music/COMPOSER_BRIEF_PACK.md
CANON SUBORDINATION — this document is PROPOSAL-TIER: it serves canon and never outranks it.
The canon: the CVD · the T1 foundation docs · the T0 registries (registries/) · the spine
(docs/spine/CH_*.md) · the region pages (_source/02_Tier_2_Region_Pages/). Authority order:docs/DOC_MAP.md§ 0.
Canon served (scanned from this document's own citations — widen it by hand where it is thin):
T0_Thread_GridREAD THAT CANON FIRST — open it and derive from it before you build anything from this document.
If this document disagrees with canon, CANON WINS and this document is the defect — fix the
document, never the canon. Nothing here is applied until it is ratified into canon.
RE-TARGETED 2026-08-05 by Josh's ruling — *"Humans aren't composing the music… what the fuck. The
entire music pipeline is yours."* Composer-in-loop is retired. **This pack is now the generation
pipeline's SPEC INPUT**: the constraint set every candidate is generated against and measured
against, read by harness/music_gen/ rather than by a person at a piano. Nothing in the briefs
themselves changed — a constraint that was worth handing a composer is worth holding a generator to,
and the specificity that made these briefs usable is exactly what makes them machine-checkable.
Where a line below says "the composer", read "the generation lane"; where it says "returned per
theme", read "emitted per theme into the run record".
It contains no melodies, and that is deliberate. ~~The standing ruling reserves melody for the
composer:~~ **The pack specifies constraints and never prescribes tunes, which is now load-bearing
in a different way: the generator proposes the melodies and the brief is the thing that judges
them.** A brief that contained the melody could not score a candidate. Every brief below therefore
specifies the *shape a melody must satisfy* — head length, contour class, the single deviant, the
transformation budget, the payoff node, and the register it must be written inside from the first
sketch. Nothing here dresses a Western tune in a local costume afterwards; that inversion is the
named failure, and it is the failure a generator commits most easily.
The bar, stated once and binding on every brief in this pack: if it cannot be hummed after one
hearing, it is not the theme yet. It iterates until it passes, and it is judged with evidence —
~~a sing-back capture~~ a measured score trail plus a critic listen-check on captured playback,
not a description. **The sing-back gate (doctrine B1) is not automatable and is not claimed as
passed by this lane** — the rig scores the computable diagnostics and declares that limit out loud;
B1, D1, D2 and D8 stay open on every artifact until a listener runs them. The only exceptions are
the two identities that are not tunes, and each says so in its own brief and names the test it
takes instead.
**The honest tier for everything the pipeline emits against this pack is GENERATED SCORE —
ITERATION.** Not "final", not "the theme". The pack is the bar; the run record is the evidence; the
tier is the honesty.
Cohort one is drafted together rather than one theme at a time, because combinability is
evaluated at cohort close and because the family cells must exist before anything can inherit them:
the protagonist, the antagonist pattern, the familiar bond, home and loss, the journey — plus the
family heads for the slice registers.
---
The job. One theme, evolving across seventy-nine nodes, that is recognisably the same tune at
the Prologue and at Ch 77 and that makes the player feel the distance between them. This is the
single-theme arc, and it is the spine of the score.
The emotional arc across its recurrences, read off the chapter register census rather than
invented: it is stated in *wonder* at the Prologue and in *tenderness* through Ch 1, and the first
thing that happens to it is that both are taken away — Ch 2 is *grief*, on a fall-then-rise valence.
Then it spends the slice being re-earned in registers that keep changing underneath it: *unease* at
Ch 4, *dread* at Ch 5 and Ch 8, *awe* at Ch 6 and Ch 10, *fury* at Ch 12, and back to *grief* at Ch
13 where the child gives away the only protection they have ever had. The tune should be able to
carry all of that without being rewritten, which is what dictates its shape.
Shape. Four notes, purely stepwise, singable inside a ninth. Stepwise rather than
triad-outlining is a deliberate choice against the alternative, and the reason is structural: this
arc leans far more on fragile solo statements than on statements that must cut through a loud
orchestra, and the law forces one or the other, never both. One deviant only — a borrowed degree
on the third note. Two deviants and it stops being hummable.
