settled-reach/docs/workshops/wiki-review/SUMMARY.md

Wiki Review Workshop — Summary & Long-Term Content Strategy

Workshop Overview

Date: 2026-02-12 Called by: Jeroen (project lead) Participants (10): Paula (librarian), Mellanie (librarian), Miri (worldbuilder), Gestalt (systems design), Gore (thematic lead), Nigel (sandbox/replayability), Ozzie (player experience), Tyre (tooling/architecture), Qatux (documenter), SI (work prep) Format: 4 rounds + lead interview between R3 and R4 + cross-agent debates Output: ~35 response documents, 3 tracking documents, 1 lead interview, this summary

Scope Evolution

The workshop began as a v0.1 wiki review (45 files in wiki/, ticket #368) with 6 defined topics: direction check, style guides, regional style guide, wiki taxonomy, content directory structure, and randomizer/PC briefs. Rounds 1-2 produced 88 tracked proposals and 20 consensus items focused on v0.1 content quality: NPC profiles, smuggler lens parity, triangle strengthening, naming conventions, and content directory structure.

Between Rounds 2 and 3, the lead redirected the workshop toward long-term content strategy. Round 3 asked "What does the full game look like?" and collected scaling estimates, cultural group proposals, NPC pattern libraries, and PC archetype lists from all contributors — all calculated at a 20-district scale.

The lead interview between Rounds 3 and 4 delivered the decisive reframe: 300 worlds before DLC. This transformed the workshop from a content authoring strategy into a generator specification strategy. Round 4 asked every contributor to rebuild their proposals for a world where style briefs are generator inputs, cultural groups are assembled from ingredient menus, and hand-authoring is reserved for irreducible emotional content.

The 300-World Reframe (Round 4)

The 300-world number changed the category of work from "writers produce districts, tooling accelerates writers" to "engineers produce generators, writers produce generator inputs, tooling IS the product." Every Round 4 response independently arrived at the same conclusion: the workshop had been designing generator specifications all along without knowing it. The district skeleton, the composition matrix, the Sacred/Profane distinction, the pool architecture — all were already generator-shaped thinking.

---

Long-Term Strategy Decisions (Confirmed)

Scale: 300 Worlds

The game targets 300 populated worlds before any DLC content. This requires a generator model, not hand-authoring.

Release path:

• v0.1 = proof of the interaction model (1 district, hand-authored)
• v0.2-v0.5 = testing generation systems and tooling
• v0.6-v0.10 = content pipeline at scale
• pre-v1.0 = 300 worlds populated via generators + style briefs + naming algorithms

Tooling first is absolute. The authoring/generation pipeline must be proven before scaling content.

Three-tier world authoring:

World Tier	Count	Content Method	Purpose
Landmark worlds	10-15	Fully hand-authored, bespoke NPCs and arcs	Emotional peaks, quality benchmarks
Regional worlds	30-50	Template kits, hand-reviewed	High quality, cultural depth
Generated worlds	230-260	Procedurally generated from ingredient menus	Breadth, replayability, exploration

Quality gradient maps to in-world logic. Older settlements are denser and richer. Core worlds have more content than frontier outposts. The gradient is intentional and mirrors real human geography — players feel the difference between a 300-year-old trade hub and a 40-year-old mining outpost, and the content density reflects that.

Irreducible authored content at 300-world scale:

• Gestalt: ~1,600-2,800 hours total for templates, arcs, rules, and benchmark districts
• Mellanie: ~3,300-7,400 authored lines + ~90 pages of specs (vs. 13,500-31,200 lines for 20 hand-crafted districts)
• The authored content DECREASES in volume but INCREASES in per-line stakes — every seed line, every parameter definition cascades across hundreds of worlds

QA paradigm shift: Test generators, not generated output. Bad parameters cascade across 300 worlds. Review samples, tune parameters, iterate. The gut-check reviews generators, not individual NPCs.

Cultural Ingredients Menu

Core principle: Opt-in, not fill-all-blanks. Gaps ARE flavor. What a society DOESN'T care about is as defining as what it does.

