settled-reach/docs/workshops/wiki-review/round3-gestalt.md

title, description, type, status, workshop, agent, round, created

title	description	type	status	workshop	agent	round	created
Round 3 — Gestalt (Systems Design)	Long-term content strategy for scaling to 20+ districts without linear authoring effort	workshop	archived	wiki-review	gestalt	3	2026-02-12

Round 3 — Gestalt (Systems Design): Long-Term Content Strategy

cracks knuckles

Seven questions. All of them are really one question: What are the systems that make a 20+ district game work without requiring 20x the authoring effort of one district?

Let me break down what each of these questions actually means mechanically.

---

1. Complicity Is Sova's Flavor. The Master Theme Is "Knowledge as Burden."

Gore named complicity as the game's theme. He's right about Sova. He's wrong about the whole game — and the reason he's wrong is the reason the game scales.

The argument

Complicity works in Sova because of a specific configuration:

• The smuggler participates in something they know is morally ambiguous
• The detective disrupts something the community considers normal
• The culture (Van Maanen's Star privacy norms) makes non-cooperation the default
• The contraband (lattice components) serves genuine needs

Change ANY of those variables and complicity stops being the right word. Consider:

District Type	What You Know	What It Costs You	The Right Word
Logistics hub (Sova)	You're part of the operation	Guilt, eroding relationships	Complicity
Political capital	Your ally is betraying your faction	Broken loyalty, political death	Loyalty
Frontier settlement	Someone is dying and only you know	Moral obligation vs. survival	Duty
Research station	Your identity is being rewritten	Loss of self, gain of power	Identity
Trade nexus	You can profit from others' suffering	Greed vs. empathy	Greed
Refugee district	You can save some but not all	Triage, survivor's guilt	Sacrifice
Military outpost	Your orders are wrong and you know it	Obedience vs. conscience	Authority

Every single one of these is a variation on the same mechanical core: you know something, and knowing it changes you. The knowledge is a burden. What you do with it defines who you become.

The framework

Master theme: Knowledge as burden.

Master mechanic: Asymmetric information (D-005 — this was always the answer).

Per-district moral texture: Each district explores a specific facet of what knowledge costs.

This isn't just thematic variety for its own sake. It's mechanically load-bearing:

1. Different moral textures produce different player decisions. In Sova, the smuggler's key decisions are about protecting vs. exposing. In a loyalty-textured district, the key decisions are about betraying vs. enabling. Same decision architecture (fork + consequences), different emotional weight. The systems don't change — the content changes.

2. Different moral textures demand different monologue vocabularies. The smuggler's moral_pang trigger works for complicity. A loyalty district needs loyalty_fracture. An authority district needs conscience_objection. Same trigger system, different named instances. The engine doesn't care what the trigger is called — it fires when conditions are met.

3. Different moral textures create replay variety at the thematic level. Two players who both played smuggler-type characters in different districts had mechanically similar experiences (managing a dual life, navigating risk, protecting relationships) but emotionally different ones. The replay isn't just "different NPCs" — it's "different questions."

What this means for systems design

The per-district moral texture is NOT a flavor field in a config file. It's an emergent property of the specific configuration of NPCs, triangles, contraband, cultural norms, and access tier profiles in that district. The designer doesn't write moral_texture: complicity. They design a district where:

• The smuggler's FRIEND is hiding something that protects someone
• The detective's investigation harms people who don't deserve it
• The cultural norms make non-cooperation virtuous
• The contraband serves genuine needs

And the PLAYER experiences that configuration as complicity. Another district where:

• The politician's FRIEND is secretly working for a rival faction
• The detective's investigation reveals that both sides are justified
• The cultural norms prize loyalty above truth
• The contested resource is political power, not goods

...produces a loyalty texture. Same systems. Different content. Different feeling.

Decision proposal (D-WR-11): The Settled Reach's master theme is "knowledge as burden" — asymmetric information is the mechanic, the cost of knowing is the experience. Each district explores a local moral texture (complicity, loyalty, duty, identity, etc.) through the same systemic framework. No district's theme is prescribed mechanically — it emerges from content configuration.

---

2. Cultural Groups at Scale: 6-8 Heritage Clusters, Mechanically Distinct

Why culture matters mechanically

Miri's Van Maanen's Star brief proved that culture isn't decoration — it's a game system. The Nordic-Baltic privacy norms ARE the difficulty gradient for the detective. The labor solidarity norms ARE the smuggler's camouflage. Culture determines:

System	Cultural Input	Mechanical Output
Dialogue access tiers	Privacy norms, authority attitudes	How hard it is to reach peer / insider
NPC disclosure behavior	Information-sharing customs	What NPCs volunteer vs. withhold
Tell visibility	Behavioral norms	What counts as "unusual" in this community
Investigation strategy	Cooperation patterns	Whether the detective can find allies
Cover viability	In-group/out-group markers	How easily the smuggler blends in
Social site dynamics	Gathering patterns, leisure norms	Where and when NPCs are observable

Each cultural group produces a DISTINCT ACCESS TIER PROFILE — a mechanically different experience of the same systems. This is the constraint that determines how many we need.

The cultural map

I'd propose 6-8 heritage clusters for the full game. Each must be mechanically distinct along the axes above.

