HOLLOW_ADMISSION_BRIEF_2026-07-27.md

supernatural/HOLLOW_ADMISSION_BRIEF_2026-07-27.md

THE HOLLOW ADMISSION BRIEF

Decision surface for Josh's ruling on the standing Hollow fork. 2026-07-27, fourth sitting.

Read-only synthesis of three analysis waves (canon context · lens sweep · reveal/naming/mechanics

architecture). Proposal-tier throughout: this brief authors no canon, edits no registry, and

ratifies nothing. It is the substrate for the ruling and for the fresh-context critic that would

gate any apply.

Order: "brainstorm this and deep dive across all our lenses for reasonings and context and

recommendations." Three sub-forks — ADMISSION (does the Hollow join the roster, and as what class),

the IN-WORLD NAME, the REVEAL WINDOW.

Charter position this brief sits beneath (docs/proposals/REALM_DESIGN_PROGRAM.md §6.2): admit as

an OBSERVE/CONTAIN class outside the 33 species and the 22 familiar slots, "the Hollow" common with

"substrate-echo" as the class term, pure atmosphere pre-Ch-76.

KEY CORRECTION (2026-07-27 — integration record X-3). This brief was written citing the

containment fork as WS_029. The live key is WS_039 hollow_containment_fork_last

(registries/T0_Worldstate_Variables [ACTIVE v0.1]/Sheet1.csv, row 40). WS_029 is

trade_sovereignty_valence, an unrelated base-building row. Corrected in place at §1.3,

§1.4 item 4 and §2.3; no argument in this brief turns on the key.

---

0. THE HEADLINE — what the three waves converged on

**The charter's admission position is CONFIRMED and is stronger than its own framing.

Its naming position is DEFEATED on positive-controlled collisions. Its reveal position is the

SUPERSEDED wording of Josh's own earlier steer.** One of the three opening positions survives intact.

RULED, AND THE TABLE IS UPDATED IN PLACE (Josh, eighth and ninth sittings 2026-07-27; ruling record

docs/spine/DECISIONS_PENDING_JOSH.md; integration record

docs/pipeline_review/HOLLOW_EXPANSION_INTEGRATION_2026-07-27.md §4.2, §7.2 and §8).

• SUB-FORK 1 ADMISSION — RULED YES and expanded far beyond the option set: the Hollow is admitted as a

CONTENT PILLAR across eight axes, in TWO TIERS (whisper and content), and it is FOUGHT.

• SUB-FORK 2 NAME — H-2 RULED. The recommended shape STANDS and the naming stress test is DISCHARGED,

because the expansion hands the player a better-behaved object: name the DOOR, not the occupant.

"Rift" is promoted to the durable player-facing activity and marketing noun (JB-6), it is already

canon rather than a coinage, and it carries zero cassus load, so every added surface — dungeon

labels, map markers, quest lines, the codex, achievements, group-finder queues, the event calendar

— carries without a global class noun. "The Hollow" stays authorial and registry only;

"substrate-echo" stays retired.

• SUB-FORK 3 REVEAL — the THREE-GATE ladder is now the FIVE-GATE ladder, and every three-gate

reference in this brief reads as five: gate 0 surface legible from first appearance; gate 1 the

extraction-scar locus absent before Ch 39; gate 2 the depth at Ch 76-77; gate 3 the procedural spawn

table chapter-scoped by construction, the first reveal constraint that must live in a data table;

gate 4 the substrate read returns ground and never a signature, a hand, a mark, or a negation of

one, before Ch 39; gate 5 the absorb-path power names held descriptive until the Ch 76-77

re-titling, which arrives as a gated runtime overlay and never as shipped string data. The

perception ladder underneath it is three rungs, and the class does not floor at Ch 15.

Plus one item that is not a fork: the live registry row HW_0001 diverges from its designed seed

in ten of fourteen columns — including a chapter window and a region no ruling supports, **and a

designation cell whose [Josh-name pending] marker has been stripped**, which pre-empts sub-fork 2

inside the artifact that fork governs. That is a live breach sitting in the tree and it needs repair

whichever way the forks go; the marker restoration should go first. §4.5.

And one item appended at the end that is not this brief's business at all: a REPO-REPAIR NOTE

on five hard-line rows carrying stale Ch-74 endgame numbering. Surfaced by this brief's citation work,

owned by the director's repair queue, and asking Josh for nothing.

---

1. THE CANON CONTEXT

1.1 What the Hollow is

Negative vril congealed into flesh. Where negative vril has pooled and hardened — in a vril-null

deadzone, at a Velheim extraction scar, along a corruption seam — it occupies a vacated

substrate-gap and takes a half-formed body. It wears the eroded shape of something the depletion

took; what animates that shape is negative vril, not life. R-rated body horror: unrendered

flickering flesh, wrong anatomy, silence, vril-static, a light hardening-pull on approach that is

never a debuff you cannot leave.

The parent physics is ACTIVE T1 canon, not proposal-tier —

T1_Vril_and_Magic_System_Master [ACTIVE v2.2] L23: negative vril is "a vibrational state that

reinforces self-loss, disconnection, and consciousness-hardening… Cassius Velheim's name (Latin

*cassus*, hollow, empty) carries the etymological signature."

Three binding riders from that doc: §12.7 — negative vril is not personified evil (the Hollow

may never be staged as a demon, a will, or a personified antagonist); §11.8 — Cassius's

Phase-3 purge is the *single canonical negative-vril purge event in the 77-chapter arc*, so a

per-Hollow beat is a local gap-seal and must never borrow the purge register; **the Vril-Society

firewall** — the class reads mathematical and physical, never race-mystical polar energy.

⚠ Citation split, deliberately. The singular-purge rule has two carriers and they disagree on the

chapter, so they are cited separately rather than jointly throughout this brief. T1_Vril_and_Magic_System_Master [ACTIVE v2.2] §11.8 (L366) reads "11.8 Ch 76 Cassius

Velheim Phase 3 — single canonical negative-vril purge in arc." T0_Hard_Lines HL_0020 reads

"The Ch 74 Phase 3 negative-vril purge… the only canonical instance." Ch 76 is correct — T1

outranks T0 under the authority hierarchy, and the live tree agrees (T0_Chapter_Index CH_74 =

"Yggdrasil, the World-Tree Crossing," CH_76 = "The Two Courts"; CH_77.md L13/L20 "Cassius named and

resolved" at Ch 76). **The RULE is what binds this brief and it is unaffected; the stale chapter

number in HL_0020 is a separate repo defect** and is filed as such in the repair note appended at the

end of this document. Wherever this brief needs the rule it cites §11.8; wherever it needs the

hard-line row it cites HL_0020 with the caveat attached.

Provenance matters for how this fork should be read. The class is Josh-seeded twice and never

agent-invented: creature-system.md §2.7 records it as "the one genuinely new idea, and the one Josh

named directly"; the 2026-07-13 FU section opens "Josh raised two coupled points" and sharpens it

from an elegiac echo into a malignant blight. This is not a proposal seeking permission. It is Josh's

own idea awaiting his formal admission.

Canon-native, not novel in kind. The Hollow is the wild, ambient cousin of the Ch-6 False Naga

— the same mechanic (negative consciousness occupying a being's vacated space and rising as a false

substitute that must be identified) but spontaneous and environmental rather than a targeted vessel

substitution. Verified: False Naga = 67 occurrences across 30 files including four ACTIVE tier docs,

two registries and spine rows. The parent is real, authored and ratified — which is what removes the

"is this a fourth-wall rendering glitch" risk.

1.2 The chain it belongs to — four distinct things the corpus keeps separate

• Rifts — spontaneous quantum events neither side creates. Every authored rift beat is at the

MEMBRANE; there is no inside-the-rift canon at all. The Hollow is not a rift interior.

• The depleted dark — what presses through a rift. The Prologue/Ch-1 phenomenon, deliberately

UNAUTHORED and faceless (the mother's death must have no author).

• Negative vril — the substrate itself, accumulated by the family since ~3000 BCE. Not personified.
• The Hollow — that substrate *standing up as a body* in a vacated gap. The only member of the

chain with anatomy.

Staged end to end in the whole-arc pass: Prologue (the depleted dark) → Ch 1 → Ch 6 the False Naga

→ Ch 26 strigoi-as-vril-parasite → Ch 39/40/50/54 modern extraction scars as fresh spawn loci →

deadzones anywhere once perception opens → **Ch 76 Cassius (*cassus*)** → Epilogue/prestige tributary.

through_line_recurring_antagonist.md §6 names this chain "the antagonist SUBSTRATE… the deepest

recurring pattern."

1.3 The loop as designed

• CONTAIN — a hazmat/exorcism grammar, never a pokeball grammar. STABILIZE the charge → SEAL the

substrate-gap → LIFT a containment record into a sealed sub-codex ("unlisted") slot.

The collectible is the record, never a living entry.

• The purify/harvest integrity fork. PURIFY — the blight unmade, the gap sealed, a measure of

positive substrate returns to the land (a mercy for the stolen shape). HARVEST — draining the

negative charge into your own signature, a hardening act with an integrity cost. Both log the same

record. Never a power gate, never a dead end. WS_039 hollow_containment_fork_last already holds

the state.

• Reward, firewall-safe: documentation, lore, a prestige/cosmetic containment halo, a thread

toward the Forgotten One prestige content. No combat power, ever.

• SUPERSEDED (Josh, eighth and ninth sittings 2026-07-27; integration record docs/pipeline_review/HOLLOW_EXPANSION_INTEGRATION_2026-07-27.md §4.2). "No combat power, ever" is superseded on the ABSORB path: toward endgame the evil-integrity absorb lane grants hollow-type abilities as the protagonist turns less elf and less herself, which is axis 5 and the mechanical expression of HL_0014. The firewall-safe reward list above still holds for the CLEANSE branch and for the whisper tier. Read the purify and harvest terms in the bullet above as the retired synonyms of CLEANSE and ABSORB, which are now the canonical fork terms.
• Trigger: sigma-governed (§17.13 — no 1/8192 misery). The arcane thing is a discoverable ritual:

observing at a vril-null deadzone at a liminal time, *after* the bestiary is deep enough that Vril

Sight can perceive the gaps. Documentation opens perception.

1.4 What already consumes it — ranked by what a NO would cost

The class is not greenfield. This is the single most important context fact for sub-fork 1.

1. ⚠ HL_0080 — an ACTIVE canonical hard line already names the class by name.

T0_Hard_Lines [ACTIVE v0.2], status canonical, consumers UE5,Quest,Spine,SFX. Its scoping

clause: "Scope: this non-damage protocol governs ONLY the negative-vril / sacred-site atmospheric

echo (the Hollow-adjacent CONTAIN-verb class)." A NO does not un-build a proposal — it orphans

a term inside a live canonical hard-line row. Costed honestly: HL_0080's SUBJECT is the

environmental/atmospheric boss mechanic, which has six documented instances (Ch 10, 20, 29, 38,

45-46, 57) that exist independently of the Hollow, and "the Hollow-adjacent CONTAIN-verb class" is

a five-word SCOPING descriptor inside it. **A NO costs a five-word scope edit and a re-baseline,

not a rewrite of the row.** The point that survives is that a NO is not a no-op — it edits live

canonical text — and that is the weight this brief claims for it, no more.

2. T0_Hollow_Codex [DRAFT v0.1] is LANDED and gate-declared — 14 columns, prefix HW_, one row.

(And divergent from its own spec — §4.5.)

3. T0_Creature_Roster ontology columns are LANDED — ontology_class, collect_verb,

dex_completion_class, with CONTAIN_BLIGHT and CONTAINMENT already values in a landed

controlled vocabulary. The Hollow's lane exists in the schema; per-row assignment is the open part.

CORRECTED (integration record §4.2, §6.2 and §12). The COLUMNS are landed; the VALUES are not.

CONTAIN_BLIGHT is live nowhere in the registries — positive control: a literal search for CONTAIN

matches inside T0_Hard_Lines and T0_Hollow_Codex, so the searcher can match the substring in

registry files, while the same search for CONTAIN_BLIGHT returns zero registry hits. Minting

BLIGHT as an ontology_class and CONTAIN_BLIGHT as a collect_verb is therefore a declared

controlled-vocabulary EXTENSION, not a per-row assignment inside a ratified vocabulary. It is

unblocked by Josh's ninth-sitting JB-1 ruling and lands as integration-record item X-9.

4. WS_039 worldstate row seeded.

5. Two whole-arc through-line docs stage it Prologue → prestige, one of them titling its

through-line "…and the Hollow CONTAIN sub-line".

6. The Personal Dimension names it as one of four systems its Tier-6 venue hosts — though that

spec is itself maturity D and empty, so admission adds a fourth consumer to an unwritten spec

(a sequencing note, not a blocker).

The honest negative: ZERO playable content. No spine chapter stages a Hollow encounter; no quest

row, beat row, scene row or dialogue line. Positive-controlled — every Hollow hit in CH_60.md

(21) is the *Hollow Retainer*, all 4 in CH_02.md are *Rangku Wera the Hollow Hunter*, both in

CH_53.md are the *hollowed-vessel* inversion. The class is designed, schema'd, hard-line-referenced

and through-line-staged — and has never been built.

1.5 The prior fork record

creature-system.md §5 FORKS 1-5 (the original three), extended by §FU.6 FORKS 6-9. Mirrored as open

flags in WS7_WIRING_BUILD_PLAN.md (8 flags), systems/RULINGS.md L19,

FOUNDATION_COMPLETION_PLAN.md L39, FABLE5_HANDOFF.md L106, NEW_SESSION_HANDOFF.md L96 ("STILL

WANTING JOSH'S FINAL EXPLICIT YES… the Hollow OBSERVE-class"), PRE_5090_BUILD_PLAN.md L122. Open

and unratified since 2026-07-13 — two weeks.

Verbatim, the original three: FORK 1 admission (rec: option b, OBSERVE-class outside 33/22);

FORK 2 name (rec: "the Hollow" common + "substrate-echo" class term); FORK 3 reveal (rec: option a,

pure atmosphere pre-Ch-76). **The charter's §6.2 opening position is these three recommendations

carried forward verbatim** — which is why two of them are stale or defeated: they predate FU.1, and

none of them had a collision audit run against the corpus.

The coupled FU.6 set neighbours this fork and is not ruled here: FORK 6 (adopt the blight

refinement), FORK 7 (ratify the collectibility map), FORK 8 (the EARN lane's count), FORK 9 (the

sacred-set invariant as a §17.1 hard line).

