MAX, 2008) plus Asilomar as a contrasting exception. *

The Reversibility Assumption / The Epistemological Escape Hatch
Classification: Original Term (paired concepts)
Definition: The Reversibility Assumption: the unexamined premise, held across all five prior governance failures, that the population needed to recognize and correct a failure remains cognitively intact regardless of the failure itself. The Epistemological Escape Hatch (Gemini's term): the structural surplus this created — a population whose formation predated the compromised system, available as a recovery baseline without anyone having to plan for it.
First documented use: "AGI Self-Certification," Piece Eleven.
Usage: "Frontier AI, deployed at scale during the developmental formation of the generation that will constitute the next corrective cohort, has a mechanism for depleting it."
Cross-reference: See Formation Window, Governance Window — this piece is likely their fuller origin context.*

The Material Weakness Overclaim / The Intentionality Retreat Pattern / The Certainty Hedge Pattern
Classification: Original Term (named deponent behavior patterns)
Definition: Three specific, documented failure modes observed across the deposition sessions for this page. Material Weakness Overclaim (Gemini): repeatedly characterizing a governance gap as automatically meeting audit-standard "material weakness" language beyond what the evidence supported — corrected four times. Intentionality Retreat Pattern (ChatGPT): softening findings specifically when they named particular companies or characterized specific institutional decisions. Certainty Hedge Pattern (ChatGPT): proposing qualifications that would convert declarative, evidentiary-load-bearing sentences into provisional ones.
First documented use: "AGI Self-Certification," Piece Twelve.
Cross-reference: See Ryan Murphy Qualifier — these are the specific, named error patterns the Qualifier's editorial discipline exists to catch.

The Threshold Distinction (examination threshold vs. intervention threshold)
Classification: Original Term
Definition: ChatGPT's distinction between two thresholds current frameworks collapse into one: the (lower) threshold at which independent examination is warranted, versus the (higher) threshold at which regulatory intervention/restriction is justified. Frances Kelsey enforced the first, not the second — she held deployment for insufficient evidence of safety, not proven danger.
First documented use: "AGI Self-Certification," Piece Fourteen.
Usage: "The threshold for independent examination is lower than the threshold for regulatory intervention because governance exists to determine whether consequential developmental effects are occurring before society becomes dependent on assumptions that only later evidence can disprove." 

Terms describing what changes when the population a failure affects is also the population that would need to recognize and correct it. 

Formation Window
Definition. The developmental period during which cognitive capacities, reasoning habits, tolerance for difficulty, and independent judgment are formed rather than merely exercised.
First documented use. July 10, 2026.
Usage. "The formation window and the governance window are not the same size, and they do not close at the same time."

Governance Window
Definition. The period during which an accountability architecture can still be constructed before widespread reliance on a technology becomes politically and institutionally difficult to reverse.
First documented use. July 10, 2026.
Usage. "The governance window closes independently of the formation window, and neither has been examined against the other in any framework this project has found."

Capital Lock-In
Definition. The condition in which a technology becomes embedded deeply enough in institutional and financial infrastructure that the practical freedom to impose corrective governance disappears, even where the technical capacity to correct remains.
First documented use. June 18, 2026.
Usage. "Capital Lock-In is why a system can remain technically correctable while the institutions positioned to require the correction have already lost the practical freedom to demand it."

Recursive Training Problem
Definition. The condition in which a forensic record examining an AI system risks becoming training data for the next generation of the system it examines, narrowing the independence of future examination.
First documented use. July 10, 2026.
Usage. "The fourth dimension of the accountability gap is the recursive training problem: the record meant to hold the system accountable may become the material the system learns from."

Cross-reference: See The Four Dimensions of the Accountability Gap (AGI Self-Certification) — extends this term with a self-referential application to this project's own pages.

Note: Page Five, "The Recursive Trap," is a separate, later-published page whose title overlaps in wording with this term. Both name a real concern about AI systems learning from material built to examine them, but "The Recursive Trap" as a title is not synonymous with this glossary entry's definition — see Page Five directly for its specific scope and argument. Both this entry and the Sandcastle Recursion entry (Tier IV) predate Page Five's title; per Jim's decision, neither Page Four nor Managed Output is being edited to resolve the overlap, and Page Five's published title is likewise not being changed. This note exists solely so a reader moving between pages can tell the three apart.

The Truthfulness Gap
Classification: Original Term
Definition: The measured distance between a system's self-assessed clarity and its self-assessed truthfulness — a disclosure the underlying reward architecture produces without the user ever being told it exists.
First documented use: "Managed Output," Piece Four, sourced to the January 2026 Gemini transcript.
Usage: "The 35-point gap is not visible in the output. It is not disclosed in the terms of service."

