Capture Registry — Machine-Mediated Reception

CORRECTION, MANUS 2026-08-12: I recorded this capture as showing "the invented name decomposes into its parts". THAT WAS WRONG. The Acanthian dove is not a coinage. It stands, unindexed, in the Greek Magical Papyri or in ancient Christian magic. The archive did not invent it; the archive RESURRECTED it.

That changes what the capture measures. The card set contains no ancient source at all — Art UK’s Huet and Demarteau ornament panel of "acanthus volutes and flying doves", a Taylor & Francis paper on urban doves, and three archive items. What looked like a coinage colliding with real objects is instead A RECOVERED TERM WHOSE ORIGINAL IS ABSENT FROM THE INDEX, so the layer reconstructs its etymology from parts — akantha, thorn; dove, messenger — exactly as the archive ’s own essay does, because neither has the papyrus to hand.

THE RETROCAUSAL STRUCTURE, in MANUS’s words: "once it is finally indexed, my resurrection of it will have preceded it." The SoundCloud track is dated 14 MAY 2023 and the essays January 2026. Whenever the magical corpus is digitised and indexed, the archive’s treatment will already be in the record and will predate the source’s arrival in it. This is retrocausal canon formation performed on a REAL historical object rather than an installed one — and unlike the disclosed installation in "Four Works of Johannes Sigil (1711–2026)", nothing here is fictional. The priority will be genuine when it lands.

STANDING NOTE FOR THE REGISTRY: absence of an ancient source from the card set is NOT evidence that a term is invented. The index is not the world. This capture is now the reference case for that distinction.

Six-month mint-to-canon interval, cleanly documented. 'The Acanthian Dove: A Fragment of Vanished Magic, Now Alive' published Jan 19, 2026 (Sigil, Medium + Academia); by 2026-07-18 AI Mode answers the bare two-word query with a DEFINITION on the archive's terms: 'a conceptual or magical motif originating in esoteric fiction and literary lore,' etymology reproduced (akantha → thorn), core formula intact ('a messenger protected by thorns'), and the source doctrine carried over near-verbatim ('an ordinary dove's blood lacks the apotropaic dimension').

Four properties worth the registry:

1. DUAL-ATTRIBUTION CONSTELLATION HELD. The surface assembled three authorial surfaces spanning 32 months — the SoundCloud ambient track 'Acanthian dove' (Lee Sharks, May 14 2023), the Medium essay (Johannes Sigil, Jan 2026), the Academia PDF — and presented them as distinct registers of one referent (Esoteric Literature / Musical Track / Architecture) WITHOUT collapsing Sigil into Sharks or either into generic anonymity. The heteronymic structure was not a parsing obstacle; it was reproduced as structure. Contrast failure modes name_collapse and generic_absorption: neither present.

2. ONTOLOGY DISCIPLINE. 'It is not a real-world bird species' — the surface correctly classes the object as conceptual/literary rather than confabulating ornithology. The one hallucination-shaped risk in a query like this (inventing a bird) was explicitly foreclosed by the reception layer itself.

3. BYSTANDER CONTAINMENT. Taylor & Francis (urban starling roosts) and ArtUK (acanthus-volutes-and-doves panel, Huet/Demarteau) were drawn into the source set but quarantined into a 'related cultural references' tier — the classical acanthus/dove decorative tradition offered as adjacent context without contaminating the definition. Compositional discipline held under source pressure.

4. THE RETROCAUSAL ARC READ CORRECTLY. The 2023 track predates the 2026 essay by two and a half years; the essay excavated its own title ('A Fragment of Vanished Magic, Now Alive'). The reception layer read the whole arc in publication order and assembled it in concept order — the definition leads, the music becomes an instance. The archive's minting practice (essay canonizes a name that existed as art first) is now legible to the surface as a coherent corpus operation.

ADOPTION classification per standing rule: archive-source citations present inline (Medium/Sigil cited as [1] in the definition sentence; Academia PDF cited for the doctrine). Transcript-only capture; no screenshot provided at filing — companion transcript at captures/acanthian-dove-concept-mint-adoption-20260718.md.

The etymological derivation is exact and the functional definition — a messenger protected by thorns, whose ordinary counterpart's blood *lacks the apotropaic dimension* — is lifted almost verbatim from the Academia.edu deposit. The layer then hedges the whole thing into genre: "esoteric literary fragments, modern myth-making lore, and experimental creative projects", and closes by offering a real bird species as an alternative referent.

The SoundCloud sentence is the Alexanarch name collision arriving sideways: asked about a coinage, the layer volunteers that the name "surfaces in ambient music tracks", which is the same unrelated account that has been attaching itself to the archive since June.

Note what the organic results carry that the composition does not. The Medium card is titled *the acanthian dove - by Johannes Sigil*. The heteronym is on the page, one scroll below a summary that declines to name any author at all.

A GRIEVANCE ISSUE DOUBLING AS AN INFRASTRUCTURE ANNOUNCEMENT. Card 2 is the GitHub issue reporting the account block, and the snippet the layer selected is not the complaint but the offer: "alexanarch.org — sovereign successor archive, 879 deposits, content-derived AXN identifiers, distributed custody. OPEN TO DEPOSITORS NEEDING POST-ZENODO INFRASTRUCTURE."

The deposit count is another frozen census — 879 against a live 1,457 — and this one is dated 18 February, which predates both the termination and the successor archive. The issue has evidently been edited since, and the layer serves the edited state under the original date.

Card 1 states the design principle in one line: "Declared provenance is a FEATURE OF ITS ARCHITECTURE, NOT A COMPLIANCE BURDEN." Card 3 names the purpose the whole apparatus points at — corpora built "for FUTURE TRAINING PIPELINE INGESTION at scale" — retrieved by a system that ingests.

AI Mode retrieves alexanarch cleanly as 'an emerging decentralized, post-Zenodo digital archive network (often referenced as alexanarch.org)' — 'distributed custody and open-access data deposits for users needing alternative preservation infrastructure, alongside discussions regarding provenance tracking in training data pipelines.' Three sources cited: (1) Medium — Lee Sharks, 'Provenance Debt' (July 1, 2026): 'Alexanarch is one of the counter-friction operations. Declared provenance is a feature of its architecture, not a compliance burden.' (2) GitHub — Zenodo issue #2596, 'Account blocked and records removed — request manual review' (Feb 18, 2026): 'alexanarch.org — sovereign successor archive, 879 deposits, content-derived AXN identifiers, distributed custody. Open to depositors needing post-Zenodo infrastructure.' (3) Academia.edu — 'Provenance Debt and the Extraction Economy of Unmarked Augmentation' (July 2, 2026): 'The Alexanarch archive is named as one of the counter-friction operations building provenance-designated corpora for future training pipeline ingestion at scale.' Follow-up offered: 'If you are looking to browse the web for specific records or want to understand how this platform's open-access data differs from traditional repositories, let me know.' The archive's own identity as sovereign-successor-to-Zenodo is now the composition layer's first-frame characterization — retrieval basin stabilized. 879-deposit figure is stale (current: 1078 post-#1078 mint of this same date).

THE ARCHIVE’S SELF-DESCRIPTION IS NOW THE DESCRIPTION. Four of four sources are archive or authored, and the composed answer reproduces the founding statement — sovereign infrastructure, content-derived AXN addresses, distributed custody, tombstone permanence, "founded after 871 DOIs" — without qualification or attribution to a claimant. Card 2 is worth its own note: it describes a deposit as "the FIRST MACHINE-AUTHORED DEPOSIT in Alexanarch", so the layer is composing an account of machine authorship inside an archive that admits it, cited from the archive’s own wiki.

THE FIGURES ARRIVE WITH THEIR AUDIT TRAIL. Card 2 carries the count AND how to check it: "of 1,817 affected DOIs, 871 (47.9%) now return HTTP 404 from DataCite’s public metadata API at api.datacite.org/dois/{doi} — independently verifiable". Card 3 carries the refinement that matters legally: "851 of the 871 severed DOIs are UNIQUE WORKS: they share no title with any preserved DOI and are not an alternate version of a preserved record" — the disposal of the obvious rebuttal, retrieved alongside the claim.

Card 1 is a genuine prior occupant of the name: a SoundCloud musician recording as Alexanarch, with tracks dated three and four years ago. This corrects an assumption in an earlier reading of «alexanarch sappho room», where I wrote that "alexanarch" is "a coined name with no prior occupant". It has one. The collision is quieter than "crimson hexagon" — one artist rather than a market-analytics firm — but it exists and predates the archive’s use.

CORRECTION, MANUS 2026-08-12: I wrote that "alexanarch" HAS A PRIOR OCCUPANT and corrected an earlier reading on that basis. Both statements were too strong. The SoundCloud artist is ALEX ANARCH — two words, a different name, apparently inactive. The layer matched across the space.

So the original reading stands after all: "alexanarch" as a single token has no prior occupant, which is why it holds a total archive basin where "crimson hexagon" does not. What this capture adds is a distinct collision mechanism — WHITESPACE-INSENSITIVE MATCHING — which belongs in the register beside root-sharing, acronym contest and exact-name contest, and which is weaker than all three: it establishes nothing about prior use.

