Darkness Falls — North Star, Player Journey, and Repair Path

Status: VISION LOCKED 2026-07-25 — owner-approved direction; implementation remains phased
Read this when: planning DF campaign/onboarding/endgame/season work, prioritizing repair, or deciding whether a new system belongs in the game.

Current mechanics: index.md
Competitive eligibility: seasons.md
Endgame ship state: endgame.md
Balance / telemetry surfaces: game-balance.md
Class distinctiveness work: class-identity.md
Detailed systems: progression.md, npcs-quests.md, world.md, monetization.md

---

The promise

Darkness Falls is the voice RPG where a player builds one character from a frightened survivor into a named defender of a wounded world. Exploration finds the danger. Combat tests the build. Loot and character choices change what the player can survive. The world remembers the result, and the endgame turns personal strength into a reason to return with other players.

The game MUST feel deep without making the player speak through a menu tree. Repetition belongs to automation. High-value choices belong to the player.

Product identity

Darkness Falls is:

• A long-form character RPG, not a short voice diversion.
• A dangerous world explored one spoken move at a time.
• A loot and build game in which equipment, class choices, the Star Map, trophies, and Mastery create different answers to the same threat.
• A personal campaign that opens into repeatable solo and asynchronous social endgame play.
• A fair seasonal competition in which money and account age do not improve ranked performance.

The design does not need more systems merely to look deep. It needs the existing systems to produce a journey the player can understand, care about, and finish.

Explicit non-goals (until the matching phase is complete)

These are intentional refusals, not oversights:

Non-goal	Why
A second parallel progression meta (new currencies, prestige tracks, account-wide power ladders)	Systems already outrun meaning; more ladders widen the “statistic to raise” problem.
New regions solely to fill level bands	Missing story connections are a campaign-map problem, not a map-size problem.
Live coordinated multiplayer combat as a launch promise	Party scaffolding exists; shared encounters/rewards are incomplete. Promise only what ships.
Pay-to-win or pay-for-pace in Current Season	Fairness is the seasonal product; convenience power stays Standard-labeled.
Soft-lock monetization after death or a missed drop	Loss-state pressure burns trust faster than it earns revenue.

Current diagnosis

The exploration, combat, loot, hideout, equipment, and advancement systems already form a strong progression machine. The player can keep finding another statistic to raise or another zone to enter.

The missing piece is shape. The world expands, but it does not yet build enough pressure toward a clear dramatic conclusion. Class identity arrives late (see class-identity.md). Many cleared locations have no reason to matter again. Levels 70–100 ask for sustained effort without enough authored change. The proposed endgame has the right three parts — Rifts, Sieges, and Mastery — but not one fully closed loop (see endgame.md for ship-vs-gap detail).

The repair therefore starts with meaning and completion, not more breadth.

The player journey

The exact quests, dialogue, and named antagonists remain content work. This five-act shape is locked. Existing zones and systems MUST be arranged around it before new regions are added.

Act	Levels	Player belief	Required movement
I — Survive	1–10	“I can stay alive here.”	Learn movement, combat, loot, the hideout, and one unmistakable class mechanic. End with a named threat and a reason to leave the starter region.
II — Belong	11–39	“This is my character, and these people know me.”	Connect major regions through recurring NPCs and causal quest chains. Boss victories change later dialogue, access, or danger instead of ending as isolated checks.
III — Choose	40–49	“I know what kind of fighter I am becoming.”	Make Promotion the answer to a campaign crisis. The class choice MUST change play, not merely rename the same statistics.
IV — Confront	50–99	“I stopped the first catastrophe, but the damage spread.”	Use the Dark Lord confrontation as the initial campaign climax (near the start of this act, not the end). Paragon identity, aftermath, and escalating fractures carry the player toward level 100.
V — Defend	100+	“My strength matters to the whole world.”	Abyssal Rifts provide personal escalation, World Sieges provide communal purpose, and Mastery supports lasting Standard progression.

The level-50 climax is a conclusion, not an ending screen. It closes the first threat and causes the fractured-world problem that justifies levels 51–100 and the endgame.