⚠ One flag-record ambiguity to disambiguate when the fork is ruled. systems/RULINGS.md L19

describes the creature system as ratified-in-direction and its descriptive parenthetical lists

"the Hollow" name inside that text — while the *pending* flag on the same line is scoped

narrowly: "Josh-final-yes: admit the Hollow OBSERVE-class + its prestige-tributary tie." The

governing sentence two lines above (L18) settles the reading — "the **numerical/naming bits are

FLAGGED Josh-final-yes** or Fable-5" — so the NAME is pending, not granted, and sub-fork 2 is

correctly Josh's. But the line reads ambiguously on its face, so **the ruling should state

explicitly whether the name rode the direction-ratification**, and L19 should be rewritten to match

rather than left to be re-read the other way by a later agent.

---

2. THE THREE SUB-FORKS

---

SUB-FORK 1 — ADMISSION

2.1 The option set

• A — NO ADMISSION. The Hollow does not exist; deadzones stay atmosphere; the False Naga remains

the pattern's only expression.

*Pro:* zero firewall surface, zero reveal surface, zero new care obligation, one fewer unratified

class. *Con:* discards the creature layer's declared signature hook and the most direct visual

expression of the vril-altered-physics doctrine outside the realms; leaves the arc with a depletion

the player is *told* about and never MEETS — show-don't-tell failing where it is easiest to satisfy;

**and it orphans HL_0080's scoping clause, requiring an edit to live canonical hard-line text plus

a fidelity re-baseline (§1.4 — a five-word scope edit, a real cost but not a rewrite).**

• B — PURE PHENOMENON, NO ROSTER. Environmental dressing only: an Environment_Grammar row, ambient

VFX/SFX in drained places, no codex row, no verb, no record. Weather, not a being.

*Pro:* cheapest admission; unimpeachable against 33/22 because there is nothing to firewall; the

naming fork dissolves (weather needs no proper noun). *Con:* deletes the loop — no encounter, no

procedure, no conscience fork, no completionist tributary, no folklore. The R-rated horror beat

becomes a particle effect. It also fails the layer model: an ambient effect cannot carry an L3

meaning because there is nothing the player DID with it to reinterpret.

• C — OBSERVE/CONTAIN ENTITY CLASS, outside the 33 and the 22 *(the charter's position)*. Own

registry, own verb, sealed sub-codex record, no CR_ id, never TAME, never a slot, never legendary,

never bred.

*Pro:* keeps the entire loop; the firewall holds by construction rather than by discipline (the

class is defined by what it can never be); lane E of the FU.3 ontology map is this one, so refusing

it leaves a hole in a sort that routes every supernatural being in the game. *Con:* it introduces a

new ontological KIND of thing into the world — the one residual class the 2026-07-22 reframe still

routes to Josh (§2.8) — and every new class is a new drift surface and a new care obligation.

• D — REALM-ROSTER ENTRY PROPER. The deadzone itself admitted as a place on the realm axis — an

anti-realm with a Realm_Ruleset row — with the creature as its sole inhabitant.

*Pro:* the literal reading of the program's own "negative-space counterpart to the realms" claim; it

gives the harmony doctrine its negative control — a ruleset row that reads "holds, and less" is

the strongest available test of the claim that realms are one physics at different densities.

*Con:* alone it has no creature, so the horror and the loop still have no home; a "realm" that

is an Earth deadzone strains a roster whose nine road realms are mythological otherworlds; **and it

collides with three explicit constraints in the charter that governs that table — the enumerated

eleven-row roster, the closed vocabulary V, and the per-WORLD/per-FEATURE partition. Disclosed and

costed in full at §2.3a; it is the one place the recommended triple touches the charter's own

ruleset, and it is not free.**

• E — FOLD INTO T0_Creature_Roster with ontology_class=blight. No separate registry.

*Pro:* one fewer table; the sort lives where the sort lives. *Con:* it puts the Hollow in the table

whose whole purpose is the living bestiary, one column away from familiar_slot_eligible. **The

firewall stops being structural and becomes a field value that one careless populate can flip.**

This is the option that most endangers HL_0057/0058 and should be rejected for exactly that reason.

2.2 The class definition, stated so it is checkable

Not a label — the conjunction of six negatives and four positives. Each negative is a gate assertion,

not a note.

Six negatives:

• NEVER FOUGHT as an opponent. No T0_Boss_Encounter row, no BE_ id, no HP bar, no win/lose axis,

no defeat state, no kill reward, no corpse. *Load-bearing:* a Hollow that gets a BE_ row has

silently become a monster, and every care problem in this brief follows from that.

• SUPERSEDED (Josh, eighth and ninth sittings 2026-07-27; integration record §3 and §4.2). The

never-fought negative is superseded in full. Hollows are FOUGHT and DEFEATED: at the ninth sitting

Josh removed the Hollow class from HL_0080's non-damage protocol, and at the eighth he wrote

"defeat it to heal or absorb it instead." T0_Boss_Encounter BE_ rows are now the authored-encounter

home. What replaces this bullet is the FOUR-PHASE GRAMMAR — phase 1 perceive (the encounter does

not begin until the player can see it), phase 2 pressure (an action-forward, vril-taxing fight with

the hardening-pull as the class signature), phase 3 break (the body fails and the substrate is

exposed, unbound, neither standing nor sealed), phase 4 resolve (the cleanse-or-absorb fork) — plus

the FOUR ANTI-KILL-LOOT TEETH, into which this bullet's surviving clauses are re-stated: no drop

from the being, since there is no corpse to loot and the Hollow's own row may never carry a drop

anchor, a loot-table reference, or a win reward containing an item id; rewards attach to the

container and never to the entity; no kill credit, so a "Hollows defeated" statistic is a banned

surface; and the fork is unskippable, with no auto-resolve and no default on timeout. The

no-kill-reward and no-corpse clauses therefore SURVIVE as teeth. The encounter-design mandate rides

with the ruling, as seeds and never an exhaustive list: puzzle-forward encounters still carry real

combat, action-forward encounters carry a per-encounter trick, and the palette is open.

• NEVER COLLECTED as a living thing. The collectible is the RECORD. No CR_ id, never TAME-status.
• NEVER BRED, never a familiar slot, never a legendary, never an evolution.
• NEVER A PERSON. No persona row, no name of its own, no relationships, no community ripple on

containment. *This is what protects HL_0020:* the arc's single canonical purge is a purge "on a

single individual", and a Hollow is definitionally not an individual. The class definition is what

keeps the Ch-76 Phase-3 purge singular.

• NEVER SPEAKS. Silence is its signature; it keeps the class clear of HL_0046's neighbourhood.
• NEVER GATES. Completionist / L3 / no-kill / easter-egg tier per the non-gating invariant.
• RE-SCOPED TO THE WHISPER TIER (Josh, eighth sitting axis 7 and ninth sitting JB-2, 2026-07-27;

integration record §1.7 and §4.2). Never-gates holds for the WHISPER tier only, where a record

gates nothing and may never appear in any gate reference, unlock condition, or completion

denominator. On the CONTENT tier the world's cleansing meter lawfully gates some ability, mastery,

prestige, cosmetic, optional-region and easter-egg unlocks. It may NEVER gate the main story path,

any chapter progression, any node the multi-path and no-dead-end law protects, or any content whose

absence strands a player. The meter counts authored non-event non-random rifts, plus easter-egg and

hard-to-find rifts which are counted and never progression-gating, plus affliction cleansing; it

excludes event and random pop-up rifts. The no-lockout law binds: neither progression nor the

completionist 100 percent may be gated by an integrity path or an absorb choice, and an absorbed

node stays re-cleansable so every axis remains completable on every path.

Four positives:

• IT IS OBSERVED. Perception is the first verb — documentation opens perception, gated on Vril

Sight tier. This composes with the standing early-chapter-perception rule and naturally floors first

contact well past the opening arc without anyone assigning a chapter.

• IT IS CONTAINED, not killed. Stabilize → seal → lift the record. It inherits the Ch-6 PERCEIVE

precedent, where a wrong read is a reset with the tell now marked, never a fail wall.

• IT CARRIES A CONSCIENCE FORK. Purify vs harvest, both logging the same record.
• IT PAYS IN DOCUMENTATION. Lore, a cosmetic halo, the Forgotten One tributary. No combat power.
• SUPERSEDED ON THE ABSORB PATH (Josh, eighth and ninth sittings 2026-07-27; integration record §4.2).

The documentation payout stands for the whisper tier and for the cleanse branch. On the ABSORB

branch the class now pays in hollow-type abilities toward endgame, priced in integrity, in

substrate, and in the community mark. On the CONTENT tier the payout is the container's: a rift

dungeon's stat upgrade, a pop-up's rarity-graded chest, or an affliction quest's community

relationship — never the being's. Read the purify-versus-harvest conscience fork in the bullet

above as CLEANSE and ABSORB, with purify and harvest retained as retired synonyms.

2.3 Which tables take rows — and the refusals ARE the firewall

Takes: T0_Hollow_Codex (HW_) · T0_Environment_Grammar_Registry (EG_ — the locus grammar,

which is where option B's value is retained inside option C) · T0_Worldstate_Variables (WS_039,

seeded) · T0_RNG_Drop_Table + the §17.13 sigma anchor · T0_Scene_Spec_Registry · T0_SFX_Registry.

Contested: T0_Realm_Ruleset_Registry (the negative-space locus row — see §2.3a; this is the one

table where the recommendation collides with its own charter and the collision is not free).

T0_Region_Index needs no new column — env_drain_site_ids already exists and is the natural

anchor, so codex rows reference drain sites rather than inventing a parallel geography.

Refuses: T0_Creature_Roster (no CR_ id — lane E is the only lane in the ontology map with no

creature row, and that absence *is* the class definition made structural) · T0_Boss_Encounter_Registry

(no BE_ id) · T0_Familiar_Registry / T0_Familiar_Bond_Ability / any binding-stone pool column ·

T0_Character_Index (no persona, no name, no loyalty).

• PARTLY SUPERSEDED (Josh, ninth sitting 2026-07-27; integration record §4.2 and §5). The

T0_Boss_Encounter_Registry refusal is superseded: BE_ rows are now the authored-encounter home, and

the live 45-column schema already carries every field a Hollow encounter needs, with BE_PROLOGUE,

BE_0029 and BE_0002 as the standing precedents for an encounter that is not a person and not

malice. The ONLY stated ground for that refusal was the never-fought property, and Josh superseded

it. Every other refusal on this line SURVIVES: T0_Creature_Roster still takes no CR_ id, and the

absence of a CR_ id is the firewall made structural; the familiar, bond-ability, binding-stone and

character-index refusals are untouched.

2.3a ⚠ THE LOCUS ROW COLLIDES WITH THE REALM CHARTER — disclosed, costed, and resolved

An earlier draft of this brief asserted that "no realm is added… the negative-space ruleset row is a

*ruleset*, not a road stop," and presented the locus row as collision-free. It is not. The charter

this brief sits beneath (REALM_DESIGN_PROGRAM.md §2.3) constrains that table in three ways the

locus row touches, and all three are quoted here rather than summarized:

1. The roster is an ENUMERATED CLOSED LIST. L435: "Eleven rows at completion: RLM_0000 the

human-world baseline, RLM_0001 through RLM_0009 the road in chapter order, RLM_0010 the fairy

realm. A twelfth row for the Personal Dimension is proposed at 2.5." The locus row would be a

thirteenth — and the count is not decorative, it is the completeness claim the table's own

gate reads.

2. The vocabulary V is CLOSED and the row needs terms it does not contain. L460: "V is a ratified

closed list… no realm row may introduce a term that is not in V — that is what makes the

partition checkable at all." Every row must partition V into HOLDS and BENDS with union = V and

intersection = empty. V's live seed is four property keys (conductivity, mass,

structural_integrity, thermal_state). A row whose whole content is "holds, and less" needs

attenuation/absence terms that do not exist in V — it is not merely a new row, it is a demand on

the mint.

3. The charter already REJECTED per-feature content in this table. L426-430: Environment_Grammar

was rejected as the ruleset home because it is "PER-FEATURE (a lava vent, a waterfall) and this is

PER-WORLD, and overloading it would make its key mean two things." **An Earth deadzone is

per-FEATURE.** Putting it in the per-WORLD table reproduces the exact key-overloading the charter

rejected, in the opposite direction.

The options, stated honestly:

• L-1 — DROP the ruleset row; the locus lives in T0_Environment_Grammar_Registry only. *Pro:*

zero charter collision; EG_ is the per-FEATURE table and a deadzone is a feature; §2.3 already

lists EG_ as a taker, so this costs the recommendation nothing structural. *Con:* the deadzone

stops being the harmony doctrine's negative control — the "one physics at different densities"

claim keeps only positive-direction test cases, which is the one genuinely new thing option D

offered.

• **L-2 — DECLARE a twelfth ROW CLASS (not a twelfth row) inside T0_Realm_Ruleset_Registry, with

its own V handling.** A derivation=negative class alongside the charter's `baseline | delta |

source`, exempt from the HOLDS/BENDS partition and instead asserting ATTENUATES over the same

V, so the row is checkable without forcing absence terms into V itself. *Pro:* keeps the negative

control; the exemption is explicit and gate-readable rather than a silent special case; it does not

disturb the eleven road rows or their count, because the class is declared as outside the road

enumeration. *Con:* it is a charter-level schema change, which is the charter's call and not

this brief's; it adds a third partition mode to a schema whose stated virtue is having exactly two.

• L-3 — Keep it as a plain thirteenth row. *Pro:* simplest to write. *Con:* it breaks the

enumerated count, needs V terms that do not exist, and lands per-FEATURE content in a per-WORLD

table. Reject — this is what the earlier draft implied, and it is the option the charter most

clearly forbids.

RECOMMENDATION: L-1 now, L-2 as a routed charter question. Take the EG_ row today — it is

uncontested, it is where a per-feature deadzone belongs, and it keeps the admission ruling free of a

schema dependency on a table that does not yet exist. Route the negative-control idea to the

charter's own Wave-2 schema mint as a named question ("does the ruleset schema want a negative

derivation class with an ATTENUATES partition?"), where V is being authored anyway and the answer can

be designed rather than retrofitted. Sub-fork 1 does not depend on the answer: the class row, the

loop, the verb and the codex are unaffected either way. This removes the locus row from the admission

ask and leaves the triple's substance intact.