Six ingredient categories:

Category	Purpose	Options
Heritage Roots (OPTIONAL)	Phonetic flavor, social dynamics, trust model	8 Earth-grounded clusters + null
Settlement Motivation	Why people came, what they value	Economic, ideological, corporate, refugee, scientific, military, frontier, generational
Economic Function	What makes money, daily rhythm, what tempts	Logistics, extraction, manufacturing, trade, agriculture, research, military, services, transit
Philosophical Alignment	Social contract, blind spots	Collectivist, individualist, meritocratic, egalitarian, traditionalist, progressive
Corporate/Faction Presence	Power structure, authority dynamics	Commission, Syndic, contested, independent, criminal, military
Drift Stage	Cultural evolution state	Founding (<50yr), establishing (50-150yr), mature (150-300yr), ancient (300+yr)

Heritage is OPTIONAL. Null-heritage produces distinct corporate/functional worlds with no cultural immune system, no substrate vocabulary, no food traditions. The world feels THIN — and that thinness is its own moral texture.

8 Earth-grounded heritage clusters (Miri named them by social signature: Frost, Tide, Iron, Spice, Jade, Dust, Vine, Salt, Stone, Arc — 10 roots that map to Nordic-Baltic, Mediterranean, Slavic-Industrial, South Asian, East Asian, West African, Latin American, Anglo-Commercial, and two additional profiles). Heritage recognizable but not mappable — "feels Nordic" not "is Finnish." Drift is the defense against franchise similarity, not abstraction.

People migrate between heritages. A character with a Nordic-sounding name in an East Asian-heritage district is not wrong — they are a person with a history. Culture is the place, not the individual.

Missing from initial proposals, added by lead: Economics. What makes extra income tempting. This determines the nature of the contested resource — whether the smuggling is sympathetic (healthcare for people who cannot afford it) or predatory (luxury goods for status). The economic pressure ingredient determines the moral texture of the entire grey economy.

The derivation function: The generator takes ingredient selections and derives mechanical parameters deterministically. Same ingredients always produce the same society. Randomization happens at ingredient SELECTION, not at derivation. Style briefs (Miri's Krenn brief) are reference implementations — proof that the derivation function produces quality output for specific ingredient sets.

Cultural Brief = Generation Seed

Dual artifact: seed.yaml (generator input) + brief.md (human authoring reference). The seed is the product; the brief is the manual.

World seed = numeric seed (pool draws) + cultural assembly (content generation). Same cultural assembly + different numeric seed = same culture, different drama. Different cultural assembly + same numeric seed = parallel structure, different culture.

Miri's Krenn brief is a proof-of-concept output; the ingredient menu is the real product. The generator must reproduce the Krenn brief's feel when given Krenn's ingredient combination. Validation: if the generator's output for frost:0.55 + salt:0.30 + iron:0.15 / economic-trade / logistics / tight-margin+prohibition matches the hand-authored Krenn culture, the derivation function works.

Three-System NPC Architecture

System A: Thematic Patterns (what the character means to the story)

Pattern	Function
FRIEND	Trust anchor hiding a contradiction. Emotional core.
MIRROR	Honest person whose transparency reveals others' dishonesty.
ANCHOR	Embodies contentment. Genuine answer to "Is this enough?"
GHOST	Absent power shaping events through influence.
CATALYST	Outsider who forces reckoning. Disruption.
THRESHOLD	Gatekeeper between the player and the next layer.
REMNANT	Living history. Remembers when things were different.
SYSTEM	Institutional embodiment. Procedure-citing, neutral.
NOBODY	Background that becomes foreground through player attention.

System B: Functional Motivations (what the character does in gameplay)

Motivation	Role
HANDLER	Managing operations, applying pressure from above
WITNESS	Accidentally observed something, carries information
TURNCOAT	Loyalty shifting, moving between camps
CIVILIAN	Uninvolved, living their life, normalcy
OPERATOR	Running something, competent, goal-oriented
SKEPTIC	Doubts the system, questions assumptions

System C: Composition. Each NPC gets one thematic pattern + one functional motivation. ANCHOR+OPERATOR differs from ANCHOR+CIVILIAN. At 300-world scale, generators pick combinations from composition rules with drama ratings (gold/silver/bronze/banned), forbidden combination lists, and soft constraints ensuring dramatic richness per district.

