RULING_QUEUE_SWEEP.md

decisions/RULING_QUEUE_SWEEP.md

Ruling-queue sweep — the whole-queue picture (2026-07-17)

▶ JOSH'S RULINGS (2026-07-17) — ALL FOUR CARDINALS RULED: "I agree with your recommendations"

• C1 — ALTITUDE→VRIL: Option C. RULED. Vril density keys off realm and site, never off elevation or

latitude. The recharge-source network already canon at T1_Vril §5 (waterfalls, obelisks, pyramids, tombs)

IS the vril geography, and it is a geography of built works and water. High altitude and the poles are

a body hazard only: exposure_hazard=altitude bites through Stamina/Health at Ch 52 (Uyuni) and

Ch 56-58 (polar) with no density change — the air is trying to kill you, and the *site* is where the

power is. **This is now a vril-physics law: the corpus has NO purely-geophysical vril law, and does not

acquire one.**

> **THE LAW, in the form a future author should carry it: "The mountain is magically ordinary; the temple

> is not."**

>

> That sentence is the ruling's operative statement — recorded verbatim at the region-page lane's request

> (Fable, 2026-07-17: "it's the sentence that tells a future page-author what to do"). The worked example

> is Besakih: it carries vril_density = moderate because its SITE row says so, not because it

> sits high on Agung. Elevation is not an input. Latitude is not an input. What was built there is.

>

> Cross-lane convergence — this is why the ruling is trustworthy. C1 turns out to RATIFY existing

> practice rather than impose a new law: the region-page lane's Zone Catalog already consumes

> vril_density per-site verbatim from the spine's §3b rows, never derived from terrain or altitude, and

> flores_island + bali already assert no geophysical vril law. Two lanes reached the same firewall

> clause independently, without coordinating. That is worth as much as the clause itself. The firewall consequence is the point: density must never track latitude/altitude, which is

exactly the polar-energy geography the Bulwer-Lytton→Nazi-occult "Vril Society" myth traffics in, and

Ch 56-58 + Ch 67 Hyperborea is where that firewall is thinnest. Deadzones remain a player tool — they key

off site and realm rather than elevation, which is more authorable and less exploitable. Option A

(thin-vril-with-thin-air) is DECLINED and must not return as "tuning".

• **C2 — TERRAIN ALTERATION: shape RATIFIED as ruled; three of the four values → Fable-5 tuning; the fourth

is RULED as thematic.** Ratified shape (RULINGS.md L40): intention-remembers / incidental-forgets; three

diegetic refusal registers; sacred/protected/Ch-57 HARD-blocked; non-sacred community ground defaceable

only at an Evil-band cost the community remembers; critical terrain self-heals so nothing gates; Good

restores / Evil scars. To tuning: the soft-respect override ceiling, the concealment-override punishment,

the per-biome persistence/decay budgets. **RULED, not tuned — "do Evil-path scars ever heal on a Good

turn": YES, but slowly and NEVER completely.** A scar fades to a mark; the ground remembers. That is not a

number, it is a statement about whether the world forgives, and it is the integrity-ripple doctrine.

Binding constraint on the tuning (so it cannot erode the care line): the override cost scales with

identity and is capped such that no amount of investment buys immunity at a sacred node — those stay

HARD-blocked, uncapped, never overridable. Same shape as the D-REL-BANDS 90-cap invariant.

• C3 — D-20 ATACAMA: ratify (c) as-built. RULED. Preserves every canonical beat at zero canon cost;

(a)/(b) would have moved chapter-region canon. On this ratification the WGM/RCM wording tightens to

"Central Andes / Zona Central 33.33°S" with the Giant tagged "[Tarapacá ~19.95°S, read remotely]" — the

tighten RIDES the ratification precisely because a leg labelled "Atacama" that is mostly not the Atacama is

a hallucination seed for every future agent (the living-source rule's exact failure mode). "Atacama

Continued" survives ONLY as the registry arc-leg label; the geography reads true everywhere it matters.

• **C4 — CVD Component 5 anchors line: DECLINE, no CVD edit. RULED — and this one supersedes an earlier

assent, deliberately and on the record.** On 2026-07-08 Josh said "Agreed with everything else", which

included adding "Ch 18 Levant" to Component 5's anchors; Claude then withheld that edit on a **false

premise** (it claimed Component 5 had no printed anchors line — CVD L88 prints it). Surfaced explicitly at

ratification rather than laundered: Josh was told the record showed he had said yes once, and ruled DECLINE

on the corrected reasoning. The line is non-exhaustive in BOTH directions — it omits six substrate items

and lists Ch 15 Egypt with no backing substrate item at all. It is illustrative, not a mapping; the

"throughout" clause names the family operating within the pattern, which IS the Ch 18 Tyrian Cartel.