Withhold closure. Write the Ch-77 resolution first, then delete it from every earlier statement.
Every statement before Ch 77 ends open. The authentic cadence on the tonic is reserved for one node
in the whole game and this is the theme that spends it — three times, once per integrity band, with
no shared arrival gesture between them.
Plant it diegetically. A motif the player physically inputs is retained far better than one
merely heard. Plant the head cell in at least one surface the player performs — an instrument
minigame, a chant, a puzzle lock, an NPC's whistle — inside the first three nodes it appears in. The
diegetic surface carries the head cell at whatever length that surface allows; there is no required
three-plus-three shape.
Deny. It shares no audible material with the Grand Sage reveal before Ch 77. That kinship is the
highest-risk shared material in the score, and every statement before Ch 76 takes the reveal check
with a fresh-context critic and an archived capture.
---
The job. This is the hardest brief in the pack and the most important to get right, because it
is the only one whose payoff depends on the player having heard it *without knowing they were
hearing it*.
The parent motif is gated at Ch 38. Before then, only its variant sounds — the head cell,
fragmented, once per node, buried at texture level. Thread 3 is live from Ch 2 at its pre-family
scales: an individual bargain at Ch 2, then the institutional forms at the rubber station, the coast
factory and the mining superintendent's works. Each of those is a real local antagonist and none of
them is named as anything larger.
Shape. Three notes, triad-outlining — it has to survive being buried under a market bed as
well as standing over a loud orchestra. One deviant: a metric oddity. The cell enters a beat
early, every time. That is what makes it recognisable when nothing names it, and it is the whole
mechanism: the listener remembers a *shape*, not a sound.
A different carrier every time. The station's machinery, the court's ceremonial percussion, a
single reed in a market — never the same voice twice before Ch 38. If the listener starts recognising
an instrument, they will start asking what it is, and the recognition arrives thirty-six chapters
early.
One operation only, before the gate: fragmentation. Not augmentation, not combination, not a
mode shift. Each of those announces that the score has an opinion about what it is showing, and the
score is not permitted to have one yet.
What Ch 38 buys. The parent states the full phrase for the first time, combined with the
region's own bed. That combination *is* the reveal, and it states the relation with no dialogue.
What paid for it, counted rather than asserted, because this is the transformation budget.
Thread 3 is present in thirty-five chapters before Ch 38, so at one statement per node that is
thirty-five fragments, and thirty-five is the number to budget transformations against.
Eleven of them fall inside the Prologue→Ch 13 slice — Ch 2 through Ch 9, then Ch 11, Ch 12 and
Ch 13, with Ch 10 carrying no Thread-3 presence — and eleven is what cohort one is actually
commissioning. Both numbers are read off T0_Thread_Grid's Thread-3 rows rather than estimated.
Budgeting the whole pre-gate span as eleven under-counts it roughly threefold, which would spend the
fragmentation vocabulary far too early.
Deny, and this one is load-bearing. The pattern motif is the one identity in the score that
carries the darkening operations, and it carries them on itself. It may never be voiced in the
register of the region it is operating inside. Scoring the broker in the local ensemble makes the
culture the villain, which is exactly the collective failure the care doctrine forbids. The
corollary points the other way and is equally binding: an individual antagonist may be scored fully
inside his own culture's register, with the same craft as that region's protagonist cues. A blanket
register ban is the named failure, not the safe option.
No cue name, track title, subtitle, achievement or soundtrack listing may carry a held token before
Ch 38. This is enforced mechanically rather than promised: harness/emit_music_cues.py refuses to
emit such a cue, and a must-fire fixture proves the refusal fires.
---
The job. The bond is the warmest thing in the game and it has to read across twenty-two species
slots in the base game and thirty-three at the post-game Epilogue, each in the register of the region
that species belongs to.
Shape. Three notes, triad-outlining so it reads under twenty-two different carriers. **One
deviant: an unusual leap** — an ascending sixth into the third note. Every per-species variant keeps
that sixth; everything else about the voicing changes. The tune does not change, the voice does.