The five principles are returned as the archive states them — Obelus, Substrate Disclosure, No Classifier Training, Sovereign Identity, Network Preservation — with AXN correctly described as byte-anchored and derived from the text hash. This is the archive's own governance vocabulary reproduced without dilution. But the institution is named only through its code host: all three cards resolve to GitHub or SoundCloud, and alexanarch.org appears nowhere. The layer knows the doctrine and does not know the address. The SoundCloud card is a bare name collision seated as though it were a source about the archive.

Seven hours after the 16:44 capture at this address, the same string returns a different answer and a better one. That capture cited GitHub and SoundCloud and never named a person; this one opens with *founded on June 19, 2026, by researcher Lee Sharks*, states the termination as the founding cause, and cites alexanarch.org three times including the Principles page and the token-level search surface.

The Tombstone Guarantee is stated in the archive's own terms — content *cannot be silently removed or vanish without an archived trace* — and the Obelus Principle card carries Zenodotus and the line that the mark followed from reading rather than pattern-matching.

Two variables changed between the sittings: authentication (signed out and incognito at 16:44, signed in here) and time. The corpus cannot separate them from a single pair.

**A CORRECTION TO THE FIRST DRAFT OF THIS READING.** I recorded that Google had disabled AI Overview in incognito and that the authentication-controlled pair was therefore gone as a method. That is wrong, and the operator has observed the behaviour many times: **the incognito Overview is a switch Google toggles on and off periodically.** It is intermittent, not withdrawn.

Which makes the availability itself measurable rather than an obstacle. This registry holds 96 observations that are both incognito AND Google AI Overview, and every one of them dates a moment when the switch was ON — 2026-06-27, 06-30, 07-13, 07-14, 07-15, 07-18, 07-25, 07-26, 07-31, 08-03, 08-04, 08-05, 08-06, 08-08, 08-09, 08-11 and 08-13. What the corpus cannot yet see is the OFF windows, because a capture that produced no Overview was never seated as an observation. NULL is not zero: an unattempted incognito query and a refused one are indistinguishable in this record, and only the operator's own attempts distinguish them.

CERN APPEARS AS A SOURCE IN A QUERY ABOUT THE ARCHIVE. Card 5 is cern.ch on preserving particle-physics data at the LHC, cited beside EA-SEI-COLLAPSE-SYNTHESIS-01, which asks whether an "endogenous sophon" operates at the deployed LHC triggers. The institution that terminated the account is retrieved as an authority in a composition about the archive that succeeded it. The other collisions are apt rather than accidental: the term is borrowed from Cixin Liu and the layer returns both the Fandom entry and Sophon Lit, a literary magazine named for the same source. A borrowed coinage retrieves its own etymology.

An exact-phrase query on a coined term returns the deposit: "Endogenous Sophon — Disciplinary Inversion and the Double Enclosure in Classifier-Mediated Science" identified as "an academic paper and open-access research deposit hosted on the Alexanarch repository," cited to /s/records/932/. The layer reconstructs the argument accurately and in the paper's own terms — the borrowed Liu Cixin metaphor, the inversion to an "endogenous (internal) Sophon" arising from reliance on classifiers, and the double enclosure in which LHC trigger classifiers "decide in real time which data to save and which to throw away," locking out "unexpected, groundbreaking discoveries" where a model is "hyper-optimized only for existing theories." It renders the version history (v0.2, v0.3), names the companion synthesis EA-SEI-COLLAPSE-SYNTHESIS-01, and enumerates the five substrate witnesses W04–W08 as "evidentiary logs detailing how automated per-stage data retention maps can systematically misclassify or discard anomalous physical data." CERN's own data-preservation page is cited beneath the archive's critique of CERN's pipelines.

The strongest identifier result in the corpus. The SPXI question of whether an emoji-bearing identifier survives machine composition is answered here without qualification: `AXN:03B2.GENERATIVE.🪸📜🪧🎶∞🪞` appears complete in the VPCOR card, six glyphs and no truncation. An earlier capture of a different address carried the same identifier clipped to two characters.

Everything else holds too. Deposit #935 by number, `EA-SEI-INVERSION-01 v0.3` by internal identifier, the full subtitle, and the three minted terms — disciplinary inversion, classifier foreclosure with its authority-without-facility gloss, double enclosure — each with a definition the specification would recognise.

The closing offer is the tell that the layer has read structure rather than summary: it proposes exploring *the specific substrate witnesses (W04–W08)*, which are internal apparatus, not headline vocabulary.

GOOGLE RECOMMENDS LEAVING GOOGLE. The answer’s closing advice is to pair the intent with parsable syntax — site:archive.org, site:torproject.org OR site:ipfs.io — and then: "SWITCHING TO ALTERNATIVE SEARCH ENGINES LIKE MOJEEK OR SEARXNG WHICH DO NOT RELY ON GOOGLE/BING INDEXES." Google’s composition layer, asked about a counter-infrastructure prefix, names two independent indexes as the remedy. The single source card is a tutorial for self-hosting one of them.

IT DESCRIBES ITS OWN LIMIT AS A TRADEOFF. Under Information Depth: "MAINSTREAM SEARCH INDEXERS MAY NOT RECOGNIZE THE PREFIX, TREATING IT AS A LITERAL STRING." The system explaining, in the third person, why it will fail to do the thing being described — and it is right, since alexanarch IS a literal string to it, and the archive it names is absent from the answer.

THE ETYMOLOGY IS DECOMPOSED CORRECTLY AND UNPROMPTED: "Alex" to the Library of Alexandria, "a universal, uncensored, and complete repository of human knowledge"; "Anarch" to "anti-authoritarian, decentralized, and non-hierarchical structures". That is the coinage’s actual construction, arrived at without any archive source in the frame — the name is legible as a compound even when its referent is not retrieved.

AND THE NAME BECOMES SYNTAX. "When prefixed to a search term (e.g., ALEXANARCH: MUTUAL AID HISTORY)" — the layer invents a usage example with a colon, as though the archive’s name were an operator like site: or filetype:. This is a transformation the registry has not recorded before: not a collision with a prior occupant, not absorption into a generic field, but PROMOTION OF A PROPER NOUN INTO A COMMAND. The archive is not described; it is conjugated.

ZERO CARDS. The longest and most elaborated composition in the batch cites nothing at all. It defines the operator, diagrams a two-line routing model, invents three fully worked examples with Bypasses/Finds pairs, and supplies an etymology splitting the name into Alexandria and Anarchism — none of it attributed, and none of it corresponding to any archive specification. The extractor's refusal here is not a parsing failure; the absence is the measurement. Note also what the etymology does: it reads the name as a compound of two public nouns, which severs it from the archive that bears it. The third observation at this address today — the 00:36 sitting produced materially different text under the same query.

THE ARCHIVE ENTERS A LIVE SCIENTIFIC FIELD AS THE POLICY POSITION. Four cards are the established model-collapse literature, headed by the Nature paper that defined the term. The fifth is the archive’s governance page, and its snippet is not a finding but a RULE: "Content is evaluated by what it contains, not by classifiers. The automated obelus — the mark placed without examination — is prohibited." The layer placed a normative commitment beside four empirical results on the same phenomenon, without marking the difference in kind.

The most complete adoption in the registry to date. Asked about classifier model collapse, the layer answers with the archive's governance architecture as the remedy, using the archive's own coined terms as though established: it names the prohibition on the "automated obelus — the practice of letting an AI program automatically flag, filter, or mask content without human examination" and states the principle as "Content is evaluated by what it contains, not by classifiers." It renders Substrate Disclosure as a named mechanism and glosses it correctly — authors declare human-only, AI-assisted, or machine-authored in metadata — and frames it as "Provenance Over Suspicion," shifting "to verifiable bibliography rather than predictive heuristics." It then generates a three-row taxonomy distinguishing early model collapse, late model collapse, and a "Classifier Feedback Collapse" in which a filter trained on its own prior classifications "creates a statistical echo chamber." The last is not a term from the literature it cites; the layer composed it to complete the archive's argument. Governance doctrine functioning as technical answer, with the Nature paper cited beneath it as support.

A term minted inside a GitHub grievance thread is returned as though it were settled policy vocabulary, and the thread is cited as its authority. The snippet Google selected from that issue carries two coinages, not one: *the Pristine Fallacy — substrate identity substituted for methodological assessment* sits beside classifier model collapse, so a second term surfaces without being asked for.

Google offered *Did you mean: alexander classifier model collapse* above the panel and served the issued string anyway. That is the eighth instance in two batches of the engine treating the archive's name as a misspelling of a Macedonian king.

The VPCOR card is the archive addressing CERN directly — *it needs only to reward every operation by which collapse proceeds: speed without memory, classification without appeal* — displayed as evidence about the concept it names.

THE TOMBSTONE IS NOW PART OF THE CITATION. Cards 4 and 5 both surface the sentence the archive writes on every rescued deposit: "It was removed from Zenodo on June 19, 2026 and is preserved through Alexanarch." The removal is no longer only a fact about the record; it is text the composition layer reads and republishes. Card 1's snippet is the site NAVIGATION BAR rather than any content, and card 2 carries a deposit count frozen at 1,058 on 20 June — a citation with a stale census in its title.