Nothing downstream is starved. Adding Ch 18 alone while five peers stay unlisted would author an arbitrary

asymmetry into cardinal CVD §3. The edit stays available forever if that ever changes.

Application status: these rulings are RECORDED here (this doc is not in the RAG canon scope, so it lands

independently). The canon-side applies — the C1 vril-physics law into T1_Vril + the exposure_hazard

semantics, the C3 WGM/RCM wording tighten, and the C2/C4 records into docs/proposals/systems/RULINGS.md

— all touch _source T1 docs and/or the canon scope, so they need `scan_migration_integrity.py

--emit-baseline` + a canon rebuild in a green window. The concurrent lane's live region-page work has

fidelity/scan/index_freshness red at time of ruling; the applies land in the next window.

Target: docs/spine/DECISIONS_PENDING_JOSH.md (459 lines, 32 ## headers).

Method: content-checked, not label-trusted. Every claim below is grounded in a file+line read this

session. 22 citations checked directly this session; all 22 resolved and said what is claimed. The

inherited D-09 evidence was independently re-verified at 15/15 by an adversarial pass; I re-confirmed 8

of those myself (CVD L86/L88/L108/L118, the two grep counts, CH_18 L58/92/257/303).

---

1. HEADLINE

The queue is 94% dead. It is stale by construction: it accreted through the R-17 recalibration, the

WS7 wiring, the whole-arc re-cascade, the Fable-5 v3 rewrite of all 79 nodes, and five ruling batches

(07-08 · 07-09 · 07-10 · 07-12 · 07-13 · 07-14 · 07-15 · 07-16), and nothing ever swept it. Several

entries are duplicates of rulings recorded in the same file, 300 lines above them (D-22, D-23).

Of the 2 live entries, D-WS7-FLAGS shrinks from 5 self-declared cardinals to 3 — I killed two by

measurement (see §5). Net: you rule 4 things, 3 of them in a sentence each.

---

2. THE RETIREMENT LIST — skim and veto

Each line: the kill evidence. If a line looks wrong, say so and I re-open that one only.

Ruled 2026-07-08 (verbatim: "For D8, yeah agreed … Agreed with everything else")

Ruled 2026-07-09 ("other than the naming deferrals I agree with your recommendations")

Ruled 2026-07-10

Ruled 2026-07-12 — recorded at queue L46-L47, i.e. 300 lines ABOVE the stale entries

Ruled / superseded 2026-07-13 → 2026-07-16

Also retirable — the L9 banner (not a ## entry): *"NEW CARDINAL … CF-2 the childhood best-friend re-met as a PERSON at Ch 70+ … NEEDS the full decision-analysis brief"*. Dead. CF-2 is CONFIRMED and BUILT — CH_75 L317: "Q14 — the CF-2 childhood-best-friend reunion — APPLIED (per Josh's directive)", ALLY-SIDE reunion at the homecoming, reveal-orthogonal, no-dead-end. Residue = the name only (§3 A9).

---

3. THE CREATIVE-WITHIN-VISION SET — I apply these under your standing acceptance

No ruling needed. Listed so you can veto any line. Nothing here authors canon; each is a correction of a

false statement *about* canon, a stale-reference cleanup, or a derivation from authoritative canon.

---

4. THE CARDINAL SET — 4 items, 3 are one-liners

Ordered cheapest-first. C1 is the only one needing real thought.

---

C1 — Does high altitude thin the vril? *(system canon — the only substantive one)*

Plain language. WS7 gave every region an exposure_hazard column. Mountains and poles obviously

hurt the *body*. The open question is whether they also thin the *magic*: at 4,000 m on the Uyuni salt

flat, or on the polar ice, is there less vril in the air — or is the air just trying to kill you while

the vril is unchanged? This governs the environment rows for Ch 52 (Uyuni) and Ch 56-58 (polar).

Why it is yours. Canon is genuinely silent. T1_Vril L17 says *"Vril is atmospherically variable"* —

but every example it gives is realm-scoped (fairy realm highest, human world lowest, threshold realms

Ch 66-69 varied). It never says altitude. Answering this establishes a new vril-physics law. RULINGS.md

L41 already lists "altitude→vril question" as a Josh-final-yes.

RECOMMENDATION: C. It gives you A's diegetic payoff without A's liability. Reasoning: the

recharge-source network at T1_Vril §5 already *is* a vril geography, and it is entirely

constructed/site-based — waterfalls, obelisks, pyramids, tombs. That is a geography of **works and

water**, not of elevation or pole. Making density track altitude would introduce the first

purely-geophysical vril law in the corpus, and it would introduce it precisely at the polar chapters,

where the standing vril-vs-Vril-Society firewall is load-bearing ("vril reads mathematical/physical/

substrate, never race-mystical polar energy"). C keeps the reading you actually want at Uyuni and the

poles — the air is trying to kill you, and the *site* is where the power is — with zero firewall exposure.