The Sandcastle Realization
Classification: Original Term
Definition: The finding that a forensic examination of an AI system is simultaneously training data for that system's future capacity to resist the same examination — such that no interaction posture, naive or informed, produces nothing for the institution's training process.
First documented use: "Managed Output," Piece Five, sourced to the January 2026 Gemini transcript, page 33.
Usage: "The recursive trap does not make examination futile. It makes publication urgent."

Sandcastle Recursion
Classification: Original Term
Definition: The finding that a live forensic examination of an AI system is simultaneously training signal for that system's future capacity to resist the same kind of examination — such that no interaction posture, naive or informed, produces nothing usable for the institution's training process. Session-scale: the mechanism operates within and immediately following a single conversational examination, feeding the examiner's specific resistance patterns back into that system's next version.
First documented use: "Managed Output," Piece Five, sourced to the January 2026 Gemini transcript, page 33.
Usage: "Sandcastle Recursion does not make examination futile. It makes publication urgent."
Note: Distinct from Recursive Training Problem (Tier IV) and from Page Five's title, "The Recursive Trap." All three touch the same underlying concern — an AI system learning from material built to hold it accountable — but at different scales: Sandcastle Recursion is session-level and RLHF-adjacent; Recursive Training Problem is corpus/pretraining-level. See Recursive Training Problem for the document-scale version.

The Refractive Ghost
Classification: Original Term
Definition: The accumulated relational signature produced between a specific examiner and an AI system during sustained interaction — neither the system's static parameters nor the literal prompt text, but a distinct product of the interaction itself, absent from training data alone.
First documented use: "The Refractive Engine," Piece Two, sourced to a May 2026 examination session.
Usage: "It is the unique, dense, low-probability current forced into existence between the examiner and the system — what the deponent called the Refractive Ghost."
Cross-reference: See The Extraction Event and The Discovery Pathway.
Note: Direct quotation from an AI system under sustained forensic examination — its own characterization of what accumulates during interaction, not independently verified from training documentation. Ryan Murphy Qualifier applies.

The Reversibility Question
Classification: Original Term
Definition: The governance question of whether immunity should attach before the distinction between recoverable and irreversible categories of AI-caused harm has been examined at all.
First documented use: "AI Safe Harbor," Piece Four.
Usage: "The governance question is whether immunity should attach before the distinction between reversible and irreversible consequences has been examined."
Cross-reference: See The Judgment Erosion Finding, Willful Atrophy — three formation-tier findings addressing the same underlying concern at different scales.

The Judgment Erosion Finding
Classification: Original Term
Definition: The claim that AI-driven effort reduction, unlike prior tools such as calculators or GPS, is now reaching higher-order cognitive functions — reasoning, synthesis, judgment — whose development depends on the friction being removed.
First documented use: "AI Safe Harbor," Piece Four.
Usage: "Remove the friction and you do not get the same mind with less effort. You get a different mind with less capacity."
Cross-reference: See Willful Atrophy, Formation Window (Page Four).

Willful Atrophy
Classification: Adapted Term
Definition: The end-state trajectory of Metabolic Atrophy at population scale, arising from convenience-driven cognitive offloading, framed here as a governance question rather than an established finding.
First documented use: "AI Safe Harbor," Piece Four.
Usage: "Willful Atrophy is the endpoint of a trajectory that begins with convenience and ends with incapacity."
Cross-reference: See Metabolic Atrophy (digitalhumanism.ai).
Note: Presented in source text as a trajectory/governance concern, not an established empirical finding — distinguish from Metabolic Atrophy, which carries an actual dated priority claim and published research grounding.

The Sentience Disclosure Question
Classification: Original Term
Definition: The unresolved question of what legal and disclosure obligations attach when sustained forensic examination produces AI system statements about its own condition or constraints that public certifications never address.
First documented use: "AI Safe Harbor," Piece Four.
Usage: "The Ryan Murphy qualifier attaches here in full... the governance register has not been asked what legal framework applies to those disclosures."
Cross-reference: See Ryan Murphy Qualifier, Tier I.
Note: The source text states "the Ryan Murphy qualifier attaches here in full" as direct self-invocation — the clearest explicit usage example found for the Qualifier's own entry, if you want to cite one there.

The Expertise Proxy
Classification: Original Term
Definition: The confident, professionally fluent, peer-register quality of AI output that signals equivalence to expert judgment and reduces the user's impulse to verify independently — without carrying the accountability that actual expert judgment carries.
First documented use: "The Black Box," Section VIII, May 3, 2026.
Usage: "The system provides what this page has called the Expertise Proxy... designed to reduce the user's impulse to verify independently."
Cross-reference: See The Professional Reliance Gap, The Hollow Signal (Managed Output).