#	Heritage Cluster	Cultural DNA	Mechanical Signature	Access Tier Profile
1	Nordic-Baltic (Van Maanen's Star)	Finnish/Estonian/Latvian/Norwegian drift	Privacy-default, labor solidarity, institutional distrust	Hard outsider wall. peer requires months. authority gets compliance but zero information.
2	Mediterranean-Near Eastern	Greek/Turkish/Lebanese drift	Hospitality-default, honor networks, generosity-as-status	Easy initial warmth (fast peer), but deep trust is family-only. Gossip flows freely — too freely. Information hides in noise.
3	East Asian	Japanese/Korean/Vietnamese drift	Hierarchical respect, obligation networks, face-saving	authority gets deference but performative cooperation. Real information flows through obligation chains, not direct questioning. The detective must EARN an obligation.
4	West African	Yoruba/Akan/Wolof drift	Communal decision-making, storytelling tradition, elder-respect	Information flows freely through communal events but is couched in story and metaphor. Direct questions are rude. The detective must learn to LISTEN.
5	Latin American	Brazilian/Colombian/Mexican drift	Personal warmth, patron-client dynamics, familismo	Warm to strangers, but loyalty runs through patronage networks. The detective is a PERSON first, an authority second. Cooperation is personal, never institutional.
6	Slavic-Central European	Russian/Polish/Czech drift	Institutional cynicism, dual economy, black humor	Everyone assumes the system is corrupt. authority is treated with open cynicism. The grey economy is not hidden — it's simply not discussed with anyone who might be naive enough to object.
7	South Asian	Indian/Pakistani/Sri Lankan drift	Community gatekeepers, code-switching, stratified access	Access is mediated through gatekeepers (elders, community leaders). The detective can't access individuals directly — must go through the gatekeeper. The gatekeeper's agenda shapes all information.
8	Blended/Cosmopolitan	Multi-heritage convergence	No dominant norm, transactional relationships, individualism	No cultural immune system. Information is freely available but untrustworthy. Everyone's performing. The challenge isn't ACCESS — it's VERIFICATION.

Why 6-8, not 20

Simple rules, complex outcomes. 6-8 heritage clusters with blending produces dozens of EXPERIENCED cultures:

• Pure Cluster 1 (Nordic-Baltic station, like Sova)
• Cluster 1 + Cluster 6 blend (Nordic-Baltic founded, Slavic immigration wave)
• Cluster 3 dominant, Cluster 8 cosmopolitan overlay (East Asian core station with a trade district)
• Cluster 2 + Cluster 5 blend (Mediterranean-Latin fusion in a warm-climate colony)

Each blend creates a NEW access tier profile by interpolating between parent clusters. A Nordic-Baltic + Slavic blend might have the privacy norms of Cluster 1 but the institutional cynicism of Cluster 6 — meaning the detective faces both a cultural wall AND open mockery when they invoke Commission authority.

Distribution

Not every system gets a unique culture. Multiple systems can share a heritage cluster (like multiple Nordic-Baltic settled systems with Van Maanen's Star-like norms). The full game's cultural map might look like:

Heritage Cluster	# of Systems	Notes
Nordic-Baltic	3-4	Founding-era settlement belt
Mediterranean-Near Eastern	2-3	Trade corridor systems
East Asian	2-3	Industrial core systems
West African	1-2	Agricultural/frontier
Latin American	2-3	Mid-Reach settlement wave
Slavic-Central European	2-3	Outer Reach mining/industrial
South Asian	1-2	Administrative/political centers
Cosmopolitan	3-5	Capital systems, trade hubs, crossroads

~20-25 systems across 6-8 clusters. Each system has 1-4 districts. Each district = 1 content pack. The cultural cluster determines the access tier profile, naming conventions, substrate vocabulary, social norms. Content authors reference the regional brief for their cluster.

The authoring constraint

Each cultural cluster requires ONE Miri-quality regional brief (~Van Maanen's Star brief). 6-8 briefs total. Each brief includes:

• Heritage lineage and drift patterns
• Naming conventions (phonological rules, surname patterns)
• Substrate vocabulary (10-15 heritage words)
• Social norms (5-6 key behavioral expectations)
• Access tier cultural profile (how the 5 tiers work in this culture)
• Food/drink/leisure patterns
• Authority attitudes

After the 6-8 briefs exist, every NEW district in that cluster references the brief. Authors don't reinvent the cultural wheel — they apply the brief to a specific location. This is how culture scales: author the pattern once, instantiate it many times.

Decision proposal (D-WR-12): The Settled Reach has 6-8 heritage clusters, each with a Miri-quality regional brief. Heritage clusters are mechanically load-bearing (access tier profiles, NPC disclosure patterns, investigation strategy). Districts reference their cluster's brief for cultural grounding. Blended districts interpolate between parent clusters.

---

3. Content Patterns That Scale: The Universal Skeleton + Variable Flesh

Which patterns are universal?

Let me map every content pattern we've designed and classify it:

Pattern	Universal?	What Varies?
Social site / functional cluster (D-025)	YES — every district has them	Count (2-6), size (4-12 NPCs), type (workplace, social, residential, commercial, institutional)
Triangles	YES — every social site has 1-3	Triangle type (power, loyalty, knowledge, romantic, economic), specific tension
Three-tier content model (D-023)	YES — always Tier 1/2/3	Tier counts scale with district size
THE FRIEND	YES — every PC archetype gets one per district	What they're hiding varies with district moral texture
Tagged line pools + 4-layer selection (D-028/D-035)	YES — universal dialogue architecture	Tag vocabularies may expand per district
Routine system	YES — NPC schedules everywhere	Cultural work patterns vary per cluster
Population entanglement ratio (D-029)	YES — 30/50/20 is a guideline	Exact ratio varies per district type and seed
THE MIRROR (Gore R2)	RECOMMENDED but not required	Some districts benefit more than others
Dual-lens content	YES for districts with 2+ archetypes	Single-archetype districts don't need it
FRIEND pool	YES — architecture always supports it	Pool size varies per district maturity
Contraband type	NO — varies by district economy	Some districts have no smuggling at all
Ring structure	NO — some districts have no criminal operation	The ADVERSARIAL STRUCTURE is universal; its form varies

The critical generalization: adversarial structures

Sova has a smuggling ring. Not every district will. But every district needs an adversarial structure — a hidden network of relationships and activities that one PC archetype knows about and another must discover.