Act IV is not one bland hill

Act IV spans fifty levels. The experience law “every ten-level band must change the experience” still applies. Until the campaign map is written, treat these as required intermediate beats (names and zones TBD in open decisions):

Band	Must add or resolve (minimum)
50–59	Dark Lord confrontation + immediate aftermath the player can hear (NPC, access, or world state).
60–69	First major fracture: a new threat type or region consequence unlocked by the victory.
70–79	Paragon depth or Mastery-adjacent preparation that changes combat decisions, not only numbers.
80–89	A returnable region or faction with a persistent consequence from earlier acts.
90–99	Explicit endgame on-ramp: the player is pointed at Rifts (and later Sieges) with a clear reason.

If a band only raises HP/damage/enemy numbers, it fails the North Star even if content volume is high.

Experience laws

1. The player always knows why the next action matters

On return, the game MUST state where the character is, what immediate danger or objective remains, and one useful next action. A larger command list is not orientation.

Return briefing minimum (voice): one sentence location → one sentence objective or threat → one spoken next step. Example shape: “You are at the hideout. The Watch still needs you at the north gate. Say teleport to return, or star map to spend a skill point.”

2. Identity arrives before investment

Every base class MUST demonstrate a distinct mechanic by level 5 and a distinct play pattern by level 10. Promotion at 40 deepens that identity. It cannot be the first time the class feels different.

3. The world has memory

Major victories MUST change something the player can hear later: an NPC response, an available route, a settlement state, an encounter pool, a repeatable objective, or the next threat. More coordinates without consequences do not make the world feel larger.

4. Automation handles repetition, not judgment

Auto-attack, auto-loot, and other assistance may remove spoken busywork. They MUST leave the player responsible for a few consequential decisions: when to spend a limited resource, change position, commit to a class action, retreat, or prepare for a known threat.

5. Every progression band changes the experience

No ten-level band may exist only to increase health, damage, and enemy numbers. Each band MUST add or resolve at least one named objective, relationship, class capability, world consequence, or endgame preparation.

6. Earned competition stays earned

Current Season is a fresh race. Paid entitlements, purchased power, and power inherited from another season or character MUST NOT improve ranked performance. The exact eligibility contract and ship gate live in seasons.md.

7. Voice time is a design budget

Server latency can be healthy while the game still feels slow if responses are long, duplicated, or option-heavy. Prefer short spoken confirmations, non-duplicated options, and progressive detail on ask (“status”, “look”) rather than dumping the full state every turn. A “deep” system that forces a long menu tree fails this law even if the underlying mechanics are correct.

Social and endgame direction

Parties currently provide useful social scaffolding, but the game should not promise conventional shared combat until shared encounters, rewards, and party actions work end to end.

World Sieges are the best first social destination because asynchronous contribution fits voice play. A player can help the community without coordinating a live session. Rifts remain the first repair priority, however, because one complete solo endgame loop is the foundation on which Siege rewards and Mastery should rest.

The completion order is:

1. Rifts: enter, escalate, clear, fail, reset, reward, and return.
2. Mastery: earn, allocate, and consume every shipped perk in the relevant gameplay calculation.
3. Sieges: schedule, join, contribute, complete, pay rewards, and show the community result.

Do not invent a fourth endgame pillar until these three proofs pass. Current gap inventory: endgame.md § Pending / Not Wired.

North-star goals

These are product outcomes. Implementation plans may choose different code or content, but they MUST prove the same result.

Area	Goal	Proof
Opening	A new player understands movement, danger, the hideout, and the next objective in the first session.	First-session path telemetry plus replay review shows no unexplained dead end. Claims: see act-1-class-signatures.md § Capability Contracts (obj/return/preserve).
Class identity	Every class has a distinct mechanic by level 5 and a distinct play pattern by level 10.	Paired class scenarios produce different useful decisions from the same encounter state.
Campaign	Every ten-level band adds or resolves a named objective, relationship, capability, consequence, or endgame preparation.	The canonical campaign map has no progression-only band (including Act IV sub-beats).
World memory	Every major campaign region has at least one persistent consequence and one reason to return.	Region checklist plus repeat-session playtest.
Combat agency	Automation removes repetition while each serious fight presents at least one consequential player decision.	Boss and elite encounter review identifies the decision and its different outcomes.
Endgame	Rifts form one complete, repeatable loop before another endgame layer expands.	Two clean runs can enter, advance, fail or clear, receive rewards, reset correctly, and return.
Social	A player can make a useful asynchronous Siege contribution and hear the collective result.	Full create-to-reward Siege scenario.
Fair competition	Entitlements, prior trophies, Mastery, and purchases produce zero change in Current Season combat state, progression rate, loot rate, or ranked metrics.	Deterministic entitlement-on versus entitlement-off parity suite.
ISP trust	Every paid promise has a live consumer, a visible result, and a regression guard.	Purchase-to-effect test and refund-safe audit path for every product.
Measurement	Progression stalls, fight outcomes, deaths, purchases, and entitlement effects can be measured per player journey stage.	Valid fight-level outcomes and level-band cohort reports.

