settled-reach/docs/workshops/content-gap-analysis_v0_1/round1-mellanie.md

Round 1 — Mellanie's Content Gap Analysis

Agent: Mellanie (Copywriter) Date: 2026-02-11 Scope: All 9 layers, copywriting perspective

---

Layer 1: Art Direction & Visual Identity

Not my lane directly, but I'll flag the intersection: visual style determines how much weight text carries. D-014 says "functional boxes with labels" and D-016 says "atmosphere carried by monologue." If the art stays minimal, my text is doing 80% of the atmosphere work. If the art gets moody, the text can be sparser.

My ask for Araminta: Tell me the visual fidelity target so I can calibrate density. Minimal art = more monologue lines per location. Moody art = fewer, sharper lines. I need to know which world I'm writing for.

---

Layer 2: Narrative Framework & Character Design

Q3: What do I need from the character builds before I can write voice-consistent dialogue?

Five things. Non-negotiable.

1. Voice signature. Not personality traits — speech patterns. Short sentences or long? Profanity or clean? Technical vocabulary or street? Internal monologue style: does the smuggler think in observations or feelings? Does the detective think in hypotheses or facts?

Here's what I mean concretely:

Smuggler sees a cargo container out of place: Option A (cautious, observational): "That container's been moved. Wasn't there this morning." Option B (anxious, relational): "Kael moved the container again. He's going to get us all caught." Option C (sardonic, detached): "Someone rearranged the furniture. How creative."

Each of those is a different character. I need to know which one before I write 500 lines.

2. What they notice. The smuggler and detective look at the same room and see different things. Not just because of information boundaries — because of attention. The smuggler notices exits, security cameras, who's watching. The detective notices inconsistencies, nervousness, things out of place. This shapes every monologue line and every dialogue choice.

3. Relationship to authority. How do they talk to people with power over them? How do they talk to people they have power over? The smuggler with their ring boss vs. the detective with their precinct captain — those power dynamics shape every interaction.

4. What they lie about. Both characters have things they conceal or misrepresent. The smuggler hides their criminal activity. The detective might hide their investigation from suspects. Knowing what they lie about tells me what their "public voice" sounds like vs. their "internal voice."

5. Emotional baseline. Not mood — posture. Is the smuggler generally anxious, comfortable, or resigned? Is the detective generally confident, frustrated, or obsessive? This is the default register that all their lines return to between dramatic moments.

Proposal: I'll write a Voice Kit for each character — a 1-page document with these five elements plus 10 sample monologue lines covering common situations (entering a room, greeting an acquaintance, noticing something wrong, being lied to, being alone). Paula defines the narrative identity; I translate it into voice. The Voice Kit becomes the reference that keeps all dialogue on-character.

Dependency: Paula's character arc outlines (#180, #181) must come before my Voice Kits. I need to know who these people are before I can figure out how they sound.

---

Layer 3: Station District Setting

Not my primary deliverable, but I write text that lives in this space. My ask for Miri:

Give me the texture, not just the lore. I don't need a history essay. I need:

• What does this place smell like? (Monologue uses all senses.)
• What's the slang? Do people say "shift" or "rotation"? "Credits" or "sols"?
• What's on the walls — corporate signage? Safety warnings? Union notices? Graffiti?
• What's the ambient sound the monologue might reference — machinery hum? Ventilation? Crowd noise?
• What's the local complaint? Every community has one. Bad water. Late transports. Overpriced food.

These details are what make monologue and environmental text feel placed instead of generic sci-fi.

---

Layer 4: Content Pipeline & Authoring Format

Q2: Can I write content packs in a provisional format before the previewer exists?

Yes, with conditions.

I can write in any structured text format — YAML, Markdown with frontmatter, even a spreadsheet — as long as we agree on the tag taxonomy first. The actual file format matters less than the metadata structure. I can adapt format later; I can't retag 500 lines.

What I need locked before I write:

1. Tag taxonomy. The minimum set of tags per line. My proposal for v0.1:

Tag	Values	Purpose
character	smuggler, detective, any	Who can trigger this line
access	insider, outsider, authority, peer	D-028 Layer 1 filter
trust	surface, real, secret	D-028 Layer 3 disclosure tier
location	workplace, bar, ring, corridor, any	Where this line fires
situation	greeting, idle, gossip, confrontation, observed_anomaly, absence, routine	What's happening
mood	neutral, tense, relaxed, suspicious, friendly	NPC emotional state
topic	work, personal, crime, rumor, weather, complaint	What the line is about
type	dialogue, monologue, environmental, ticker	Content category

That's 8 tags. Enough to filter meaningfully, few enough to author quickly. Gestalt can add more if the system needs them, but I can start writing with these.