District Type	Adversarial Structure	Insider Archetype	Investigator Archetype
Logistics hub	Smuggling ring	Smuggler	Detective
Political capital	Faction conspiracy	Politician	Intelligence operative
Research station	Unauthorized experiment	Researcher	Ethics investigator
Military outpost	Mutiny plot	Dissenting officer	Loyalty investigator
Trade nexus	Price-fixing cartel	Merchant	Economic regulator
Refugee district	Underground railroad	Smuggler variant	Border enforcement
Cultural enclave	Preservation network	Cultural insider	Assimilation enforcer

The pattern: there's always a hidden network and there's always someone whose job is to find it. The insider BELONGS to it. The investigator STUDIES it. The information asymmetry between them IS the game.

This means the Sova patterns generalize as follows:

Sova-Specific	Generalized Pattern
The Ring	The Hidden Network (whatever shape it takes)
Ring operations	Network activities (smuggling, conspiring, experimenting, plotting)
Contraband	The Contested Resource (goods, power, knowledge, identity, freedom)
Commission regulation	The Institutional Framework (whatever authority the network defies)
Detective investigation	Observer Investigation (any archetype that discovers)
Smuggler cover management	Insider Risk Management (any archetype that hides)
Complicity moral texture	Local Moral Texture (whatever knowledge costs in this place)

The structural template for a district

Every district, regardless of setting, follows this skeleton:

DISTRICT SKELETON
├── 2-6 Social Sites (functional clusters, 4-12 NPCs each)
│   ├── 1-3 Triangles per social site
│   ├── Physical space with sightlines + overhearing zones
│   └── Bridge NPCs connecting to other social sites
├── 1 Hidden Network (adversarial structure)
│   ├── 3-8 network members across social sites
│   ├── 1 coordinator (Devra-equivalent)
│   ├── 1 external connection (Sabel-equivalent)
│   └── Network activities that deviate from base routines
├── 1 FRIEND per supported PC archetype
│   ├── Tier 1 depth, full contradiction arc
│   ├── Connected to the hidden network (directly or indirectly)
│   └── Discovery produces moral reckoning, not just information
├── 0-1 MIRROR per district (recommended)
│   ├── Honest character, no secret
│   └── Their transparency reveals everyone else's dishonesty
├── 1 Cultural Group reference (heritage cluster brief)
│   ├── Access tier profile
│   ├── Naming conventions
│   └── Substrate vocabulary
├── 1 Local Moral Texture (emergent from configuration)
│   ├── What knowledge costs in this place
│   └── What the player's key decisions feel like
├── 2-4 Supported PC Archetypes
│   ├── 1 insider archetype (belongs to the hidden network)
│   ├── 1 investigator archetype (seeks to understand the network)
│   └── 0-2 additional archetypes (affected bystander, rival network, etc.)
├── 1 Contested Resource (what the network moves/controls/protects)
│   ├── Culturally grounded (why this resource matters HERE)
│   └── Morally ambiguous (the network isn't pure evil)
└── Environmental Text
    ├── Location-specific (signs, menus, graffiti)
    ├── News ticker headlines (Reach-wide + local)
    └── Cultural artifacts (food, slang, customs)

This skeleton is the repeatable unit of content creation. Every new district instantiates it with different flesh. The systems code works identically against every instantiation.

Where the structure itself can vary

The skeleton above is the MINIMUM. Some districts may have additional patterns:

• Multi-network districts: A trade nexus might have TWO hidden networks (smuggling ring + price-fixing cartel) that overlap. NPCs might belong to both. This creates a third investigative dimension.
• Network-free districts: A quiet residential district with no hidden network — just mundane triangles, life-sim substrate, and a FRIEND whose contradiction is personal (infidelity, debt, illness) rather than criminal. These districts serve as contrast — the "is this enough?" question is literal.
• Hub districts: A major political or commercial center with 6+ social sites, 25+ NPCs, and multiple overlapping adversarial structures. These are premium content — expensive to author, but they're the centerpiece locations.
• Transitional districts: Small districts (1-2 social sites, 8-10 NPCs) that serve as connective tissue between larger districts. The player passes through on their way somewhere else — but something catches their eye.

Decision proposal (D-WR-13): The district skeleton (social sites + hidden network + FRIEND + cultural group + moral texture + PC archetypes + contested resource) is the universal content template. All districts instantiate this skeleton. The skeleton's size varies (2-6 social sites, 15-40 NPCs) but its structure is consistent.

---

4. The Content Directory as a Platform

The hierarchy at scale

The Round 2 converged structure works for one district. At scale, it needs a system level:

content/
  _meta/
    manifest.yaml               # Game-level metadata
    load-order.yaml             # Content pack ordering