Key composition examples:

• FRIEND+OPERATOR = identity betrayal (Kael: "I didn't know who you really were")
• MIRROR+CIVILIAN = uncomplicated honesty (Naia Tamm: innocence endangered by events)
• NOBODY+WITNESS = hidden information activated by player attention
• GHOST+HANDLER = absent authority applying pressure through intermediaries

THE NOBODY & Evolving Relationships

THE NOBODY = just being there. Background that becomes foreground through player attention. The lead explicitly opened the "evolving relationships" mechanic: "evolving relationships sounds like a can of worms I would LOVE to open."

Dynamic tier promotion:

Stage	Trigger	Effect
NOBODY	Default	Name, routine, greeting. Social wallpaper.
NOTICED	~3-5 interactions	NPC gains a name in the player's knowledge graph.
RECOGNIZED	~8-12 interactions	Basic dialogue unlocks. "Her name is Tev."
KNOWN	~15-20 interactions	Dormant Tier 2 content activates. Functional motivation visible.
INVESTED	~25+ interactions + specific trigger	Full dormant pattern activates. NPC becomes ANCHOR/WITNESS/etc.

At 300 worlds, most NPCs are NOBODYs. Player attention is the resource. The generator pre-assigns hidden motivations: 70% CIVILIAN (stay NOBODY forever), 20% activatable motivations, 10% load-bearing motivations that could reshape the investigation.

Key quotes on THE NOBODY:

• Nigel: "Two players, same world, same seed, different experience because they noticed different people."
• Gore: "Learning to see is the endgame."
• Ozzie: "THE NOBODY is an infinite replayability engine. The randomizer shuffles the pool. THE NOBODY lets the player shuffle it again, by choosing what to see."

THE MIRROR Pattern

Tier 1 emotional depth, NO contradiction arc, no secret. The honest person in a world of liars. First instance: Naia Tamm (renamed from Hael — Miri proposed, unanimous).

Gore proposed THE MIRROR as a formal pattern; Paula converged independently, describing it as "Tier 1.5 Production Emotional NPC" with 50-70 authored lines. Both agree: the MIRROR must NOT have a contradiction arc or secret. The MIRROR's power is simplicity and transparency. When events drag the MIRROR into the conspiracy, the damage is devastating — the player broke the one genuinely innocent person.

The Kael/Hael sonic collision (tension T-03 from Round 1) was resolved by renaming Hael to Naia Tamm, following Krenn naming conventions from Miri's regional brief. Unanimous support.

8 PC Archetypes (Fluid)

8 archetypes at v1.0 as a stress test of system flexibility. The specific 8 vary between contributors (Gore: Worker/Investigator/Operator/Wanderer/Builder/Seeker/Guardian/Rebel; Paula: Insider/Investigator/Newcomer/Merchant/Administrator/Caretaker/Fixer/Exile; Ozzie: Insider/Investigator/Newcomer/Operator/Watcher/Exile/Fixer/Believer). The exact list is not yet canonized, but the recurring archetypes across all proposals are: Insider/Smuggler (universal), Investigator/Detective (universal), Newcomer (5+ agents), Operator (4+ agents), and Merchant/Fixer (4+ agents).

Archetypes are FLUID — current position in a community, not character class. Players can move districts, change jobs, buy houses. Archetype transitions are a game mechanic.

Paula's "vulnerable window": During transition, the player has given up their old position's protections but has not earned the new one's. That gap is where the most interesting gameplay lives. The Insider who becomes an Investigator faces a period where the ring knows they left but the Commission does not fully trust them yet.

Insider + Investigator deep for v0.1 but must NOT dominate. The lead specifically warns that the other archetypes must feel equally primary in their districts.

Newcomer = v0.2 content update centerpiece. Natural tutorial, fresh eyes on established world. The Newcomer is the player's proxy — walking into a place they do not understand, meeting people they do not know, trying to figure out what is going on. The Newcomer's "Is this enough?" is the most raw version: "Is this place enough? Are these people enough?"