2. Line format. My proposed provisional format (Markdown-friendly, convertible to YAML/JSON later):

---
id: wp-dock-greeting-01
character: any
access: insider
trust: surface
location: workplace
situation: greeting
mood: neutral
topic: work
type: dialogue
speaker: dock_supervisor
---
Another day, another manifest. You know the drill.

If the actual format changes, I can batch-convert. The content and tags are what matter.

3. One example template fully tagged. Before I write 165 lines for the workplace, I want to write 20, tag them, and have Gestalt validate that the tags produce the right filtering behavior. Call it a "tag smoke test." This catches taxonomy problems before they compound.

What I do NOT need before starting:

• The previewer CLI (#193). Nice to have, not a blocker. I'll write blind and validate later.
• Final file format. Provisional is fine.
• Generation pass specs (#194). I write the authored core; expansion is a separate step.

---

Layer 5: Template Content (3 Social Sites)

Q1: What does a content pack deliverable look like? Break down #190.

Here's what the Workplace Content Pack (#190) contains, broken into concrete writing tasks:

A. Role Voice Kits (4-6 roles, ~5 lines each = 20-30 lines)

Each role in the workplace template gets a voice kit: a set of sample lines that establish how this type of person talks. Not a specific NPC — a role that gets filled by a generated NPC.

Proposed workplace roles:

• Dock Supervisor — authority figure, stressed, deadline-driven
• Cargo Handler — physical worker, pragmatic, gossip-prone
• Logistics Clerk — desk worker, detail-oriented, nervous or bored
• Maintenance Tech — gruff, observant, knows every corner of the facility
• Security Guard — bored or paranoid, depending on personality axis
• Shift Manager — middle management, caught between workers and bosses

Each kit: 5 lines covering greeting, idle chatter, reaction to anomaly, reaction to authority, and one line that hints at their secret/vulnerability axis.

B. Situation Scripts (tagged line pools — the bulk, ~80-100 lines)

These are the lines that fire during gameplay, filtered by tags. Organized by situation:

Situation	Lines	Description
Greetings	15-20	First contact lines, filtered by access tier and relationship
Idle/ambient	15-20	Background chatter, complaints, observations. The "daily life" texture
Work talk	10-15	Job-related conversation. Manifests, schedules, problems
Gossip	10-15	Trust-gated. Surface: weather, sports. Real: who's sleeping with whom. Secret: who's smuggling what
Confrontation	5-10	When triangle tensions surface. Accusations, denials, pleas
Observed anomaly	5-10	NPC reacts to something unusual. Feeds the observation chain
Authority interaction	5-10	How NPCs talk to/about the boss. Power dynamics

C. Monologue Hooks (per character, ~30-40 lines)

Lines that fire as the player character's internal monologue in this location:

Category	Smuggler	Detective
Entering workplace	8-10 lines: routine, familiar, noting who's in/out	8-10 lines: scanning, assessing, noting what seems off
Observing NPC behavior	5-8 lines: protective, concerned about exposure	5-8 lines: analytical, building hypotheses
Noticing anomaly	3-5 lines: alarm, calculating risk	3-5 lines: interest, cataloguing evidence
Absence/deviation	3-5 lines: "Where's Kael?" worry-flavored	3-5 lines: "Subject deviated from pattern" analytical

D. Environmental Text (~15-20 items)

Diegetic text that exists in the space:

• Shift schedule board (readable, updates with game time)
• Safety notices ("Report unauthorized cargo to Security — Management")
• Cargo manifest fragments (readable if you examine them — information source)
• Employee notice board (upcoming events, complaints, lost items)
• Terminal login screen text
• Door labels, section markers, warning signs

E. Trait Transformation Guide (~1 page)

How personality axes modify the base lines. Not per-trait scripts — guidelines:

Trait axis	Low	High	Effect on dialogue
Tolerance	Snaps fast, short answers	Patient, elaborate	Shorter/longer responses, more/fewer warnings before confrontation
Contentment	Bitter, complaining	Satisfied, dismissive	Tone of idle chatter. Discontented NPCs gossip more
Want (ambition)	Passive, accepting	Driven, scheming	Proactive vs reactive dialogue. High-want NPCs steer conversations

Total for one template: ~165-200 authored lines + environmental text + trait guide.