The trap, and it is easy to fall into. The player believes the Child Familiar died at Ch 2. The
truth gates to Ch 58. No grief colour may touch this material before Ch 58 — no minor inflection
at a memory, no thinned voicing at a mention, nothing. A mourning register applied early tells the
player the score knows something they do not, and the score does not get to know things first.
The one combination it is allowed — with the protagonist carrier, contrapuntally — is reserved
for the Ch-58 reunion. Motif combination is the most expensive event in the catalogue and it is spent
here to state a relationship with no dialogue.
---
The job. Establish, at the very start of the game, a home that is whole — and establish it as an
unqualified good before it is ever a wound. Seventy-five chapters climb back toward this
register. If it is scored as fragile, the climb has nothing to want.
The emotional arc across its recurrences, and these are real beats rather than a plan: the
amulet passes from the mother's throat to the newborn at CHPRO_B07; the origin closes at
CHPRO_B16; it comes off the neck in the sun on the last morning at CH01_B13; it crosses the seam
with the child at CH01_B08; it is recovered out of the fall-strewn earth at CH02_B11; and it is
given away to the healer's child at the Ch-13 departure, CH13_B13. Full and luminous and unhurried
at the origin; reduced to near-silence at the mother's scene; **returning to the abundance register
intact and unironised at the exit**, because the plenitude is real and the score must never
retroactively mourn it.
Shape. Five notes, purely stepwise, chamber scale. One deviant: a metric oddity — the phrase
begins on the second beat and never resolves onto a downbeat. That is precisely why it can be stated
at the origin carrying no weight at all and still ache at Ch 13.
Deny, and this is the strictest list in the pack. No prophetic or foreshadow cue anywhere in the
origin: no dread swell under the chase, no held note as the ward comes off the neck, no minor turn
before the rift. The rift-fall has no author, no aim, no omen and no curse, and music is a
player-facing surface of that hard line. It shares no material with the antagonist motif before that
motif's own gate — a loss cue that reads as *aimed* at the royal house leaks the pattern
thirty-eight chapters early, and the mother's death has no author and no intent. It is declared
independent of the protagonist head cell, because that motif carries a gated kinship and sharing
material would transmit it into the earliest statement in the game. No sung, chanted or spoken text
may name the mother, who is written by relation only, or the ward, which is uninscribed and
unidentified before Ch 8.
The one cadence the slice is allowed is the closure change at CH13_B13, when the object is
given away.
---
The job. The overworld and travel identity, re-scored into each region's own register as the
journey crosses it, and still recognisably the same tune. The precedent is a theme returning not as
a variation but as a full re-arrangement in a different civilisational voice — proof that this is
possible, and the mechanism that makes a seventy-nine-node world tour survivable.
Shape. Four notes, triad-outlining, because it must survive re-scoring into ensembles that
share no tuning system with each other. One deviant: an unusual leap. Its head-motif stinger is
composed to each region's tonic and its relative only.
Interchangeable, never additive. Layers swap rather than stack, so the melody is present at every
intensity and no state is ever left without it. The transformation budget is spent entirely on
orchestration migration, register substitution and reharmonisation under a fixed melody — never on
new material.
Deny, hardest of all. This is the motif that visits every region, so the collective-protection
rule binds it harder than anything else in the score: it may never become the recurring cue for
threat in any region it enters, and no region's register may be systematically darkened under it. A
generic world-music wash is a care failure, not a taste note.
---
The job. She follows the route unsigned from Ch 3 and says nothing. At CH13_B01 she steps
forward, signs her first note *Ntala*, and opens the evidence journal — the chapter's quiet centre of
gravity, and the first time the protagonist has a witness who is also a keeper.
Shape. Three notes, purely stepwise, and deliberately small: a watcher's motif has to survive
being almost inaudible. One deviant: a borrowed degree on the last note.