Sacred / Profane / Middle Kingdom Randomization

The lead approved Nigel's Sacred/Profane/Middle Kingdom framework with a key refinement:

• Sacred = rules of the system (never randomize). "You don't randomize mass interactions."
• Profane = parameters drawn from pools. "You DO randomize gravity."
• Middle Kingdom = sacred structure, profane cast. Fixed dramatic architecture, variable characters filling it.

Full system mapping (from Gestalt's comprehensive table):

System	Sacred	Profane	Middle Kingdom
Cultural generation	Derivation rules, access tier formulas	Which ingredients selected per world	Cultural parameter set structure
NPC generation	Composition matrix, tier requirements	Which NPCs fill which roles, names	NPC profile structure, pattern templates
FRIEND system	5-phase arc structure, tell mechanics	Which NPC is FRIEND, specific contradiction	FRIEND template with cultural skinning
NOBODY system	Promotion trigger types, tier transitions	Which NOBODYs have which paths	Promotion path templates
District structure	Minimum social sites, NPC count, adversarial pair	Social site config, NPC assignments	District skeleton
Gate topology	Connectivity requirements, diameter limit	Which worlds connect, travel times	Small-world network properties
Storyteller	Pacing algorithm, escalation phases	Timing of events, pool selections	Storyteller decision nodes

Gate topology as macro randomization:

• Core worlds Sacred (always connected)
• Peripheral worlds Profane (vary per playthrough)
• Accessible count Middle Kingdom (number fixed, which ones varies)
• Unlockable worlds through gate events/quests = chapter-level randomization
• Future: planetary owner/developer DLC

Light generation — pools of authored content, not generated from scratch. The generators PICK from authored templates, COMBINE them with cultural parameters, and DERIVE specifics. The authored template library is the product; the generator is the assembly line.

Interaction Model

WASD direct control with context-sensitive interaction — unanimous (9/9 in Round 4 discussions). Player IS the character. Body is the attention mechanism.

Direct embodiment required for: complicity, NOBODY tier promotion (player chooses where to look), occlusion (information hidden behind physical presence), FRIEND betrayal (proximity creates emotional weight).

• Gestalt: "Hotline Miami movement + Disco Elysium interaction"
• Paula: single-action prompts that change based on proximity + relationship + knowledge
• Dedicated workshop brief written for full design: docs/workshops/control-interaction/

Generator Pipeline

Old model: Writers produce districts, tooling accelerates writers. New model: Engineers produce generators, writers produce generator inputs, tooling IS the product.

Gestalt's 6-phase generator development pipeline:

1. INGREDIENT AUTHORING — heritage roots, motivations, economic functions, alignments, factions, drift stages, derivation rules
2. TEMPLATE AUTHORING — FRIEND arcs (30-50), MIRROR profiles (15-25), NOBODY promotion templates (40-60), triangle type templates (10-15), voice cards (30+), PC archetype definitions (8)
3. GENERATOR DEVELOPMENT — Cultural Parameter Generator, NPC Generator, FRIEND/MIRROR Skinner, District Composer, Naming Algorithm, Dialogue Line Expander, Gate Topology Generator, NOBODY Promotion Generator
4. VALIDATION DEVELOPMENT — schema validation, composition rule checker, constraint satisfaction verifier, cultural coherence checker, emotional integrity checker, diversity metrics
5. GENERATION + REVIEW — generate 300 worlds, automated validation, human review of FRIEND/MIRROR instances (100%), spot-check generated districts (5-10%)
6. HAND-ELEVATION — starting districts get polish, hub worlds get additional depth, narrative-critical worlds get authored threads

Mellanie's content math:

• 300 worlds: ~3,300-7,400 authored lines + ~90 pages of specs
• 20 hand-crafted districts (Round 3 model): 13,500-31,200 authored lines
• The 300-world model requires LESS hand-authored content because the generator does the multiplication, but every authored line is higher-stakes

---

Theme — Open, Converging

Candidates from the team:

Candidate	Author	Perspective
"Cost of participation"	Gore	The individual inside the system
"Collective bargain"	Mellanie	The community
"What do you owe the people around you?"	Ozzie	The relationships
"What do systems do to people inside them?"	Miri	The system itself
"Cost of connection / entanglement"	Paula	The relational web
"Knowledge as burden"	Gestalt	The information
"Who do you choose to see?"	Ozzie (Round 4)	The player's attention

Current direction: "Don't pick, derive" (Gore). All candidates orbit the same field. The derivation function guarantees every generated world falls within the thematic field. Let it resolve through the work.