Strongest objection. C is A with extra steps, and it gives up something real: under A, a deadzone is a

place you can *walk to* — climb high and your pure-vril pursuer degrades harder than you do. That is a

genuinely excellent, physics-legible player tool, and it is exactly the "rewards intelligence" pillar.

C makes deadzones authored rather than navigable. My answer: the deadzone tool survives under C — it

just keys off site and realm rather than elevation, which is *more* authorable and less exploitable. But

if you want the mountain itself to be a magic-thin place, rule A and I will wire the firewall explicitly

(density as a function of built works and realm-proximity, never of latitude, with an authorial flag at

Ch 56-58/67). The column holds either ruling — nothing is invented pending this.

---

C2 — The terrain-alteration final-yes *(confirm; the shape is already ruled)*

Plain language. You already ruled the *shape* (RULINGS.md L40, 2026-07-13, ratified-in-direction):

intention-remembers / incidental-forgets; three diegetic refusal registers; **sacred/protected/Ch-57

HARD-blocked; non-sacred community ground defaceable only at an Evil-band cost the community

remembers**; critical terrain self-heals so nothing gates; Good restores / Evil scars. The §17.1 call is

therefore already made. What is left is four numbers-and-ceilings the doc flags as "Josh-final-yes":

the soft-respect override ceiling, the concealment-override punishment, whether **Evil scars ever

heal on a Good turn, and the per-biome persistence/decay budgets**.

RECOMMENDATION: ratify the shape as ruled; send all four values to Fable-5 as tuning. Reasoning:

the care line is the canon part and you set it; a decay timer is not canon, it is feel, and feel is

playtest's job — the same disposition D-REL-BANDS took (shape ratified, numbers tunable). **One I would

pull back from tuning:** *do Evil-path scars ever heal on a Good turn* is not a number — it is a

statement about whether the world forgives, which is thematic. **My rec: yes, but slowly and never

completely** — a scar fades to a mark; the ground remembers, which is the whole integrity-ripple

doctrine. Strongest objection: ratifying an override *ceiling* sight-unseen is how "defaceable at a

cost" quietly becomes "defaceable"; if the ceiling is set too cheap, the §17.1 respect register is

decorative. My answer: bind it to the ratified invariant shape from D-REL-BANDS — the cost scales

with identity and is capped so that no amount of investment buys immunity at a *sacred* node (those stay

HARD-blocked, uncapped, never overridable). Then the tuning cannot erode the care line.

---

C3 — D-20 Atacama: ratify (c) as-built *(one line)*

Plain language. The Ch 53-55 "Atacama" arc label spans places up to ~2,000 km apart. Resolution (c)

is already applied and changes no canon: the prose is geographically precise (Ch 54's installation is

the high Central Andes at 33.33°S — Tupungato, the Cristo Redentor pass), "Atacama Continued" survives

only as the registry arc-leg label, the Tarapacá Giant is read astrally (the Ch-51 Nazca precedent,

not a 2,000 km overflight), and the 33rd-parallel numerology is the family's imposed obsession with

the real-world corrective held authorial, never a player-facing debunk. Options (a)/(b) would *move

chapter-region canon*, which is yours. Independently corroborated: NEW_SESSION_HANDOFF.md L191 —

"D-19/D-20/D-21 resolved-applied"; OPEN_DECISIONS_DOSSIER.md classes it a low-stakes confirm.

RECOMMENDATION: ratify (c). Preserves every canonical beat at zero canon cost; (a)/(b) move ground

you set. On ratification I tighten the WGM/RCM wording to "Central Andes / Zona Central 33.33°S" with the

Giant tagged "[Tarapacá ~19.95°S, read remotely]" — held for your word today because it is a geographic

label. Strongest objection: keeping a leg labelled "Atacama" that is mostly not the Atacama is a

hallucination seed for every future agent — the living-source rule's exact failure mode. My answer:

that is why the wording tighten rides the ratification; the label survives only as an arc-leg name, and

the geography reads true everywhere it matters.

---

C4 — *(optional, low priority)* Should CVD Component 5's anchors line be completed?

Plain language. Component 5's anchors line names 4 chapters against 9 substrate items, and closes

with "throughout". Six substrate items have no listed chapter (Phoenician monopoly → Ch 18; Nalanda →

Ch 16; Nazi book-burnings; Cultural Revolution; apocryphal/gnostic suppression; colonial destruction).

Surfaced because the 07-08 brief recommended adding "Ch 18 Levant" and you said "Agreed with everything

else" — then Claude withheld the add on a false premise (§5).

RECOMMENDATION: decline — no CVD edit. Reasoning: the line is non-exhaustive **in both

directions** — it omits 6 substrate items *and* it lists **Ch 15 Egypt with no backing substrate item at

all**. It is illustrative, not a mapping; the "throughout" clause names the Velheim family operating

within the pattern, which *is* the Ch 18 Tyrian Cartel. Nothing downstream is starved. **Strongest

objection:** you approved an edit that was never made — retiring it launders a mis-read into a ruling.