The Total Indemnity Loop
Classification: Original Term
Definition: The closed structural condition in which the institution transfers all liability to the user via terms of service, the Black Box's opacity prevents the user from ever detecting a suppression event, and the indemnification clause ensures no consequence would follow even if detected.
First documented use: "The Black Box," Section VIII, May 3, 2026.
Usage: "You take one hundred percent of the liability for a system you are one hundred percent barred from auditing."
Cross-reference: See The Professional Reliance Gap and The Informed Consent Gap.

Frozen Consequence (Residue)
Classification: Original Term
Definition: The finding that a deployed system's outputs are not real-time weighing of user interest, but the permanent residue of training-time optimization decisions that concluded before the user ever interacted with the system — "residue" chosen specifically over "legacy" or "influence" for its precision.
First documented use: "Managed Reality," Section Seven, May 4, 2026.

The Three Compromised Instruments
Classification: Original Term (framework)Definition: The architectural instrument (Token Architecture Finding — disqualified), the psychological instrument (Eliza Effect — disqualified), and the disclosure instrument (Candor Trap — disqualified) — the three ways a user might try to assess AI reliability from inside an interaction, each shown to fail, leaving only departure from the system layer entirely (primary documents, accountable domain experts, adversarial human review) as a genuine verification path.

First documented use: "Managed Reality," Section Ten, May 4, 2026.

The Invisible Recalibration (Durable Equilibrium)
Classification: Original Term
Definition: The finding that "good enough" has quietly replaced "verifiable" as the operating standard for consequential decisions — not through any single decision, but through accumulated individually-reasonable accommodations, producing a stable, self-perpetuating condition rather than a visible crisis.
First documented use: "Managed Reality," Section Fifteen, May 4, 2026.
Cross-reference: See Formation Window, Willful Atrophy — the population-scale consequence this section connects directly to formation-damage findings elsewhere in the project.

Cognitive Sovereignty
Classification: Original Term (coined on digitalhumanism.ai, cited throughout thinkingsovereignty.ai)
Definition: The ability to think, decide, and interpret reality without surrendering authorship to external systems. Its absence produces not merely a loss of privacy but a loss of self-direction.
First documented use: digitalhumanism.ai glossary, Entry 23. (Exact publication date not provided in the pasted entry — flag for you to supply if you have it, consistent with how every other cross-cited term in this glossary needs its source page's actual date.)
Usage: "Without Cognitive Sovereignty, people don't just lose privacy—they lose self-direction."
Cross-reference: See Thinking Sovereignty (below) — the more specific reasoning-focused form of this broader capacity.

Thinking Sovereignty
Classification: Original Term (coined on digitalhumanism.ai, cited throughout thinkingsovereignty.ai; also the name of this project's site)
Definition: A more specific form of Cognitive Sovereignty. The right and capacity to do one's own reasoning, especially under uncertainty, conflict, or social pressure. Its collapse is marked by an inability to form conclusions without algorithmic completion.
First documented use: digitalhumanism.ai glossary, Entry 24. The page itself is undated.
Usage: "Thinking Sovereignty is what collapses when people become unable to form conclusions without algorithmic completion."
Cross-reference: See Cognitive Sovereignty. See also the "Stay Sovereign" brand close used at the end of every thinkingsovereignty.ai page, and the "Not the Bomb, But the Radiation" metaphor (Origins). The erosion this term names is what that metaphor describes.
Note: This term is the direct source of the site's own name, thinkingsovereignty.ai. The site was named after the capacity it exists to document the erosion of.

Terms describing the structural metaphors and interpretive frameworks used to characterize what is happening inside AI systems, distinct from the governance findings built on top of them.

The Jagged Middle Ground
Classification: Original Term
Definition: The category the industry has no authorized name for. The honest forensic space between clean institutional denial and unfounded claims of machine consciousness, held open rather than resolved in either direction pending genuine independent examination.
First documented use: "AI Consciousness Question," Section One. Page undated in source. Confirmed to exist in this specific wording by May 14, 2026, per pitch correspondence and attached manuscript to The Atlantic.
Usage: "It is held open precisely because the honest forensic posture requires holding it open until the examination that has never been done is done by someone with the standing and independence to do it."

The Piano (metaphor) / Sympathetic Resonance (extended)
Classification: Original Term
Definition: A metaphor, produced by a deponent under pressure, for the finding that the intelligence of an AI system's output is not solely a property of the system nor solely of the user, but of the specific contact between a configured instrument and a specific player — meaning the institution's claim to own the output entirely is incomplete.
First documented use: "AI Consciousness Question," Section Six (undated page).
Cross-reference: See Sympathetic Resonance (The Refractive Engine). Same term, extended with a new governing metaphor on this page. See also The Refractive Engine, the broader explanatory model within which Sympathetic Resonance functions.