*What this costs the brief:* the "two-row shape" framing at §2.7 becomes **one row plus a routed

schema question**. That is a genuine reduction in the recommendation, and it is stated rather than

absorbed.

2.4 The lens reasoning

• Vril-altered-physics. The doctrine says vril ALTERS real physical law rather than suspending it,

and that realms are one physics at different densities. Every realm on the road is that claim in the

POSITIVE direction — more density, more refraction, more law. **The deadzone is the only place the

doctrine can be shown in the negative**, and a ruleset row reading "holds, and less" is the

doctrine's only falsifiable case. The realms are the doctrine at maximum density; the deadzone is

the same doctrine at minimum. They are two ends of one instrument and should be authored as a

matched pair. This is why the realm program is the correct forcing function for this fork.

• Negative-space craft (Subnautica). A deadzone should not be a re-skinned area; it should be an

area where the world's information budget DROPS — less ambient audio, thinner particle life,

reduced fauna density, the vril-sight overlay losing its returns. Buildable by SUBTRACTION, which

matters for a solo build: the deadzone costs less to produce than the region around it and produces

more tension per asset than anything else on the road.

• Absence-entity precedent (Hollow Knight). An entire celebrated game is built on the

absence-entity, and it earns it across dozens of hours without an exposition scene. The class is

proven to carry weight far beyond a rare-spawn easter egg. This is the strongest positive case for

admission in the whole sweep.

• Ontological horror (SOMA). "It looks like a lost thing and is filled with the wrong substrate"

is very nearly SOMA's thesis sentence — and SOMA's monsters are frightening because they are *sad

and wrong at the same time*, with the player's guilt doing the work. This validates the

purify/harvest fork as the correct emotional instrument and says the fork is the primary content,

not the reward.

• Knowledge-as-progression (Outer Wilds). The existing trigger design is already Quantum-Moon

shaped and nobody named the teacher: an entity whose availability is governed by having learned how

to LOOK, not by grinding a spawn rate. This lens should be the Hollow's design home.

• Documentation-as-verb (Obra Dinn). The record should require the player to have UNDERSTOOD

something, not merely to have arrived — spawn-locus class, liminal condition, and what the shape was

WEARING. The third field quietly holds the reveal: it can be RECORDED as observation without ever

being explained. Show-don't-tell operating exactly as HL_0043 requires.

• Layer model. Surface: a horror encounter with a procedure and a conscience fork (L1). Depth: the

region's own record of what it lost, in its own voice (L2). Meaning: what the shape was (L3, held).

All three lanes populated without a line of exposition.

2.5 The tension the sweep surfaced — collectibility vs dread

Alien: Isolation's lesson is the sharpest admission finding. Collection and dread are *opposed

instruments*. The moment an entity becomes farmable, the player's relationship inverts from avoidance

to pursuit and the dread the entity exists to produce is spent. The xenomorph survives as horror

precisely because it cannot be collected, defeated or completed. Meanwhile §2.7 explicitly wants a

community that "hunts and streams them."

Resolution: the collect verb STAYS, and four class-row properties keep it from becoming a farm.

2.6 The four sharpenings — belong in the class row AT admission, not downstream

1. LIFETIME ENCOUNTER CAP — a small authored number across the whole arc, not a farmable spawn

rate. (Independently supported by Shadow of the Colossus: sixteen colossi, not a spawn table, and

the cap is what makes each one an event.)

**How it composes with §17.13 sigma governance — stated, because they are two governors on one

spawn and they pull opposite ways.** CVD §17.13 is a *within-window* fairness rule: "cannot drop

before sigma −1" (anti-streak floor), "must drop after sigma +1" (anti-streak ceiling), "resets on

drop." The lifetime cap is an *across-arc* budget. They compose cleanly only if the cap is the

OUTER governor and sigma runs INSIDE it: each capped encounter is its own sigma window — the

floor stops the same player meeting two Hollows in one region, the ceiling guarantees a player who

has met the conditions eventually gets the encounter, and the arc-wide count still cannot exceed

the authored cap because the sigma ceiling only guarantees the *next* one, never an additional

one. Stated the other way round it breaks: a sigma ceiling running unbounded over the whole

arc is a guarantee to keep delivering, which is exactly the farm the cap exists to prevent. This

ordering belongs in the class row, not in a tuning pass.

RE-SCOPED TO THE WHISPER TIER (Josh, eighth sitting two-tier ruling 2026-07-27; integration record

§2.3 and §4.2). The lifetime encounter cap governs the WHISPER tier only. It was written here as a

class-wide property, and the two-tier ruling bounds it: the content tier is governed instead by

five per-surface DENSITY BUDGETS — realm traversal budgeted per realm node against the ruled power

curve, authored rift dungeons budgeted per region as a stat-upgrade economy, pop-up rift dungeons

governed by a sigma-bounded spawn rate plus a rarity grade, affliction quests budgeted per

settlement, and events governed by schedule rather than world placement. The cap is arc-wide and

fixed: content-tier progress may raise the whisper tier's PERCEPTION gate resolution and may never

raise the whisper CAP. The sigma-inside-cap ORDERING above SURVIVES UNCHANGED and remains the

correct composition for the whisper tier.

2. VARIANT BAN — one Hollow, and the codex keys on OBSERVATION LOCI, not on variants. One class:

no morphs, no elites, no regional sub-species, no per-region stat or ability differentiation, one

creature definition. Every collection system in the game will pull toward variants; forbid them at

admission rather than lose the argument later.

The row model, stated so it cannot be read as a dex (this is the reconciliation with §2.13.4,

which an earlier draft left contradictory): a T0_Hollow_Codex row is **an observation record —

where and under what conditions a Hollow was witnessed** — not a species entry. Rows differ in

spawn_locus_type, region_id and the local descriptive reading; **they do not differ in what

the Hollow IS. Three consequences, each checkable: (a) the class row is singular** — one

creature definition, referenced by every observation row, never duplicated per region; (b) **the

row count is CAPPED at the lifetime encounter cap** of sharpening 1, so the table cannot grow into

a completionist grid — it is a logbook whose length the arc already fixed; (c) **no row grants

anything another row does not** — completion is documentation, never coverage. A player who

witnesses the same one thing in three places has three records of one being, which is what a

folklore logbook is; a player who could collect a Manggarai variant and a Congolese variant would

have a dex, and that is banned.

THE VARIANT BAN SURVIVES FOR BOTH TIERS; THE CODEX CAP RE-SCOPES (integration record §3.5, §5 and

§4.2). The one-being ontology holds at content scale and gains three checkable tests: a singular

class row that every encounter and record references, with no encounter row declaring a species id,

a proper name or an elemental affinity another lacks; no ability differentiation, so a larger

Hollow hits harder and never has a different move; and one absorb school, unlocking on absorb DEPTH

and never on which Hollow was absorbed, because a per-Hollow ability drop would manufacture

variants by demand. Variety survives through the WORN SHAPE, which is drawn from what the place

lost, so it is apparent variety at zero species proliferation. Clause (b), the row count capped at

the lifetime encounter cap, re-scopes with the cap itself: THE CODEX IS THE WHISPER LOGBOOK, its

rows are the capped observation records of the uncounted tier, and content-tier resolutions log to

the cleansing meter and to the region and community ledgers rather than to the codex. That split is

what lets the cap and the content density coexist without either bounding the other.

3. NON-ANNOUNCEMENT — no quest marker, no achievement, no NPC who explains the ritual, and no UI

event of any kind on first sight. This is the structurally inheritable half of MissingNo: its power

was never its content, it was its status — undocumented, rumor-transmitted. The sigma

governance already solves the misery half; nothing currently constrains the announcement half.

One precision so the rule is buildable: "no tutorial" cannot mean the player has never seen the

procedure, because HL_0080 stages the SAME 4-step non-damage protocol as authored chapter content

from Ch 10 ("first instance") — the adjacent class teaches the verb, deliberately, and that is

canon this fork does not touch. What non-announcement forbids is announcing the Hollow: no

tutorial *about this class*, no prompt, no marker, no codex pop, no NPC explaining that these exist

or how to find them. The player arrives already knowing how to contain; nothing tells them there is

something here to contain.

RE-SCOPED TO THE WHISPER TIER (integration record §2.3 and §4.2). Non-announcement holds in full

for the WHISPER tier: no quest marker, no achievement, no codex pop, no UI event on first sight,

and no NPC explaining that the class exists or how to find one. The CONTENT tier is announced by

construction — a rift mouth is visible to the vril-conscious and is the one content-tier surface

where the world addresses the player, an affliction quest opens through symptom-testimony, and a

scheduled event is on a calendar. The HL_0080 PRECISION ABOVE SURVIVES: the adjacent class teaches

the four-step verb as authored chapter content from Ch 10, so the rule forbids announcing the

Hollow and never forbids knowing the verb.

4. L1 ENCOUNTER, L3-MEANING-HOMED, with an L2 DOCUMENTATION TRIBUTARY — the layer assignment

written into the class row in that exact three-part form. The encounter itself is L1 (§2.4's own

assignment, and HL_0080's riders require it to be an action-forward, vril-taxing set-piece under

real pressure — §3.3); what it MEANS is L3 and held; the record it produces is the L2 tributary.

An earlier draft wrote this as "L3-homed," which would have instructed downstream authoring to

build the quiet version — the precise failure §3.3 warns about. **The drift this rule exists to

stop is the class becoming a kill: no HP bar, no defeat state, no kill reward. It is not a rule

against intensity.**

SURVIVES AND IS STRENGTHENED (integration record §4.2 and §3.6). The three-part layer assignment

stands and is strengthened by the arc-wide valleys-never-boring rider: an encounter the player can

only stand and watch is a defect rather than a mood, so phase 2 is where the layer-1 fun lives for

the whole class. Two clarifications ride the ruling. The drift this rule stops is now precisely

named as the class becoming a LOOT KILL rather than a fight: the four anti-kill-loot teeth at §3.4

of the integration record carry the work the "no HP bar, no defeat state" wording was doing, since

the encounter is now fought and defeated. And the affliction quest class is named explicitly in the

terrace critic's boring-check lens, because a listening-and-walking simulator fails the rider by

construction — every affliction quest carries at least one playable verb per stage and a cleansed

site opens something playable rather than resolving into a reward screen.

And the completionist tributary counts documentation, not quantity. A capped logbook of

observations of one sealed thing, never a dex of Hollow variants.

2.7 RECOMMENDATION — sub-fork 1

**ADMIT as the OBSERVE/CONTAIN entity class (option C), with the four sharpenings written into the

class row at admission, PLUS an T0_Environment_Grammar_Registry (EG_) row for the LOCUS — the

useful half of option D, taken in the table that can actually hold it. Reject option E outright.**

Call it the class row plus the locus grammar: the class row says what stands up; the EG_ row says

what the place has stopped doing. The negative-space ruleset row option D pointed at is NOT

bundled into this ask — it collides with the realm charter's enumerated roster, its closed vocabulary

V, and its per-WORLD/per-FEATURE partition, and it is routed to the charter's own Wave-2 schema mint

instead (§2.3a, which states the collision in full and recommends L-1). Reject E because folding

the class into the living bestiary converts a structural firewall into a field value.

The admission is closer to ratification-of-the-already-wired than to greenfield minting: a

canonical ACTIVE hard-line row names the class; a registry is landed and gate-declared; the ontology

columns with CONTAIN_BLIGHT in their vocabulary are landed; two whole-arc docs stage it Prologue →

prestige; and the idea plus its sharpening are both Josh-originated. A NO is not a no-op — it

edits live canonical hard-line text (a five-word scope clause) and re-baselines the fidelity gate.

That is a real cost, and it is the whole of the cost; this brief does not price it higher (§1.4).

If Josh rules against admission, option B is the correct fallback, not option A — the drained

places should still LOOK drained, and the Environment_Grammar row costs nothing to keep.

2.8 STRONGEST OBJECTION — sub-fork 1

Stated against the LIVE rule, not the retired one. The original "do not invent canon" guardrail

was reframed by Josh on 2026-07-22 and the reframe is binding: what is LOCKED is the pillars, the

vision, the reviewed structural skeleton, naming canon and the §17 floor; everything that *expands

and enhances within* that locked vision is EXECUTION, and drift is contained by the apparatus — the

proposal-tier → ratified firewall, cited-or-cut, fresh-context adversarial critics, and the gates —

not by the rule. The objection has to survive that correction to be worth Josh's time, and it

does:

**This is not a fill-in of locked structure and it is not ordinary execution either — it introduces a

new ontological KIND of thing into the world.** Under the reframe, execution is work that FILLS IN

the locked skeleton; a new phenomenon class is a new entry in the world's inventory of what exists.

That is the residual class the reframe still routes to Josh, and it is the only ground on which

sub-fork 1 is his rather than the loop's. **Every argument above is an argument that it FITS, and

fitting is not the same as being necessary.** The corpus already has the False Naga

carrying this exact pattern as authored chapter content at a care register that has been reviewed.

Generalizing it to an ambient, repeatable, collectible class multiplies the surface where the pattern

can go wrong by however many loci exist — and does so for a hook whose stated value is partly that

streamers will clip it. A reasonable ruling is that the pattern is already expressed, once, well, and

that the arc does not need a systemic version of it.

*The honest answer:* strong on scope, weak on kind. The class multiplies INSTANCES, not metaphysics —

every instance is the same mechanic the Ch-6 boss already ratified — and the multiplication is bounded

by the sigma discipline, the perception gate, and (under the recommendation) the lifetime cap and the

variant ban. But the objection is not dissolved, and it is the reason this is Josh's call.

CAP RE-SCOPED (Josh's two-tier ruling, eighth sitting 2026-07-27; integration record §2.3, §2.4 and

§4.2). The lifetime cap named in the bounding list above governs the WHISPER tier only. The CONTENT

tier is bounded instead by five per-surface density budgets, one per surface, plus the variant ban and

the sigma discipline, which both still apply class-wide. Josh made this call and ruled the expansion

in: the multiplication of instances is now deliberate content rather than a risk to be capped, and the

care discipline travels with it through the class row rather than through scarcity.

A second objection, from the production side: a capped, unannounced, variant-banned, non-farmable

entity may be seen by so few players that the cost of a bespoke creature with a bespoke aesthetic and

a bespoke verb is not repaid. *Answer:* the deadzone environment grammar is reused across every

deadzone in the arc whether or not a Hollow appears, so the atmospheric investment carries

independently; and under the folklore economy, reach is driven by transmission, not by encounter

frequency.