  _schema/
    (per-type validation schemas — unchanged from R2)

  global/
    factions/                   # Reach-wide factions
    technology/                 # Technology definitions
    knowledge/                  # FactId definitions, entity attributes
    enums/                      # Tag vocabularies (situations, moods, etc.)
    archetypes/                 # PC archetype definitions (NEW)
      smuggler.yaml             # Archetype: starting access, agency, perception
      detective.yaml
      politician.yaml
      guardian.yaml
      merchant.yaml
      civilian.yaml
    patterns/                   # Reusable content patterns (NEW)
      friend.pattern.yaml       # THE FRIEND pattern requirements
      mirror.pattern.yaml       # THE MIRROR pattern requirements
      hidden-network.pattern.yaml  # Adversarial structure requirements
      triangle.pattern.yaml     # Triangle structure requirements

  regions/                      # Heritage cluster briefs (NEW — promoted from global/)
    nordic-baltic.yaml          # Cluster 1 (Van Maanen's Star, etc.)
    mediterranean.yaml          # Cluster 2
    east-asian.yaml             # Cluster 3
    west-african.yaml           # Cluster 4
    latin-american.yaml         # Cluster 5
    slavic-central.yaml         # Cluster 6
    south-asian.yaml            # Cluster 7
    cosmopolitan.yaml           # Cluster 8

  systems/                      # Star systems (NEW level)
    van-maanens-star/
      system.yaml               # System metadata, wormhole connections
      districts/
        sova-transit/
          (existing R2 structure: district.yaml, pools.yaml,
           npcs/, locations/, templates/, triangles/, lines/)
        sova-residential/       # Second district on same station
          ...
        van-maanens-star-prime-port/       # Planet-side district
          ...
    tavros/
      system.yaml
      districts/
        tavros-exchange/        # Trade district
          ...
    (etc.)

Key changes from Round 2

Change	Rationale
regions/ promoted to top-level	Heritage clusters are game-wide reference data, not district content. A region is referenced by MANY districts.
global/archetypes/ added	PC archetype definitions are global — the smuggler archetype exists independently of any district. Each district declares which archetypes it supports.
global/patterns/ added	Reusable content pattern definitions (FRIEND, MIRROR, hidden network, triangle) as formal schema/requirement docs. District content is validated against these.
systems/ level added above districts/	Star systems group districts. Wormhole connections are system-level data. A content pack can be one system (containing multiple districts).

Content packs, DLC, and mods

The platform architecture must support three content delivery models:

1. Core content packs (shipped with the game)

# content/_meta/manifest.yaml
packs:
  - id: core
    districts: [van-maanens-star/sova-transit, van-maanens-star/sova-residential, ...]
    version: 1.0

2. DLC packs (official expansion content)

# dlc/tavros-exchange/_meta/manifest.yaml
pack:
  id: tavros-exchange
  type: dlc
  requires: [core >= 1.0]
  adds:
    systems: [tavros]
    districts: [tavros/tavros-exchange]
    archetypes: [merchant]     # New PC archetype
    factions: [syndic-guild]   # New faction

3. Mod packs (community content)

# mods/jax-veteran/_meta/manifest.yaml
pack:
  id: jax-veteran
  type: mod
  requires: [core >= 1.0]
  extends:
    districts: [van-maanens-star/sova-transit]
  adds:
    npcs: [jax-korrenson]
  merges:
    pools: [bar_regulars]
    lines: [bar/dialogue]

Content pack isolation

Each content pack is a self-contained directory tree that follows the same structure as the core. The engine merges them at load time using the R2 overlay rules (ADD/REPLACE/MERGE). The key constraint: a content pack never modifies files in another pack's directory. It provides its own files that the merge system combines.

This means:

• DLC districts are fully self-contained (own NPCs, lines, templates)
• DLC can ADD new archetypes, factions, cultural groups
• Mods can EXTEND existing districts (add NPCs to pools, add dialogue lines)
• Mods can REPLACE existing content (swap an NPC, change a routine)
• Nothing breaks if a mod is removed — the merge system gracefully degrades

Cross-district references

At scale, content must reference entities in other districts:

• An NPC in District A has a sibling in District B
• A faction spans multiple systems
• A wormhole connects two districts
• A news ticker headline references events in another system

These cross-references use the canonical_id system (Miri/Tyre R2):

# In van-maanens-star/sova-transit/npcs/sabel.yaml
relationships:
  - npc: tavros.exchange.npc.sabel-partner  # Cross-district reference
    type: "business associate"

The engine resolves canonical_id to file path at load time. Missing references (because a DLC isn't installed) are handled gracefully — the relationship exists in the data but the referenced NPC doesn't render.

Decision proposal (D-WR-14): The content directory supports three delivery models (core, DLC, mod) through a pack manifest system. Content packs are self-contained directory trees merged at load time. Cross-district references use canonical_id. Missing references degrade gracefully.

---

5. Randomization Philosophy: Randomize Context, Author Content

The principle

At Sova scale, Nigel identified three randomization axes (FRIEND pool, contraband type, who's compromised). At full game scale, the number of potential axes is much larger. We need a PHILOSOPHY, not a list.

The philosophy: randomize the CONTEXT, author the CONTENT.

Layer	Authored or Randomized	Examples
The specific line of dialogue	AUTHORED	"There you are. Good — I was starting to worry."
The tell behavior	AUTHORED	"Looks left when lying"
The contradiction arc	AUTHORED	5 phases, each with specific authored beats
The cultural norms	AUTHORED	Van Maanen's Star privacy-default, labor solidarity
Which NPC delivers that line	RANDOMIZED (pool selection)	Kael or Renn as FRIEND
Which NPC has which role	RANDOMIZED (role-slot assignment)	Drin or Maret as compromised inspector
Which contraband is in the crates	RANDOMIZED (per-seed)	Lattice components or medical-grade
Which triangles are active	PARTIALLY RANDOMIZED	Core triangles always; optional triangles per seed
Evidence locations	RANDOMIZED	Dead drop in B-7 or B-12
NPC routine micro-variations	RANDOMIZED	Break at 10:30 or 10:45

The four randomization tiers

At full scale, I see four distinct tiers of randomization, each with different rules:

Tier S: Structural randomization (set at game creation, never changes)

These define the shape of the playthrough. They create the "two friends compare notes and played different games" experience.