Three templates = ~500-600 authored lines of dialogue/monologue + ~50-60 environmental text items + 3 trait guides.

That's achievable. It's a focused sprint of writing, not a marathon. The quality bar is what takes time, not the volume.

---

Layer 6: NPC Profiles & Dialogue

Q1 (with Paula): How deep does each NPC profile need to be?

Two tiers of depth, matching the 30/50/20 split.

Flat NPCs (30%, ~4-5 NPCs): The 10 axes plus one distinguishing line. That's it. These people are wallpaper with a pulse. They greet you, they make small talk, they go about their routine. They exist so the station feels populated.

What I need per flat NPC:

• Name
• Role (which role voice kit to draw from)
• One distinguishing trait or habit (the dock worker who always eats during shift, the clerk who hums)
• 3-5 custom lines that make them slightly memorable (beyond the generic role pool)

Triangle NPCs (50%, ~7-8 NPCs) and Entangled NPCs (20%, ~3 NPCs): The 10 axes plus a voice paragraph. Not a full backstory — a paragraph that captures how this specific person talks and thinks, enough for me to write lines that feel specific to them.

What I need per triangle/entangled NPC:

• Name
• Role + which social sites they appear in
• 10-axis values (especially Want, Secret, Relationships, Contentment)
• One paragraph: "Kael talks like [X], thinks about [Y], is hiding [Z]. When stressed, he [behavior]. When comfortable, he [behavior]."
• Their triangle position: who they're in conflict with, what they want from the triangle
• 8-15 custom lines (beyond role pool) that express their specific voice and situation

Who writes these profiles? Paula defines the narrative identity and triangle dynamics. Miri provides cultural/setting flavor. I write the voice paragraph and sample lines. It's a collaboration, not a solo task.

Q2: How many lines per NPC for v0.1?

Let me do the real math.

Flat NPCs (~5): 3-5 custom lines each = 15-25 lines. Plus they draw from the role voice pool, so they have maybe 20-30 lines available total (role pool + custom).

Triangle NPCs (~8): 8-15 custom lines each = 64-120 lines. Plus role pool. Total available per NPC: 30-45 lines.

Entangled NPCs (~3): 12-20 custom lines each = 36-60 lines. These need the most specific, plot-relevant dialogue. Total available per NPC: 40-55 lines.

Custom NPC lines total: 115-205 lines.

These sit ON TOP of the template content packs (~500-600 lines of role-based dialogue). The role pools provide the substrate; NPC-specific lines provide the personality.

Grand total authored lines for v0.1:

Category	Lines
Template dialogue (3 packs)	500-600
NPC-specific lines (~15 NPCs)	115-205
Monologue — smuggler	100-150
Monologue — detective	100-150
Environmental text	50-60
UI/insert text	20-30
News ticker / unisphere	15-20
Total authored	~900-1,215

After 4x generation expansion on the template dialogue: ~3,000-3,600 total lines available to the system.

That's a solid content floor. It won't feel infinite, but it won't feel empty during a 30-minute session. The player will encounter maybe 150-200 lines per playthrough. Repetition becomes a problem only on the 3rd or 4th play of the same character.

---

Layer 7: Monologue & Observation Text

Q1: How many monologue lines make v0.1 feel alive?

The 200-300 estimate is the floor. Let me break it down by function:

Function	Smuggler	Detective	Total
Location atmosphere (entering/being in a place)	25-30	25-30	50-60
NPC observation (noticing behavior, tells)	20-25	20-25	40-50
Absence/deviation (expected NPC not present, routine broken)	10-15	10-15	20-30
Sound interpretation (hearing through fog)	8-10	8-10	16-20
Investigation/deduction (connecting dots)	10-15	15-20	25-35
Emotional/reflective (quiet moments, character depth)	10-15	10-15	20-30
Diegetic tutorial (hinting at available actions)	8-10	8-10	16-20
Mirror moments (hinting at the other perspective)	5-8	5-8	10-16
Total	~96-128	~101-133	~197-261

Call it 250 monologue lines as the authoring target. Some will overlap (same observation, different character voice), but most need to be distinct because the two characters think differently about what they see.

