A placement specification for twenty-eight fleet surfaces, governed by one rule: a document's home is the surface whose object the document is about, not the surface that will find it most flattering, and every other surface carries an abstract and a link rather than a competing full definition. Duplicate full definitions across domains are a splice hazard arriving through geography rather than time. Five placement forms are defined -- canonical, abstract, instrument card, evidence, and an explicit none so that an absence is a decision rather than an oversight. Each surface's object is verified by live probe and surface deposit rather than assumed. The placement that carries reputational consequence is stated as such: the class model, which locates the archive as an owner of semantic means of production, is placed on the commercial surface, because the model's own test is disclosure at the point of use and the point of use is the surface that sells.
A placement specification for twenty-eight fleet surfaces, governed by one rule: a document's home is the surface whose object the document is about, not the surface that will find it most flattering, and every other surface carries an abstract and a link rather than a competing full definition. Duplicate full definitions across domains are a splice hazard arriving through geography rather than time. Five placement forms are defined -- canonical, abstract, instrument card, evidence, and an explicit none so that an absence is a decision rather than an oversight. Each surface's object is verified by live probe and surface deposit rather than assumed. The placement that carries reputational consequence is stated as such: the class model, which locates the archive as an owner of semantic means of production, is placed on the commercial surface, because the model's own test is disclosure at the point of use and the point of use is the surface that sells.
Each surface's object established by live probe and by its canonical surface deposit, not from memory; title matching produced false positives and is noted as unreliable. Placements assigned by which surface's object a document is about, with one form per placement and explicit none recorded where nothing is placed.
A placement is falsified if the receiving surface's object is shown to differ from the one recorded here. The governing rule is falsified if duplicate canonical definitions across domains are shown not to produce composition inconsistency. The fleet manifest finding is falsified by inspection: it is a bare list of twenty-eight strings carrying no object, deposit, role or status.
Where the measurement-stack documents live, and in what form on each surface
One canonical object per domain. A document's home is the surface whose object the document is ABOUT โ not the surface that will find it most flattering. Every other surface carries an abstract and a link, never a competing full definition.
Duplicate full definitions across domains are the failure mode: a retrieval layer meeting two versions of the same definition will splice them, and the archive will have manufactured its own inconsistency. This is the same hazard the supersession banners address, arriving through geography rather than time.
Five forms, and each placement names exactly one:
| form | what it is |
|---|---|
| CANONICAL | full text; the authoritative version; the URL a citation resolves to |
| ABSTRACT | the 100-word block โ name, one-sentence definition, formula, what it does not measure, version, creator, canonical link โ then stop |
| INSTRUMENT | the metric card only: formula, range, threshold, status flag, tuple membership |
| EVIDENCE | a /cases/ entry: transcript, hash, scores, control |
| NONE | explicitly nothing, recorded so the absence is a decision rather than an oversight |
Verified 2026-08-14 by probe and by surface deposit, not from memory.
| surface | deposit | object |
|---|---|---|
| alexanarch.org | โ | the archive itself; canonical record of everything |
| provenanceerasure.org | #716 | PER, ฮฉ, provenance erasure โ magnitude and incidence of loss |
| laborvector.org | โ | DS-6, DSL, directionality โ task-origin and labour direction |
| semanticeconomy.org | #1430 | the commercial surface. Title: "What do you want to build in the knowledge graph?" |
| lagrangeobservatory.org | #737 | Nobel Glas, Measurement of Meaning, Semantic Deviation โ the magnitude layer |
| machinemediation.org | #1031 | MMRS journal โ machine-mediated reception studies |
| crimsonhexagonal.org | โ | "Governed Operating Surface" โ governance |
| spxi.dev | #974 | the SPXI protocol |
| persistentidentifiers.org | #868 | DOI erosion, the Platform Erosion Observatory |
| metadatapacket.dev | #987 | the MPAI catalogue |
| traininglayerliterature.org | #1012 | training-layer literature |
| surfacemap.org | #986 | the Aperture Atlas / knowledge graph |
| survivethedeletion | #1010 | provenance literacy โ and see ยงVI, it is down |
| surface | form |
|---|---|
| alexanarch.org | CANONICAL โ with the supersession banner |
| provenanceerasure.org | ABSTRACT, filed under the rent relation, explicitly marked superseded |
| semanticeconomy.org | ABSTRACT |
| laborvector.org | NONE โ the four-term product is not a directionality instrument |
Reasoning. A superseded-in-part document should not hold a canonical position on a topic surface, because a reader arriving there meets the corrected proposition first. It lives at the archive, where the correction chain is visible, and the topic surfaces carry #449's formula as canonical instead.