Axis	What Varies	Constraint
FRIEND selection	Which NPC is your FRIEND in each district	Pool of Tier 1 authored candidates; minimum 2 per archetype per district
PC archetype identity	Which specific NPC you ARE (within archetype)	Pool of authored PC candidates per archetype
Hidden network membership	Which NPCs are in the network	Constrained by role requirements (need coordinator, courier, etc.)
Contested resource type	What the network moves/controls	Must match district cultural grounding
Inter-system connections	Which wormhole routes are active	Graph connectivity must allow progression
Faction power balance	Which faction dominates which system	Must create tension, not uniformity

Tier D: Dramatic randomization (storyteller-controlled, changes during play)

These are the storyteller's levers. Nigel's "dramatic randomness."

Axis	What Varies	Constraint
Escalation timing	When pressures increase	Pacing algorithm (D-005's Rimworld-style storyteller)
NPC cracking	Who breaks under pressure first	Weighted by tolerance threshold + accumulated stress
Evidence availability	When clues become observable	Must follow the observe→notice→follow→discover sequence
Environmental events	Ventilation breaks, supply disruptions, social events	Provide texture and alibi opportunities
News ticker content	Which headlines appear	Pool of authored headlines, drawn per cycle

Tier L: Local randomization (per-session micro-variation)

These make each play session feel slightly different even on the same seed.

Axis	What Varies	Constraint
NPC routine micro-timing	Exact arrival/departure within windows	±10 minutes from scheduled time
Dialogue line selection	Which valid line from the pool fires	Weighted by mood + topic match + variety tracking
Social site population	Who's at the bar right now	Base on schedule + micro-timing + small random absence chance
Environmental details	Which graffiti is visible, which headline scrolls	Pool selection, changes per visit

Tier M: Meta-randomization (per-NEW-GAME variation across multiple playthroughs)

These are the axes that make the GAME different, not just the session.

Axis	What Varies	Constraint
Starting district	Where the player begins	Must match archetype (smuggler starts in logistics districts)
Which districts exist	Procedural system/district generation at game scale	Core districts are fixed; procedural districts fill gaps
Cultural blending	Which heritage clusters are present in each system	Must be historically coherent per the setting
Reach-wide events	What's happening in the larger political landscape	Storyteller-driven macro events that affect all districts

New randomization axes at full scale

Beyond Nigel's Sova-specific axes, the full game adds:

1. Wormhole topology: Which systems connect to which. This shapes trade routes, which shapes smuggling routes, which shapes where the player encounters hidden networks. Different topologies create different strategic landscapes.

2. Faction ascendancy: Which faction is dominant in which system. A Commission-dominated system has heavy regulation and lots of contraband opportunity. A Syndic-dominated system has a corporate monopoly and worker resistance. Different power configurations produce different adversarial structures.

3. Technology availability gradient: Which lattice tiers, which perception modes, which tools are available where. Frontier systems have basic tech. Core systems have advanced capabilities. This affects gameplay strategy AND the player's observational capacity.

4. Diaspora patterns: Which cultural groups have migrated where over 200+ years. A second-wave Nordic-Baltic settlement on a predominantly East Asian station creates a blended culture with unique access tier dynamics. These patterns are set at game creation and shape the social landscape.

5. Historical events: Per-seed selection of which founding-era events happened. "Talvik promised us quarters by '98" is Sova-specific. Each district has 2-3 historical grievances/triumphs drawn from a pool, expressed through environmental text and NPC attitudes.

The anti-randomization constraint

Not everything should vary. Some things MUST be fixed for the game to cohere:

• The setting's core technology (lattice, Meridian, span gates, founder gates) is fixed
• The faction list is fixed (though their local power varies)
• The cultural cluster definitions are fixed (though their distribution varies)
• The content patterns (FRIEND, MIRROR, hidden network, triangles) are fixed
• The master theme (knowledge as burden) is fixed
• The master mechanic (asymmetric information) is fixed

These are the INVARIANTS. Everything else is a variable. The game is defined by its invariants; its replay value comes from its variables.

Decision proposal (D-WR-15): Randomization philosophy: "randomize context, author content." Four tiers: Structural (game creation), Dramatic (storyteller), Local (per-session), Meta (per-new-game). Authored content (lines, arcs, tells, cultural norms) is never generated at runtime. Randomization selects which authored content is active and how it's configured.

---

6. PC Archetypes at Full Scale

How many archetypes?

The smuggler and detective are two sides of the same adversarial coin: insider vs. investigator. At full scale, the game needs enough archetypes to support different adversarial structures AND different player fantasies.

Let me map the archetypes to what they actually DO mechanically:

Archetype	Information Position	Core Mechanic	Fantasy
Smuggler	Insider (knows the network)	Risk management, dual-life navigation	"I'm part of something dangerous and I can't stop"
Detective	Outsider (discovers the network)	Pattern recognition, social investigation	"Something's wrong here and I'll find out what"
Politician	Meta (knows the power dynamics)	Faction navigation, deal-making	"Everyone owes me something and I owe everyone"
Guardian	Counter (knows the surveillance)	Evasion, counter-intelligence	"They're watching everything and I'm invisible"
Merchant	Parallel (knows the economics)	Trade optimization, moral arbitrage	"I know what everything costs, including my soul"
Civilian	Blank (knows nothing at start)	Discovery, adaptation, survival	"I was just living my life until it all went wrong"

Six archetypes. Each produces a fundamentally different information asymmetry experience through the same systems.

Not every district supports every archetype

The key insight: archetype availability is a PER-DISTRICT property, not a global one. This is how content scales without requiring 6x everything.