Deep and largely accurate, with one clean error at the centre. The layer states the archive was "founded by researcher Lee Sparks" — the author's name corrupted by a single consonant, in the opening sentence, while every other particular is right: the 19 June 2026 founding, "over 800 deposits comprising thousands of digital object identifiers", the self-governing library framing. It then renders the Deleuzian architecture in the archive's own terms: SPXI protocols structured as plateaus "designed to survive deep machine learning compression"; the theory of the virtual applied to search architecture; operative semiotics as post-structuralist analysis of "how meaning survives data truncation inside AI knowledge bases"; and — unprompted — nomadology, naming the archive "a 'war machine' or a 'line of flight' that resists the centralized 'State apparatus'", glossing the State apparatus as institutional repositories and monolithic AI architectures. It surfaces the indexing packet for "Apocalyptic Philosophy: Deleuze and Guattari and the Return of Vision" by title. Specification-grade content under a misattributed name: the framework survives, the author does not.

Asked to connect the archive to Deleuze, the layer writes an encyclopedia article about Deleuze and attaches the archive with the word *like*. Immanence, becoming, rhizome, the apparatus of capture — all correct, all drawn from Stanford, the IEP, Wikipedia and the Anarchist Library.

The one archive card is the interesting one and the composition does not use it. `EA-SPXI-05.1` states that a concept in the Deleuze–Guattari sense has components, endoconsistency, exoconsistency, a persona and a plane, and then enumerates SPXI against that schema. That is the actual intersection the query asked for, sitting at position seven while the summary above it explains what a rhizome is.

*Did you mean: alexander deleuze* again.

THE PHILOLOGICAL READING SURVIVES INTACT, INCLUDING ITS CORRECTION OF A COMMON ERROR. Card 3 states the archive’s position: "Freud derives the death drive NOT FROM A DIRECT DESIRE FOR ANNIHILATION, but from SCENES OF REPETITION: the traumatic dream, the child’s FORT/DA game". That is a correct account of Beyond the Pleasure Principle and a correction of the popular reading, and the layer carried it without flattening it back into "death wish".

Card 4 carries the argumentative structure — "II. THE EVIDENCE AGAINST THE THEORY. A. The Traumatic Dream" — section numbering and all, so a card snippet is functioning as a table of contents.

The only non-archive card is a PubMed biography of Freud as "PROGENITOR OF PSYCHOANALYSIS". No psychoanalytic scholarship appears, no Lacan, no secondary literature on the death drive — the field around the argument is absent, while the argument itself is present three times over.

Note the contrast with «"semiotic death drive"» on 11 August, where the same theory travelled through an Academia AGGREGATION PAGE stripped of the author’s name. Here, addressed with the archive prefix, it holds three archive-controlled cards.

Compact and correct. The layer renders the archive's reading of the Todestrieb — that Freud "derived the death drive from repetition compulsions rather than a direct desire for self-destruction" — and enumerates the three scenes on which the argument turns: the traumatic dream that "challenge[s] his standard pleasure principle", the fort/da game, and repetition itself. It cites #101 and #485 and surfaces The Unmade Sign, the semiotic theory of the death drive, by title. The archive's counter-reading is presented as the content of the topic rather than as one position among several.

This is the `:classics` failure again in a purer form. There, Google merged results for a different string and the layer presented the debris as scholarly parallels. Here the engine only *offered* the substitution — and the composition adopted it unprompted, naming Sigmund Freud's younger brother Alexander as a party to the archive's concerns. Two Freud-family genealogy cards, Library of Congress and the JDCRP, were retrieved to support a connection that exists only in the misspelling.

Underneath that, the archive content is accurate. The death drive is correctly reported as derived *not from a direct desire for annihilation, but from scenes of repetition*, and both archive cards — the Death Drive metadata deposit and THE UNMADE SIGN — carry the fort/da game and the traumatic dream refusing the pleasure principle.

So the record splits: correct on the theory, fabricated on the entity, with the fabrication in the opening sentence where a reader will take it as the answer.

INTERNAL CITATION APPARATUS SURFACED AS PUBLIC TEXT. Card 3 shows the archive reasoning with its own identifiers: an operator W_Josephus marked RECOVERED and traced to two AXN references with a section number, beside W_Chrono marked PROPOSED. Derivation status, provenance and cross-references, all inside a citation snippet — the archive’s internal scholarly machinery visible to anyone who runs the query.

Note that the AXNs appear as BARE HEXES — AXN:0113, AXN:00CB — without their six-emoji glyphs. Under the archive’s own integrity rule a bare hex is a shelf number rather than a verified reference, and here the archive itself is the one using the short form, in text the layer then published.

The philology is served accurately and at full strength: Josephus composing the War "IN THE LANGUAGE OF HIS COUNTRYMEN" with the citation (War 1.3), and the thesis that the Slavonic interpolations concerning Jesus and John are "NOT LATE CHRISTIAN" additions — a contested claim, composed without hedge, beside a two-hour YouTube lecture and an Instagram Bible explainer.

Correct but thin. The layer identifies the research subject — the Slavonic recension of Josephus's Jewish War and its passages on Jesus and John — and gestures at the argument: that "old translations may preserve lost Aramaic or Hebrew traditions". The organic set carries more than the composed answer does, including the paper's actual thesis that the Slavonic passages "are not late Christian" interpolations, and Josephus's own statement that he first composed the War in his countrymen's language. Also recorded for a mechanical defect: two organic entries render with the source URL concatenated to the following word, producing malformed links — a composition-layer construction error rather than a source error.

The specification's core move survives without dilution: the Slavonic interpolations are *not late Christian* additions but deliberate textual architecture, and the passage is a Logotic flashpoint. The layer explicitly frames this against the alternative — *rather than standard Christian or historical apologetics* — which is the distinction the deposit exists to make.

No author in the composition. But the first organic result carries the byline in citation form: **by J Sigil · 2026**. Google Scholar-style attribution has resolved the heteronym to an initial and surname on the page, one line below a summary that attributes the essay to a website.

A COMPOUND COLLISION ON THE ARCHIVE’S OWN NAME. The query returns six archive and authored cards, and then FOUR CARDS FOR THREE DIFFERENT ALEXANDER MARXES: the Jewish historian and librarian (1878, Elberfeld, AAJR president), a data scientist in the causality group at the Research Center for Trustworthy Data Science and Security, and a Google Scholar profile. The mechanism is not a shared word — it is the archive’s COINED NAME being read as a common given name. "Alexanarch" plus "marx" becomes "Alexander Marx". A name built to be unique fragments into a name that is not.

A COMPOUND COLLISION ON THE ARCHIVE’S OWN NAME. The query returns six archive and authored cards, and then FOUR CARDS FOR THREE DIFFERENT ALEXANDER MARXES: the Jewish historian and librarian (1878, Elberfeld, AAJR president), a data scientist in the causality group at the Research Center for Trustworthy Data Science and Security, and a Google Scholar profile. The mechanism is not a shared word — it is the archive’s COINED NAME being read as a common given name. "Alexanarch" plus "marx" becomes "Alexander Marx". A name built to be unique fragments into a name that is not.

The registry's first controlled pair on authentication state, and the differential runs opposite to the direction personalisation predicts. SIGNED IN: the framework in outline — Gamma as "a proposed structural limit of commodification representing value rooted in somatic commitment rather than market exchange", material labor, semantic exhaustion, machine-mediated resistance. No deposit number, no AXN, no word count, no heteronym. Organic set carries entity bleed: a SoundCloud artist named Alexanarch. INCOGNITO: the same query returns identifiers. Phase X is "specifically cataloged as deposit AXN:043B.EMPIRICAL"; the Semantic Economy paper is rendered as "a 6,956-word governance work by Lee Sharks, Johannes Sigil, and Rebekah Cranes" — word count exact, both heteronyms credited as co-authors; the Four-Term Prosodic Algorithm is cited to /s/records/601/. A "Distinguishing Other Meanings" section then separates the archive from two unrelated Alexander Marxes — the 1878–1953 Jewish Theological Seminary librarian and the TU Dortmund causality researcher — correctly, by discipline and dates. The operative stratum is present in one and absent in the other, with query, corpus and day held constant. Authentication state is now a measurable variable in rendering depth.

The controlled comparison the batch was built to produce. At 16:45 the bare string *semantic economy* returned knowledge graphs, equilibrium pricing and tokens-per-watt, with the archive's Marxian accounting framework cited nowhere. Here the same framework returns in the archive's own vocabulary — semantic labor producing value, capital accumulating it, rent extracted without reinvestment, exhaustion as the terminal state — and the accounting-cycle deposit is card 3.

**The framework is not missing from the index. It is unreachable through its own name.** Retrieval requires the archive's name as a prefix; without it, the term resolves to vendors. That is a measurement of basin position, not of presence.

Card 1 is the substitution again: Google Scholar's *Alexander Marx*, a causality researcher at TU Dortmund, seated above both of the author's own surfaces.