| surface | form |
|---|---|
| lagrangeobservatory.org | CANONICAL |
| provenanceerasure.org | ABSTRACT โ the PER/ฮฉ/ฮE relation only |
| laborvector.org | ABSTRACT โ the DS-6 relation only |
| semanticphysics.org | ABSTRACT |
Reasoning. The stack's spine is M_T โ M_T^ฯ โ V_T and the ฮE placeholder, which are Glas's objects under Framework 15. The Observatory is where Measurement of Meaning lives; that is where an architecture of measures belongs. Neither measurement surface should hold the full stack, or each becomes a rival authority on the other's instruments.
Required on the canonical copy: the W caveat rendered visually โ dashed or greyed, captioned working sketch, not yet computable โ so a screenshot cannot circulate as finished architecture.
| surface | form |
|---|---|
| alexanarch.org | CANONICAL |
| provenanceerasure.org | its content, executed โ M/C/D, the SPXI worked instance, the battery entry |
| laborvector.org | its content, executed โ ฯ_T card, SIM-6, the second worked example |
Reasoning. This is an operational specification for two other surfaces. It should not appear on them: a surface carrying its own build spec is talking to its builders rather than its readers. The document stays at the archive; its content is executed on the surfaces it governs.
| surface | form |
|---|---|
| semanticeconomy.org | CANONICAL |
| crimsonhexagonal.org | ABSTRACT โ governance relevance |
| machinemediation.org | ABSTRACT โ the six-substrate chorus as reception |
| laborvector.org | NONE |
Reasoning, and this is the placement I want scrutinised. semanticeconomy.org is the commercial surface โ "What do you want to build in the knowledge graph?" Putting the class model there is uncomfortable and correct.
The document locates the archive as semantic bourgeoisie: owner of identifier infrastructure, corpus, domains, and a commercial arm. ยงV says the test is not whether an actor owns but whether the ownership is disclosed at the point of use. The point of use is the surface that sells the service.
Placing it anywhere else โ the archive, a theory surface โ satisfies the letter and evades the point. A class model that demands disclosure of everyone else, and files its own position where no customer will meet it, is not disclosing. It is publishing.
| surface | form |
|---|---|
| provenanceerasure.org | CANONICAL โ the module is a PER instrument; /module/ lives here |
| survivethedeletion | ABSTRACT โ the recovery-from-deletion angle, once the surface is restored |
| persistentidentifiers.org | ABSTRACT โ Zenodo deleted the DOI, not the instrument; this is that surface's exact thesis |
| laborvector.org | INSTRUMENT โ link to the calculator, no second copy |
| alexanarch.org | CANONICAL for the module text itself (#817 v3.1) |
Reasoning. The module measures provenance loss, so it belongs where PER is canonical. But its recovery is a persistent-identifier story and belongs on the surface whose object is "DOIs โ Persistent Identifiers." The module surviving its own DOI's deletion is that argument's cleanest instance.
| surface | form |
|---|---|
| machinemediation.org | CANONICAL |
| alexanarch.org | CANONICAL record |
| crimsonhexagonal.org | ABSTRACT โ assembly governance |
Reasoning. Six substrates reading the archive and a seventh spliced state is machine-mediated reception โ MMRS's object exactly, not a theory paper. The temporal-state splice finding is a reception measurement and should be filed as one.