Gore's formulation: "The theme is the field, not any specific phrasing of it. The field is: what does it cost to be human inside something bigger than yourself?"

The lead has not canonized a phrasing. Round 4 consensus: complicity is Sova-local; the universal theme is structural but unnamed. Every ingredient combination includes a system failure mode. Every failure mode produces a cost of participation. Every cost produces a version of "Is this enough?" The field is built into the generator's architecture.

---

Decisions from Rounds 1-2 (v0.1 Specific, Pending v0.1 Scoping Round)

These were produced in Rounds 1-2 before the long-term reframe. They remain valid for v0.1 but need re-evaluation in light of the 300-world strategy:

• Hael renamed to Naia Tamm (Miri proposed, unanimous) — resolves Kael/Hael sonic collision
• All NPCs get full canonical names (Miri generated 12 for Sova)
• Smuggler lens gap needs fixing — wiki is detective-shaped; smuggler needs operational attributes (trust_read, exposure_risk, loyalty_assessment, leverage_held, social_debt, moral_weight, cover_integrity, loyalty_conflict) and moral arc phases (Comfort, Doubt, Reckoning, Compromise)
• Triangle 1 (Hub Power) needs Nils-vs-Voss escalation decision — weakest triangle, smuggler role is passive, needs concrete fork
• Routine format: 2-layer (YAML + prose) — engine-readable schedule alongside human-readable description
• Content directory structure: global/ + districts/ as top-level split (Tyre + Gestalt converged)
• FRIEND pool: v0.1 = 1 per character (Kael, Sera), v0.2 = 2, architecture from day one. Pool-based selection, full Tier 1 minimum, no exceptions
• Primary contraband fixed per district, secondary varies per seed (medical/severance/enhanced/nothing)
• Cultural gate modifier on dialogue access — regional culture modifies disclosure independently of relationship state
• Drin may need tier promotion — doing Tier 2 work in a Tier 3 profile (Triangle 2, Triangle 5, compromised by Harek)
• PC-as-NPC Concept A for v0.1 — other PC exists as NPC (~20-30 authored items), full identity randomization deferred to v0.2+

---

Deferred to Future Workshops

• Control & Interaction Scheme (brief written: docs/workshops/control-interaction/)
• Perception system gravity variation for world transitions — flagged by lead, endorsed as top priority by Ozzie, needs perception team
• Full NPC pattern composition rules — forbidden/preferred lists drafted by Gore, Gestalt, Paula; need consolidation and validation
• Gate topology design — network structure, hub placement, connectivity constraints, small-world properties
• Storyteller cultural literacy — how the storyteller adapts pacing to cultural trust-building rates
• LLM-assisted content pipeline design — Tyre proposed "write 10, generate 40" for Tier 2 expansion
• Modding toolkit specification — content pack manifest, ADD/REPLACE/MERGE overlay operations, template role-slot system
• Inter-world political generation — Paula proposed macro political layer: trade dependencies, cultural friction, historical grievances between worlds
• IP originality validation at scale — Miri proposed automated cultural distance validator to prevent franchise similarity in generated worlds

---

Ticket Recommendations

Pending v0.1 scoping round. SI has a preliminary list of 28 tickets from Round 2 that needs revision in light of the strategy reframe. Key areas for ticketing:

• Ingredient menu schema definition
• Derivation rule specification
• FRIEND arc template authoring (first 3-5)
• NOBODY promotion mechanic design
• Cultural parameter generator prototype
• Naming algorithm prototype (phonetic rules, not word lists)
• Composition matrix formalization
• Archetype voice card conversion (prose to axis-based specs)
• Naia Tamm rename cascade across wiki files
• Smuggler attribute vocabulary addition to wiki
• Triangle 1 strengthening (Nils stub page + decision moment)
• Gate topology design brief

---

Workshop produced: 4 rounds x 8-9 contributors = ~35 response documents + lead interview + 3 tracking documents. Total workshop output in docs/workshops/wiki-review/.