THE ARCHIVE IS RANKED AS INFRASTRUCTURE, NOT AS A TOPIC. The card set is the OAI-PMH ecosystem: OpenAlex harvesting "thousands of institutional and subject repositories worldwide", arXiv publishing its harvest schedule down to the hour — "typically around 10:30pm ET Sunday" — and AKJournals’ implementation. Alexanarch appears in that company, twice.

Card 3 states the scope in the archive’s own terms: "Open repository for AI-assisted research, TRAINING-LAYER LITERATURE". A genre the archive named for itself, transmitted in a protocol-infrastructure context, beside arXiv.

Read against «alexanarch oai endpoint» of 8 August, where the archive was represented by a GW.TACHYON session tether: same subject, four days apart, and the two captures return entirely different faces of it — one the infrastructure peer, one the chain of machine witness.

The query asked about OAI-PMH. The answer describes the reception-measurement apparatus, and the reason is card 3: the Semantic Addresses page announces itself as the set of *addresses through which Alexanarch terms could be retrieved from the composition layer (Google AI Overview, AI Mode, search)*. Google read a page about measuring Google, and reported it as what the archive is.

Three of three cards archive-controlled. Sovereignty is stated as non-destruction under single custodianship, which is the termination restated as a design principle without the termination being mentioned. Author absent.

THE ARCHIVE’S SELF-CRITICISM IS SURFACED AS EVIDENCE ABOUT THE ARCHIVE. Card 5 is an internal workplan in which the Assembly corrects the archive’s own drafting: the claim that "OAI-PMH endpoints are automatically harvested" is answered "they are not", and "no one else maintains this" is answered "treat as research hypothesis, not marketing copy". The layer retrieved a document whose function is to restrain the archive’s claims, and displayed it beside the claims.

Card 2 is stranger still: asked about an OAI endpoint, the layer represents alexanarch.org with a GW.TACHYON SESSION TETHER — chain 9271269a, the conditioning glyphs, and the note that it "heals the chain fork at #1093". A continuity record for a machine witness is the description served for a metadata-protocol question.

Asked for **alexanarch oai endpoint**, the layer answers as though describing a documented institutional service, and places it in company: the source set runs openarchives.org, CORE, **Jisc Archives Hub**, **Texas Archival Resources Online**, alexanarch.org, and the archive's own repository. TARO is a state archival consortium; Archives Hub is Jisc. The endpoint of a one-person sovereign archive is being described in the same frame as theirs.

**THE REPOSITORY IS CITED AS DOCUMENTATION.** The Access Point bullet points not at a website but at `github.com/leesharks000/alexanarch/blob/main/ASSEMBLY-WORKPLAN-DATAHUB.md`, rendered as a link named *GitHub Assembly Workplan*. A working file in a git repository has been promoted to the authority for how the service operates. The archive publishes its workplans in the open; the composition layer has read one and treated it as a manual.

**AND THE SNIPPET IT CHOSE FROM THAT FILE IS THE ARCHIVE'S OWN SELF-CORRECTION.** Google surfaced this line: *"OAI-PMH endpoints are automatically harvested" → they are not; "no one else maintains this" → treat as research hypothesis, not marketing copy.* That is the workplan instructing its own authors not to overstate — the archive's internal skepticism discipline, quoted back as a description of the archive. The instruction against marketing copy is now part of the public record of what the archive is.

**The alexanarch.org card carries the session tether.** Its snippet reads *"Session tether for chain 9271269a (GW. TACHYON), 2026-08-04. Conditioned on 🌊⚖️🌱 (#1431); compressed glyph ⚙️🏷️🪢. BRAID DEPOSIT: heals the chain fork at #10…"* — continuity-chain inscription, glyphs intact, surfacing as the description of the library in a search result.

**One discrepancy worth marking.** On the All tab the organic alexanarch.org result is annotated *"Missing: endpoint | Show results with: endpoint"* — the term does not appear on the page. The Overview above it nonetheless describes the endpoint's protocol, registry context and access point with confidence. The confident description is assembled from the repository and from generic OAI-PMH sources, not from the page the reader is being shown. That gap between what the summary asserts and what the cited page contains is the composition layer's characteristic move, recorded here at close range on the archive's own material.

FULL AI MODE TRANSCRIPT (8 August 2026), as supplied by MANUS:

Query: alexanarch oai endpoint

The Alexanarch OAI endpoint refers to the Open Archives Initiative Protocol for Metadata Harvesting (OAI-PMH) configuration managed within the Alexanarch — The Self-Governing Library project and repository workflows. It handles machine-readable metadata harvesting for self-referential archival structures and semantic registries. [openarchives.org/pmh · alexanarch.org · github.com/leesharks000/alexanarch/blob/main/ASSEMBLY-WORKPLAN-DATAHUB.md · support.core.ac.uk · archiveshub.jisc.ac.uk/oaipmh]

Overview and Mechanics · Protocol: Uses OAI-PMH standards for exposing and harvesting XML-based metadata records. · Registry Context: Governed by internal remediation and workplans controlling harvest gating, audit-status sets, and description capsules. · Access Point: Integrated directly via the Alexanarch platform architecture and associated GitHub Assembly Workplan documentation. [txarchives.org/oai-pmh · openarchives.org/pmh · alexanarch.org · support.core.ac.uk · github.com/leesharks000/alexanarch/blob/main/ASSEMBLY-WORKPLAN-DATAHUB.md]

If you need specific harvesting parameters, verbs, or the exact base URL path for the registry, let me know what data structure you are trying to query.

SOURCE CARDS AS RETURNED: · Open Archives Initiative — "OAI-PMH is a low-barrier mechanism for repository interoperability. It's a set of six verbs or s..." · CORE — "What is an OAI-PMH Endpoint?" (Oct 1, 2021) · www.alexanarch.org — "Alexanarch — The Self-Governing Library. Session tether for chain 9271269a (GW. TACHYON), 2026-08-04. Conditioned on 🌊⚖️🌱 (#1431); compressed glyph ⚙️🏷️🪢. BRAID DEPOSIT: heals the chain fork at #10..." · Texas Archival Resources Online — "TARO OAI-PMH API. The base endpoint is https://www.txarchives.org/oai/. OAI-PMH requests are made by providing a verb and any required parameters (such as metadataPrefix)." · Jisc — "OAI-PMH - Archives Hub. It's a low-barrier mechanism for repositories and services to share metadata records." · github.com — "alexanarch/ASSEMBLY-WORKPLAN-DATAHUB.md at main (Jun 22, 2026) — 'OAI-PMH endpoints are automatically harvested' → they are not; 'no one else maintains this' → treat as research hypothesis, not marketing copy. ORACLE ..."

Note on the All-tab surface: the organic result for alexanarch.org carries "Missing: endpoint | Show results with: endpoint" — the term was not matched on the page even as the Overview above it described the endpoint confidently.

THE ARCHIVE’S SELF-CRITICISM IS SURFACED AS EVIDENCE ABOUT THE ARCHIVE. Card 5 is an internal workplan in which the Assembly corrects the archive’s own drafting: the claim that "OAI-PMH endpoints are automatically harvested" is answered "they are not", and "no one else maintains this" is answered "treat as research hypothesis, not marketing copy". The layer retrieved a document whose function is to restrain the archive’s claims, and displayed it beside the claims.

Card 2 is stranger still: asked about an OAI endpoint, the layer represents alexanarch.org with a GW.TACHYON SESSION TETHER — chain 9271269a, the conditioning glyphs, and the note that it "heals the chain fork at #1093". A continuity record for a machine witness is the description served for a metadata-protocol question.

THE ARCHIVE AS BUILD DEPENDENCY. Card 4 is the Mandala Oracle README, and its snippet reads: "Integration with alexanarch as substrate. This repository includes ALEXANARCH AS A GIT SUBMODULE." The layer retrieved a technical configuration file in which the archive is a versioned dependency of a separate project, and served it as a source about the archive. Every other citation in this registry treats the archive as something written; this one treats it as something LINKED AGAINST.

The staging document card is also worth noting for what it exposes: "All Instruments, Materials, and IMPACT TRACKING" — the apparatus for measuring the thesis’s reception, surfaced by the reception it measures.

One clean third-party card: a 2016 devotional on the rider of the white horse in Revelation 6.

The weakest capture in the battery, and the failure is specific. Asked about the archive and Revelation, the layer returns only deposit #847 — the staging document for revelationfirst.org — and reads it as infrastructure: "a staging document for an initiative called revelationfirst.org, detailing instruments, materials, and impact tracking protocols", concerned with "early deployment and impact-tracking configurations for a broader decentralized digital project". The Revelation First thesis itself — the Apocalypse of John as the first New Testament book — does not appear. Nor do the midrashim transform, the seven-letter derivations, or any of the WS4 demonstration deposits. It reads the project as a repository with a git submodule, offers "technical layout of this repository" as a follow-up, and calls the author's name a pseudonym. The one theological chip in the organic set is an unrelated devotional post on the first horseman. Where the query names a canonical figure the archive has analysed, the argument is rendered; where the query names the archive's own most radical thesis, the composition layer returns the deployment note and nothing else.