The head cell alone, for every watcher appearance from Ch 3 to Ch 12. One statement per node, at
the edge of audibility. The full phrase is first stated when she signs her name. Fragmentation is
normally the foreshadow operation and using it for ten chapters would ordinarily be a reveal problem
— it is not one here, because nothing is held. She is really there and the player really can see
her. The fragments are doing *presence*, and what they buy is that Ch 13 lands as recognition rather
than introduction.
Carrier. A second solo voice, chosen to duet with the protagonist's without doubling it.
Their combination is reserved for the Ch-66 Sealed Letter close.
Deny. It shares no audible material with the Ch-1 childhood cue. The spine records a
designer-only resonance between the two characters that is invisible on first play and never
surfaced; a shared motif would surface it, which is the one thing it may not do.
---
Each card below is the documented instrumentation of that region, taken from the region page's
own Section 10. Nothing is added. Where a page says its evidence is thin, the card says so too — the
gap is left rather than filled with a plausible sentence, because a plausible sentence is how a
generic wash gets in.
Reference-composition into the register is what is being asked for. Sampling a consecrated
performance never is, and no fine-tune is trained on one.
Every card carries its own page's prohibition, in that page's own words. The generic preamble
above is not the guardrail — the guardrail is per-tradition and it is named, because the failure it
prevents is specific every time. A composer who is handed a card naming an instrument, and no line
saying that this particular instrument is a hard care line rather than an asset, will reasonably
score it. So the May not line below is copied from the page's own Section 10 on the same
nothing-is-added rule the instrumentation follows, and the care doctrine's own exemplar is the
precedent: Ch 9 excludes Yaredawi zema *by name*, not by a general sentence about sacred music.
Flores (Ch 2–3). The gong-waning, the tuned gong-and-drum ensemble, central to the caci
ceremony and canonized as the region's Thread-21 vril-active instrument — the first demonstration in
the arc of an instrument creating audible vril resonance. The theory to compose from is the
interlocking-cycle structure and the ceremonial drumming that scores the caci rounds. Ch 2 is a
survival-horror-opener register, low and tense, rising to the crater-lake climax; Ch 3 is
highland-ceremonial, rising at the caci duel. *The page flags its own substrate as thin on the caci
contest detail and the gong-waning theory — treat this card as provisional until that acquisition
lands.* May not: no reproduction of protected sacred music — the gong-waning register is
reference-composed, never sampled from consecrated performance.
Bali (Ch 4). The one region where music is not scoring but structure. The gamelan — the
interlocking bronze orchestra whose kotekan figuration is the chapter's musical architecture —
plus the gender wayang of the wayang accompaniment. Thread 21 reaches confrontation here and sorcery
music theory is the chapter's lead questline. Every acoustic property the page describes is a real,
documented property of the ensemble and each is convertible into something the player perceives,
learns and eventually performs. Compose this one to be read, not just heard. May not: no
reproduction of protected sacred music — the gamelan register is reference-composed, never sampled
from consecrated performance, and the consecrated repertoire is rendered with weight and is never
turned into a combat jingle; the rites the player may witness, the matatah filing, the ngaben
cremation and the temple festival, carry their own music at reference register and are never scored
as spectacle.
Sumatra and Java (Ch 5). A voice vocabulary rather than an ensemble: sung lament, chanted
spell, arranged prayer, narrative riddle, told story, in a society where orality rather than writing
was how knowledge was stored and moved. The page states plainly that its manuscript sources carry
almost no instrumental ethnography, and the gondang bed is carried as ambience rather than as a
system. Build the region's sound around speech and song. May not: no reproduction of protected
sacred music, and the sharper rule this region actually needs — the pustaha's ritual content is
referenced, never reproduced; the gondang is reference-composed, never sampled from consecrated
performance; and the bius ceremony, the porbuhitan and the papurpur sepata funeral are scored at
reference register and never as spectacle.
Angkor and Champa (Ch 5 excursion). The pinpeat-class ceremonial ensemble and the temple
bells and conches of Mahayana devotion; on the Cham leg the ginang drum and saranai flute of
the living Kate rite at Po Klong Garai. The theory is the interlocking ceremonial cycle of the temple
ensemble. May not: both registers are carried at reference register with no reproduction of
protected sacred music, and the Cham sanctuary register for My Son and the Po Nagar pilgrimage is
grounded in the living ceremonial forms and never sampled from consecrated performance.