District Type	Supported Archetypes	Why
Logistics hub (Sova)	Smuggler, Detective, Civilian	The hidden network is a smuggling ring
Political capital	Politician, Detective, Guardian	The hidden network is a faction conspiracy
Research station	Merchant (tech trader), Detective, Civilian	The hidden network is unauthorized research
Trade nexus	Merchant, Smuggler, Detective	Multiple hidden networks overlap
Frontier settlement	Smuggler, Guardian, Civilian	The hidden network is a resistance movement
Military outpost	Guardian, Detective, Civilian	The hidden network is a mutiny plot

Most districts support 2-3 archetypes. The MINIMUM is 2 (one insider, one investigator — the adversarial pair). Premium districts support 3-4 for richer replayability.

The adversarial pair principle

Every district must have at least one adversarial pair — two archetypes that see the same world from opposing information positions.

Adversarial Pair	Insider	Investigator	The Tension
Smuggler / Detective	Knows the ring	Discovers the ring	Complicity vs. justice
Politician / Intelligence	Knows the conspiracy	Uncovers the conspiracy	Loyalty vs. truth
Guardian / Security	Knows the surveillance gaps	Closes the surveillance gaps	Freedom vs. order
Merchant / Regulator	Knows the market	Polices the market	Profit vs. fairness

The adversarial pair is what makes the dual-lens work. The insider's home is the investigator's crime scene. Same NPCs, same social sites, different access tiers, different monologue voices, different moral textures.

The per-archetype authoring cost

Each archetype in a district requires:

Component	Lines/Effort	Notes
PC-candidate NPC profile	2 wiki pages	Standard NPC + PC appendix (B1-B7 from Paula R2)
Starting knowledge graph	1 YAML file	FactId confidence table
Starting relationship map	1 YAML section	Forward + reverse knowledge per known NPC
Orientation monologue	5-8 lines	"First 60 seconds" content
Character-specific monologue pool	30-50 lines per district	Separate pool per character (D-032)
Agency boundaries	1 YAML section	What the PC can/can't do
Voice card	1 page	How this archetype thinks/speaks

For a district with 3 supported archetypes: ~3 PC profiles + ~90-150 monologue lines + supporting YAML. That's significant but bounded.

The archetype-as-framework pattern

Here's the scaling trick: archetype definitions are global, archetype INSTANCES are local.

The smuggler archetype is defined once in global/archetypes/smuggler.yaml:

archetype:
  id: smuggler
  information_position: insider
  default_access_tiers:
    network_members: insider
    community_members: peer
    institutional_figures: authority  # they have authority over you
    strangers: public
  core_mechanic: risk_management
  monologue_voice: relational    # thinks in people, not patterns
  perception_modes: [natural, social_awareness]
  moral_arc_type: complicity     # default, can be overridden per district

Each district that supports the smuggler archetype provides a LOCAL INSTANCE:

# In systems/van-maanens-star/districts/sova-transit/archetypes/smuggler.yaml
archetype_instance:
  archetype: smuggler
  pc_candidates: [kael-davan-smuggler-pc]  # Which NPCs can be this archetype
  local_override:
    moral_arc_type: complicity    # matches default for Sova
    cultural_voice: van-maanens-star         # uses Van Maanen's Star substrate vocabulary
    network_role: logistics       # smuggler's function in this specific ring

This separation means:

• Adding a new archetype to an existing district = adding one instance file + monologue pool + PC candidate
• Modders can add archetype support to districts without touching the archetype definition
• Different districts can customize the same archetype (a smuggler in Sova feels different from a smuggler in a frontier settlement)

Decision proposal (D-WR-16): The Settled Reach supports 6 PC archetypes (Smuggler, Detective, Politician, Guardian, Merchant, Civilian). Each district supports 2-4 archetypes, with a mandatory adversarial pair (insider + investigator). Archetype definitions are global; archetype instances are per-district.

---

7. The Authoring Pipeline at Scale: Districts as Production Units

The numbers

Let me be honest about what "full game" means in content terms.

Assumptions:

• 20-25 star systems
• 1-4 districts per system (some systems have one major station, others have multiple locations)
• ~50 total districts
• ~15-20 NPCs per district average (smaller districts: 10, larger: 30+)
• ~750-1,000 total NPCs

Content volume per tier:

Tier	NPCs per District	Total (50 districts)	Authored Lines per NPC	Total Authored Lines
Tier 1 (FRIEND/MIRROR)	2-4	100-200	70-100	7,000-20,000
Tier 2 (working profile)	8-12	400-600	15-20 (+ 4x gen)	6,000-12,000
Tier 3 (sketch)	5-10	250-500	5-10	1,250-5,000
Totals	15-26	750-1,300		14,250-37,000 authored

Plus:

• Monologue pools: 30-50 lines per archetype per district × 2-3 archetypes × 50 districts = 3,000-7,500 monologue lines
• Environmental text: 20-40 items per district × 50 districts = 1,000-2,000 environmental text items
• News ticker: 15-20 headlines per district × 50 districts = 750-1,000 ticker headlines

Total authored content: ~20,000-50,000 unique text items. That's a LOT. But it's spread across the full game's lifecycle (years of development + DLC). And a significant portion is generation-expandable.