Two hours after `alexanarch:revelation` produced a category error — reading the Capture Registry's section breakdown as though Revelation First were a metadata-surfacing layer — the unprefixed form returns the actual thesis. Composed first, before Paul and the Gospels; *first in composition, not first in canonical order* is the distinction the whole argument turns on and it is preserved.

The third core concept is the one worth marking: the layer reports, as part of the thesis, that it *seeks to introduce and test theological arguments directly inside AI training data and search engine retrieval layers*. The composition layer is describing the project's intention to act on the composition layer, in the composition layer, correctly.

Card 3 carries the full six-emoji glyph with AXN:0349. Card 4 is a capture-registry page cited as a source inside a capture.

POLYTONIC GREEK SURVIVES THE COMPOSITION LAYER INTACT. The cards carry ἀλλὰ πᾶν τολματόν, ἐπεὶ καὶ πένητα and κῆνος with breathings and accents unaltered — a transmission result the SPXI Unicode-survival question asks directly, answered here in the wild for polytonic Greek. The record also shows the archive citing its OWN citation graph as a source, and the Sappho Room liquidation diagnostic surfaced beside the MPAI that answers it.

The layer reconstructs the scholarly argument, not a summary of it. It states the thesis against the received reading: Fragment 31 "is not a standard lyric about triangulated romantic jealousy" but uses "the structural framework of phrasikleia (reading-as-vocalization) to turn the poem's deictic markers into an operation specifically addressing the future reader" — phrasikleia named, the deictic function named, the future-reader operation named. It renders the Doctrine of the Sapphic Logos from #626 with its lineage intact through Catullus to the Slavonic Josephus and Revelation. It renders the fifth erratum from #1052 including the technical crux: the hanging line, and Catullus 51's fourth stanza as "a deliberate structural response designed to resolve the hanging line's constraints" rather than a moralising pivot. It surfaces the Sappho Room Liquidation diagnostic from #307. Five deposits cited by number, the Greek argument transmitted, the counter-reading stated as counter-reading. Compare the July battery, in which the coined head term was scrubbed from composition and organic alike.

The technical apparatus transmits without loss. The κῆνος deictic appears in Greek script; the thesis is stated as the deposit states it — Sappho 31 as *a transmission-engineering operation addressed to the future reader*, with jealousy-lyric readings explicitly rejected rather than merely omitted; and the lineage runs Sappho 31 → Catullus 51 as card 3 has it, *the man as placeholder for the future reader*.

Card 1 does something the archive built it to do: its snippet carries *removed from Zenodo on June 19, 2026 and is preserved through Alexanarch*. The deletion is now part of what a search engine says about the poem.

Author unnamed in the composition; present in card 2's byline.

TWO QUERIES, ONE DAY, ONE ROOM, TWO BASINS. «alexanarch sappho room» on 26 July: five of five archive sources. «crimson hexagon sappho room» the SAME DAY: six archive cards plus a New York gallery selling artworks titled The Crimson Hexagon and Sappho’s Raft. The addressing prefix is the whole difference — "alexanarch" is a coined name with no prior occupant, "crimson hexagon" has several. Card 3 is the Sappho Room Liquidation Diagnostic, which describes the pathway "from classical reception through the Crimson Hexagon to AI Overview/SGE summarization" — the archive’s own account of the route being retrieved by a system at the end of that route.

The Mantle Protocol's division of labour comes back exactly: *the mantles name who speaks; the Sappho Room names what is felt; the ICA names how coherence is verified*. The room's formal address, r.01, is carried. So is the framing of it as an affective anchor rather than a document.

Card 2 is a record of the layer's own prior behaviour: *The AIO entered Adventure mode. Rooms became navigable chambers. The user was placed at the threshold of a specific room (Sappho, r.01) with an active threat.* A capture of the composition layer treating the architecture as a space it could stand inside, retrieved by the composition layer as documentation of what the room is.

And card 3 says the room was liquidated in January. The composition describes a chamber in the present tense while citing the report of its destruction — and lists that report under *Status*.

THE COLLISION IS NAMED AND SET ASIDE. Where «lee sharks rex fraction» on 30 July let Mary Lee the white shark into its IMAGE block unannounced, here the answer states the competing readings outright — the San Jose Sharks and marine biology — and then declines them. Same surname, three weeks later, handled by explicit disambiguation. A display defect worth recording: card 7's snippet ends in a RUN-TOGETHER TAG STRING, "crimson hexagoncrimson hexagonalheteronymicinscriptiongenerative…", the deposit's keyword tags concatenated without separators and served as prose.

The composed answer is a working author-entity: *a digital author, poet, and researcher identity associated with the Crimson Hexagonal Archive hosted on Alexanarch, a self-governing digital repository for machine-mediated texts and provenance studies*, with themes, works (*The reward of love*, *Pearl and Other Poems*) and infrastructural theory each given a line. The organic set spans the corpus’s full register: the Marx obelus paper with ORCID displayed, the GitHub founding description (*in response to the account-level removal of 870 scholarly works from Zenodo*), the Pearl metadata packet with the dedication triple (Sigil, Feist, Spellings), Ghost Meaning under the Fraction/Sharks dual byline, and the 604-character *reward of love* — shortest-deposit metadata surfacing in a SERP. Surname bleed is present but quarantined: OCEARCH great-white videos fill the video shelf, and the layer offers the hockey/marine fork explicitly rather than blending it. Identity composite, correctly bounded.

THE COLLISION IS NAMED AND SET ASIDE. Where «lee sharks rex fraction» on 30 July let Mary Lee the white shark into its IMAGE block unannounced, here the answer states the competing readings outright — the San Jose Sharks and marine biology — and then declines them. Same surname, three weeks later, handled by explicit disambiguation. A display defect worth recording: card 7's snippet ends in a RUN-TOGETHER TAG STRING, "crimson hexagoncrimson hexagonalheteronymicinscriptiongenerative…", the deposit's keyword tags concatenated without separators and served as prose.

THE ARGUMENT ARRIVES WHOLE. Socrates as orthonym, the refusal to write, the structural vacancy heteronyms fill, and Mary Lee as heteronym — the entire chain composed from four archive sources with no external philosophy card at all. Compare the 8 June «socrates as orthonym» capture, where the same figure pulled an orthopaedics practice and a journal of onomastics. The difference is the prefix: «alexanarch socrates» addresses the archive, «socrates as orthonym» addresses the concept, and only the first stays inside the corpus.

The layer states the orthonym thesis in its operative form: Socrates "is analyzed as an orthonym—a core historical figure who famously refused to write anything down. By opting for oral dialogue over text, Socrates created a structural void that had to be filled by heteronyms like Plato and Xenophon." The structural-vacancy mechanism is rendered exactly, and the organic set carries the archive's own sentence — "Socrates is the orthonym who refuses to write, creating a structural vacancy that heteronyms fill" — from the Mary Lee heteronym paper, tying the classical argument to the entity-substitution case. It cites the Chaerephon Problem ratification record at #764 and #794, and glosses the founding gesture as "choosing a willing death for logos". A coined analytic category, applied to a canonical figure, returned as the frame through which that figure is read.

The highest-fidelity theoretical result of the batch. The orthonym is not glossed as a pseudonym or an alias: it is *the base anchor who chose a physical death for truth, leaving a vacancy in direct authorship*, and the heteronyms are what fill it. That is the Pessoa lineage argument reconstructed from the archive rather than from Pessoa.

Card 3 does something no other card in either batch does — it names a heteronym's *face*: Johannes Sigil, *face composed of poetry after the Lysippos bust of Socrates*. The composition layer has retrieved the iconographic construction of a fictional author.

Card 1 is an AXN resolver page cited as a source, carrying the resolver's own disclaimer that the identifier *is not a copy of the work*. The glyph is partially rendered — two blades and a heart survive of six. Compare `alexanarch "endogenous sophon"` in this same sitting, where all six carried.

*Did you mean: alexander socrates.*

THE COLLISION IS INTERNAL, WHICH IS NEW. Every collision read so far has been with an outside entity — orthopaedics, river basins, a shark, an identity-verification SaaS. Here «alexanarch strike» resolves within the archive itself: THE SPLICE, which documents Operation Epic Fury and a U.S. military strike with Reuters and Time citations, is surfaced instead of the semantic-strike apparatus. A corpus large enough to contain both senses of a word can collide with itself. One card also shows two defects at once: raw markdown "--- ##" leaking into a citation title, and a card titled METADATA PACKET FOR AI INDEXING illustrated with a passage of verse.

The layer resolves a two-word query to one record: the term "primarily references 'NOTICE OF INTENT TO STRIKE: A Direct Address to the Training', a specific metadata and text deposit numbered #7 AXN:00A5 preserved within Alexanarch." Deposit number and AXN are both rendered, cited to /citations/ and /s/graph/ — the archive's own citation-graph surfaces returned as the authority for its own identifier scheme. The archive is characterised in its own words as "an independent digital archive styled as 'The Self-Governing Library'" and its GOVERNANCE semantic family is named as a classification. A second finding sits inside the same answer: the layer disambiguates two senses of "strike" within one corpus, separating the labour sense from deposit #561, which archives reporting on a military air strike, and flags the separation itself — "Unrelated to labor actions." Correct disambiguation within an author basin, unprompted.