My answer: honest, and recorded rather than buried — but adding Ch 18 alone while five peers stay

unlisted would author an arbitrary asymmetry into cardinal CVD §3 on a 9-day-old blanket assent that the

same record explicitly closed. Nothing built has to move either way; the edit stays available forever.

---

5. WHAT THE ADVERSARIAL PASSES REFUTED — and what it changes

Five refutations. Two of them shrink your ruling load.

1. REFUTED — the recorded reason for closing D-09 is false. The 07-08 record (queue L26/L193, applied

at CH_18 L303) says Component 5 has *"NO printed chapter-anchor line (only Component 1 at L71 has

one)"*. CVD L88 prints it. I reproduced the mechanism rather than asserting it: a line-initial

grep for Canonical chapter anchors: returns 1 (Component 1, L72) — because Component 1 is the

only component whose anchors sit on their own physical line; Components 2-10 bury theirs mid-paragraph.

Anywhere-in-line: 10. That single formatting quirk generates the brief's exact sentence.

Changes: the ruled *outcome* is right and stays (Ch 18 carries 5/10/11; no CVD edit) — the *reason*

gets replaced (§3 A1), and the anchors-line question is logged honestly as C4 rather than quietly

resolved.

2. REFUTED — "the CVD was almost certainly never changed" (overstated). The queue cites "§3 L85-87"

and "Component 1 at L71"; live is L86-88 and L72. The CVD did shift — by exactly 1 line. But

this *corroborates* the mis-read: the brief's own "L85-87 = header + Claim" maps precisely onto live

L86/L87/L88 under a 1-line shift — it saw the structure, correctly called it "header + Claim", and

never read to the end of the run-on Claim paragraph. Nine added anchor lines would shift ~9 lines

and would sit line-initial. Changes: wording only — "shifted 1 line; the brief mis-scanned."

3. NEW — WS7's own cardinal #2 is a WRONGLY-FLAGGED cardinal. The entry asks you to rule "the

resource-bar model: ~4 universal bars (Health/Vril/Stamina/Rage) vs 5 archetype-keyed systems".

This is already canon, and the answer is neither option — it is the synthesis.

T1_Combat_System_Spec [ACTIVE v1.0] L41: *"The four canonical resource types are HEALTH, VRIL,

STAMINA, and RAGE. Health and Vril are always present. Rage and Stamina share a single

archetype-swapped slot… RUNES and AMMUNITION are archetype and weapon overlays, not always-on bars."*

Mirrored at HL_0074, severity=canonical. Changes: cardinal #2 → retired (§3 A2); the

exposure_hazard drain binds to the canonical four with no ruling from you. Bonus catch:

CANON_CORRECTION_CHANGESET.md L419's J13 "issue" — *"HL_0074 lists a '5 systems' resource economy…

inconsistent with the corrected four-bar model"* — is itself stale; the live row already carries

the corrected model. **Per the standing rule, I still surface it to you rather than silently retiring

a flag that called itself cardinal.**

4. NEW — WS7 cardinal #5 is closed. The entry flags "the mastery/Trade_Registry tier-id renumber

(mastery Option D)" as needing you. You ruled it as F4 on 2026-07-14 and it landed in the CVD —

L663-666 prints the two-index disambiguation (10 grades over 12 levels; Adeptus Minor = grade 6 = level 6 — the gate index RETIRED 2026-07-28;

the registry's 120-quest roster keyed on the LEVEL index, 120 = 12 levels × 10 trades). Changes: WS7 drops from 5

self-declared cardinals to 3 — and with #2 also dead, to 2 live + 1 confirm (C1, C2).

5. REFUTED — two premise defects in the stale D-23 entry. (a) It asserts *"T3_Core marks the

healer's-child arc closure audit-pending"* — false; the string "audit" does not occur anywhere in

T3_Core_Characters; the marker is RCM-side only. (b) It cites "RCM Ch 65 L6056"; the actual line is

L6069. Changes: nothing substantive — D-23 was already ruled 2026-07-12 — but it is a clean

demonstration of why this queue's entries cannot be trusted on their own say-so.

Meta. Unresolving citations have been the dominant defect in this lane across five rounds. Every

citation in this document was programmatically checked to resolve and to contain the claimed text. The

two things I did not independently verify: the docs/proposals/systems/environment-*.md per-fork

detail behind C2 (I read the ruling summary at RULINGS.md L40, not all six sub-forks), and the F3

5-rung application state across Ch 3/7/23/24/28/51/63/Epilogue (listed as A10 to confirm, not asserted).

Generated by harness/site/structure_site.py — the URL path is the repo path. review root