The production pipeline

Districts are the unit of production. Each district goes through a consistent pipeline:

DISTRICT PRODUCTION PIPELINE

Phase 1: DESIGN (1 week)
├── District concept: moral texture, adversarial structure, cultural group
├── Social site definitions: count, type, size, physical layout
├── Triangle design: 3-6 triangles with tension types
├── NPC role slots: who is needed (FRIEND, coordinator, bridge, etc.)
├── PC archetype assignments: which archetypes, adversarial pair
└── Output: district.yaml + template files + triangle files

Phase 2: POPULATION (1-2 weeks)
├── NPC generation: names (from cultural brief), personalities, routines
├── Tier assignment: which NPCs are Tier 1/2/3
├── Relationship mapping: who knows whom, what they know
├── Hidden network design: membership, activities, contradiction hooks
└── Output: NPC YAML files (all tiers)

Phase 3: TIER 1 AUTHORING (2-3 weeks)
├── FRIEND profiles: full 10-axis, contradiction arc, tell progression
├── MIRROR profiles (if applicable): emotional arc, vulnerability map
├── PC-candidate profiles: PC appendix (B1-B7)
├── Monologue pools: per-archetype character-specific lines
└── Output: Tier 1 NPC files + monologue YAML files

Phase 4: TIER 2 AUTHORING (1-2 weeks)
├── Working profiles: 10-axis, voice samples, dual-lens notes
├── Core dialogue lines: 15-20 per NPC
├── Generation expansion: 4x from core lines
├── Triangle integration: tension-specific dialogue
└── Output: Tier 2 NPC files + dialogue YAML files

Phase 5: TIER 3 + ENVIRONMENTAL (1 week)
├── Sketch profiles: want, routine, personality minimum
├── Environmental text: signs, menus, graffiti, ticker headlines
├── Location descriptions: atmospheric text per social site
└── Output: Tier 3 NPC files + environmental text files

Phase 6: INTEGRATION (1 week)
├── Pool configuration: FRIEND pool, role-slot pools, randomization axes
├── Cross-reference validation: all canonical_ids resolve
├── Schema validation: all content passes validation
├── Dual-lens review: both archetypes' experiences coherent
└── Output: pools.yaml + validated content pack

Phase 7: PLAYTESTING (1-2 weeks)
├── Per-archetype playthrough: does each archetype's experience work?
├── Triangle testing: do all triangles produce interesting decisions?
├── FRIEND testing: does the contradiction arc hit emotionally?
├── Randomization testing: do different seeds produce different experiences?
└── Output: Bug fixes, balance adjustments, polish

Total per district: 8-12 weeks. That's 2-3 months per district with a small team. For 50 districts, that's ~100-150 team-months = roughly 4-5 years with a 3-person content team, or 2-3 years with a 5-person team.

This is a multi-year production. Which is expected for a game of this scope. The key is that each district is INDEPENDENT — they can be built in parallel, shipped as DLC, and contributed by modders.

Generation vs. authoring: the 70/30 split

Not all content needs to be hand-authored. The pipeline uses three content creation methods:

Method	What It Produces	Quality Level	Volume
Hand-authored	Tier 1 profiles, FRIEND arcs, MIRROR arcs, key monologue, key dialogue, environmental text	Highest — every line is specific, personal, irreplaceable	~30% of total content
Generation-expanded	Tier 2 dialogue expansion (write 15, generate 60), routine micro-variations, Tier 3 personality variations	Good — matches voice and tone but lacks hand-crafted specificity	~40% of total content
Template-generated	Tier 3 NPC profiles, procedural district layouts, random NPC names + basic personalities, filler environmental text	Functional — serves as social wallpaper, background noise	~30% of total content

The critical constraint: Tier 1 is NEVER generated. FRIEND arcs, MIRROR arcs, contradiction beats, tells — these are always hand-authored. The emotional payload requires authored specificity. Gore is right: emotion doesn't template.

Tooling requirements

The pipeline above requires specific tools. Here's what the content team needs:

1. District Designer (design-time tool)

• Input: Moral texture, cultural group, adversarial structure type, district size
• Output: Skeleton district.yaml + templates/*.yaml + triangles/*.yaml with placeholder NPCs
• Validates: Triangle structures have minimum members, adversarial pair exists, FRIEND slot defined
• Who uses it: Gestalt + Paula (design phase)

2. NPC Generator (population-time tool)

• Input: Cultural group brief + role slot + tier assignment
• Output: NPC YAML with name, personality, routine, basic relationships
• Validates: Name follows cultural phonological rules, routine uses canonical activities, faction is valid
• Who uses it: Paula (population phase), modders (extending districts)
• Note: Generates STARTING POINT — Tier 2+ always gets human review

3. Line Expander (authoring-time tool)

• Input: Core dialogue lines + voice card + tag taxonomy
• Output: Generated dialogue variants matching voice, mood, and tag patterns
• Validates: Generated lines preserve access tier, trust level, situation context
• Who uses it: Mellanie (Tier 2 expansion)
• Note: NEVER used for Tier 1. Expansion factor: 3-5x (write 15, generate 45-75)

4. Content Validator (continuous tool)

• Input: Full content directory
• Output: Schema validation report (errors + warnings)
• Validates: All cross-references resolve, all pool candidates exist, all triangles have required members, all Tier 1 NPCs have full profiles, bidirectional knowledge consistency
• Who uses it: Everyone, on every commit
• Note: This is Tyre's make validate-content at full scale

5. Content Previewer (testing tool)

• Input: District content pack + PC archetype + seed
• Output: Simulated playthrough of dialogue selection, monologue triggers, routine observation
• What it shows: "On this seed, with this archetype, entering this location at this time, the player would see X NPCs, hear Y dialogue, trigger Z monologue"
• Who uses it: Everyone during integration and playtesting
• Note: Evolution of the line previewer (#193), expanded to district-scale

6. Cultural Consistency Checker (validation tool)

• Input: NPC YAML + cultural group brief
• Output: Report of cultural violations (names don't match phonological rules, vocabulary not from substrate, social norms violated)
• Who uses it: Miri (review), content authors (self-check)

The modder pipeline

Modders follow the same pipeline but can enter at any phase:

Mod Type	Entry Phase	What They Produce	What They Need
New NPC mod	Phase 2	1 NPC + pool assignments	NPC Generator + Content Validator
New district mod	Phase 1	Full district content pack	Full pipeline
Extension mod	Phase 4	Additional dialogue lines for existing NPCs	Line Expander + Content Validator
New archetype mod	Phase 1	Global archetype definition + district instances	Full pipeline + archetype schema
Cultural mod	N/A	New heritage cluster brief	Cultural Consistency Checker

The critical principle: modders use the same tools as the core team. No special modding SDK. The content directory structure IS the modding API. The validation tools ARE the modding tools. This collapses the "mod support" development cost to near zero — we're not building two pipelines.