Measurement map (where proof should land)

Phase 0 closes instrumentation gaps. Prefer existing surfaces before inventing new stores.

Goal area	Primary data / test surface today	Gap if any
Opening / dead ends	game_log.v_real_error_log (NO_OPTIONS_RENDERED), call/session views, scenario suite	First-session funnel by act band not yet a single report
Class identity	Hermetic scenario_tests per class; combat choice traces	Paired same-encounter class scenarios largely unwritten
Campaign bands	Milestone log (LEVEL_UP, ZONE_FIRST_ENTER, quest events); content map in design	No canonical campaign map doc yet (open decision)
Combat outcomes	game_log.combat_metrics, v_combat_balance, v_death_hotspots	One-row-per-fight + victory/death repair is locally implemented; production trust begins only after deploy verification, and legacy NULL-victory rows remain excluded
Loot / economy	game_log.loot_drop_log, purchase + entitlement audits	Item vendor value repair is locally implemented; Current Season parity implementation is not shipped
Endgame Rifts/Sieges/Mastery	Handlers + services per endgame.md; scenario coverage	Clear/death/reward hooks incomplete
Fair competition	Planned parity suite; seasons contract in seasons.md	Suite + Current Season not live
ISP trust	v_purchase_audit, scenario @purchase paths, product tests	Product-by-product guard completeness varies

Repair path

Phases are sequential gates. Later phase work may be designed early, but MUST NOT ship ahead of the prior phase’s Done criteria without an explicit owner exception recorded here.

Phase 0 — Restore truth

Current status — 2026-07-25

Workstream	Status	Remaining proof
Fight/loot telemetry	Local implementation complete: Darkness Falls writes one terminal fight row; Darkness Falls and Lineage populate victory; loot logs vendor value; dashboard excludes legacy NULL outcomes. Unit suite and scenario guard 2935 are authored.	Deploy Java/dashboard, run the scenario gate, then verify one production win and death row plus one-row-per-encounter grain before declaring live truth.
Status hygiene	Core North Star, seasons, monetization, ISP review, and balance docs now separate current behavior from historical repair notes.	Continue the same single-status-block pattern when a later phase touches another subsystem doc; do not create parallel status sections.
Pricing/value	Live catalog and shared economy agree on $1.99/month. Current promise and a “hold price pending cohorts” decision are recorded in monetization.md.	Build exposure, renewal/retention, refund, and benefit-delivery cohorts before repricing.
Fairness proof	Competitive power contract and explicit paired entitlement parity plan are locked in seasons.md.	Implement suppression consumers and the planned JUnit/scenario/integration fixtures before Current Season launch.

Work

• Repair fight-level victory and death telemetry; round rows cannot stand in for completed fights.
• Define one current status block per subsystem and move historical repair notes below it.
• Reconcile subscription pricing and value assumptions across the shared economy and Darkness Falls docs.
• Add the seasonal entitlement parity test plan before any Current Season implementation.

Done when

• Live balance can be measured per completed fight (combat_metrics / balance views trustworthy).
• Every North Star metric in the Measurement map has a named data source or an explicit instrumentation gap listed in this doc.
• Current behavior and approved target behavior are visibly separated in the docs.

Phase 1 — Repair the first promise

Work

• Implement early class identity, beginning with one signature mechanic per class (class-identity.md).
• Tighten the first-session objective and return briefing (Experience law 1).
• Connect the starter region to one recurring NPC and one named threat.
• Keep free auto-loot and other accessibility basics intact.