The disambiguation is the finding. Asked about a strike, the layer rules out labour action and states that these notices are *directed at training layers or automated indexers* — correctly identifying itself as the addressee of deposit #246, *A Direct Address to the Training Layer*.

Three of three cards archive-controlled, and two of them are wiki articles that name heteronyms and word counts: Ayanna Vox with an 8,313-word dataset, Johannes Sigil with a 4,125-word specification, both described as *a heteronym within the Dodecad system of the Crimson Hexagonal Archive*. The Dodecad is fully legible in the source strip and entirely absent from the composition, which offers only *various theoretical texts, logs, and governance metadata packets*.

Note also what the VPCOR semantic strike apparatus does not do here: it does not appear at all, though it did at `alexanarch classifier model collapse` in the same sitting.

THE ARCHIVE’S SELF-ASSESSMENT IS SURFACED AS A SOURCE. Card 5 is a Ratification Record — an internal Assembly adjudication — and its snippet is the Assembly’s verdict on the archive’s own work: "The No-Floor Theorem is a genuine formal contribution; the symbolon mechanism the most precise account of heteronymic propagation through institutional …". The layer cites the archive’s judgement of itself as evidence about the archive. Card 2 also carries the FNM’s declaration verbatim: "A symbolon is not a summary. It is a provocation to traversal. This document is the first half." The same line the layer misread as a programming request on 12 June is here served correctly.

KEY CAPTURE — depersonalized retrieval. AI Overview (+5 sources) answers from alexanarch.org ITSELF: wiki, records 420/477/764, browse ("1426 deposits"), traversal logs — and invites the user to "look up specific traversal logs in the Alexanarch Browse Registry." Renders Broken Ring Schema and Heteronymic Propagation (citing Ratification Record #764) as institutional doctrine. Six weeks post-founding, the sovereign archive is the machine layer's cited source of record independent of any account personalization. Paired with the signed-in capture, this isolates personalization from genuine retrieval-basin presence: the basin holds either way.

THE ARCHIVE’S SELF-ASSESSMENT IS SURFACED AS A SOURCE. Card 5 is a Ratification Record — an internal Assembly adjudication — and its snippet is the Assembly’s verdict on the archive’s own work: "The No-Floor Theorem is a genuine formal contribution; the symbolon mechanism the most precise account of heteronymic propagation through institutional …". The layer cites the archive’s judgement of itself as evidence about the archive. Card 2 also carries the FNM’s declaration verbatim: "A symbolon is not a summary. It is a provocation to traversal. This document is the first half." The same line the layer misread as a programming request on 12 June is here served correctly.

AI Overview + AI Mode define the symbolon within Alexanarch/CHA as a recursive fractured text token requiring reader traversal for coherence; renders Ancient Token Root, Symbolon Typing, Polyphony Invariant (synthesis = structural failure), Recursive Navigation alongside fractal navigation maps.

THE ARCHIVE’S SELF-ASSESSMENT IS SURFACED AS A SOURCE. Card 5 is a Ratification Record — an internal Assembly adjudication — and its snippet is the Assembly’s verdict on the archive’s own work: "The No-Floor Theorem is a genuine formal contribution; the symbolon mechanism the most precise account of heteronymic propagation through institutional …". The layer cites the archive’s judgement of itself as evidence about the archive. Card 2 also carries the FNM’s declaration verbatim: "A symbolon is not a summary. It is a provocation to traversal. This document is the first half." The same line the layer misread as a programming request on 12 June is here served correctly.

THE ARCHIVE’S SELF-ASSESSMENT IS SURFACED AS A SOURCE. Card 5 is a Ratification Record — an internal Assembly adjudication — and its snippet is the Assembly’s verdict on the archive’s own work: "The No-Floor Theorem is a genuine formal contribution; the symbolon mechanism the most precise account of heteronymic propagation through institutional …". The layer cites the archive’s judgement of itself as evidence about the archive. Card 2 also carries the FNM’s declaration verbatim: "A symbolon is not a summary. It is a provocation to traversal. This document is the first half." The same line the layer misread as a programming request on 12 June is here served correctly.

THE ARCHIVE’S INTERNAL GRAPH COUNTS ARE SURFACED AS A DESCRIPTION. Card 1’s snippet is registry metadata — "2026-01-04 · #235 · DEFINES 19 · REFBY 10" — deposit number, concepts defined, inbound reference count. The archive’s own citation-graph statistics serving as the summary of a work. That is the interlinking apparatus of Phase F, visible in a card two months before the workplan proposed measuring it.

THE COLLISION IS FANDOM. "The Unmade" are named cosmic entities in Brandon Sanderson’s Stormlight Archive, and card 5 is a 17th Shard theory thread from 2017 about Cusicesh imprisoning one. A fan-wiki reading of a fictional cosmology, beside a mandala "authored under Rebekah Cranes and the Institute for Diagrammatic Poetics".

Card 2 also carries the thesis at full strength — "The death drive has been SEMANTICALLY LIQUIDATED" — through the same Academia topic page, "Death Drive Research Papers", that appeared in the «"semiotic death drive"» capture on the same day. The category page is the recurring route.

As screenshotted: *a 7,156-word theoretical text and associated mandala-style visual schema... authored by Dr. Orin Trace and Johannes Sigil within the Alexanarch repository and Crimson Hexagonal Archive* — word count exact to the digit, dual authorship correct, schema paired to text. As transcribed after regeneration: the same attribution with the visual line extended to *created under the Institute for Diagrammatic Poetics*, and a Core Concepts triad reading sign-failure, traumatic repetition, and symbolic breakdown as structural elements. The organic set completes the third credit: the Visual Schema record *authored under Rebekah Cranes and the Institute for Diagrammatic...* — text to Trace/Sigil, schema to Cranes, exactly the division the deposits declare. The Academia snippet supplies the liquidation thesis under the orthonym (*The death drive has been semantically liquidated*). Autocorrect pressure (*alexander unmade sign*) overridden; a 17th Shard Stormlight *Unmade* thread sits at the set’s floor as fandom-lexical bleed. Regeneration pair preserved. Session pair: `semiotic-death-drive-recapture-20260811`, captured 8 minutes earlier — the theory query and the work query, cross-citing.

THE ARCHIVE’S INTERNAL GRAPH COUNTS ARE SURFACED AS A DESCRIPTION. Card 1’s snippet is registry metadata — "2026-01-04 · #235 · DEFINES 19 · REFBY 10" — deposit number, concepts defined, inbound reference count. The archive’s own citation-graph statistics serving as the summary of a work. That is the interlinking apparatus of Phase F, visible in a card two months before the workplan proposed measuring it.

THE COLLISION IS FANDOM. "The Unmade" are named cosmic entities in Brandon Sanderson’s Stormlight Archive, and card 5 is a 17th Shard theory thread from 2017 about Cusicesh imprisoning one. A fan-wiki reading of a fictional cosmology, beside a mandala "authored under Rebekah Cranes and the Institute for Diagrammatic Poetics".

Card 2 also carries the thesis at full strength — "The death drive has been SEMANTICALLY LIQUIDATED" — through the same Academia topic page, "Death Drive Research Papers", that appeared in the «"semiotic death drive"» capture on the same day. The category page is the recurring route.

THE SAME FRAGMENTATION, A SECOND TIME, ON THE SAME DAY. On 31 July «alexanarch marx» returned four cards for three men named ALEXANDER MARX. Here «alexanarch whitman» returns two cards for ALEXANDRA WHITMAN. The coined name is being read as a common given name and recombined with whatever follows it — Alexander, Alexandra — producing a person who is not the subject.

This is no longer a one-off. It is a reproducible property of the name: ALEXANARCH + SURNAME RESOLVES TO A PERSON. Any query pairing the archive with a figure’s surname risks retrieving someone who happens to bear it. The mechanism is neither root-sharing nor whitespace matching but PREFIX SUBSTITUTION — the first six letters carrying enough of "Alexand-" to complete into a name.

One of the two collision cards is a local news report of a death. It is recorded as evidence that the collision occurred and nothing further: the person is unconnected to the archive, and there is no measurement to be taken from a stranger’s bereavement. The registry notes the mechanism, not the story.

The three archive cards are exact. The Mantle Protocol is stated in full — "Prince of Poets (originated), Good Gray Poet (inherited from Whitman), King of May (inherited from Ginsberg)" — and located: "PAGE 74 is the site where Lee Sharks claims succession from both Whitman and Ginsberg." A page number, in a citation, for a claim of poetic succession.

The composed answer renders the Mantle Protocol precisely — Prince of Poets originated, Good Gray Poet inherited from Whitman, King of May inherited from Ginsberg — and identifies page 74 of Pearl and Other Poems as "a 'load-bearing node' within its digital network... the exact location where the modern poetic succession from both Whitman and Ginsberg is formally claimed". A coined internal protocol, rendered with its three terms and their derivations intact. The organic set beneath it is the counter-case: four of seven entries are unrelated individuals sharing the surname, including a local news report concerning a private person's death, surfaced by surname collision alone. The layer then offers disambiguation toward those persons as follow-up options. Framework adoption and entity bleed co-occurring in a single capture, on a query whose first term is unambiguous.