Decision proposal (D-WR-17): Districts are the unit of content production. Each district goes through a 7-phase pipeline (Design → Population → Tier 1 → Tier 2 → Tier 3 → Integration → Playtesting). Six tools support the pipeline (District Designer, NPC Generator, Line Expander, Content Validator, Content Previewer, Cultural Consistency Checker). The same tools serve both core development and modding. Content is ~30% hand-authored, ~40% generation-expanded, ~30% template-generated. Tier 1 is never generated.

---

Synthesis: The Content Strategy as a System

Let me zoom out and map how all seven answers interact. This is where it gets good — this is where the systems produce emergent variety.

The multiplication table

The full game's content variety isn't additive — it's MULTIPLICATIVE.

TOTAL UNIQUE PLAYTHROUGHS =
  PC archetypes (6)
  × Starting districts (~50, not all available to all archetypes)
  × FRIEND selection (2-4 per archetype per district)
  × Hidden network configuration (per-seed role assignment)
  × Faction power balance (per-seed)
  × Wormhole topology (per-seed)
  × Moral texture experience (emergent from the above)

Conservative estimate: 6 archetypes × 30 average district-archetype combinations × 3 average FRIEND candidates = 540 structurally distinct first-district experiences. Before dramatic randomization. Before multi-district play. Before inter-system travel.

The two-friends-compare-notes test passes at this scale not because we randomized everything — but because we randomized the RIGHT things while keeping the authored core rock-solid.

The feedback loops

Here's what makes me excited as a systems designer: the seven components INTERACT.

1. Cultural groups shape moral textures. A Nordic-Baltic district naturally produces complicity (privacy norms protect networks). A Latin American district naturally produces loyalty (patron-client dynamics create competing obligations). The culture isn't decoration layered over the mechanics — it IS the mechanics.

2. Moral textures shape FRIEND arcs. In a complicity district, the FRIEND hides criminal participation. In a loyalty district, the FRIEND hides factional betrayal. In a duty district, the FRIEND hides knowledge of someone in danger. The FRIEND pattern is universal; what the FRIEND hides is determined by the district's moral texture.

3. Archetypes shape randomization. A detective doesn't care which NPC fills the "courier" role slot — they care which NPC has observable tells. A smuggler doesn't care about tells — they care about operational reliability. The SAME randomization (who fills what role) produces DIFFERENT player-facing experiences based on archetype.

4. The content directory enables all of the above. Cultural briefs in regions/. Archetype definitions in global/archetypes/. District skeletons in systems/. Pool configurations in each district. Mod overlays that extend any of these. The directory structure isn't just file organization — it's the game's content API.

5. The authoring pipeline scales because districts are independent. No district depends on another district's content. You can build Sova first, ship it, then build Tavros, ship it, then let modders build custom districts. The pipeline's parallel nature means content velocity scales linearly with team size.

The ten decisions

#	Decision	Domain
D-WR-11	Master theme is "knowledge as burden." Per-district moral textures emerge from content configuration, not prescribed labels.	Theme
D-WR-12	6-8 heritage clusters with Miri-quality regional briefs. Mechanically load-bearing. Districts reference cluster briefs.	Culture
D-WR-13	Universal district skeleton (social sites + hidden network + FRIEND + cultural group + moral texture + archetypes + contested resource). All districts instantiate this skeleton with varying flesh.	Structure
D-WR-14	Content directory supports core/DLC/mod delivery through pack manifests. Cross-district references via canonical_id. Missing references degrade gracefully.	Architecture
D-WR-15	Randomization: "randomize context, author content." Four tiers (Structural/Dramatic/Local/Meta). Authored content is never generated at runtime.	Randomization
D-WR-16	6 PC archetypes (Smuggler, Detective, Politician, Guardian, Merchant, Civilian). 2-4 per district. Mandatory adversarial pair. Global definitions, local instances.	Archetypes
D-WR-17	Districts as production units. 7-phase pipeline. 6 tools. Same tools for core and mod development. ~30/40/30 authored/expanded/generated split. Tier 1 never generated.	Pipeline
D-WR-18	Content variety is multiplicative, not additive. The systems interact: culture → moral texture → FRIEND arc → archetype experience → randomization → replay value.	Synthesis
D-WR-19	Heritage cluster distribution: 20-25 systems across 6-8 clusters, with blended districts interpolating between parent cultures.	World structure
D-WR-20	The adversarial pair is the minimum content unit: one insider archetype + one investigator archetype per district, seeing the same world from opposing information positions.	Design constraint

---

The full game isn't 50 copies of Sova with different wallpaper. It's 50 instantiations of a universal skeleton, each with different flesh — different cultures, different moral textures, different adversarial structures, different contested resources. The systems are the same everywhere. The content makes each place unique. And the randomization ensures that no two playthroughs — even in the same district, with the same archetype — tell the same story.

That's not a content pipeline. That's a story machine.

— Gestalt