Done when

• The Class identity proof in North-star goals passes for all eight base classes (paired scenarios).
• A clean new-player run always ends with a clear next objective (Opening proof).

Phase 2 — Build the campaign spine from existing content

Work

• Map existing zones, bosses, and NPCs onto the five acts (and Act IV sub-beats).
• Replace isolated bounty repetition with recurring quest chains where the world needs escalation.
• Make Promotion answer the Act III crisis.
• Make the Dark Lord victory cause the Act IV aftermath (band 50–59).
• Give major regions consequences and return reasons before adding more zones.

Done when

• The Campaign and World memory proofs pass (no progression-only ten-level gap; region checklist).
• A playtester can explain who or what they are pursuing and what changed after each act.

Phase 3 — Close the first endgame loop

Work

• Complete Rift tier advancement, clear detection, death reset, reward fulfillment, weekly state, boss variety, and the return path (gap list: endgame.md).
• Wire Mastery actions and effect consumers only after Rift rewards reliably feed them.

Done when

• The Rift proof in North-star goals passes twice from clean state.
• A Mastery choice changes the documented Standard calculation it promises to change.

Phase 4 — Add communal purpose

Work

• Finish Siege scheduling, contribution, completion, rewards, and result narration.
• Improve party discoverability, but describe shared play only at the level actually implemented.

Done when

• The Social proof passes: a solo voice player can contribute asynchronously and later hear what the community achieved.

Phase 5 — Launch fair seasons and tune ISP

Work

• Implement the locked competitive power contract before Current Season opens.
• Convert planned seasonal power rotations into cosmetics and collection rewards.
• Keep permanent mechanical trophy and Mastery value in Standard.
• Measure subscription and pack behavior after fairness gates exist; tune prices from cohorts, not a single historical sample.

Done when

• The Fair competition and ISP trust proofs pass (parity suite green; no ranked leaderboard consumes paid or inherited power; every Current Season offer states Standard-held vs available now).

Guardrails

Until the corresponding repair phase is complete:

• DO NOT add another broad progression system.
• DO NOT add new regions to solve a missing campaign connection.
• DO NOT launch a reward-bearing Current Season.
• DO NOT ship premium equipment, purchased trophy power, extra active trophy power, or paid seasonal combat modifiers into ranked play.
• DO NOT describe Standard record boards as fair competition if paid or inherited power affects them.
• DO NOT promise live multiplayer combat in player-facing copy beyond what is implemented end to end.
• DO NOT skip Phase 0 measurement repairs and “tune by feel” combat or economy for North Star work.

Decisions still open

This document locks the shape and fairness boundary, not every content choice. Resolve open items in priority order — each blocks a phase.

Priority	Open decision	Blocks
P0	Named campaign antagonist chain + recurring NPC cast (enough to wire Act I–II)	Phase 1–2
P0	Exact level bands and quest chains assigned to each existing region (canonical campaign map)	Phase 2
P0	Class signature mechanics and their level-5 introductions (per class)	Phase 1 — design pass ready: act-1-class-signatures.md; owner locks D1–D5 before code
P1	Act IV sub-beat zone/NPC assignment for bands 60–99	Phase 2
P1	Subscription and Blood Emerald pricing after current cohort measurement	Phase 5
P2	Whether Standard record boards remain showcases or gain normalized, reward-bearing categories	Phase 5

When an open decision is resolved, record it here (one line + date + pointer to the owning subsystem doc) and remove it from this table. Do not leave resolved choices only in chat history.

How agents should use this doc

1. Before adding DF systems or content, check Guardrails and Non-goals. If the work is breadth without advancing the current phase’s Done criteria, stop and re-scope.
2. When code and vision disagree, prefer filing repair work against this doc over rewriting the vision — unless the owner explicitly changes the vision.
3. For mechanics detail, open the subsystem doc; for seasonal fairness, open seasons.md; for endgame ship state, open endgame.md. This file is the journey contract and repair order only.
4. Success is the Proof column, not a checklist of tickets closed. Prefer hermetic scenarios and measurable cohort signals over prose claims.