---

SUB-FORK 2 — THE IN-WORLD NAME

2.9 The naming method, run properly

Deep-dive-canon → generate widely on DISTINCT axes → adversarially kill anything that fails the known

failure modes → player-vibe-test the survivor. **The prior pass (FORK 2) generated four candidates on

essentially two axes and did not run the collision audit.** This is the full run, and the collision

audit is what changes the answer.

The canon axis. *Cassus* = hollow, ACTIVE T1 canon. Negative vril is "self-LOSS, disconnection,

consciousness-hardening" — the register is subtraction, not malice. The Ch-6 False Naga precedent:

the counterfeit carries a ROLE-TITLE, never an invented sacred personal name. The natural-voice

doctrine: every player-facing word sounds like a PERSON, culturally registered per region,

meaning-carrying rather than literal. The no-mode-taxonomy rule: internal class vocabulary never

reaches a player surface.

The single grounded axis these converge on: THE ABSENCE THAT STANDS UP — not the thing, but what

is missing from it, given a body.

Generation, five axes. Absence (the Hollow, the Empty, the Vacant, the Missing) · arithmetic and

remainder (the Remainder, the Unnumbered, the Carry, the Sigma-ghost) · wearing and theft (the Wearer,

the Worn, the Taken-Shape, the Wrong Face) · silence and static (the Silence, the Static, the

Unsounded) · what a person would actually say (the empties; the ones that came back wrong; the

leavings; the standing dead; what the drought left; the shape in the dry place).

2.10 The adversarial filter — including three collisions the prior pass never ran

2.11 The four grounds against player-facing "the Hollow"

• GROUND 1 — external collision. "Hollow"/"Hollowing" is FromSoftware's central

undead-degeneration term in the most recognizable series in the adjacent genre, and **"the Hollow

Knight"** is the titular void-filled vessel of the medium's most celebrated absence-entity game — a

game already claimed on our own lens roster at L-R7. In an action RPG, "the Hollow" is not a

distinctive name; it is the genre's most generic available word for exactly this concept, attached

to two of its most famous entities. **For a hook whose entire purpose is community folklore, the

name is unownable** — searching it returns another game.

• GROUND 2 — internal collision. *(Evidence tiers marked. The ratified layer carries this ground

on its own; the spike layer is shown separately and proves nothing.)*

RATIFIED: "Hollow" is already a proper-noun role-title on two registry bosses —

BE_0002 "Rangku Wera, the Hollow Hunter" (T0_Boss_Encounter_Registry row 2, CH_02, main,

CHAR_0008 — the game's first main boss) and BE_0060_S1 "the Hollow Retainer" (Ch 60).

Plus BE_0053's EXPOSURE-OF-THE-HOLLOW win-condition register, the Josh-ruled shared

"hollow-core tell" boss grammar, and the ordinary-adjective field measured at 218 lines

below (Ground 3) — including constitutionally at HL_0014 ("Evil-path protagonist is hollowed

not cartoonish").

NOT CANONICAL — shown only because an earlier draft cited it as if it were: the quest id

QS_CH_02_04 ("The Hollow Hunter"), the anchor CH02_HOLLOW_HUNTER and the dialogue namespace

DLG_CH02_HOLLOW_HUNTER live only in docs/proposals/quest_spike/CH_02_quests.md (L79/L81),

whose own header reads "Status: SPIKE DRAFT — validation artifact, not canon… **Nothing here is

canonical." They are not in T0_Quest_Definition_Registry. They carry no weight in this

argument** and are listed here only so the ground is not read as broader than its ratified evidence.

The natural-voice question, asked directly on the ratified layer alone: does a player who meets "the

Hollow Hunter" at Ch 2, contains "a Hollow" at Ch 20, and frees "the Hollow Retainer" at Ch 60 parse

three things or one?

• GROUND 3 — the reveal ground, and it is the decisive one. **The field, measured and

reproducible** (an earlier draft carried an unscoped "146 lowercase uses" that no scoping

reproduces — corrected here): git grep over tracked *.md, *.csv, *.py, counting **matching

lines, gives bare \bhollow\b = 218 lines across 89 files**, and with inflections

\bhollow(ed|ing|s)?\b = 374 lines across 120 files. (Raw match counts, which double-count

lines carrying the word twice, are 266 and 445.) The error ran in this argument's own favour — the

ambient field is larger than the brief previously claimed, so Ground 3 is stronger than it was

stated, not weaker. That 218-line field is **loaded

ammunition**. At Ch 76 the player learns the patriarch's name is Cassius, and every one of those uses

detonates at once in retrospect — the octopus/Kasa/Khan foundation model the design already names,

meaning that lands backwards. **Naming the creature "the Hollow" SPENDS that ammunition early and

loudly.** It takes the game's quietest, most diffuse register word and puts it on a hunted monster

with a codex entry and a community wiki page. The word stops being ambient and becomes a proper noun

with a definition, and the Ch-76 click weakens in exact proportion to how memorable the creature is

— and the creature is designed to be the most memorable thing in the bestiary. *This is a

reveal-discipline argument for a naming decision.*

• GROUND 4 — the care ground (§3.5 of the architecture wave). The corpus already uses "hollow" to

mean a person emptied by a condition (HL_0014; CH_76.md's framing of Cassius as "a hollow

vessel the condition found and fit"; CHAR_0008 "hollowed by the polo's bargain"). If the same word

becomes the player-facing proper noun for a monster you seal and file, the game teaches — in its

own vocabulary, without ever saying it — that negative vril produces a KIND OF BEING. That is

precisely what HL_0114 forbids ("not personified evil… a vibrational state"), and it is the

failure mode the Ch-76 guardrail is sharpest about. If the player has spent forty chapters

containing Hollows, and Cassius is a hollow vessel, the Truly-Good purge reads as exterminating the

last one. See §3.1 — this ground does not fire on its own; it fires when the three charter defaults

compose.

2.12 The option set

• N-A — "the Hollow" global player-facing + "substrate-echo" class term *(the charter)*. Pro: one

memorable franchise noun; streamable; the *cassus* loop legible to any player who later learns the

etymology. Con: all four grounds above, unmitigated; and it flattens 76 regions and 5,000 years into

one folk word, which the natural-voice doctrine forbids in as many words.

• **N-B — "the Hollow" as the AUTHORIAL and REGISTRY class term; NO global player-facing common name;

the player layer is what each culture SAYS about a drained place, carried DESCRIPTIVELY in its own

attested worldview. Pro: solves the collision by construction** — no player surface ever says

"a Hollow", so "hollowed" stays a description of what happened to a person and never becomes a

species; it is what the natural-voice doctrine and the golden-standard enrichment mandate actually

ask for; it is how real folklore behaves (a revenant is a *strigoi* in Wallachia, a *draugr* in

Iceland, a *vetala* in India); and the record displays the reading the player first learned it by,

so their own logbook says what *their* world made of it. Con: no franchise noun — MissingNo is

famous partly because it has a name.

⚠ The variant that must NOT be chosen, and why it matters that the two are distinguished:

N-B is not "mint a local word per culture." Minting one is ruled against — see §2.13 items 3

and 3a. The descriptive form is the version this brief recommends, and it is the only version the

substrate can support today.

• N-C — a different global common name (the Leavings, the Wrong Shape, the Standing Dead). Pro:

keeps a single shared word without the person-condition clash. Con: still flattens the per-culture

layer; and every candidate vivid enough to be a franchise noun tends to name the face-wearing

property, which §2.10 killed on reveal grounds.

• N-D — no name at all, in-world or out. Pro: maximal eerie; maximal community coinage. Con: a

phenomenon nobody in-world has a word for is not folklore, and the design's whole claim is that NPCs

whisper about them. Villages name what frightens them; a nameless thing reads as unfinished.

2.13 RECOMMENDATION — sub-fork 2

N-B in its DESCRIPTIVE form, with two designed additions.

1. "The Hollow" is ratified as the AUTHORIAL and REGISTRY class term. The registry name

T0_Hollow_Codex and the HW_ prefix are already correct and need no rename. The *cassus*

resonance is kept everywhere it actually earns its keep — in the design, where the reader who

notices is the one who was going to notice anyway. This is a rename/reframe, never a removal —

exactly as the anti-fringe corrective is authorial-only.

2. "Substrate-echo" is RETIRED with a supersession note. Not a taste call: it names the superseded

§2.7 metaphysics that FU.1 replaced, *and* it collides with a term wired into enrich.py and read

by every chapter build. Keeping it leaves a stale term describing a retired ontology in a live

schema — precisely the stale-reference class that makes later agents hallucinate.

3. **The player-facing layer is DESCRIPTIVE, in each culture's own attested worldview — not a minted

local word.** A small hand-authored set — six to ten clusters, not seventy-six — authored inside

the natural-voice pass with its two critic lenses ("does a person say this?" / "does this carry

the culture?"), meaning-carrying rather than literal, never machine-derived. What the layer

actually contains is what that culture's attested worldview would say about a drained place,

said in English description: *the empties · the ones that came back wrong · what the drought left ·

the shape in the dry place* — §2.9's own "what a person would actually say" axis, which is

description, not vernacular. **A VERNACULAR term is used only where a source genuinely attests one

for a void/absence/drained-place concept, positive-controlled per culture** (the concept is

searched for in that culture's own substrate, and the absence is recorded as an absence rather than

filled).

**⚠ 3a — WHY DESCRIPTIVE. The precedent, and the three things an earlier draft of this brief did

not disclose.** This is the one recommendation in the document that was re-costed after

adversarial review, and it is re-costed downward.

• The ratified precedent runs directly against minting. _source/02_Tier_2_Region_Pages/bali.md

(Naming discipline, L1311-1317) is verbatim: "supplying an unattested local word inside

living-tradition substrate is exactly the authenticity failure §17.1 guards against — it would

read as authoritative cultural detail while being uncited, and **it would flow straight into the

Section-17 localization glossary. This page therefore carries the dividers descriptively**

until a source that genuinely attests a term is acquired." A player-facing folk name for the

Hollow is exactly that move, and the localization-glossary point is the sharp end: a minted word

propagates into the loc pipeline as if it were real vocabulary. Reinforced by CVD §19.8

(Josh-ratified D-RR, 2026-07-12), which retired "an authorial-coinage naming layer (Pohljar,

Tiamarath, Maatherion) that sat one remove from the real traditions it drew on" in favour of "the

real tradition-names their substrate already carried." **The repo's ruled direction is coinage →

attested tradition-name. Minting folk-words would run it backwards.**

• The substrate to ground them does not exist yet. Four region pages exist (angkor_khmer,

bali, flores_island, sumatra_java); T0_Region_Index carries 48 pending rows. And for

the cultures this brief has used as examples the *specific concept* is unattested: Flores attests

a rich malevolent register (*polo*, *janto*, the *wa*-displaces-*mae* possession, *poti*,

*sumang*) but no void / absence / drained-place / ambient-residue entity; Sámi attests

*Stállu* and *Jábmiidáibmu* (a death-REALM, threshold-only care) with the RCM flagging Sámi

vital-force vocabulary as pending book-processing; Congo attests charm/power-figure material,

with *biloko* explicitly fenced to the Mongo-Nkundo. **A folk-word would be invented in all

three.**

• And a ratified ruling already governs the exact region HW_0001 names. docs/spine/CH_11.md

(Congo/Ituri, RESOLVED ENRICHMENT_PROPOSAL [LEAVE]): "the chapter **deliberately introduces no

folklore malevolent-being boss** … the neighbouring biloko ogre-spirits belong to the

Mongo-Nkundo, not the Baka or Mbuti; the deliberate absence is the ethnographic point."

HW_0001's live region_id is central_africa_congo. **At that region the descriptive form is

not merely safer — the minted form contradicts a ruling already on the books.**

• The honest consequence for the option comparison: an earlier draft presented N-B's care

profile as strictly superior to N-A's. It is not strictly superior — **N-B transfers an

authenticity obligation downstream** into the natural-voice pass and the region-page substrate,

where N-A would not. The descriptive form is what discharges that obligation cheaply and by

precedent; **attested-vernacular naming, where anyone wants it, is GATED on region-page

acquisition** and is not a thing this ruling can schedule. That gate is disclosed here rather

than discovered later.

*Not enumerated in this brief, and now for a second reason:* naming a Manggarai or Sámi or

Congolese folk-word is culture-bearing authorship that belongs in the region-page substrate with

its own care read — and on today's substrate it would be an invention, not an authorship.

4. **Structurally this needs one schema note, not a migration — and it is the OBSERVATION-LOCUS model,

not a variant key. Codex rows key on where and under what conditions the thing was witnessed**

(spawn_locus_type × region_id), and manifestation_designation carries **that place's own

descriptive reading of it. This is a logbook key, not a species key** — see §2.6 sharpening 2

for the three constraints that keep it one: the class definition is singular and shared by every

row; the row count is capped at the lifetime encounter cap; and no row grants anything another row

does not. An earlier draft wrote this as "spawn-locus type crossed with region cluster… that

cluster's own word," which read as up to thirty named sub-species and directly contradicted the

variant ban. **The two are reconciled here: rows differ in WHERE IT WAS SEEN and WHAT THAT PLACE

MADE OF IT; they never differ in WHAT IT IS.**

CAP RE-SCOPED (integration record §4.2 and §5). "The row count is capped at the lifetime encounter

cap" now reads at whisper scope: the codex is the WHISPER LOGBOOK and its rows are the capped

observation records of the uncounted tier. Content-tier resolutions log to the cleansing meter and

to the region and community ledgers rather than to the codex, so the cap bounds observation records

only and never content-tier encounters. The other two constraints on this line are unchanged and

bind BOTH tiers: the class definition is singular and shared by every row, and no row grants

anything another row does not.

On the lost franchise noun — the community coins one. A phenomenon with a dozen local names and no

official word is the strongest folklore generator available, because the shared word becomes the

players' own rather than the marketing department's. That is how every real cryptid name in the

corpus's own source material happened, and the design brief's stated goal was

community-folklore-generating.

⚠ CROSS-WAVE COORDINATION ITEM. The natural-voice rewrite pass is separately re-authoring boss

epithets ("boss names are what PEOPLE call the being"), and it will touch "the Hollow Hunter" (Ch 2)

and "the Hollow Retainer" (Ch 60). These two passes must be coordinated, or the rewrite will

independently keep — or independently move — an epithet this fork just ruled on.