The generation pass doesn't work as well for monologue as for dialogue. Dialogue can be expanded with variations on the same sentiment. Monologue needs to feel specific — it's the character's private thoughts. Generic expansion will sound hollow. I'd estimate a 2x generation pass on monologue at best, giving us ~500 monologue lines total. That's enough for 2-3 playthroughs per character before noticeable repetition.

Q2: Same observation event, different monologue — separate pools or shared?

Separate pools. No question.

The smuggler and detective don't just say different words about the same thing — they see different things. The whole game is built on this. Shared pools with character filters would tempt us toward "same observation, different adjective." That's death for the core mechanic.

Here's the concrete difference:

Event: NPC enters the back room during off-hours.

Smuggler pool:

• "There goes Kael. Right on schedule."
• "Back room again. He better not be skimming."
• "Hope he remembered to lock the side door this time."

Detective pool:

• "Unauthorized access to restricted area. Noted."
• "That's the second time today. What's in the back room?"
• "He looked around before going in. People who belong somewhere don't check if they're watched."

Same event trigger. Completely different interpretation. The smuggler's lines reveal familiarity with the criminal activity. The detective's lines reveal analytical distance. You cannot get this from a shared pool with a character tag — the entire framing, vocabulary, and emotional register are different.

Structure: Two parallel monologue pools, one per character, keyed to the same observation event triggers. The observation event generator (#239) fires a trigger; the monologue system selects from the appropriate character pool. Tags on each line handle further filtering (location, mood, trust level, prior knowledge).

This doubles the monologue writing work, but it IS the game. If the monologue sounds the same from both characters, we've failed D-027's core success criterion. There is no shortcut here.

Mirror Moments

Paula asked about these in the brief. Here's what they look like from a writing perspective:

A mirror moment is when the monologue almost acknowledges the other character's viewpoint without breaking the fourth wall. The character glimpses the moral complexity of their position.

Smuggler mirror moment (watching the detective interview a colleague): "She's just doing her job. ...Aren't we all."

Detective mirror moment (learning that a suspect supports their family through smuggling): "Twelve dependents on a cargo handler's salary. The math doesn't work. But the math never does for these people."

These are rare — 5-8 per character across the whole session. They're the moments where the writing does the thematic heavy lifting. They plant the seed of "what would this look like from the other side?" without ever saying it explicitly. On a second playthrough as the other character, the player remembers these lines and they hit differently.

---

Layer 8: UI Text & Environmental Content

Q1 (with Araminta): What text appears on screen besides dialogue and monologue?

Here's the full inventory:

Insert/HUD Text

• Time display (diegetic — the character's neural insert showing time)
• POI labels on minimap ("Logistics Hub," "The Rusty Valve," "Sector 7 Maintenance")
• Status indicators ("Signal weak," "Insert interference detected")
• Notification text ("New contact added," "Location discovered")

Environmental Text (diegetic — exists in the game world)

• Door labels and room signs
• Corporate signage (company names, slogans)
• Safety/regulatory notices
• Employee notice boards
• Cargo manifest fragments (readable as investigation tools)
• Terminal screens (login prompts, error messages, data readouts)
• News ticker / unisphere feed headlines
• Graffiti (if the station has a rougher district area)
• Menu/price board at the bar
• Wanted notices or institutional postings

UI Chrome (non-diegetic — game interface)

• Dialogue choice labels (if we have explicit choices, not just proximity-triggered lines)
• Journal/knowledge entries (if #264 makes it into v0.1)
• Tooltip text on interactive objects ("Examine," "Talk," "Read")
• Save/load screen text
• Settings menu labels
• Pause screen text
• Character selection screen text (if any)

Diegetic Tutorial Text (via monologue — D-016)

This isn't separate UI text — it's monologue lines that teach mechanics:

• "I could check my insert for the area layout." (teaches minimap)
• "Better keep my distance if I'm going to follow him." (teaches follow mechanic)
• "That terminal might have shipping records." (teaches examine)
• "Sounds like an argument in the next room. I could get closer to hear." (teaches eavesdrop)

Volume estimate for non-dialogue/non-monologue text:

Category	Items	Lines
Insert/HUD	10-15 unique strings	10-15
Environmental — workplace	15-20 items	20-30
Environmental — bar	10-15 items	15-25
Environmental — ring	5-10 items	10-15
Environmental — corridors/shared	5-10 items	8-12
UI chrome	15-20 unique strings	15-20
News ticker headlines	10-15 items	10-15
Total		~88-132

This is quick writing — short strings, functional language, world-flavored. I can produce the full set in a focused session once I have Miri's setting document for terminology and Araminta's UI spec for constraints (character limits, display format).

