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>
4.6 KiB
Round 3 Tracking — Long-Term Content Strategy
Round focus: Long-term vision only. No v0.1 scoping. "What does the full game look like?"
All 9 contributors responded (8 design + SI production). Tyre's response was last.
Theme Proposals
| Agent | Proposed Universal Theme |
|---|---|
| Gore | "Cost of participation" — complicity is Sova-local |
| Mellanie | "Collective bargain" — each district has one, player inherits a position |
| Ozzie | "What do you owe the people around you?" — different voice per district |
| Miri | "What do systems do to people inside them?" — system failure modes vary |
| Nigel | Complicity scales — same question, different currencies |
| Paula | "Cost of connection / entanglement" |
| Gestalt | "Knowledge as burden" — moral texture emerges from config |
| Tyre | Theme-agnostic moral engine — theme as data, not code |
| SI | "Thematic palette over monolith" — 4-6 moral textures, districts compose |
Consensus: Complicity is Sova-local. The universal theme is structural but unnamed. All framings are facets of the same insight. Lead has not picked canonical framing yet.
Cultural Groups — Round 3 Proposals
| Agent | Count | Organizing Principle |
|---|---|---|
| Miri | 8-10 | Settlement motivation + economic function |
| Gore | 7 | Civilizational philosophy |
| Gestalt | 6-8 | Heritage cluster with mechanical parameters |
| Mellanie | 6 max | Cultural register (voice constraint) |
| Ozzie | 8-12 | Player-facing feel |
| Paula | 9-13 | Mechanical parameters |
| Tyre | ~25 eventually | Reference data with distance model |
Key tension identified: Heritage (ethnic roots) vs. settlement type (economic function) are two different axes being called the same thing.
Content Patterns
| Agent | Patterns Proposed |
|---|---|
| Gore (7) | FRIEND, MIRROR, ANCHOR, GHOST, CATALYST, THRESHOLD, REMNANT |
| Paula (5) | MIRROR, ANCHOR, SYSTEM, GHOST, CATALYST |
| Nigel (8) | FRIEND, MIRROR, HANDLER, WITNESS, TURNCOAT, CIVILIAN, OPERATOR, SKEPTIC |
Key tension: Gore/Paula think thematically (story role). Nigel thinks functionally (gameplay role). Both valid.
PC Archetypes
| Agent | Count | Specific Archetypes |
|---|---|---|
| Gore | 8 | Worker, Investigator, Operator, Wanderer, Builder, Seeker, Guardian, Rebel |
| Ozzie | 6 | Operator, Investigator, Newcomer, Fixer, Believer, Remnant |
| Miri | 2 universal + regional | Insider/Outsider + per-region specifics |
| Nigel | 5 | Smuggler, Detective, Administrator, Newcomer, Enforcer |
| Paula | 8 | Smuggler, Investigator, Merchant, Engineer, Administrator, Medic, Journalist, Newcomer |
| Gestalt | 6 | Smuggler, Detective, Politician, Guardian, Merchant, Civilian |
| Mellanie | 6 | Smuggler, Detective, Fixer, Medic, Newcomer, Elder |
| Tyre | 4 (v1.0 cap) | Not enumerated |
| SI | 4 | Not enumerated |
Recurring: Insider/Smuggler (all), Investigator/Detective (all), Newcomer (5 agents), Administrator (4), Merchant/Fixer (4), Guardian/Enforcer (3).
Randomization
Nigel's Sacred/Profane/Middle Kingdom framework received broad support:
- Sacred = world physics (never randomize)
- Profane = pool-selected per seed
- Middle Kingdom = sacred structure, profane cast
Gestalt's 4 tiers: Structural, Dramatic, Local, Meta. Tyre: "Randomize social, fix spatial." Gore: Shuffle WHY the conspiracy exists as deepest lever.
Scale
| Agent | District Count | NPCs | Key Constraint |
|---|---|---|---|
| SI | 20 | 300+ | 1-2 person-years, Tier 1 is bottleneck |
| Paula | 20 | ~460 | ~21K hand-authored lines |
| Mellanie | 20 | — | 6 cultural registers max in production |
| Gestalt | 30+ | — | 8-12 weeks per district |
| Tyre | 20 | — | 15-20 sprints platform, 9-11 sprints tooling |
Note: These were all calculated at 20-district scale. The lead subsequently reframed to 300 worlds, fundamentally changing the production model from "hand-author districts" to "build generators fed by style briefs."
Key Cross-Cutting Insights
- Regional brief is the most important document (Miri, endorsed by all)
- Template system IS the game; regional brief IS the content (Miri)
- "Randomize context, author content" (universal consensus)
- Tooling before scale (SI, Tyre, universal)
- Cultural diversity = anti-metagaming (Nigel)
- LLM-assisted Tier 2 expansion (Tyre) — write 10, generate 40
- Paired authoring teams (Mellanie) — cultural + character specialist
- Moral texture as deepest randomization lever (Gore)