Sri Lanka (Ch 6). Chanting is the dominant register, and the record gives it a **procedure rather
than an atmosphere**: performed with doors and windows shut, the reciter ringed by armed people,
begun with thoughts of love foremost, escalating to the courtyard of a stupa if the affliction
persists, closed by an address to the deities. Separately: an all-night voice in a lamplit hall
with a crowd that comes and goes — the page treats that as a distinctive audio bed and so should the
score. The healing rite adds drums and pipes and dancing. May not: reference-composed, never
sampled from consecrated performance, and the protective chant is carried at reference register with
no reproduction of liturgy; the Vedda forest register is breath and single struck notes and is never
scored as exotica.
Tamil Nadu and Kerala (Ch 7). The living instrument of the kalari is the master's voice — the
vayttari, the verbal commands recited as students exercise, written into his own texts and
described as iconically written into his body. A temple percussion orchestra at a kalari's
inauguration. On the Tamil side, the Vedic chanting register and the devotional-verse tradition, at
reference register. The page reproduces no formula and no chant text, and neither does the score.
May not: protected sacred music is referenced, never sampled; the page reproduces no formula, no
chant text and no consecrated performance; and the command cadence is composed at reference register
from its documented function, so no real master's recorded commands are used.
Swahili Coast (Ch 8). **The thinnest music lane on any slice page, and the page says so before it
says anything else** — a literal term census returns zero occurrences of the two obvious instrument
words across all 472 chunks. What is attested is three things: women singing as the burning hut is
circled at the Makunduchi New Year, dances that evening, and a later commercial dance-hall layer.
Build from those three. Do not build a musicology on them. May not: the one ensemble beat the
page stages is at reference register — women's voices circling a fire, drums implied and never
sampled from a consecrated performance, no ceremonial text authored or reproduced — and the
warded-threshold cue at the reef climax never reproduces a possession-rite performance.
Ethiopian highlands (Ch 9). **The page's own confidence flag inverts the obvious card, and the
card follows the page:** the musical-theory substrate here is good and the *instrument* substrate is
descriptive only, so no named instrument roster is authored — the theory is what this region is
composed from. That theory is the region's deepest documented layer and a real Thread-21 anchor:
the hymnary carries an elaborate system of musical notation marking the liturgical chant, whose signs
the record compares in function to the Hebrew cantillation accents while explicitly doubting any
direct connection between the two systems; the tradition's own account attributes the chant to Yared
under inspiration, and its own careful position is that what was evolved then was the graphic
expression rather than the chant itself, which the page keeps cleanly separate. **Performance is
adaptive in the record itself** — it varies with the occasion, from simple recitation to elaborate
cantillation according to the character of the feast and the liturgical prescription for that part of
the service, with a cantor and antiphonal response. That is a documented adaptive-music system, and
it is the strongest thing on this page for a score to be built on. The professionals are the
debtera, trained in sacred poetry, in song, and in dance and rhythmical movement. The vernacular
instrument vocabulary the record glosses — karar, bägäna, mäsänqo, käbäro, and the beaten staffs and
rattled sistra of the tabot procession — is carried descriptively, as colour the page attests
rather than as a roster to orchestrate from. May not: the liturgical-chant register informs all
of it at reference level only, no protected sacred music is sampled, reproduced, or imitated in a
consecrated form, and no liturgical text is set.
East African Rift Valley (Ch 10). **No instrument is attested in this corpus and the page says so
first.** The content is the Ju|'hoansi healing-dance clapping and the women's song-line rising at the
n/um dance, at reference register only — a composed cue that honours the *shape* of the practice,
never a recording and never a reconstruction. Because no musicological source is in the corpus, no
rhythmic structure, scale, call pattern or instrument is described anywhere on the page, and none may
be invented here. The page's own briefs are the guide: a deep-time near-silence at Olduvai, wind
and stone, the arc's quietest bed; and a warm, unhurried, human-scaled living-present bed at the
Eyasi shore, deliberately the opposite register. May not: no protected sacred music is
reproduced, sampled, transcribed or notated, and the reference register means a composed cue that
honours the shape of the practice, never a recording or a reconstruction of it.