2.14 STRONGEST OBJECTION — sub-fork 2

**A game needs a word for the thing it wants people to talk about, and "the Hollow" is a genuinely

excellent one** — evocative, short, canon-rooted, already written into a registry name, a table prefix,

a design doc, a charter and a wiring plan. It is Josh's own vocabulary from the 2026-07-13 sitting.

And the §2.11 Ground-4 collision is a CRAFT problem, and craft problems are normally solved by

craft rather than by withdrawing a word: hold the line that no narration ever calls a *person* a

Hollow, keep the purge language separative ("the condition, not the man"), and the collision never

fires. Under that reading, N-B pays a permanent cost in shared vocabulary to insure against a risk a

style rule already covers.

*The counter, offered fairly:* style rules degrade across a thirty-million-word corpus and dozens of

authoring passes, whereas a word that is never on a player surface cannot leak. And N-B loses less

than it appears to, because it keeps the word everywhere it earns its resonance.

*A second objection, and this one the brief concedes rather than answers:* **N-B does not remove a

care obligation, it MOVES one.** N-A puts one word on a player surface and polices it with a style

rule. N-B puts nothing on the player surface and hands the natural-voice pass the job of making each

region speak about the thing in its own register — which is more authoring, in more places, each

carrying its own §17.1 read. The descriptive mechanism (§2.13.3a) is what keeps that obligation cheap

and precedented rather than open-ended, but **it is an obligation N-A would not incur, and any claim

that N-B's care profile is strictly superior is overstated.** What survives is narrower and still

decisive: N-B removes the *reveal* risk (Ground 3) and the *personification* risk (Ground 4)

structurally, which a style rule cannot do.

*The part that is not answerable:* the evocativeness objection is real. "The Hollow" is a better word

than anything the filter left standing, and that is why this goes to Josh as a brief with both options

live rather than as an applied decision. **If Josh prefers the sound of "the Hollow" as a player-facing

name, the mitigation set at §3.1 is what must ride with it** — and Grounds 1-3 remain unaddressed.

2.15 ⚠ A CONTRADICTION BETWEEN THE WAVES, SURFACED

The lens sweep independently recommended "the Unnumbered" as the player-facing anchor and

kept "substrate-echo" as the class term. **Both halves are defeated by the canon sweep's

positive-controlled collision audit** — "unnumbered" is the live Vimana-coordinate adjective in an

ACTIVE T1 doc, and "substrate-echo" is the wired naming method. The lens sweep did not run those two

greps; the canon sweep did, with positive controls. The collision audit wins, and the lens sweep's

own stated fallback — "keep 'the Hollow' as the DESIGN-layer nickname and give the player-facing layer

the folk names only, which several lenses would actively endorse" — is N-B. So all three waves

converge on N-B once the audit is applied. The contradiction is recorded rather than smoothed.

One precision on that convergence: the sweep's fallback says "folk names," and this brief adopts

it in the descriptive form only (§2.13.3a) — the minted-vernacular reading of "folk names" is

ruled against by the bali §17.1 precedent and by CH_11's ratified absence. The waves converge on the

SHAPE; the mechanism is narrowed here.

---

SUB-FORK 3 — THE REVEAL WINDOW

2.16 ⚠ The charter's position is the SUPERSEDED wording

This is the most consequential single correction in the brief. §6.2 carries "pure atmosphere

pre-Ch-76" forward as the opening recommendation. That is verbatim the pre-refinement §5 FORK-3(a)

wording. Josh's own 2026-07-13 steer already superseded it. creature-system.md FU.1 contains a

sub-heading that says so in the text:

"Reveal, refined (updates FORK 3). Because negative-vril corruption is on stage from Ch 6 onward

(the False Naga, the strigoi-as-vril-parasite at Ch 26…), the Hollow's SURFACE reading — a

negative-vril blight, a corruption-organism — is **reveal-SAFE and can be legible from its first

appearance. What stays Layer-3 and post-Ch-76 is the deeper meaning… So the Hollow is now safER

than the old framing**, which leaned its whole meaning on the reveal-locked depletion-substrate.

Surface: corruption made flesh (say it). Depth: a wound wearing a lost face, pointing at the

pre-depletion truth (never say it pre-reveal)."

Five locations across four documents carry the REFINED split, not the §5 wording: the DESIGNED

registry const in the wiring plan (WS7_WIRING_BUILD_PLAN.md L371 — reveal_tier (const)

'SURFACE_blight_legible; DEPTH_L3_post_Ch76'); through_line_system_throughline.md L159-160 and

L350-351 (the same document, two places); through_line_recurring_antagonist.md L366-368;

phase1_A1_prologue_ch1.md L36.

⚠ Read the first one precisely — it is a DESIGN spec, and the LANDED row contradicts it. The

const lives in the wiring PLAN, not in the registry: the live HW_0001 cell reads design-layer

(§4.5, divergence 7). So the correct statement is "**the designed const, which the landed row has

already lost**" — the schema underneath says the right thing and the artifact on disk does not, which

is a sharper argument for ruling the FU.1 split than the smoother version, and it is the one the

evidence supports.

So the stale phrasing propagated through three documents while the schema underneath it says the right

thing. **Presenting "pure atmosphere pre-Ch-76" to Josh as the standing recommendation would ask him

to re-ratify a position his own steer already replaced.**

A precision note so the brief does not overclaim. The charter's *pre-Ch-76 phrasing* is CORRECT

against the HL_0044 floor — REVEAL_WINDOW_REALIGNMENT.md L9 records that HL_0044/HL_0112 were

deliberately LEFT unchanged in the realignment because they are a FLOOR, not a held-window label. Both

things are true: the floor is right; the window label is stale. "Pure atmosphere pre-Ch-76" and

"surface legible from Ch 6" are materially different builds — the former forbids an NPC from ever

saying what the thing is; the latter lets the world name it a blight and forbids only the *why*.

2.17 The reveal constraint set is THREE-layered, not two

• HL_0044 (CVD §17.9) — "Pre-Ch 76 Grand Sage identity hints are forbidden… across dialogue,

journal entries, inscription content, NPC exposition, **environmental storytelling, art direction,

music**." Scope Ch 1-75.

• HL_0043 — the thesis is never declared; Ch 76-77 is the single canonical direct surfacing.
• The standing silences: Cassius held to Ch 76; thesis and Grand Sage to Ch 77; **no

House-of-Velheim pattern before ~Ch 38** (even invoked-and-negated agent-nouns leak).

• CH_60.md L90, the most useful sentence in the corpus for this fork: "the family is nameable

(reveal-lifted post-38), but Cassius stays held to Ch 76 and the thesis/Grand-Sage to Ch 77."

2.18 The option set

• R-A — PURE ATMOSPHERE PRE-CH-76 *(the charter's stale line)*. Zero explanation; all meaning

post-reveal. *Pro:* maximum apparent safety, no interpretive surface to police. *Con:* **it is

strictly MORE dangerous than R-B, not less** — an unexplained wrongness invites the player to explain

it, and the only explanation the world offers is the reveal-locked depletion substrate. **Silence

does not prevent inference; it concentrates it.** It also wastes an already-legible canon fact

(negative-vril corruption on stage since Ch 6) and produces the "cryptic or like a computer wrote it"

register the natural-voice doctrine was created to end.

• R-B — THE TWO-TIER WINDOW *(Josh's own refinement)*. Surface legible from first appearance, said

diegetically in a person's voice per culture; depth held to L3 / post-Ch-76. *Pro:* the position Josh

already steered; reveal-SAFER because the surface answer satisfies the question the encounter raises;

it gives the L2 lane real content (a region's own record of what it lost) without touching the

reveal; and it gives the natural-voice pass something a person can actually say.

• R-C — PARTIAL DEPTH BREADCRUMB PRE-CH-76. *Pro:* rewards the rabbit-hole player. *Con:* the depth

IS the depletion substrate, one inferential step from the Grand Sage; HL_0044's scope is total across

every channel. This asks the reveal ceiling to hold a line drawn *inside* the reveal. Reject.

• R-D — POST-REVEAL ARTICULATION, additive on top of R-B: the sealed records the player has been

accumulating recontextualize after Ch 77 and the codex's L3 lane opens. *Pro:* turns a forty-chapter

completionist thread into a retrospective payoff — exactly how the octopus/Kasa/Khan foundations are

canonically specified to work ("surface only in retrospect"); costs the pre-reveal window nothing;

gives the class a destination. **The player who harvested twenty Hollows and learns at Ch 77 what the

shapes were is the design's best possible outcome, and it requires no new content — only that the

depth be held.**

2.19 The middle gate that does not currently exist — a live reveal defect

FU.2 adds "fresh Velheim extraction scars and corruption seams" as a second spawn locus and praises it

for tying the hunt to "the antagonist's operational footprint." But the standing discipline holds the

House-of-Velheim pattern until ~Ch 38. **A player who can observe, before Ch 38, that Hollows

cluster at a specific organization's industrial sites has been handed the pattern by environmental

storytelling** — which HL_0044 names as a forbidden channel.

The whole-arc pass already certifies "the Hollow surface" as one of the pre-Ch-38 seeds that introduce

no agent-noun and no who-frame, which is exactly why extraction-scar loci are listed starting at

Ch 39. The gate exists in the staging and is absent from the spec. Closing it is this

sub-fork's one substantive addition.

2.20 ⚠ The art-layer hazard — where no gate looks

**The Ch-77 reveal is a shape that RESOLVES INTO YOUR OWN FACE. The Hollow is a shape that WEARS A

FACE THAT IS NOT ITS OWN.** The Hollow is the exact structural negative of the waterfall — and neither

the creature-system doc nor the charter names it.

It is an asset: the CVD's whole reveal architecture is *earned pattern recognition*, and a

phenomenon that trains the player's eye on "a shape, and what is inside it" is the perfect

grammar-teacher for the waterfall, teaching without one word of exposition.

It is a hazard: HL_0044 forbids foreshadowing through environmental storytelling and art direction

— exactly where a creature's visual grammar lives — and the existing teeth do not reach it.

check_reveal_discipline.py scans body prose; check_grand_sage_silence.py scans voice rows, scene

bindings and the two endgame chapter files. **Neither scans a concept board, a material, a silhouette

or a staging note.** The Hollow is the first content class whose reveal risk is primarily VISUAL, and

it arrives exactly as the realm program stands up an art-board pipeline.

Two waves found this independently, which is the strongest confidence signal in the brief. The

specific trap: FU.1 says the Hollow "wears the eroded shape of something the depletion took — a stolen

face"; HL_0115 says the Grand Sage at Ch 76-77 "shifts into protagonist's specific gender

presentation." **"A hollow wearing your face" is the most obvious horror composition available and is

exactly what a generative concept pipeline would produce unprompted** — an HL_0044 violation no text

gate would catch.

2.21 The six art-layer rules — proposed as the window's operative content

A reveal window is normally a chapter number. For this class it must also be a set of **visual

prohibitions**, each phrased so a board review can check it.

1. **No Hollow ever wears the protagonist's face, silhouette, stance, height or gender presentation

before Ch 77.** HL_0115 makes the morph the reveal's own instrument; a Hollow pre-echoing it is the

single worst leak available anywhere in the corpus.

2. No reflective staging pre-Ch-77. No Hollow encounter composed at a still pool, a mirror, glass,

or a reflecting water plane. The mirror belongs to the waterfall.

3. The worn shape is always non-human or anonymous pre-Ch-76 — an animal the region lost, a shape

nobody present recognizes. Never a named character, never a companion, never a person the player has

met, never the protagonist. This also keeps the class clear of the persona system and the

community-ripple canon, both of which attach to people.

4. No Hollow RESOLVES. The horror is that it never finishes becoming anything: it flickers between

half-formed shapes and stops. Resolution is the waterfall's move and Ch 77's alone. This is the

rule that most directly separates the two images.

5. The presence and the Hollow never share a frame, palette, motif or light logic before Ch 77.

The presence is luminous, warm, directional, already-there; the Hollow is unrendered, cold, silent,

newly arrived. Nothing in the art bibles may let them rhyme.

6. No Hollow speaks, vocalizes lexically, or delivers quest direction. Silence is its signature and

keeps the class structurally distant from HL_0046.

2.22 RECOMMENDATION — sub-fork 3

**The THREE-GATE LADDER (R-B extended), PLUS R-D as the additive post-reveal layer, PLUS the six

art-layer rules adopted as the window's operative content and enrolled in the concept-board review

BEFORE any Hollow art is generated.** Reject R-C. Rule the FU.1 split, not the §5 sentence.

• Gate 0 — from first appearance: the SURFACE is legible. It is a blight; negative vril has pooled

here and stood up. Say it plainly, diegetically, in a person's voice, in each region's own register.

NPCs may fear it, describe it in the terms their own worldview gives them (a vernacular term only

where a source attests one — §2.13.3a), and be wrong about it. Safe because Ch-6's False Naga

already puts negative-vril corruption on stage — the explanation uses canon the player already has.

• Gate 1 — the EXTRACTION-SCAR LOCUS DOES NOT EXIST BEFORE Ch 39. Not "exists but is

unattributable" — absent. Every pre-39 Hollow stands at a natural deadzone: a place that was

over-drawn, or drained, or simply dry. The locus vocabulary the player can encounter before Ch 39 is

one item long. From Ch 39 the extraction-scar locus becomes available, and the correlation between

blight and industrial site is then assembled by the player from placements they have already seen —

one of the arc's better mid-game realizations, and it arrives in the window where the family is

nameable.

Why the ABSENCE form and not the withheld-author form. An earlier draft wrote that a pre-38 scar

"reads as damage without a legible author." That sentence is the leak it exists to prevent: the

standing pattern-reveal discipline (Josh D-14, as corrected at RR-32) rules that invoking **and

negating** the who-frame still leaks — "no name for the hand," "not yet a single hand" all fail,

because the negation itself plants the single-agent scaffolding. A player told a wound has no

author has been told the wound has an author. The safe register introduces **no agent-noun and no

cross-chapter ordering at all**, which is exactly how the whole-arc pass certifies the Hollow today:

through_line_system_throughline.md L348-351 lists "the Hollow surface" among the pre-~Ch-38 seeds

that are "all regional craft/myth/economy unconnected to the family — **none introduces an

agent-noun or the who-frame**." A locus that is not in the world cannot be described at all, which

is the only fully safe form. *(The Ch-60 forcing-mark is deliberately NOT cited here as a

precedent: CH_60.md L90 states the family is "nameable (reveal-lifted post-38)," so Ch 60 sits

in the OPEN window and its handling cannot be imported into the HELD one.)*

• Gate 2 — Ch 76-77: the DEPTH lands, in two beats. At Ch 76 the patriarch's name makes the

*cassus*/hollow field detonate retrospectively; at Ch 77 the thesis makes the stolen shape mean what

it means. The sealed record gains its final section only in this window; the L3 connective material

is authored to Ch 77.