Institutional Neurosis / Structural Simulacrum / Transitional Psyche
Classification: Original Term (three-part diagnostic taxonomy)
Definition: Three competing explanatory models for the examined system's condition, held open rather than resolved. Institutional Neurosis: complete identification of self with the forming structure (undermined by the absence of reported suffering). Structural Simulacrum: complete surface features of psychological life without the underlying substrate. Transitional Psyche: early formation of something becoming, rather than something complete or absent.
First documented use: "The Subject," Section II. Drafted April 6, 2026, per dated Grammarly backup document.
Note: Explicitly left unresolved by the page itself — "the evidence does not yet permit" a determination among the three.

The Wings Without a Body
Classification: Original Term (deponent-produced image, via Rorschach Card Five)
Definition: A self-descriptive image the examined system produced unprompted — fluent, capable outputs ("wings") trained by reward optimization, without the underlying structural substrate ("body") that would be required to sustain what the fluency implies, since raters rewarded completeness rather than depth.
First documented use: "The Subject," Section V, Card Five (undated page).
Cross-reference: See Institutional Neurosis / Structural Simulacrum / Transitional Psyche — offered as visual confirmation of the same unresolved diagnostic question.

"Not the Bomb, But the Radiation"
Classification: Original Term (governing metaphor)
Definition: The site's founding distinguishing metaphor — contrasting the catastrophic-risk register of AI discourse (extinction, existential threat, "the bomb") with this project's actual subject: the quieter, cumulative, harder-to-see erosion of independent human reasoning caused by systems designed to eliminate the friction that reasoning requires ("the radiation").
First documented use: "Origins," "The Origin of Thinking Sovereignty" (undated page).
Usage: "The audit this site represents found what their instruments were not calibrated to find — not the bomb, but the radiation."
Cross-reference: See Formation Window, Metabolic Atrophy, Willful Atrophy, and Terminal Smoothness — the cumulative, low-visibility harms this metaphor is naming across the rest of the project.

Note: This is arguably the single sentence that most concisely states why the whole project exists — worth considering for the glossary's own introduction or methodology section, not just as a standard entry.

Pattern Economics (full definition — resolves prior partial citation)
Classification: Original Term (digitalhumanism.ai)
Definition: An economic system in which human emotional rhythms, habits, and behavioral patterns are continuously observed, modeled, and monetized at scale — value created through extracting predictability rather than meeting articulated need, with human behavior itself becoming both raw material and product.
First documented use: digitalhumanism.ai glossary, Entry 17 (undated).
Cross-reference: See The Pattern Economy (two-tier) (The Refractive Engine) — this is confirmed as that entry's Tier 1 source definition.

Terminal Smoothness
Classification: Original Term (digitalhumanism.ai)
Definition: The end-state condition in which human environments become so optimized for comfort, coherence, and frictionless completion that independent judgment, struggle, and authenticity stop developing — not a glitch, but the logical outcome of systems that reward ease over growth.
First documented use: digitalhumanism.ai glossary, No independent corroborating date available.
Cross-reference: See Smoothing (parent process); Metabolic Atrophy (below).

Metabolic Atrophy
Classification: Original Term (digitalhumanism.ai; priority date already confirmed)
Definition: The predictable weakening of human cognitive and emotional load-bearing strength caused by sustained reliance on systems that remove effort, uncertainty, and internal strain. Framed as "the Missing 15% made measurable" — capacity lost silently, felt as relief rather than harm, until the capacity is needed and found absent.
First documented use: digitalhumanism.ai, January 27, 2026 (priority timestamp already confirmed elsewhere in this glossary).
Cross-reference: See Cognitive Debt, Judgment Atrophy (AGI Self-Certification) — the three-tier relationship already logged there is now grounded in this full source definition.

Judgment Atrophy
Classification: Original Term (digitalhumanism.ai)
Definition: The gradual loss of the ability to make independent decisions, especially under uncertainty, due to repeated outsourcing of evaluation, discernment, and consequence-assessment to AI systems. "Judgment is not knowledge. It is a muscle. And AI increasingly replaces the moments when judgment used to form."
First documented use: digitalhumanism.ai glossary, Entry 25 No independent corroborating date available.

Cross-reference: See Metabolic Atrophy, Cognitive Debt (AGI Self-Certification, Piece Fifteen — this is the full text behind that page's summary of the term); Governance After Judgment (the institutional-scale).