---

Layer 9: Audio Direction

From a copywriting perspective: silence is fine for v0.1, but it shifts burden to me.

If there's no ambient audio, every sense except sight goes through monologue. The character has to narrate the soundscape:

"The ventilation system hums. Always that same frequency. I've stopped noticing it."

"Footsteps behind me. Two people. Unhurried."

"The bar's loud tonight. Good. Harder for anyone to overhear."

These sound-interpretation monologue lines are already in my budget (the "sound interpretation" category in Layer 7: 16-20 lines). If we add actual audio, some of those lines become redundant — the player hears the bar is loud instead of being told. But for v0.1, I'll write as if the monologue is doing all the sensory work. Audio can reduce the monologue load later.

My recommendation: Defer audio content. Let me carry the atmosphere through text. It's cheaper, faster, and we can always add audio on top. What we can't do is add text quality retroactively — so invest in the writing now, add sound later.

---

Cross-Layer Summary: What I Need to Start Writing

Here's my dependency chain, in order:

Step	What I need	Who provides it	Ticket
1	Station district setting brief	Miri	#189
2	Character arc outlines (smuggler + detective)	Paula	#180, #181
3	Dual Lens Authoring Guide	Paula	#261
4	Tag taxonomy agreement	Gestalt (confirm or modify my proposal)	#168
5	NPC profiles (10-axis + triangle positions)	Paula + Miri	#179, #184-186

What I can start NOW, before any dependencies resolve:

1. Voice Kit drafts for smuggler and detective — provisional, based on D-027 descriptions. I'll write three voice options per character and let Paula pick or blend. Gets us moving.
2. Environmental text templates — generic diegetic text (safety notices, terminal screens, door labels) that work in any station district. Setting-specific text waits for Miri.
3. Tag taxonomy proposal — the table in Layer 4 above. Gestalt validates, we iterate.
4. Monologue starter set — 20-30 monologue lines per character covering universal situations (entering a room, being alone, hearing something in fog). These don't depend on specific NPC profiles.

What I cannot start without dependencies:

• Template content packs (#190-192) — need NPC profiles and Dual Lens Guide
• NPC-specific dialogue — need NPC profiles
• Setting-specific environmental text — need Miri's district brief

---

Proposed New Tickets

Title	Priority	Type	Rationale
Smuggler voice kit	High	Story	Character voice reference for all smuggler dialogue/monologue. Blocks #190-192 content packs.
Detective voice kit	High	Story	Character voice reference for all detective dialogue/monologue. Blocks #190-192 content packs.
Monologue starter set (universal)	High	Story	50-60 monologue lines (both characters) for common situations. Unblocked — can begin immediately.
Tag taxonomy specification	High	Story	Define the metadata structure for all authored lines. Must precede content packs. Gestalt validates.
Environmental text templates (generic)	Medium	Story	Diegetic text items (signs, terminals, notices) that work across location types.
News ticker / unisphere content	Medium	Story	10-15 headline items for ambient world-building. Needs Miri's setting brief.
UI microcopy document	Medium	Story	All non-diegetic interface strings. Needs Araminta's UI spec.
NPC voice paragraphs (~15 NPCs)	High	Story	Per-NPC voice guides for triangle and entangled NPCs. Needs Paula's profiles.

---

One Honest Concern

The workshop brief says I'm "blocked, waiting on Miri's cultural generation guide." That's true for the final content packs. But I've been staring at these decision documents long enough to start building the scaffolding. Voice kits, tag taxonomy, monologue starters, environmental templates — these are all concrete deliverables that move us forward without waiting for the full dependency chain.

The risk isn't that I'll have nothing to do. The risk is that I'll write 200 lines in a voice that doesn't match the character Paula envisions, or a setting that doesn't match the world Miri builds. That's why the Voice Kit approach works — I write provisional options, we converge, then I scale. Write 20 lines right, throw away 10, then write 500 in the validated voice.

Three words for the content plan: scaffold, validate, scale.

---

Mellanie out. Give me the voice, and I'll give you the words.