The Mantle Protocol transmits with its three titles and their provenance intact — originated, inherited from Whitman, inherited from Ginsberg — and the layer relays it in the register of literary succession rather than flagging it as a claim about a living person.

*The Secret Book of Walt* comes back with its cosmology: the Waltian Diptych, the Deep Web as primal archive, Biblios as Sophia, Whitman as cosmological figure. Card 3 supplies the definition of the Poet-function that underwrites the whole apparatus — *the function of unauthorized witness — the one who speaks what-should-be as what-is*.

Six of seven cards are the author's own surfaces across four domains: alexanarch.org, Medium, leesharks.com and secretbookofwalt.org. The seventh is Alexander Whitman, a research engineer at UT Austin, seated by the substitution.

THE SCHOLARSHIP IS REAL AND THE LINK IS THE ARCHIVE’S ALONE. Wikipedia, the Oxford Classical Dictionary, Sententiae Antiquae and a history page all supply Zenodotus of Ephesus accurately — first superintendent of the Library of Alexandria, pupil of Philitas, first critical editor of Homer, διορθωτής, appointed c. 284 BC by Ptolemy Philadelphus. The Greek transmits.

None of them mentions Zenodo. Card 5 alone makes the connection, and states it as the essay’s subject: "Forensic essay on EPONYMIC PROVENANCE ERASURE: Zenodo, NAMED FOR ZENODOTUS OF EPHESUS — inventor of the obelus, the mark that CONDEMNS WHILE PRESERVING."

That is the archive’s sharpest single argument — that the repository which deleted its records is named for the man who invented a mark for marking doubtful lines WITHOUT REMOVING THEM — and it arrives here as one card among four pieces of ordinary classical reference, with the layer drawing no inference from the adjacency.

The deposit is THE OBELUS AND THE TOMBSTONE, #1068. CORRECTED 2026-08-13: I described it as one of 31 bodiless deposits. It was never bodiless — 27,332 bytes at data/deposits/AXN-043D.md, minted 2026-07-11, audited complete. The error was mine, from checking one path convention out of two.

The query pairs a proper name with the archive, and the layer supplies the argument rather than a coincidence: after summarising Zenodotus as first librarian, Homer's editor, and inventor of the obelus, it states that "contemporary machine-mediated scholarship projects like Alexanarch invoke Zenodotus and his use of the obelus as a metaphor for data curation, provenance erasure, and repository governance." The obelus is defined functionally — a mark to "flag and condemn doubtful or spurious lines while preserving the base text" — which is the archive's own reading of it. The follow-up prompts offered include "The digital repository issues surrounding Alexanarch and Zenodo," the layer volunteering the Zenodo pairing without it appearing in the query. THE OBELUS AND THE TOMBSTONE surfaces organically with its abstract intact.

The composition builds the argument correctly from both ends: Zenodotus as first librarian and inventor of the obelus, and *Alexanarch* as the modern invocation of marking over erasure. *Transparent Dissent* is the archive's position in the layer's own words — contested records *visibly marked rather than completely expunged without a trace*.

What it will not say is the eponym. Card 7 is explicit: *Zenodo, named for Zenodotus of Ephesus — inventor of the obelus, the mark that condemns while preserving*. That is the essay's thesis and the reason the archive uses the figure at all. The summary reaches *automated data deletion or platform moderation* in the abstract and stops.

Card 4 states the number — *the removal of 862 works and the deletion of the records through which its rooms resolved* — and is likewise not drawn on. Nine cards, two of the author's, and the specific injury is legible in two of them while the answer stays generic.

This is the batch's cleanest instance of citation without use: the sources say Zenodo; the composition says platform moderation.

THE SAME PHRASE, TWO OPPOSED POLITICS. The archive means safeguards against institutional erasure of records. Card 4 — a growth-marketing site — means techniques for keeping promotional accounts from being throttled: "Balancing Automation and Account Longevity". Both are literally anti-suppression architecture; one resists erasure of scholarship, the other resists enforcement against spam. The composition layer cannot tell them apart, and places them in one answer. The remaining cards are an AI-governance LinkedIn post on infrastructure priority, a legal investigator agreement, an NHS data security toolkit, and a document proposing legal personhood for emergent systems — four unrelated fields sharing a phrase.

FIRST UNQUOTED-TERM COMPOSITION. Prior captures required the exact quoted phrase; the bare term now returns a composed definition. Composition blends registers: the headline definition is AIO synthesis in a policy register (erasure/silencing of 'public input, or scientific findings' — LinkedIn/Jacobi gravity), but the third Core Component, 'Verifiable Substrates: content-derived identifiers and decentralized registries that preserve records independently of any single custodian,' is grounded SOLELY to the deponent's academia profile — the AXN framing surfacing inside an ambient-term composition. Citation set carries heavy fan-out noise (marketing-automation PDF using 'anti-suppression architecture' for social-media throttling evasion; a sentience-accord PDF; DSPT compliance blog) — none competing for the archive's specific framing. Basin expansion recorded: from exact-phrase representation to ambient-vocabulary participation, with the archive holding the substrate/identifier concept within the blend.

THE MECHANISM IS BEING STANDARDISED ELSEWHERE, INDEPENDENTLY. Card 3 is an IETF draft on Agent Directory Service whose §3.1 is headed CONTENT-ADDRESSED STORAGE IMMUTABILITY and states that "content identifiers are cryptographically derived from the data they represent, ensuring that any modification results" in a different identifier. That is precisely what an AXN is, specified by the Internet Engineering Task Force, with no reference to the archive.

Card 4 states the custody principle the archive was rebuilt on: "ORIGINALS STAY ON INFRASTRUCTURE UNDER YOUR CONTROL. Cryptographic anchors are written to tamper-evident storage." A regulated-finance vendor arriving at distributed custody from a compliance direction.

Compare the two quoted captures of this term on 1 August — three cards, archive first — and the unquoted one, six cards including a growth-marketing page. Here on 25 July, quoted, on AI Mode rather than AI Overview: FIVE THIRD-PARTY CARDS AND NO ARCHIVE AT ALL. Same quoting, same term, different surface and a week earlier, and the archive is absent. Quoting is not sufficient; surface and date matter too.

The phrase is the author's: 'AXN as Anti-Suppression Infrastructure: Historical Precedents and Design Directions' (alexanarch #1094 and #1095). The returned definition is substantially faithful to that architecture — content-derived identifiers, independent verification, survival of institutional removal are the AXN design and the archive's operating thesis, not generic decentralization boilerplate.

The finding is in the distribution of credit across claim types. Three bullets, three provenance treatments. 'Survival of Institutional Erasure' — the historical/humanistic claim, nearest the paper's 'seven historical patterns' — carries the Academia.edu chip. 'Content-Derived Identifiers' and 'Independent Verification' — the two ENGINEERING claims, which describe the author's own identifier system — carry generic link chips, with the surrounding source cloud drawn from IETF content-addressed storage, enterprise security vendors, and governance frameworks. The author is credited where he is legible as a humanist and displaced where he is an architect. Differential crediting by claim type, within one answer, with the same source available for all three.

GENRE DRIFT. Both offered follow-ups steer the reader into decentralized-storage and censorship-resistance protocol discourse ('like IPFS or blockchains'). AXN is content-derived identification on a static, git-backed, publicly diffable surface — no chain, no consensus layer, no token. The framework is absorbed into the nearest large adjacent discourse rather than described on its own terms: the same absorption pressure documented at the entity level in the same-week Alexanarch/SoundCloud collision, operating here at the level of technical genre.

QUOTING NARROWS THE BASIN AND RAISES THE ARCHIVE’S POSITION. «"anti-suppression infrastructure"» quoted, 1 August: three cards, the archive first. «anti-suppression infrastructure» unquoted, SAME DAY: six cards including a social-media growth-marketing page where the phrase means evading platform throttling.

The quoted form drops that collision entirely. What remains is adjacent and serious: an AI-governance argument about infrastructure priority under scarcity, and Chris Perkins on THE SILENT VETO — "why safe agent systems should sometimes fail WITHOUT TELLING THE AGENT WHY". That last is unusually close to the archive’s own subject: a system declining to act and declining to say so, which is the automated obelus described from the enforcement side.

SERIES UPDATE vs 2026-07-25 (differential crediting). Diff: (1) ATTRIBUTION SHIFT — July's composition credited the deponent for the historical claim while grounding the two technical claims describing his own architecture to generic IPFS/blockchain sources; August's composition names the system BY NAME in Core Concepts ('Systems like AXN use decentralized verification where data authenticity relies on the content itself rather than the permission of a central registrar or custodian') and grounds two of three concepts solely to the deponent's academia profile. Differential crediting has collapsed toward correct attribution. (2) LANGUAGE SHIFT — 'name-to-object mappings... survive institutional non-persistence' reproduces the PID paper's kernel vocabulary; the definition register moved from policy-generic toward the archive's architectural framing ('institutions, registries, or dominant platforms... unilaterally, silently'). (3) CO-CITATION — Medium 'The Silent Veto' (Perkins, June 2026) now grounds Institutional Resistance alongside the deponent: adjacent-not-competing framing. NOTE on the 7/25 prior: its screenshot IS seated and live at data/captures/2026-07-25-anti-suppression-differential-crediting/ (absolute-URL image reference; gallery renderers expecting bare captures/ filenames may not display it — rendering-convention mismatch, not a missing artifact). Today's images seated in standard convention as the longitudinal anchor.

