board.md

assets/boards/flores_island/BRD_0001/board.md

Flores Island — the region establishing board

GENERATED by harness/boards/render_board.py from docs/art_boards/MANIFEST.csv, PLATES.csv and this board's content.json. Do not hand-edit; edit the sources and re-render. The web page at board.html is the review surface; this file is the in-repo mirror and the home of the pick line.

The ask

Pick one accurate reading of Flores and one amplification of it, and mark which surrounding recommendations get built. The accurate reading sets the palette, the silhouette grammar and the material vocabulary that every later Flores asset inherits without another board. The amplification sets what the engine is allowed to add on top, and the rule binding it is that reality is the floor.

Panel A — the accurate depiction

What the source tradition itself says this place looks like. Three genuinely distinct readings, not three renderings of one idea. Every option is a three-plate sheet: silhouette and architecture, palette and light, one signature detail. Every plate carries exactly one rights value, and none of them may be generated — this is a living tradition, so Panel A is Lane R by law.

A1 — The land is already owned

Flores as its cosmology holds it: already populated. The Manggarai name them ata pale-sina, those on the other side. Spring head, tree, doorway and centre-stone are addressed; the visitor is not.

A2 — The ladder of stone and fire

The Sunda Arc's vertical island: karst at Liang Bua, forest, ash on Inerie, colour-shifting lakes on top. The Lio name one Tiwu Ata Polo, the lake of the wicked dead: altitude is moral ground.

A3 — The worked fan

Flores as its people built it. Settlement radiates from the compang and the lingko lodok fields fan to the valley floor below, Manggarai adat: customary land tenure and irrigation, worked today.

Panel A recommendation — A1

Chapter 2 is the arc's survival opener. The protagonist is seven years old, alone, and has no dialogue at all — the spine's own everything-woven section names the danger explicitly, that Flores must never read as a blank survival wilderness. A1 is the only reading that fixes that with the camera rather than with a line, and it is the best-cited layer in the whole substrate: three independent ethnographic sources agree on it. It also pre-teaches the Ch-3 mechanic. The land contract, ngoéng-de-tana, asks the land what it wants before the land is used; a player who has spent a chapter learning to see which points are addressed already understands why that contract exists when it is finally named. And A1 does not cost the other two: the ladder and the fan are both subjects that A1 photographs.

Strongest objection — A1's grammar is made of small things, and an establishing board is supposed to establish a silhouette readable at a hundred metres — that is the art bible field that most determines whether a region reads as itself from a distance. A1's read lives at three metres. Take it and the board has ruled a detail language and left the skyline unruled, which is precisely what A3 would have given. If the answer is that we want both, the honest form of that is an amendment on the pick line — A1 with A3's plan-view law folded in as the distance read — not a second board.

Panel B — the enhanced version

Reality is the floor. Every option here is a strict superset of an accurate option: it amplifies scale, light, density, motion, verticality or legibility, and it contradicts nothing on Panel A. Each carries what was amplified, what stayed true, the specific play it buys, and the altered law in one clause. An amplification with no play claim is decoration, and decoration is the first symptom of slop.

B1 — The addressed landscape, made legible at distance

B2 — The vertical read

B3 — The fan at work

Panel B recommendation — B1

It is the only option whose amplification is a strict superset of the recommended Panel A rather than of a reading we did not take, and it is the only one that buys a mechanic the chapter is actually missing. Chapter 2 has no dialogue, no map and no marker; navigation-by-attended-point is a real answer to that, and it costs nothing in canon because every point it lights is a point the tradition already lights. It also has the cleanest audit surface of the three: what was amplified is contrast, air and motion, and a reviewer can check that claim against any accurate plate in ten seconds.

Strongest objection — B1 amplifies almost nothing a good photograph could not already show, which makes it the least enhanced of the three and arguably gives the 5090 nothing to do. B3 is the option that actually buys a new world state — the restored pre-colonial forest edge — and therefore the option that most justifies having a generative pipeline at all. If the answer to that objection is that the first board should be conservative while the pipeline calibrates, that is a defensible reason, but it is a scheduling reason, not an art reason, and it should be said out loud rather than smuggled into the recommendation.

Panel C — the surrounding recommendations

STORY SUPPORT

SIDE QUESTS

EASTER EGGS

EXPLORATION, ADVENTURE AND ACTION HOOKS

NEARBY IMPORTANT ZONES

Panel C recommendation — C-SQ-1. One marked recommendation for the panel, and it is the card that converts the picked visual law into playable behaviour in the same chapter it is established. If Panel A and Panel B are taken as recommended, this quest is what stops the addressed-landscape grammar from being decoration.

Strongest objection — C-NZ-1 has a better claim to being marked, because it is the only card that changes what gets BUILT rather than what gets played — three zones inherit one palette ruling from it, and it is already measured and ruled ground. The reason it is not marked is that it is not really a recommendation at all: it is a consequence of the pick, and it applies whichever Panel A option is taken.

The pick line

The pick is FILLED and the board is PICKED — the lines below are the recorded ruling, and the ratified reading lives in the art bible the manifest row names.

Visual register

generation_eligible PARTIAL — Panel A is Lane R only. Panel B may run Lane P, generation over a Lane-R underlay, on the six standard-care zones. The five elevated-care zones can never dispatch to Lane G at all: the composer refuses and emits a Lane-R plate request instead (CONCEPT_ART_PROGRAM 4.4).

Free visual grammar

Named visual exclusion

Plate state

Every option on this board is fully specified and decidable today. No plate has been produced yet, and this section says exactly why rather than showing something that pretends otherwise.

Panel A — PENDING ACQUISITION — 9 plates, Lane R

Panel B — PENDING GENERATION — 6 plates, Lane P

Panel C — COMPLETE — 11 cards, no generation required

Critic record

Cultural authenticity — GO WITH FIXES

Physics coherence — GO WITH FIXES

AAAAA de-slop (the unlabeled-relabel test) — GO WITH FIXES

Reveal discipline — GO

Canon coverage — what this board does not carry

62 things docs/spine/CH_02.md and docs/spine/CH_03.md name between them; 5 are on this board and 57 are not. GATE 39 holds the residue — a canon-named thing neither carried nor named here — at zero.

Added beyond the ruling, for veto

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