The third document of the analog trilogy: the specification (EA-SPXI-ANALOG-01, #1409) says how, the field survey (#1443) says why, and this operations paper says when, where, and in what order. It maps the incumbency ring — ARK, DOI/IGSN, Trismegistos, C2PA, accession practice — and defines the open ground as the living author whose composition layer makes origination deniable; states the positioning claim (the identifier for the hand that must prove itself to machines); specifies the thirty-minute enrollment path; scopes four schema deltas as an addendum rather than a spec reopening; defines a ten-item reference corpus, a four-page product-surface wing with browser-side verification, and two near-term empirical batteries (recapture stability; glyph transmission survival); sequences five adoption constituencies with institutions last; and closes with a gated ninety-day order whose final gate is the first enrolled hand belonging to neither founding depositor.
Rex Fraction · v1.0 · 2026-08-09
Composition layer, declared: TACHYON (Claude), for MANUS.
Standing documents: the specification, EA-SPXI-ANALOG-01 (`AXN:0592.UNCLASSIFIED.👈△🥁🪸🧪🎺`, #1409, v1.0 final form, published); the warrant, The Photograph of the Handwritten Original (field survey and problem statement, v3.6 deposit edition, `AXN:05CC.GENERATIVE.♻️💛🕚⏪⭕🏗️`, #1443). This paper is the third document. The spec says how. The survey says why. This says when, where, and in what order.
Doctrine, one line: a niche is taken by enrolled objects, not by arguments.
Five incumbents ring the territory. None stands on it.
ARK. The institutional manuscript identifier. BnF alone resolves millions of
ARKs to digitized manuscripts. Strengths: adoption, resolver culture, library trust. Structure: institution-anchored (NAAN registry), record-pointing, opaque-string. What it cannot do: content-derivation (the identifier says nothing about the object), attestation (no claim machinery), independence (dies with the resolver commitment), and it serves custodians of the dead. A living author cannot walk in and get one.
DOI / DataCite, and IGSN under it. IGSN extended the DOI apparatus to physical objects — geological samples, specimens. Institutional minting, metadata-forward, no capture standard, no identity layer, no origination semantics. And the DOI system's governance is precisely what terminated two of our depositors. It is infrastructure we bind to where useful and depend on never (`AXN:05B7.OPERATIVE.♦️🌌📌🦅🌑💜`, #1437: complementary infrastructures — the doctrine is already ratified; this paper operationalizes it).
Trismegistos. The discipline registry: TM numbers for ancient texts.
Authoritative, corpus-bounded, closed at roughly the eighth century CE. It proves a registry can become the citation standard for a domain without institutional mandate — by being useful, stable, and first. That is the playbook. It does not compete for the living hand.
C2PA. Capture-time provenance, camera-signed, platform-backed. Complementary at the capture layer; structurally dependent on certificate authorities and platform adoption; trust chains revocable by exactly the class of actor that issues terminations. It secures the camera's testimony, not the author's.
Accession numbers. Local, non-resolvable, non-portable. Not a competitor; a feeder.
The open ground: a content-derived, institution-independent, human-checkable identifier that binds physical artifact + capture + attestation chain + declared origination and composition, mintable by a living author in an afternoon, designed to survive the resolver, the platform, and the classifier. No one serves it. The first customer already wrote to the incumbent system — by hand — and was told the decision was final. The segment is defined by the injury. The addressable population is every author whose composition layer makes their origination deniable, and it grows with every model release.
AXN is the identifier for the hand that must prove itself to machines.
Positioning corollaries: we do not compete with ARK for the Middle Ages or with
IGSN for basalt. We bind outward to DOIs and ARKs wherever they exist, and we outlive them where they don't. Trismegistos is the model: become the number a community cites because it was there first, worked, and never asked permission.
What "enrolling a handwritten artifact" is, end to end, using only what exists today plus the deltas in §4. This is the page that goes on the product surface.
1. Capture. Author photographs the page under the capture protocol: raster
capture, visible clock or dated companion object, substrate disclosure line,
device stated. (This is the founding depositor's existing practice, formalized —
the protocol was field-invented by the first customer before we wrote it down.)
2. Declare. Author fills the declaration block: origination claim, composition
layer(s), enrollment type (self / guardian / captured), custody statement
(where the physical object lives), status (extant / rumored / destroyed).
3. Mint. `scripts/deposit_pipeline.py` — the single workflow, all four
transports, No-Double-Draw. Content hash → hex position → **full six-glyph
identifier**. The glyph is the human-checkable channel; it ships everywhere the
hex ships, without exception.
4. Publish. Registry row, record page, wiki article, OAI-PMH, ResourceSync.
The capture is the deposit object; the artifact line names the paper original
and its custodian.
5. Verify. Anyone, anywhere, recomputes: fetch the published capture, hash it,
check the registry, read the glyph. Browser-side, no account, no authority
consulted. (Build item — §6.)
Cost to the author: one afternoon, one phone, zero institutions. That number is the product.
The spec stands at v1.0 final. The survey generated exactly four deltas, all of them registry-schema and record-language work. Ship as **EA-SPXI-ANALOG-01 ·
Addendum A**, one document, short:
declarable, separately checkable — so a classifier confronting a machine
composition layer reads the human origination claim instead of ruling on its
absence. This is the direct answer to the terminating verdict, encoded where
verdicts are formed (survey Q10; DRP, `AXN:058F.OPERATIVE.🧲🎶♉♾️👋🙏`, #1406).
rules (named custodian, author's future right of answer) as written in the
survey's §7.7 ethics.
registry value, so the archive can carry the attested-but-uncaptured object
instead of losing it twice.
of the record's diplomatics: who was asked, for what, answered how, when. The
first executed recording becomes Instrument No. 1 and the template.