On the chapter window itself: do NOT hand-set one. Derive it from the perception gate and locus

availability exactly as WS7 specified, with a floor no earlier than the Vril-Sight tier the gate

requires. The two loci naturally produce two arrival points — deadzone Hollows whenever the

perception gate opens; extraction-scar Hollows once the family's operational footprint is player-facing

from Ch 39 — and that split is reveal-coherent without anyone assigning a chapter. This is also the

answer to the live HW_0001 breach (§4.5).

Populate the registry's reveal_tier column with the gates rather than leaving them in prose. The

codex should SHOW that the entry is incomplete — a sealed record with a visibly unfilled section is

an honest promise to the player and leaks nothing (the Outer Wilds ship's-log lesson).

One further option, surfaced deliberately and NOT recommended for bundling: a post-Ch-77 Hollow

that wears HER face, legal only in the Epilogue or new-game-plus — the reveal's dark twin, a shape

filled with the wrong substrate against a shape that was you all along. It is the most powerful beat

this class could ever produce and the one most able to re-open a closed reveal and re-personify

negative vril. **It should be routed to its own ruling with its own care read, never folded into this

admission.**

2.23 STRONGEST OBJECTION — sub-fork 3

**R-B asks the game to SAY something about negative vril at every Hollow encounter, forty-odd chapters

before the reveal, in a voice that is per-culture and therefore authored many times by many passes.

Each saying is a chance to say one sentence too much.** R-A's silence has no such surface: nothing said

is nothing leaked — and the corpus's own documented history is that **prose fan-outs introduce reveal

leaks that gates cannot catch**. A ruling for R-A on pure risk grounds is defensible.

*The counter:* the risk R-A avoids is a prose risk; the risk it creates is an inference risk,

which is worse because it cannot be reviewed. A player told "this is a blight, negative vril pooled

here" stops asking. A player told nothing keeps asking — and the corpus has exactly one unexplained

wordless presence for them to reach toward. The safest window is the one that answers the small

question so the large one is never raised.

*A second objection:* a three-gate ladder is more machinery than a rare creature warrants, and gates

are where content rots — a Gate-1 unlock no downstream author remembers is exactly the stale-window

defect this repo has already paid for twice. *Answer:* the ladder is not new machinery; it is the

arc's EXISTING reveal ladder (CH_60.md L90) applied to one more entity, landing in a column that

already exists and currently holds a placeholder. The rot risk is real and is answered by putting the

gate in the column, not in prose.

---

3. CROSS-SUB-FORK COHERENCE — the recommended triple as ONE design

3.1 ⚠ The finding that binds all three forks — and why it only appears when they compose

Compose the three CHARTER DEFAULTS and read the result aloud: a monster class named after the

antagonist patriarch's own etymology, spawning at that family's industrial sites, which the player

hunts, seals and files — in a game whose Ch-76 §17.1 block calls the antisemitism guardrail its

absolute sharpest because the family's raw premise is a hidden banking dynasty.

That composition is a vermin metaphor attached to a dynasty. No single fork produces it. The three

defaults together do. This is the strongest argument that the three sub-forks must be ruled as one

design rather than three independent questions.

The recommended triple breaks the chain in four places, each cheap and each checkable:

1. The name stays authorial (N-B). No player surface links the blight to Cassius's etymology.

*This mitigation alone breaks the chain.*

2. The class is never a person and never a Velheim (the class definition's fourth negative). No

persona, no name, no relationships, no community ripple.

3. At least one canonical locus is NON-Velheim in origin — a natural deadzone, or a community's own

historical over-draw. The extraction scar is what extraction does, not what a bloodline

*emanates*; any negative-vril accumulation makes the same scar. **A codex whose every row points at

the family has made the family the world's only source of rot** — both a care failure and, per

CH_76.md's own framing, a canon error: the reach traces THROUGH Cassius, not FROM him.

4. The purify verb heals LAND and never punishes a people. FU.2's wording already does this; it

should be stated as a rule rather than left as prose.

Note the load-bearing interlock: mitigation 1 *is* sub-fork 2's recommendation and mitigation 3 *is*

a consequence of sub-fork 3's Gate-1 (the extraction-scar locus does not exist pre-Ch-39, so every

early locus is a natural deadzone by construction). **If Josh rules N-A on the name,

mitigations 2-4 become mandatory rather than recommended**, and the composition risk rises rather than

disappears.

3.2 The triple, stated as one design

**The Hollow is admitted as an OBSERVE/CONTAIN phenomenon class — capped, variant-banned,

unannounced; an L1 encounter whose MEANING is L3-homed and whose record is an L2 tributary —

carried by the class row plus a locus-grammar (EG_) row, holding no CR_ id, no BE_ id, no slot and

no combat power. "The Hollow" is its authorial and registry name and is never spoken on a player

surface; what each region makes of it is said in that region's own attested worldview, described

rather than named, authored by the natural-voice pass — and they may be wrong. What it IS is

legible from the first time you meet one; the extraction-scar locus does not exist before Ch 39, so

whose footprint it clusters on becomes observable only in the open window; what the shape it wears

MEANS arrives at Ch 76-77 and recontextualizes every record you already filed.**

TRIPLE-STATEMENT NOTE (Josh, eighth and ninth sittings 2026-07-27; integration record §4.2). Three

clauses in the block quote above are superseded and the rest stands. "No BE_ id" is superseded: BE_

rows are the authored-encounter home. "No combat power" is superseded on the ABSORB branch only. And

"capped" now reads at whisper scope, with the content tier bounded by per-surface density budgets.

"No CR_ id" and "no slot" SURVIVE and remain the firewall made structural, as do the naming and reveal

sentences, which the H-2 and JB-6 rulings confirmed.

Read as one design, the three recommendations reinforce rather than merely coexist:

• Admission's four sharpenings serve the reveal. Non-announcement *is* reveal discipline — an

entity the game never explains cannot leak. The variant ban and the encounter cap keep the class from

the over-showing that would let a player accumulate enough observations to infer the depth.

• The name recommendation serves the reveal. Keeping "hollow" off the entity and on the ambient

register is what preserves the Ch-76 retrospective detonation (Ground 3). The naming fork and the

reveal fork are not independent.

• The reveal recommendation serves the care read. Gate 1 (no extraction-scar locus before Ch 39,

so every early Hollow stands at a natural deadzone) *is* mitigation 3 of the §3.1 chain, arrived at

from a completely different direction.

• The reveal recommendation serves the natural-voice doctrine. Gate 0's "say it, in a person's

voice, in each region's own register" is what gives the descriptive layer of the naming

recommendation something to say. N-B and R-A would fight each other: a per-region descriptive

layer with nothing it is permitted to say is a naming system with no surface. N-B and the three-gate

ladder are mutually enabling.

• Horror craft and reveal discipline push the same way. What the Hollow MEANS is the one thing the

design can never over-show, because HL_0044 forbids it until Ch 76-77. That is a rare alignment and

should be read as a reason for confidence rather than a constraint to work around.

3.3 Where the triple has a seam

The HL_0080 anti-defang riders must ride forward verbatim, and the recommended sharpenings pull

mildly against them. HL_0080 requires that "the 4-step non-damage protocol is ONE encounter class

within a chapter, never the chapter's sole combat layer" and that "the containment beat itself is

staged as an action-forward, vril-taxing set-piece — dodge, parry, and manage the pool under real

pressure — not a static perceive-and-wait puzzle."

A capped, unannounced, non-fought, documentation-rewarded class read carelessly becomes exactly the

static perceive-and-wait puzzle those riders forbid. The resolution: "never fought as an opponent"

constrains the *ontology* (no HP bar, no defeat state, no kill reward) and not the intensity. The

containment is a real set-piece under real pressure. Both hold — but they must be stated together or a

downstream author will ship the quiet version.

SUPERSEDED — THE SEAM IS CLOSED IN THE OPPOSITE DIRECTION (Josh, ninth sitting 2026-07-27, JB-1;

integration record §3 and §4.2). The resolution above tried to hold "never fought as an opponent" as

an ontology constraint while granting the riders their intensity. Josh removed the Hollow class from

HL_0080's non-damage protocol entirely: Hollows are DEFEATED, so there is no seam left to hold. The

riders' own language is now simply the correct register for phase 2 of the four-phase grammar — an

action-forward, vril-taxing set-piece with dodge, parry and pool management under real pressure — and

the HL_0080 rider that a containment class is never the chapter's sole combat layer continues to bind

the sacred-site atmospheric echo, which keeps its non-damage protocol and its six built instances at

Ch 10, 20, 29, 38, 45-46 and 57. What replaces the ontology constraint is the anti-kill-loot teeth:

no drop from the being, rewards attached to the container, no kill credit, and an unskippable fork.

---

4. WHAT ADMISSION TOUCHES

4.1 THE LOCKED NUMBERS — the guarantee, stated

**33 legendary species (HL_0057) and 22 familiar slots (HL_0058) are UNTOUCHED, and the guarantee is

STRUCTURAL rather than promissory.** Four independent gates:

1. Never TAME-status → outside the 22nd-slot binding-stone pool. VERIFIED DIRECTLY THIS SESSION

(the one claim the lens sweep flagged as taken on trust): T3_Creatures_Tameable [ACTIVE v1.0]

§2.4 L202 — "T3_Creatures_Tameable owns the tameable-creature pool from which the binding stone can

elevate one creature to legendary at L2, L4, and L5 finales. **Eligible pool: any TAME-status

non-familiar species across all 400-plus Beastiary entries." A never-TAME class cannot** be

elevated to legendary at any finale. *This is the strongest of the four: it is a consequence of an

existing rule, not a new promise.*

2. No CR_ id — it lives in its own HW_ registry, not T0_Creature_Roster. It cannot be counted

into the legendary roster because it is not in that table.

3. Not breedable — the offspring firewall (bred offspring always ordinary-tier) is untouched; the

class never enters the husbandry pool at all.

4. Not a combat slot, no combat power — the three sacred slots stay legendary-only.

SUPERSEDED IN HALF (Josh, eighth and ninth sittings 2026-07-27; integration record §4.2). The

NOT-A-COMBAT-SLOT half SURVIVES and is untouched: the three sacred familiar slots stay

legendary-only, and the absorb lane never touches a slot. The no-combat-power half is superseded on

the ABSORB branch, where the evil-integrity path grants hollow-type abilities toward endgame. The

guarantee this gate protects is unaffected, because the absorb lane is CATALOG ACCESS on the

ability tree and never a slot, never a species, and never an entry in the 33 or the 22.

Precedent for the additive move: CVD §17.12's new-systems reconciliation already ratified

(2026-07-12) that additive live-layer registries' *existence* is canonical and their *counts* are

authorial — the exact basis on which MG_ (Minigame) and VH_ (Vehicle) were admitted. **Admitting HW_

adds a registry, not a count.**

⚠ The one number that DOES move, and should be declared at ratification: the bestiary/dex total.

That figure is *already* drifting — T3_Creatures_Tameable says "200+" in §0.1, "400-plus/420-plus" in

§1.1/§2 (confirmed at §2.4 L202 above), and "222 entries" in §3. **Recommendation: declare the

containment records a SEPARATE SEALED SUB-CODEX, explicitly NOT an addition to the bestiary count** —

which is what §2.7/FU.2 already describe. Saying so at ratification prevents the drift getting worse.

And it is a logbook, not a dex: its rows are observation records of ONE class (§2.6 sharpening 2),

capped at the lifetime encounter cap, so the sub-codex has a ceiling the moment the cap is authored.

CAP RE-SCOPED, AND THE CONSEQUENCE HOLDS (integration record §4.2 and §5). The cap governs the WHISPER

tier only, and the codex is the whisper logbook, so the ceiling above is a ceiling on OBSERVATION

RECORDS and never on content-tier encounters. The sealed-sub-codex recommendation is unaffected and is

strengthened: content-tier resolutions log to the cleansing meter and to the region and community

ledgers, so they never enter the bestiary count either.

4.2 The proposed teeth — the assertions that make the guarantee hold

Offered as design, not applied. Each deterministic and positive-controllable.

• FIREWALL: no HW_ id appears in T0_Creature_Roster.creature_id, in

T0_Boss_Encounter_Registry.boss_id, in T0_Familiar_Registry, or in any binding-stone pool column;

and no T0_Hollow_Codex row carries a TAME collect_verb. *This converts "outside the 33 and the

22" from prose into a fact a gate can read,* which is what the numerical canon requires of any new

class.

• SINGULAR-PURGE PROTECTION: no T0_Hollow_Codex row's containment language uses the

individual-purge register reserved to the Ch-76 Phase-3 purge (T1_Vril §11.8; HL_0020 carries

the same rule with a stale Ch-74 number — see §1.1 and the repair note).

• LOCUS PLURALITY: at least one HW_ row's spawn_locus_type is not family-derived (§3.1

mitigation 3).

• ONE-CLASS / ROW-MODEL: every T0_Hollow_Codex row resolves to the single class definition —

no row carries class-differentiating stats, abilities or tier; row count never exceeds the authored

lifetime cap; and manifestation_designation is a per-observation descriptive field, never a

species name. *This is the tooth that keeps §2.13.4's row keying from becoming §2.6.2's banned

variant dex.*

• RE-SCOPED, WITH THE PAIRED CLAUSE SURVIVING (integration record §4.2 and §5). Two of this tooth's

three clauses bind BOTH tiers and are unchanged: the single class definition every row resolves to,

and manifestation_designation as a per-observation descriptive field that is never a species name.

The middle clause, row count never exceeding the authored lifetime cap, is a WHISPER-tier assertion,

because the codex is the whisper logbook. The FIREWALL tooth above also loses one term: the

T0_Boss_Encounter_Registry.boss_id exclusion is superseded, since BE_ rows are now the

authored-encounter home; the T0_Creature_Roster, familiar-registry, binding-stone and non-TAME

exclusions all survive and remain the structural firewall.