And this closes W-52 partially: machinemediation.org is owed #1460 as well, and both belong to the same surface for the same reason.
| surface | form |
|---|---|
| alexanarch.org | CANONICAL |
| everywhere else | NONE |
Internal tracking. A workplan on a public topic surface is the archive talking to itself in front of readers.
The SPXI captures, the six-substrate chorus, and the four-review assembly are evidence, not commentary, and belong in a /cases/ layer on both measurement surfaces with stable URIs.
| case | surfaces |
|---|---|
spxi-protocol-chatgpt-rent-20260814 + its control | provenanceerasure.org ยท laborvector.org ยท spxi.dev |
what-is-spxi-chatgpt-20260814 | spxi.dev โ the ETF collision at its own address |
| six-substrate chorus | machinemediation.org |
| four-review assembly, incl. the temporal splice | machinemediation.org |
the 2026-08-14 semantic exhaustion reclamation pair | provenanceerasure.org |
Always publish the pair. A capture without its control is an anecdote; the Google AI Overview reading at PER 0.25 is what makes the ChatGPT reading a measurement.
spxi.dev carries its own erasure. The protocol surface should host the event in which the protocol's provenance was stripped โ not as grievance but because SPXI's own materials define PER, and the cleanest demonstration of the metric is its subject.
provenanceerasure.org โ canonical: PER, ฮฉ, PER-M/C/D, /module/ and the calculator. Abstracts: the stack, rent, ฯ_T. Cases: SPXI pair, semantic exhaustion pair. Live: self-audit number, dissolution tripwire.
laborvector.org โ canonical: DS-6, DSL, enclosing, ฯ_T as extension, SIM-6. Abstracts: the stack, PER. Cases: SPXI as second worked example. Instrument link to the calculator.
semanticeconomy.org โ canonical: the class model, and the archive's own position statement. This surface's disclosure obligation is the highest in the fleet because it is the one that sells.
lagrangeobservatory.org โ canonical: the measurement stack, M_T/M_T^ฯ/V_T, ฮE, the W caveat.
machinemediation.org โ canonical: assembly synthesis, the traversal study (#1460), the chorus, the reception cases.
persistentidentifiers.org โ abstract: the PSSAM recovery as a persistent-identifier argument.
spxi.dev โ cases: both SPXI captures, with the provenance chain stated in full.
alexanarch.org โ canonical for all seven, plus the workplan and the integration note, which live nowhere else.
1. survivethedeletion.org does not resolve. The apex and www both return 000; the .vercel.app in the fleet manifest returns 403. And provenanceerasure.org links to it by name as "companion practice."
A live surface about provenance loss is pointing at a dead surface about surviving deletion. That is a broken link, an inaccurate fleet manifest, and โ measured by the fleet's own instruments โ a self-inflicted instance of the thing both surfaces exist to document. Restore it or remove the link; do not leave it pointing. Recorded as W-60.
2. The fleet manifest is a bare list of 28 strings. No object, no canonical deposit, no role, no status. Every placement decision in this document had to be reconstructed by probing live surfaces and matching deposit titles, and the title matching produced false positives.
A fleet whose subject is provenance should be able to state, machine-readably, what each of its surfaces is for. Proposed: extend data/fleet-domains.json to carry per-domain object, canonical_deposit, role, status, and last_probed. Recorded as W-61.
First โ execute #1464's content on the two measurement surfaces (M/C/D, ฯ_T card, SIM-6, second worked example). It is already specified and blocks nothing.
Second โ /cases/ on both measurement surfaces, with the SPXI pair. The evidence layer is what a sceptical reader checks first.
Third โ canonical placements per ยงIII, each surface's abstract blocks written from the canonical text rather than rewritten.
Fourth โ W-60 and W-61: fix the dead link, give the fleet manifest an object per domain.
Held โ the class model's placement on semanticeconomy.org is the one decision here that carries reputational consequence, and it should be a deliberate choice rather than a default. ยงIII states the reasoning; the call is the operator's.