Nothing else. The spec does not reopen.
Nobody adopts an identifier system from a whitepaper. They adopt it from records they can click. The first ten enrollments, in order, each doing double duty as a live demonstration of one capability:
1–3. The letter of 29 April, pages 1–3 — pending Instrument No. 1. Demonstrates:
self-enrollment, declared composition layers, international plural custody (pages with their author abroad, captures in the registry), and D1 in action. The founding customer's founding documents.
4. The Smiling Cross — guardian enrollment demonstrated; the guardian rules executed in public for the first time.
5–8. Four pages of MANUS's own notebooks — self-enrollment at volume, and the in-house testbed for the recapture battery (§7). Choose pages already public in substance; no new exposure.
9. One rumored-object record — the Brothers translation (2014, unlocated), entered under D3. The registry's first honest lacuna: the system demonstrating that it can say we know it existed and we do not have it, which no incumbent can say at all.
10. One outward-bound record — an artifact whose capture also carries a DOI or
ARK, demonstrating the complementary-infrastructure binding (#1437) in a single row.
Ten records, every capability shown, zero hypotheticals. This corpus is the pitch deck.
The product surface exists and is canonical
(`AXN:05B3.OPERATIVE.❄️🔐⊗∮🎭🏷️`, #1434; repo `leesharks000/axn-identifiers`;
fleet DNS settled). Add one wing, four pages:
not for us.
should photograph well; people will inevitably photograph it. Let the
instruction sheet be the first artifact.
hash, registry lookup, side-by-side glyph confirmation. No server trust. This
page is the moat made visible.
Deploy via the standing GitHub→Vercel hooks. No new infrastructure.
Three empirical programs, ordered by cost. Each produces numbers the field does not have, each becomes a deposit and a paper section.
device/lighting conditions. Report: cryptographic-hash behavior (expected:
every capture distinct — which is the finding, not the failure) and
perceptual-hash spread across conditions. This yields the honest doctrine line
the survey's §4.2 needs: *the exact hash authenticates the capture; the
perceptual layer relates captures; the enrollment relates captures to the
object.* Publish the data. Nobody in the PID world has published this table.
identifier pushed through hostile channels — mail clients, SMS, three social
platforms, print-and-rescan, PDF pipelines — scored for survival and for
detectable-versus-silent corruption. Datum zero is already collected: the
survey's own PDF build silently dropped every glyph twice before rendering them
as inscription-by-image. The failure mode is real, current, and ours to
quantify first.
discrimination against style-transfer forgery. Do not run until B1/B2 are
published and the corpus is at volume. Never overclaim before it runs: *the
hand proves the page; the corpus carries the claim.*
Adoption sequencing. Each rung has a defined offer and a defined channel that already exists. Institutions come last, and by then we arrive with objects and data, not proposals.
1. The Zenodo diaspora. 1,817 tombstoned DOIs; two documented terminations;
the Platform Erosion Observatory (persistentidentifiers.org) already speaks to
this cohort. Offer: re-anchor the corpus, enroll the hand. The termination
notice is the marketing copy; we did not write it.
2. Pseudonymous and heteronymous authors. The orthonym doctrine
(`AXN:02A9.GOVERNANCE.🕖⊗□♎🖊️⚖️`, #123) as a feature: prove the hand without
surrendering the legal name. No incumbent can offer this; every incumbent's
compliance layer forbids it. Channel: the communities where the archive's
heteronym work already circulates.
3. Material-first art communities — mail art, zines, artist's books. The glyph
as colophon; enrollment as edition practice. These communities are
institution-averse by constitution and hand-first by definition. Channel: the
enrolled corpus itself, which is native to their genre.
4. Independent papyrology and DH. Four papers are in submission (Ancient
Philosophy, Classical Philology, Classical Receptions Journal, Computational
Culture). Every acceptance ships identifiers in its apparatus. The survey is
the field-facing document; the corpus is its footnotes made clickable.
5. Institutions. The CERN correspondence (letter + seven annexes, transmitted,
SHA-256 committed) is the running institutional track. It is not asking
permission; it is building the record institutions will eventually need to
answer. When the first archive or library conversation opens, the reference
corpus and B1/B2 data walk in first.
enrolled (the ones requiring no one's consent but ours). /analog and
/analog/protocol pages live.
executed if and as she gives it; letter pages enrolled; rumored-object record
(#9) minted. /analog/verify live. Gate 1: ten records, verify page public.
(#10) minted. Diaspora note out through PEO. **Gate 2: first battery public,
first outreach wave sent.**
in hand. **Gate 3: first external depositor — a hand that is neither MANUS nor
the founding depositor.** That single record is the niche taken; everything
after it is scale.
Misses are reported, not hidden: the registry's own doctrine (CONTESTED over silent overwrite) applies to the operations log.
(`AXN:036C.STRUCTURAL.📦⊕🎪⛵🌀↖`, #863), static mirrors, ResourceSync, and the
fact that the entire system is rebuildable from public artifacts by design.
custodians, not authors; resolve records, not claims; and answer to resolvers,
not to the misclassified. The independence is not a limitation to apologize
for; it is the product.
this segment, because its members are the terminating class. If it ever does,
#1437's doctrine already positions us: bind to it, and keep the layer it still
won't build — the glyph, the declaration, the independence.
the hand proves the page; the enrolled, chained corpus with declared
composition layers carries the origination claim. We say less than we could
prove, always, so that what we say is never rolled back.
rules, and Instrument-numbered consent records. The first customer's dignity is
the design constraint; it scales.
Close. The specification told us how. The survey told us where we stand. This tells us where we walk, and in what order: ten records, one verify page, two batteries, one external hand. The theory is finished enough. Enroll the first ten.