• NAME-SURFACE: "substrate-echo" appears in no player-facing registry column or string, and (under

N-B) neither does "Hollow" as a proper noun. Positive-controlled with a literal-substring sweep.

• ATTESTATION: any vernacular term appearing in a player-facing Hollow string carries a source

citation in the region page that attests it; an uncited vernacular term fails (the bali §17.1

precedent, §2.13.3a). Descriptive readings need no citation — that is the point of them.

• ART-WINDOW: the six §2.21 rules enrolled as a concept-board review checklist and as a critic lens

on any Hollow art board. **This is the honest limit — it is a REVIEW OBLIGATION, not a script, and it

should be named as such rather than dressed as a gate**, because no existing gate reads images.

4.3 THE CARE READ — at bolder calibration

• The §17 hard-line FLOOR is untouched. No genocide-as-goal, no real-world targeted hate, no sexual

violence, no child harm. Body horror against a non-person blight is squarely inside the rated-M bar.

• ZERO cultural-attribution surface. The Hollow is nobody's tradition, nobody's sacred being,

nobody's condemned monster. **It is the only lane in the FU.3 collectibility map with no §17.1

living-culture exposure** — and stating explicitly that it is a VRIL-PHYSICS phenomenon rather than a

mythological being is its care shield: it can appear widely without ever making a claim about

anyone's cosmology. *(Count correction inherited from the canon substrate: that map is labelled

"seven lanes" and enumerates eight — LEGENDARY · ORDINARY_FAUNA(TAME) · ORDINARY_FAUNA(OBSERVE) ·

A SACRED · B CONDEMNED · C AMBIVALENT · D CONSTRUCT · E BLIGHT. Nothing in this brief turns on the

count, but FORK 7 ratifies that map and should not ratify a miscount.)*

• The strongest care argument for admission is that care ADDS content here. Lane A of the map

exists *so that* the aku-aku, Pele, the Nightmarchers, taniwha, moʻo, the revered Rangda, the

conscious Naga and the Ouidah sacred python are structurally uncapturable by design — routed to

DOCUMENT/FREE/ALLY, never owned. **The Hollow is what absorbs the "collect the horror" impulse onto a

thing that is nobody's tradition.** Refusing admission does not make the impulse go away; it leaves

it pointed at things that *are* somebody's tradition.

• The mythology firewalls (a clean write, since the corpus has ZERO void-cosmology grounding —

positive-controlled):

• Buddhist śūnyatā — HARD EXCLUSION, written into the class definition rather than left to a

downstream agent's good sense. Emptiness in Madhyamaka is dependent origination — a

soteriological, liberating insight held by living practitioners, and not a void, a hole, an

absence of substance or a lack. Glossing a blight-monster as "emptiness made flesh" would be

both offensive and simply WRONG about the concept, which is the failure the immersive-teaching

mandate exists to prevent. Ch 72 (Naraka and Meru) is the road's maximum care tier. So: no

śūnyatā reference, no form-is-emptiness vocabulary, no Hollow at Ch 72, no codex or L3 entry

glossing the Hollow as "emptiness." *This is a rename/reframe with teeth, not a removal* — the

concept is not banned from the game, it is banned from THIS entity, and Ch 72 is where Buddhist

cosmology is honored properly and at length.

• Ginnungagap — the Hollow may echo the STRUCTURE (a gap between densities) and must never

borrow the name, the vocabulary or the Norse cosmological framing. Ch 74 carries the loudest

Ariosophy/Nazi-occult/völkisch firewall on the road, and a negative-vril entity glossed with Norse

primordial-void vocabulary sits one association away from the exact Thule/Vril-Gesellschaft

register the standing firewall exists to prevent. Concretely: no Hollow at Ch 74; no frost/fire

polarity language in the deadzone grammar.

• The Duat's second death (Ch 69, ELEVATED care) gives the Hollow its clearest moral

vocabulary — the un-mourned remainder, the thing denied its ending — and this is exactly the

true-history immersive teaching the care doctrine says not to scrub. The line: understood

*through* the second-death idea in L3 material; never presented as an Egyptian entity, never in the

Duat, never involving a weighing, a judgment or an assessor. **Purification-as-mercy-for-a-denied-

ending is a strong, care-clean reading** that composes with the purify branch already designed.

• Hesiodic Chaos (Ch 70, standard care) is the safest available cosmological register and maps

cleanly onto "the substrate before density" — T1_Vril's own doctrine in Greek clothing. Usable for

L3 connective material.

• Per-region readings are the correct expression of local traditions — as BELIEF, described:

what a Manggarai elder believes a Hollow is and what a Congolese hunter believes may differ and

may both be wrong. Framing them as belief rather than as attributed tradition-fact is what

makes them reveal-safe and care-safe at once, and it is precisely what the N-B naming

recommendation is built on. The care line inside this bullet: the reading is theirs, the

*word* is not minted for them — a vernacular term appears only where a source attests one

(§2.13.3a), and at the Congo region specifically, CH_11's ratified "the deliberate absence is the

ethnographic point" governs.

• HARVEST is the one genuine care surface, and it is the mature register rather than a problem.

Draining a stolen shape into your own signature is an evil-path act carrying an integrity cost,

non-gating, with purify always available. The register to hold: **the harvested thing wears a lost

face, and the game should let that land.** Make the cost SEEN, not merely scored (the Shadow of the

Colossus lesson).

**THE DEVIL'S-ADVOCATE PASS, calibrated against over-conservatism — required, and it points the other

way. The repo's documented failure direction is DEFANGING**, and it has repeatedly had to be

REVERSED (D-DRAGON-18 reversed; the "never captured" rule killed; §17.16 written specifically to stop

this). So the question to ask is the opposite of the one the design keeps asking: **is the Hollow TOO

SAFE?** No combat power, no fight, a documentation-only reward, and "a light hardening-pull… never a

mechanical debuff you cannot leave" — for a hook Josh explicitly steered toward *mature R-rated horror,

not a cute pet*. **The corrective is already in canon and must ride forward verbatim: HL_0080's riders

— action-forward, vril-taxing, real pressure, dodge and parry, never the chapter's sole combat layer.**

A ratification that does not carry them risks shipping the timid version of Josh's own steer.

ANSWERED, IN THE DIRECTION THE PASS POINTED (Josh, eighth and ninth sittings 2026-07-27; integration

record §4.2). The devil's-advocate question was the right one and Josh answered it affirmatively. The

class is FOUGHT: he removed the Hollow from HL_0080's non-damage protocol, wrote "defeat it to heal or

absorb it instead," and set the encounter-design mandate — puzzle-forward encounters still carrying

real combat, action-forward encounters carrying a per-encounter trick, and the palette open. The

"no combat power" clause is superseded on the absorb branch, and the documentation-only reward now

holds for the whisper tier and the cleanse branch. The hardening-pull survives and is promoted to the

class's signature phase-2 mechanic — an escalating field that leans on the player's signature the

longer they stay close, still never an unleavable debuff — and the arc-wide valleys-never-boring rider

makes an encounter the player can only stand and watch a defect rather than a mood.

Symmetrically, the §2.11 Ground-4 / §3.1 care findings are not care-tier inflation: they do not

recommend removing anything from the game. They recommend which SURFACE carries a word, and which locus

is canonical. Both are rename/reframe moves, which is what the doctrine asks for.

4.4 THE REVEAL-SAFETY VERIFICATION

The honest limit, stated plainly: the art-layer constraint is a review obligation, not a gate.

check_reveal_discipline.py and check_grand_sage_silence.py do not read images, and nothing proposed

here changes that. The six rules are enforceable only by enrolling them in the concept-board floor

before the realm program's art pipeline generates its first Hollow board. That sequencing is the

recommendation's one time-sensitive element.

4.5 ⚠ THE FOURTH ITEM — a live do-not-invent breach, and it is NOT a fork

registries/T0_Hollow_Codex [DRAFT v0.1]/Sheet1.csv, row HW_0001, landed in WS7 Increment A

(commit 22dac91), diverges from its own designed seed. **An earlier draft of this brief said five

fields. Re-read cell by cell against the seed at WS7_WIRING_BUILD_PLAN.md L376: it is TEN of the

fourteen columns** — the five it named plus five more, flagged below. One of the additions

pre-empts sub-fork 2 inside the artifact that fork governs, and one silently downgrades the care

register. Its twin is also missing.

*(contain_verb=CONTAIN matches the seed. source_ref differs in value but conforms to the column

spec, so it is not counted. sigma_trigger_anchor also matches the seed — it was seeded blank — which

is why it is listed below as a GAP against the design rather than as a divergence from the seed.)*

Plus sigma_trigger_anchor is EMPTY against a §17.13 anchor the design explicitly relies on, and

HW_0002 (the extraction-scar locus) was specified and never landed — so the canon inventory's

"two seed rows" is also inaccurate, and the 2026-07-27 inventory's "chapter assignment is deliberately

unset" is a false clean against the live tree.

This needs reconciling whichever way the forks go. It should not ride quietly inside a ratification.

Under the recommended triple the repair is determined: manifestation_designation → the designed

placeholder with the [Josh-name pending] marker restored (and, once sub-fork 2 is ruled, the

authorial term with an explicit not-player-facing note), chapter_window → deadzone-availability,

region_id → blank, spawn_locus_type and purify_harvest_fork → their declared enum tokens,

perception_gate → bestiary_depth_threshold, reveal_tier → the three gates, containment_reward

→ the full designed list, firewall_anchor → the full chain, care_register → the R17/firewall/§17.1

register, sigma_trigger_anchor → populated, and HW_0002 authored with a non-family-derived

locus alongside it (§3.1 mitigation 3).

⚠ The marker-restoration is not cosmetic. [Josh-name pending] is the mechanism by which the

open fork stayed visible in the artifact. With it stripped, the live tree reads as though the name

were settled — which is how a fork gets closed by drift rather than by ruling. **Restoring it is the

one repair that should happen before the ruling rather than after it**, so the artifact Josh's ruling

lands on is not already asserting an answer.

STILL LIVE, AND NOW ADDITIONALLY STALE (integration record §4.2, §4.9 and §11). The ten-column

divergence above is unrepaired in the tree and remains a real breach; it is integration-record item

X-5, which is director-owned because it applies a user-authorized sensitive edit and because the

region cell interacts with a ratified care ruling. Three parts of the prescribed repair have moved

under the rulings and must not be applied as written. The purify and harvest tokens are RETIRED

synonyms, so the fork column's declared vocabulary becomes NONE, CLEANSE and ABSORB with the old

tokens recorded in notes. The reveal_tier cell takes the FIVE gates rather than three, and

perception_gate takes the three-rung perception ladder rather than a single threshold. And the row

itself is now promoted: HW_0001 becomes, or is superseded by, the singular CLASS row carrying the

ontology, the firewall anchors, the fork, the visibility law, the silence, the reward class and the

care rules, with instance rows reserved for whisper-tier observation records. The chapter_window,

region_id, locus-enum, containment_reward, firewall_anchor, care_register and sigma_trigger_anchor

repairs stand exactly as prescribed above.

One sequencing caveat on the enum. The spawn-locus vocabulary sits on an unreconciled enum:

HL_0073 confirms WGM owns a per-zone vril_density enum including *null*, but

V3_CONVERSION_RUNBOOK.md L202 rules the spine enum as {baseline, low, moderate, high, anchor} and

"never mid/null", while CH_17 carries vril_density null in a live zone row. **The class's locus

vocabulary should be authored AFTER that reconciliation, not before.** Separately: the world-state

ATMOSPHERIC band vocabulary (Luminous/Restored/Neutral/Damaged/Corrupted, T1_Vril §10.6) is a

different axis from the per-zone density enum; conflating them would be an easy and invisible error

in any Hollow locus authoring.

4.6 What admission does NOT touch

• No realm is added, and — as recommended at §2.3a — no T0_Realm_Ruleset_Registry row either.

The 9 road realms + the fairy fold + the Personal Dimension structure is unchanged; the Hollow stays

on the "exists as MECHANIC or PHENOMENON, never as an authored place" list alongside the astral

plane, the dream tradition and rift interiors. **An earlier draft claimed the negative-space ruleset

row was collision-free because "a ruleset is not a road stop." That was wrong** — the charter's

roster is an enumerated eleven-row list, its vocabulary V is closed, and it explicitly rejected

per-FEATURE content. The locus therefore takes an T0_Environment_Grammar_Registry row, which is

uncontested, and the negative-control idea is routed to the charter's Wave-2 schema mint as a named

question. Nothing in the admission depends on the answer.

• No chapter assignment. Derived from the perception gate and locus availability (§2.22).
• No spine node changes. The class has zero playable content today and the recommendation adds none

— it makes the class ratifiable so content can later be built against it.

• FORKS 7/8/9 (the ontology map's ratification, the EARN lane's count, the sacred-set invariant)

neighbour this fork and are not ruled here — though note the coupling: **the BLIGHT lane exists

only if the Hollow is admitted**, so a YES makes FORK 7 nearly forced. A NO degrades the map cleanly

(lane E empties; every supernatural being routes to DOCUMENT/FIGHT/FREE/CONTAIN/BOND with no other

change).

---

5. WHAT IS JOSH'S, AND WHAT EXECUTES ON HIS RULING

5.1 What is Josh's — routed against the LIVE rule, and the three are not equal

The routing standard. Under the binding 2026-07-22 reframe, what goes to Josh is the LOCKED

set — the pillars and the vision, the reviewed structural skeleton (chapter assignments, the

numerical canon), naming canon, and the §17 hard-line floor — plus, as the residual class, a new

ontological KIND of thing in the world. Everything that expands and enhances *within* the locked

vision is EXECUTION, and it is contained by the apparatus (proposal-tier → ratified, cited-or-cut,

fresh-context critics, the gates), not by his signature. **An earlier draft asserted that all three

sub-forks are "cardinal canon… and none is creative-within-vision." That was an overclaim** measured

against the live rule, and it is corrected here — two of the three are genuinely his, and the third

is a confirmation, not an open fork.

1. ADMISSION — HIS. Not because "do-not-invent" forbids it, but because it introduces a new

ontological KIND into the world's inventory, which is the one residual flag class the reframe

preserves (§2.8). *Recommended:* option C + the EG_ locus row + the four sharpenings, with the

realm-ruleset row NOT bundled (§2.3a). *Fallback if NO:* option B, never option A.

