settled-reach/docs/workshops/wiki-review/round4-tracking.md

Round 4 Tracking — 300-World Generator Reframe

Round focus: Apply the lead's decisions from the interview to all workshop topics. Reframe everything for 300-world generator-input scale.

All 9 contributors responded (8 design + SI production).

---

The Reframe

Every agent independently arrived at the same insight: the 300-world directive doesn't scale up Rounds 1-3 proposals — it transforms them from content authoring strategies into generator specifications. The shift:

Before (20 districts)	After (300 worlds)
Writers produce districts	Engineers produce generators
Style guides for humans	Parameter specs for algorithms
Hand-authored content	Authored INPUTS, generated OUTPUT
Quality via per-line review	Quality via per-parameter review
Cultural briefs as documentation	Cultural briefs as generation seeds

---

Per-Agent Key Contributions

Gore (Themes & Endgame)

• System failure mode as 7th ingredient category (what's broken here — generates moral texture)
• "Philosophical alignment" → "dominant social contract" — reframe from ideology to implicit deal
• "Don't pick, derive" — universal theme emerges from generator derivation, not authorial choice
• THE NOBODY is the most important pattern at 300-world scale — most NPCs start as NOBODYs
• Three-tier world authoring: 10-15 landmark (hand-authored), 30-50 regional (template kits), 230-260 generated

Gestalt (Systems Design)

• 8 proposed decisions (D-WR-21 through D-WR-28) — most comprehensive mechanical spec
• Full composition matrix for thematic patterns x functional motivations (valid/forbidden/excellent)
• THE NOBODY tier promotion spec — promotion triggers, dormant content paths, YAML structure
• Generator interaction map — how 8 generators feed each other
• Derivation rules as Sacred content — ingredient combinations → mechanical parameters

Ozzie (Player Experience)

• "Who do you choose to see?" as player-action framing of universal theme
• Cheers moment as universal first-visit emotional beat (sits down at bar, someone's kind)
• Generator validation = emotional quality gates — review generators, not 4,500 NPCs
• NOBODY + Newcomer = "two invisible people seeing each other" — strongest scene template
• Gravity as perception — walking animations change with planetary gravity, fastest "you're somewhere new" signal

Miri (Worldbuilding)

• 10 heritage roots with phonetic + social signatures (Frost, Tide, Iron, Spice, Jade, Dust, Vine, Salt, Stone, Arc)
• Blend weights — societies pick 1-3 roots with weights summing to 1.0
• Society seed YAML — full world_seed.yaml spec with all ingredient categories
• Naming algorithm spec — phonetic rules, not word lists; syllable assembly from heritage parameters
• NOTE: Heritage root naming (social signatures vs Earth-grounded names) — lead pushed back on abstract naming. Revised to Earth-grounded clusters in discussion but the mechanical specs remain valid.

Paula (Librarian)

• Null-heritage worlds fully specified — corporate/functional stations with no cultural depth, "thinness as moral texture"
• 8 heritage roots (Earth-grounded naming: Nordic-compact, Mediterranean-warm, East-formal, etc.)
• Quality gradient = in-world logic — Exemplar/Polished/Standard/Frontier maps to older=denser, core=richer
• Vulnerable window during archetype transitions — lost old protections, haven't earned new ones
• 8 PC archetypes defined with full generator template specs
• Authored content estimate: ~1,600-2,800 hours irreducible for 300 worlds (vs 13,500-31,200 for 20 hand-crafted)

Nigel (Sandbox & Replayability)

• Sacred/Profane/Middle Kingdom applied to generators — Sacred=grammar, Profane=vocabulary, Middle Kingdom=sentence structure
• Economic pressure profile YAML spec — temptation_source, temptation_justification as load-bearing fields
• world_seed schema — {numeric_seed, cultural_assembly: {heritage, motivation, economy, philosophy, ...}}
• Full ingredients→mechanics mapping — every ingredient category mapped to mechanical outputs + replayability contribution
• Gate topology as Sacred/Profane — connected graph (Sacred), which worlds connect (Profane)

Tyre (Technical Architect)

• Architectural pivot — content directory becomes generator input repository; content loader becomes content generator
• Pool + overlay + canonical_id triad — addressing scheme for mixed authored/generated content
• Two-tier content addressing — authored content has stable IDs, generated content has seed-derived IDs
• Generation pipeline architecture — ingredient authoring → template authoring → generator dev → validation → generation + review → hand-elevation
• Simulation tier implications — 300 worlds means most are State-saved or Ungenerated at any time

Mellanie (Copywriter)

• 8 voice parameter categories — sentence rhythm, formality gradient, substrate density, profanity pattern, naming address, silence rules, emotional expression style, environmental text register
• Writer → tool-writer — "At 20 districts I'm a writer who uses tools. At 300 worlds I'm a writer who writes tools."
• Quality control shift — per-parameter review, not per-line review
• Authored content estimate: 3,300-7,400 lines across all content types for 300-world scale
• Blending rules — heritage substrate interactions ("never blend profanity across heritage pools")

SI (Project Manager)

• Production pipeline reframe — 6-phase generator development pipeline
• New bottleneck: generator quality, not content volume
• Release path: v0.1 (interaction proof) → v0.2-v0.5 (generator testing) → v0.6-v0.10 (content pipeline at scale) → pre-v1.0 (300 worlds)
• Generator acceptance criteria: "generated output indistinguishable from hand-authored at Tier X"

---

Key Convergences

1. Generator-first is unanimous — every agent independently reframed their R3 proposals as generator specifications
2. Cultural ingredients menu converged across all agents — 6-7 categories, opt-in, absence-as-signal
3. Sacred/Profane/Middle Kingdom adopted as universal architectural language for generators
4. THE NOBODY recognized by all as the most important pattern at 300-world scale
5. Tooling first, content second — universal agreement on development priorities
6. Quality gradient as in-world logic — older/denser worlds are naturally higher quality (more authored polish)
7. ~500-1000 authored templates produce 300 worlds — the authoring is front-loaded into generator inputs

Key Tensions

1. Heritage root naming: Miri uses abstract social signatures (Frost, Tide, Iron). Paula + lead prefer Earth-grounded names (Nordic-compact, Mediterranean-warm). Mechanical specs compatible either way.
2. Heritage root count: Miri proposes 10, Paula proposes 8, Gore proposes this is the wrong question (derive from ingredients, don't enumerate)
3. 7th ingredient: Gore proposes "system failure mode" as an additional category. Others keep to 6.
4. Universal theme: Still open. Ozzie's "Who do you choose to see?" gained traction as player-action framing. Gore says "Don't pick, derive."

Proposed Decisions (Gestalt)

#	Decision	Status
D-WR-21	Cultural ingredients menu (6 categories, generator-composed)	Needs lead confirmation
D-WR-22	Three-system NPC architecture (9 patterns x 6 motivations + composition matrix)	Needs lead confirmation
D-WR-23	8 PC archetypes with fluid transitions as game mechanic	Partially confirmed by lead interview
D-WR-24	Sacred/Profane/Middle Kingdom maps to all game systems	Confirmed by lead interview
D-WR-25	District skeleton survives as generator output specification	Needs lead confirmation
D-WR-26	6-phase generator development pipeline	Needs lead confirmation
D-WR-27	Authored content at 300-world scale (~1600-2800 hours)	Needs lead confirmation
D-WR-28	Gate topology generation with Sacred/Profane constraints	Needs lead confirmation