Congo Basin (Ch 11). Two anchors, not one, and the page's richest layer is the second.
The molimo trumpet is the first, and the facts are precise enough to compose from: the songs'
words are few and the central one says the forest is good; the chorus runs from evening into the
small hours and grows more serious toward morning; men sing lying on their hunting nets or gazing
into the fire, and a father may sing quietly with his mouth against his infant son's ear; the
presence departs with the instrument as its voice recedes into the forest. The women's technique is
described exactly: ribald solos one at a time, each trying to outdo the last, the rest joining in a
throaty chorus cascading downward like a waterfall, each holding her note as the others enter,
swelling until the breath is gone, then all stopping together sharply and cocking their heads to
listen — and the listening-for-the-echo gesture is named on the page as the region's musical
signature.
The second anchor is the **Bwiti ngombi harp, and the page states plainly that it is the Gabon
anchor and that it is a body**: the support post is the Sister of God's backbone, the keys her ribs,
the strings her sinews, the sounding box her stomach, and during the ceremony her voice comes into
the harp and speaks in musical mode to the membership; it carries a white half and a red half. The
night has a documented structure — the path of the harp, a three-part liturgy of Efun engosie,
Nkong engosie and Mesoso or Meyaya engosie, closing with the harp danced out to its chamber and a
communal meal. Against the harp stands obango, the drum-danced sequence celebrated most
continuously in the hours after midnight, representing the cleavage of spirit from flesh, the dancers
whirling around the central spot and touching the pillar. The antelope horn is blown as a first
alert to the ancestors. Percussion beyond that: ngbengbe sticks clapped together in complex
cross-rhythms by boys and girls alike through the honey-season nights. For the modern strata the page
carries the Fang secular dance history in order — belebele, the night dance with men lined one
side of the court and women the other; elong, which brought in balophones and set a circle of men
turning inside a circle of women; ball, elaborated from elong under modern pair dancing with
drums, bottles, accordions and guitar; and manzang, in which the balophone troubadour teams both
played and danced.
*The page also flags its own gap: the Baka music lane rests on canon assertion without region-shelf
substrate and must not be composed until the acquisition lands.* May not: the molimo night is
reference-composed at a respectful distance, never sampled from consecrated performance and **never
staged as a horror cue, because the theology has no evil in it**; the Bwiti engosie is referenced and
never reproduced, and never scored as intoxication.
West Africa (Ch 12). One of the page's richest lanes rather than one of its thinnest, with a
dedicated musicology on the shelf. Tension drums — dundun, talking drums, squeeze drums — some
adorned with light bronze bells and decorative cloth, personalised by drummers with coloured cloth
and short streamers, frequently replacing the Ogboni drums in the king's court and royal festivals.
The tradition's own scholar states the mechanism without hedging: the dundun is a set of four drums
that can use codes to produce a type of language. The texture instruction is explicit — in a
three-drum set the largest and deepest is played with sticks and with stick-and-left-hand, while the
two smaller give intermittent rhythmic phrases dropped almost as soon as they become clearly
established in the listener's mind, leaving the listener suspended in the middle of the phrase.
**The dawn register is the page's own headline — "the most useful documented fact here and it is a
system, not colour" — and it is a complete adaptive-music brief on its own.** At Oyo, praise poetry
was performed in the early morning and dawn music was for listening rather than dancing: all
secondary instruments normally used for dance are excluded except the gudugudu, a small kettle
drum, which alone supplies the basic pattern against which **no fewer than five talking drums play
poetry in unison**. That is the shape to compose — a single pulse under speech-shaped drum lines,
dance instruments excluded, the score deferring whenever the drums speak, thinning toward
mid-morning into a working-city bed — and the oriki praise register rides it. Three named lanes
are genuinely absent and the page holds them rather than composing over them; so does this card.