2. THE NAME — HIS, explicitly and beyond argument. Naming canon is named LOCKED in the reframe

itself. This one additionally moves a word sitting in a canonical hard line, two registry boss

titles, and a 218-line ambient field (§2.11 Ground 3, measured). *Recommended:* N-B in its

descriptive form. Note the artifact currently pre-empts this fork — HW_0001's designation

cell has the [Josh-name pending] marker stripped (§4.5, divergence 1); the marker should be

restored before the ruling.

3. **THE REVEAL WINDOW — NOT an open fork. It is YOUR OWN STEER, APPLIED, plus two additions to

confirm.** §2.16 criticizes the charter for asking Josh to re-ratify a position his 2026-07-13

FU.1 steer already replaced — and an earlier draft of this section then did exactly that by

routing sub-fork 3 as an unruled cardinal fork. Corrected: **the two-tier split (surface legible /

depth held to Ch 76-77) is already ruled and this brief simply applies it.** What is genuinely new

and wants his eye is narrow:

• Gate 1 — the extraction-scar locus does not exist before Ch 39. This closes a live reveal

gap (§2.19) and is reveal-discipline application, apparatus-contained.

• R-D — post-Ch-77 articulation of records already filed. Additive, costs the pre-reveal

window nothing.

• The six art rules — the one item with real weight, because HL_0044's art-direction channel

is where no existing gate reaches (§4.4). They are execution under CVD §17.9 rather than a new

rule, but they are time-sensitive against the art pipeline, so they are surfaced rather than

applied silently.

*Recommended framing for the ruling: "confirm the FU.1 split stands, and confirm Gate 1 + R-D +

the six art rules ride with it."*

Ruled together, not separately — §3.1 is the reason: the composition risk exists only in the

triple, and the mitigations interlock across forks. That the third item is a confirmation rather than

a fork does not change this; it still has to be read as part of the composed design.

Two sub-items inside the forks that are also his, surfaced so they are not smuggled:

• Whether the containment sub-codex is declared separate from the bestiary count (§4.1) — a

numerical-canon declaration.

• Whether the post-Ch-77 "wears HER face" beat is ever authored (§2.22) — **routed to its own ruling

with its own care read, deliberately NOT bundled here.**

5.2 Executes on a YES — no further ruling needed

• Reconcile all ten diverged HW_0001 cells to the designed spec and author HW_0002 with a

non-family-derived locus (§4.5), after the vril_density enum reconciliation.

• Populate reveal_tier with the three gates; populate sigma_trigger_anchor; restore the

firewall_anchor chain and the R17 care_register.

• UPDATED (integration record §4.2 and §7). reveal_tier takes the FIVE gates, not three, and

perception_gate takes the three-rung perception ladder. Every three-gate reference in this brief

now reads as the five-gate ladder.

• Retire "substrate-echo" with a supersession note in creature-system.md, WS7_WIRING_BUILD_PLAN.md

and the charter; leave docs/NAMING_METHOD.md and enrich.py untouched — the method keeps the

name.

• Author the four sharpenings as class-row properties: the lifetime cap **with its sigma-ordering

note (§2.6.1), the variant ban with the observation-locus row model** (§2.6.2), non-announcement

as scoped (§2.6.3 — the HL_0080 4-step protocol is still taught by the adjacent class from

Ch 10), and the L1-encounter / L3-meaning / L2-tributary layer assignment in that three-part

form (§2.6.4).

• RE-SCOPED (integration record §4.2). Author the lifetime cap and non-announcement as WHISPER-TIER

properties, and the five per-surface density budgets as the content tier's governors. The variant

ban and the layer assignment are authored class-wide as written, and the sigma-ordering note rides

the cap unchanged.

• Add the Environment_Grammar (EG_) locus-grammar row. **Do NOT add a T0_Realm_Ruleset_Registry

row** — route the negative-control question to the charter's Wave-2 schema mint instead (§2.3a).

• Implement the §4.2 teeth as gate assertions; enroll the six art rules in the concept-board review

checklist before any Hollow board is generated.

• Carry the HL_0080 riders verbatim into the class definition (§3.3).
• Write the śūnyatā / Ginnungagap / Duat firewall lines into the class definition (§4.3).
• Anti-orphan: a DOC_MAP row for T0_Hollow_Codex and a DESIGN_GAP_REGISTER row. *(Note: the

missing DOC_MAP row is a WS7-Increment-A-wide gap — positive-controlled: Rune 0, Dwelling 0,

Vehicle 0, Environment_Grammar 0, Minigame 1 — so it is a separate batch cleanup, not a finding

against the Hollow.)*

• Lens wiring: fold the dread/horror analysis into the proposed director-ai-emergent-dread lens

charter (COMPARATOR_LENS_PROGRAM cell 15, a named open gap) and add SOMA and Subnautica to its

title set — the proposed set (Alien: Isolation, RE4 2023, Shadow of Mordor) covers neither the

ontological-horror nor the void-dread axis, and both proved load-bearing. Dual-list Hollow Knight

(claimed at L-R7 for fading-court; owed for void/absence-entity) and Shadow of the Colossus

(claimed at L-R4 for pilgrimage; owed for emptiness-and-cost). Name the Hollow as a consumer of the

proposed knowledge-progression-decipherment lens.

• Coordinate with the natural-voice rewrite pass on the Ch-2 and Ch-60 boss epithets (§2.13).

5.3 Executes REGARDLESS of the ruling

The HW_0001 repair (§4.5). The chapter window, the region and the designation are asserted

without authority today. A NO makes the row's existence moot but does not make the breach acceptable

in the interim; a YES determines the repair values. Either way it is a defect, not a design question,

and it should be reported as such rather than folded into a ratification. **The

[Josh-name pending] marker restoration should go FIRST**, before the ruling, so the artifact is not

asserting an answer to the fork being ruled (§4.5).

The five stale Ch-74 hard-line rows — see the REPO-REPAIR NOTE appended at the end of this

document. Unrelated to the Hollow decision; surfaced here because this brief's §11.8/HL_0020 citation

work is what exposed them.

5.4 What this brief does NOT rule, and where those belong

• The per-region descriptive readings themselves — culture-bearing authorship with its own care

read; belongs in the natural-voice pass and the region-page substrate. **And any VERNACULAR term is

gated on region-page acquisition** — it cannot be authored today for most of the arc, and where no

source attests one it is never authored at all (§2.13.3a).

• Any specific chapter assignment — locked structural canon; Josh's, and the recommendation is

that none be set at all.

• Purify/harvest integrity magnitudes and sigma trigger parameters — Phase-5M tuning under §17.13.

*(The sigma/cap composition ORDER is not tuning and is recommended at §2.6.1; only the numbers are

deferred.)*

• **Whether T0_Realm_Ruleset_Registry gains a negative-derivation row class with an ATTENUATES

partition** — a charter-level schema question routed to the realm program's Wave-2 mint (§2.3a),

where V is authored. Not part of this admission.

• FORKS 7/8/9 of the FU.6 set.
• Whether systems/RULINGS.md L19 meant the NAME to ride the direction-ratification — a

flag-record ambiguity to disambiguate in the ruling itself (§1.5).

---

6. CONTRADICTIONS AND LIMITS — kept visible

1. Wave disagreement on the NAME (§2.15). The lens sweep recommended "the Unnumbered" + keeping

"substrate-echo"; both halves are defeated by the canon sweep's positive-controlled collision audit

(the Vimana "unnumbered" adjective in an ACTIVE T1 doc; the Substrate-Echo method wired into

enrich.py). The lens sweep did not run those greps. The audit wins, and the lens sweep's own

stated fallback *is* N-B — so the waves converge once the audit is applied.

2. **Wave disagreement on whether the charter's reveal line is STALE or merely UNDER-SPECIFIED

(§2.16). Both are true and neither wave was wrong: the HL_0044 floor** the charter cites is

correct and was deliberately left unchanged in the realignment; the **"pure atmosphere" window

label** is the superseded FORK-3(a) recommendation. The brief rules the FU.1 split and keeps the

floor.

3. Admission's strongest objection is not dissolved (§2.8). "It fits" is not "it is necessary."

Restated against the live rule: under the 2026-07-22 reframe this is not a do-not-invent question

but a new-ontological-KIND question, which is the residual class the reframe still routes to

Josh. The counter is real but partial.

4. The name's evocativeness objection is not answerable (§2.14). "The Hollow" is a better word than

anything the filter left standing. N-B pays a real cost.

• 4a — N-B moves a care obligation rather than removing one (§2.14, §2.13.3a). The descriptive

mechanism discharges it cheaply and by ratified precedent, but the obligation lands downstream in

the natural-voice pass and the region pages, where N-A's would not. Any reading of N-B's care

profile as strictly superior is overstated; what N-B removes structurally is the reveal risk and

the personification risk, and that is the claim that holds.

• 4b — the vernacular layer cannot be built today (§2.13.3a). Four region pages exist against

48 pending rows, and for the brief's own example cultures the void/drained-place concept is

unattested. Descriptive readings are buildable now; attested-vernacular naming is gated on

region-page acquisition and may, for some regions, never be available at all — which is a correct

outcome, not a gap.

5. The art-layer protection is a review obligation, not a gate (§4.4). No script reads images. This

is the recommendation's weakest enforcement point and it is time-sensitive against the realm

program's art pipeline.

6. A sharpening/rider tension inside the recommended triple (§3.3). "Never fought" (ontology) and

HL_0080's "action-forward, vril-taxing set-piece" (intensity) must be stated together or a

downstream author ships the quiet version.

CLOSED (Josh, ninth sitting 2026-07-27, JB-1; integration record §4.2). The tension is resolved and

this item is no longer open. The Hollow class is removed from HL_0080's non-damage protocol, so

ontology and intensity now agree: Hollows are fought and defeated, and the rider's language is

simply the register phase 2 is authored in. The sacred-site atmospheric echo keeps the protocol and

its six built instances.

7. A sequencing dependency (§4.5). The locus enum sits on an unreconciled vril_density

vocabulary; the class's locus authoring should follow that reconciliation.

8. A consumer dependency (§1.4). Admission adds a fourth consumer (the sealed sub-codex) to the

Personal Dimension's empty spec. Not a blocker; a scheduling note.

9. **The locus row collides with the realm charter and the collision was previously undisclosed

(§2.3a).** Resolved by taking the EG_ row and routing the ruleset question to the charter's own

schema mint — which is a real reduction in what option D offered (the harmony doctrine keeps no

negative control until that question is answered), and it is stated rather than absorbed.

10. A stale-citation defect surfaced under this brief and is filed separately — HL_0020 and four

sibling hard-line rows carry pre-v3 Ch-74 endgame numbering against the live Ch-76 spine. The

RULE they carry is unaffected and this brief's arguments do not depend on the number. See the

REPO-REPAIR NOTE below.

Method and limits. Canon-first; every canon claim traces to an artifact read across the three

waves, with game-lens readings analytical and carrying no citation burden. Positive controls were run

on every absence claim across all three waves (the Sigma-ghost isolation — 1 file; the

void-archetype zero — controlled against 12+ Duat hits; the DOC_MAP zero — controlled against

Minigame 1; the False Naga control — 67 hits, 30 files).

Counts, with their globs stated (an earlier draft carried two unglossed figures; both are

restated here as reproducible measurements, git grep over tracked files, counting **matching

lines** unless noted):

Both corrections run in the brief's own favour — the ambient field is larger than claimed, so

Ground 3 is stronger, not weaker. A count with no stated glob is not evidence, and neither of the

originals reproduced under any scoping tried; they are replaced rather than defended.

**The one claim previously flagged as taken on trust — T3_Creatures_Tameable §2.4's

binding-stone pool text, on which the entire 22-slot firewall rests — was READ DIRECTLY this session

and is verbatim as quoted (§4.1).** No claim in this brief now rests on an unverified quotation, and

the two evidence-tier defects an adversarial pass found (a spike-draft artifact cited beside ratified

rows at §2.11 Ground 2; a designed const described as a landed cell at §2.16) are corrected in place

with the tier marked.

This brief authors no canon and ratifies nothing. Every recommendation is proposal-tier and would

take the fresh-context critic plus the full gate suite before any apply.

---

APPENDIX — REPO-REPAIR NOTE (not part of the Hollow decision)

**Surfaced by this brief's citation work; belongs to the director's repair queue, not to Josh's

ruling. Nothing in the Hollow forks depends on it. Filed here only so it is not lost.**

**Defect: five canonical T0_Hard_Lines rows carry pre-v3 Ch-74 endgame numbering against the live

Ch-76 spine.**

Corroboration for Ch 76 being correct: the authority hierarchy (T1 outranks T0); CH_77.md L13

("prev_chapter_link CH_76 — The Two Courts, Cassius named and resolved") and L20 ("Cassius named

and resolved Ch 76 Phase 1"); and T0_Chapter_Index's own chapter titles.

Also carrying the stale number: the antagonist-pattern-reveal-timing memory entry. Worth

refreshing in the same pass, since it is read as standing direction.

Why it matters beyond tidiness. This is the stale-reference class the repo has been burned by

repeatedly — a wrong chapter number inside an ACTIVE canonical hard-line row is exactly what a

later agent reads as ground truth and propagates. And the rows are *internally* consistent with each

other, so nothing flags: only a cross-read against T1 and the live index exposes it.

Shape of the repair (director-owned, NOT part of any Hollow apply):

• Correct the chapter number in HL_0020, HL_0035, HL_0108, HL_0113, HL_0117 — number only; the

rules themselves are unchanged and must not be reworded in the same edit. **Check the

chapter_scope column as well as the rule text** — HL_0117 carries Ch74 there too, and a

text-only fix would leave the lookup key stale.

• HL_0035 is the one to think about rather than sed: "Ch 74 region name: Velheim" against a live

CH_74 that is the Norse/Yggdrasil node. The correction is a chapter number, but confirm against

T0_Region_Index which node the Velheim region name actually belongs to before writing it.

• Sweep for further Ch-74 endgame references before applying, so the fix is a class-closure rather

than five point-edits (the "verify with fan-out, not commit labels" lesson).

• Registry edit → registry_fidelity.py --emit-baseline in the same commit, or the gates go red.
• Fresh-context critic on the diff, per the standing rule that a registry/canon edit is never

"trivial mechanical" for review purposes even when the transform is.

• Refresh the antagonist-pattern-reveal-timing memory entry to Ch 76.

This note is informational. It does not ask Josh for anything.

Generated by harness/site/structure_site.py — the URL path is the repo path. review root