jpmschweitzerandClaude Opus 4.6
0100b33635
Long-term content strategy workshop: 300-world generator model, cultural ingredients menu, three-system NPC architecture (9 patterns x 6 motivations), Sacred/Profane/Middle Kingdom framework. 9 agents across 4 rounds plus lead interview establishing the production path from hand-authored Sova to generated 300 worlds. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
119 lines
8.1 KiB
Markdown
119 lines
8.1 KiB
Markdown
# 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 |
|