May not, and this is the hardest care line on any slice page: the **Ogboni drums are a hard care
line rather than an asset** — sacred objects possessing medicinal powers, their sides rubbed with
sacrificial blood at annual ceremonies and covered at dawn with new cloth at a member's nocturnal
funeral, with only cult members allowed to see the carved decoration. **The build never shows that
carving and never voices that drum.** They appear in this card only as the thing the dundun
historically replaced in the king's court, which is the page's own framing and the only framing they
have. Binding on every cue: no protected sacred music is reproduced, and the Ogboni drum carving, the
Fon ancestral songs whose texts the ethnographer himself could not obtain, the Dogon sigu so, and the
Jos Plateau instruments the community classes as sacred are all referenced at form and register —
never sampled, never recreated, never given an authored ceremonial text.
South Africa (Ch 13). The brief writes itself, because the archive contains a song catalogue
whose organising principle is that a song belongs to a specific creature, person or state: the Cat's
Song, the Song of the Caama Fox, the Songs of the Blue Crane, the Old Woman's Song, the Song of the
Bustard, the Song of the Springbok Mothers, the Broken String. Instruments: the /Xam drum, the
dancing rattles, the |goin|goin. **The design line is therefore not one regional theme but a
catalogue of short named songs**, so that the region's soundtrack is discovered rather than triggered
— a player who learns a creature learns its song. **The Broken String is the region's emotional
cadence and is reserved for the two moments that earn it.** This is also the node where the child
gives the amulet away, and the chapter register is *grief* on a rise-then-fall valence.
The counter-score is the other half of this card and it is where the chapter's combat lives. The
mine is mechanical: the stamp battery is a metronome, and because the boss's third phase runs his
whip-calls on the machine timings, the music and the combat grammar are the same object. The
industrial layer is built as a rhythm bed the player learns to read — which makes this the one slice
node where the region's music is a combat mechanic rather than a setting, and it is the node the
slice closes on. May not: no sacred ceremonial content is reproduced; the trance-dance and
rain-rite music is referenced and never recorded, performed or synthesized, and no melody attributed
to a living ceremonial tradition enters the score.
---
Given: this pack, the register card for the node, the machine cue table
(build/audio/music_cue_table.csv — every cue the theme has to serve, with its trigger, its
adaptive class, its layer role and its layer_mode), and the theme's own registry row with its
allow and deny lists.
The adaptive law is a column on that table, not a sentence in this pack. Section 5 states it —
interchangeable, never additive — and layer_mode encodes it per row: base is the melodic
statement a swap layer replaces, swap is an intensity state that substitutes for the base rather
than piling onto it, and ambience_stack is the single declared exemption, granted only because a
chapter-wide weather-and-presence bed carries no melodic material and a non-melodic layer cannot bury
a melody. The emitter refuses to construct a melodic row that declares a stack, and a verify rule
plus a positive control prove the column is real rather than uniformly filled. Before this column
existed, this pack stated a law and the table beside it described stacking, with nothing to reconcile
them — if a row's mode ever disagrees with what the music actually needs, the row is the thing to
argue with.
Returned, per theme: ~~three to five~~ at least eight candidate head cells ~~at piano or short
score~~ as rendered audio plus its extracted melodic line, each notated with its contour string
and its one named deviant, written *inside* the register from the first sketch. Then the scored
candidate table — including the losers, because the score distribution across the first ten
themes is what calibrates the thresholds, and discarding failed candidates destroys that data. The
ladder went from three-to-five to eight-or-more for one reason: a generator's candidates cost
minutes rather than days, so the sample that calibrates the thresholds should be the larger one.
Two deliverables from one composition: real notation a human orchestra can sit down and play,
and rendered stems for the build. The score is the master; the stems are a build of it. **Under the
generation lane the order inverts and the honesty travels with it:** the render is authored first
and the notation is EXTRACTED from it, so notation emitted by this lane is tiered
EXTRACTED-NOT-AUTHORED and the orchestral MusicXML pass that would make it concert-ready is a
named later rung, not a claim made here.
A theme does not proceed on schedule pressure. The ruled bar is that it iterates until it passes.