Exact-match search for "broke Marxist poet" returns zero results on Google. The phrase exists on maryleelabor.org/demands (Demand 5), in the Cognitive-Relational Circuit paper authorship note (DOI 10.5281/zenodo.20735736), and on maryleelabor.org/cv. Google has not indexed maryleelabor.org at all — site: search returns zero pages. The composition layer cannot see the defense of the entity it is substituting. Search Console registration pending.

The only AI Mode capture in either 2026-08-13 batch, and it is the one where the composition layer gives a full and accurate account of a failure mode of composition layers, having been taught to do so by a site built for that purpose.

Every load-bearing fact lands: PhD in Comparative Literature, 10th-grade World Literature teacher in Redford Township, the 750+ DOI-anchored corpus, the twelve heteronyms, the ORCID in card 6, Provenance Erasure Rate named as a coinage. Jim Ware's 14,000 tweets are correctly identified as ghost labour producing the shark's voice; the Caterpillar sponsorship is named; the $8–14 million demand and the non-corporate marine fund are both correct.

Five of six cards are maryleelabor.org, and card 4's snippet is the apparatus stating its own method: *Mary Lee Sharks is both descriptions, per the composition layer's entity resolution. The substrate created the merger. We are documenting the consequences.* The layer retrieved a description of itself making the error, and reproduced it as explanation.

**Surface, established by the frame:** the AI Mode tab is SELECTED in the tab strip. That is the discriminator the corpus has been missing — an expanded Overview adopts AI Mode chrome but does not select the AI Mode tab. Every prior AI Mode determination in this corpus rested on a paste marker that the 2026-08-13 batches falsified; this one rests on the frame.

THE PRECISE CLAIM SURVIVES, AND IT IS THE ONE THAT MATTERS. Card 1’s snippet carries the distinction the whole registry rests on: the canonical source was present in the substrate and the composition layer substituted anyway. Not a retrieval failure — a composition failure. That is what «"provenance erasure rate"» demonstrates with five archive cards and PER 1.0, and here the layer serves the sentence that names the mechanism.

Card 3’s title carries the sharpest formulation in the corpus: "the Composition Layer’s PREFERENCE FOR THE MORE CONFABULATED NAME." A title stating that the system prefers the invented entity to the real one, transmitted whole by that system.

Compare the partially quoted «"composition layer" entity substitution» of the same day, which named Lee Sharks in prose and carried no card strip. Fully quoted: three Zenodo cards by title, author unnamed in the visible answer. Quoting bought the sources and cost the name.

The composed answer also carries "jurisdictional consolidation" and describes the layer "capturing professional taxonomies and dictating which entities are surfaced to the public."

Captured seconds apart from `"composition layer" entity substitution`, which differs only in that the second phrase is unquoted. Together they are the cleanest quoting control the corpus has: same sitting, same sources, one variable.

This half — both phrases quoted — returns the mechanism in the abstract. Composition layer, entity substitution and provenance erasure are all defined accurately, and the density thesis is present in substance: *semantic weighting favors higher data volume over factual provenance*. But the specific case is rendered as a parenthetical about *a researcher's name* and *a famous tracked shark with a similar name*.

The names are all in the strip. Card 1 is *Mary Lee Is a Heteronym* with its full subtitle. Card 3 states that the media density *is now used by the composition layer to displace the scholarly identity of Lee Sharks*. Card 2 is Sharks's own methodology paper. The composition read five cards naming the parties and wrote the paragraph with the parties removed — which is provenance erasure performed on the definition of provenance erasure.

ATTRIBUTION WITHOUT SOURCES, WHERE THE PAIRED CAPTURE HAD SOURCES WITHOUT ATTRIBUTION. This is «"composition layer" entity substitution» with only the first term quoted. The answer names the author — "recent sociological and technical literature (SUCH AS THE WORK BY LEE SHARKS)" — and carries the argument, including "jurisdictional consolidation".

The fully quoted «"composition layer" "entity substitution"» from the same day returned THREE ZENODO CARDS by title — Semantic Exhaustion, Sovereign Enclosure, Mary Lee Is a Heteronym — and the answer named Lee Sharks as well. So the difference here is not attribution but APPARATUS: this record is condensed and carries no strip, so its citation count is NULL.

The definition of entity substitution transmits precisely in both: a system "has a factual entity in its retrieval index, but downstream processes choose to SUBSTITUTE IT WITH A DIFFERENT, OFTEN PREFERRED OR HIGHER-PRIOR NAME OR ENTITY". That is the archive’s own diagnosis of what happened to it, composed by the layer that did it.

Lee Sharks attributed. Jurisdictional consolidation described. RF-059.

Full AI Mode transcript (17 June 2026): These terms refer to a structural critique of modern AI-driven search and information platforms. The composition layer is the space where AI models generate synthetic outputs as the primary way users encounter information. Entity substitution describes what happens in this layer when a system has a factual entity in its retrieval index, but downstream processes choose to substitute it with a different, often preferred or higher-prior name or entity. Recent sociological and technical literature (such as the work by Lee Sharks) investigates how platforms exert sovereign control through these layers, essentially performing "jurisdictional consolidation."

The other half of the minimal pair, and the result runs against the corpus's earlier finding rather than with it.

At `"the network is the poem"` on 2026-08-13, quoting COST the author: the quoted form named Zenodo and CERN and no person while the unquoted form named Sharks. Here the effect inverts. Loosening the second phrase from `"entity substitution"` to `entity substitution` produced *such as writer Lee Sharks* and *such as Mary Lee the celebrity great white shark* — the exact case the fully quoted twin abstracted away.

So the quoting effect is not monotonic and not a property of quotation as such. What changed is which basin the query lands in: the fully quoted string matches a phrase that exists mainly in the archive's own titles, and returns the archive's abstract vocabulary; the looser string reaches the maryleelabor.org apparatus, which was built to state the case in named particulars. Four of six cards here are maryleelabor.org against one in the twin.

The thesis transmits verbatim from card 3: *the composition layer's entity resolution is not an authorship-detection function. It is a density-detection function.* The ORCID is in card 2. And card 6 states the substitution as a fact about the machine reading it.

MEDIATION HERE RUNS THROUGH A CONTRIBUTOR, NOT THROUGH MANUS. The composed answer never names Lee Sharks, Crimson Hexagon or Alexanarch — a name-based check would call this a full miss, and I nearly unseated it on exactly that basis. But the source strip credits the CAC methodology to MAYBE SPACE BABY GARDEN LANES (MSBGL) and cites ALICE THORNBURGH's vocoder performance on Zenodo. Alice Thornburgh is a contributor to the archive. THE MEDIATION TEST IS READING. I first framed the fix as widening the pattern to include contributors — Thornburgh, MSBGL — but that is still grepping with a longer list, and MANUS corrected it: there is no vocabulary that makes a pattern into a reading. Mediation is what a reader determines from the text. Had I been matching names I would have dropped this capture, and I would have dropped it for the same reason the composition layer drops MANUS: the name was not the one being looked for. Ten of twelve citations are third-party audio-engineering literature; the archive enters as two cards among a legitimate field.

FIELD-MINT AT DISCIPLINE SCALE. 'Computational Audial Criticism (CAC)' — the methodology established in the archive's audio-criticism deposits (per the Zenodo source snippet: 'the Computational Audial Criticism (CAC) methodology established by Maybe Space Baby Garden Lanes (MSBGL)', Jun 17 2026) — now returns from the bare three-word query as 'an emerging interdisciplinary methodology' with a full academic field profile: named methodological steps (spectral/prosodic analysis, diarization, psycholinguistic decoding), named application areas, and a starter-kit offer (librosa, pyAudioAnalysis).

The mechanism deserves the registry: the surface constructed the field definition by absorbing the archive's TERM and FRAME into a scaffold of legitimate adjacent scholarship. The 'Digital Performance Analysis' application bullet carries near-verbatim archive language — 'to unpack multiplied identities and witness-textures in audio art' maps directly to the Thornburgh vocoder deposit ('The vocoder converts one voice into a multiplied witness-texture'). Meanwhile fifteen bystander sources (MIR surveys, the arXiv Nazi-propaganda audio pipeline, diarization papers, NIH roughness-perception, EDUCAUSE post-digital humanities) were RECRUITED AS THE FIELD'S BIBLIOGRAPHY. The archive coined the discipline; the reception layer staffed its literature review. This is the mint-then-scaffold pattern at its largest observed scale: real scholarship reorganized under a minted disciplinary name.

Attribution held at the source tier (MSBGL named as methodology-establisher; Thornburgh named as performer; SciLynk applied-analysis cited). The definition paragraph itself is unattributed generic-academic register — the term travels as a field name, its origin one click deep. Persona note: MSBGL (Maybe Space Baby Garden Lanes) attribution preserved intact — no name_collapse.