Project-M/Docs/Vault/01_Vision/Pillars.md

tags, status, updated, permalink

tags	status	updated	permalink
vision pillars	draft	2026-05-29	gamevault/01-vision/pillars

Design Pillars & Locked Decisions

The few non-negotiables every system must serve. Keep this short; promote only decisions that are truly locked.

[!info] Direction: LANTERN (locked 2026-07-13, DR-048_Lantern_Adoption_Full_Pivot). The mechanical pillars below still hold (Unity DOTS + Netcode, server-authoritative, 2–4 co-op, skill-expression combat, persistent base, automation-as-progression). What changed is the fiction + three subsystems: the fiction is now Identity_Lantern (not the Awakening Engine); combat sharpens to SoD-grade manual-aim skillshots (the soft auto-target era ends); the world becomes a seed-pinned pocket-graph where light is territory; the build layer becomes suit-frames + cultured Sparks + wild mutations. See the LANTERN locked decisions at the bottom.

Pillars

Combat is the primary verb; base + automation braid around it (locked 2026-06-08, DR-028_Combat_Primary_Verb_Depth_First). The fusion of all three is the identity — but as ONE braided loop with a single primary verb, never three co-equal, independently-deep modes. Roadmap: Path_to_Fun.

1. Action-ARPG combat — the primary verb — twin-stick, controller-first (Hades / Risk of Rain 2 / Deep Rock Galactic feel); skill expression over stat-checks (dodge, read-and-react, a real ability kit). The moment-to-moment the other pillars serve.
2. Co-op base power fantasy — braids in — 2–4 friends defend and grow a shared home base (V Rising / The Riftbreaker feel); the base is what you fight for and the siege's stakes.
3. Automation as progression — braids in — self-running production that makes what you fight with (charges / munitions / turrets / upgrades) and that sieges threaten; rewards setup, not grind (The Riftbreaker / Core Keeper).
4. Server-authoritative & deterministic — input-only clients, client prediction; the simulation is the source of truth. Co-op is non-negotiable, so every mechanic stays server-authoritative.

Locked decisions

• Tech stack: Unity DOTS (Entities) + Netcode for Entities — server-authoritative, input-only clients, client prediction. Established 2026-05-29_Project_Setup.
• Control (locked 2026-05-30): WASD movement, controller-first, directional / twin-stick aim with soft auto-target in-arc. 2026-05-30_M1_Player_Slice
• Multiplayer (locked): small co-op 2–4, client-hosted listen-server (BinaryWorlds + in-proc IPC; not the experimental SingleWorld host mode), PvE.
• World (locked): persistent buildable home base + instanced/procedural expeditions. Sharpened 2026-06-17 (DR-037_Procedural_Expedition_Spine_Two_Classes_Persistent_Meta): the procedural Expedition region is the required combat SPINE (where the bulk of the fight happens); the base is the persistent HUB that never resets ("Mission-as-Sortie": Deep Rock Galactic / Cult of the Lamb). A "run" is a re-seeded Expedition region in the SAME world, not a new world or a base reset.
• Automation (locked): progression accelerator — self-running production chains; data model designed to grow toward full logistics.
• Combat-primary + depth-before-breadth (locked 2026-06-08): combat is the primary braided verb; no new system until one braided loop is genuinely fun, and every milestone ends with a play/fun-gate (not just green tests). DR-028_Combat_Primary_Verb_Depth_First · roadmap Path_to_Fun.
• Classes + persistent meta (locked 2026-06-17, DR-037_Procedural_Expedition_Spine_Two_Classes_Persistent_Meta): two classes — Warrior (melee anchor) + Ranger (ranged, manually aimed — not auto-fire, which would break skill-expression). Class identity = the primary-attack archetype; run-acquired meta-buffs ride the existing StatModifier/TimedModifier pipeline. A persistent meta-progression layer (SaveData v6) funds permanent growth at the hub; run combat-buffs are run-scoped. Reverses DR-031_Base_Mining_Loop_Cohesion's expedition pause (a real playtest — Scratch Notes 6152026 — showed the single arena goes stale).

LANTERN direction — locked decisions (2026-07-13, DR-048_Lantern_Adoption_Full_Pivot)

These sit on top of the mechanical pillars above (which are unchanged). Full design: Identity_Lantern; roadmap: Roadmap_Lantern_Slice.

• The one rule: what the light holds, stays; what the dark holds, changes. Applied to the world = persistence; to the body = death; to the kit = builds; to the story = stakes.
• World = seed-pinned pocket-graph (light is territory). Permanence is spendable: lit pockets persist, dark pockets re-knot. The hardest netcode problem — Lantern_World_Model_Spike (★review-first).
• Combat feel = SoD-grade manual-aim skillshots (telegraphs, dashes, cooldown kits; soft auto-target cut). Suit-frames = chassis/class; Sparks (abilities-as-organs, cultured) graft into sockets; wild mutations are the augment layer (paced thresholds, evaporate at homecoming, Tallow fix-slots).
• Death = the split: flame returns cheap (kit works within minutes); cargo + fixed mutations + raw organs strand with the suit — recovery runs are optional value-runs, never mandatory kit-repair.
• Session-only burn (the Lamp burns only while a session runs — no offline punishment). Hosting = Wicks travel; worlds are saves (Valheim contract). Loss ceiling = pockets, never a shelf.
• The slice's falsifiable test: does pinning a pocket feel like winning, and does a stranded suit feel worth a second descent? — the only gate that matters.

Related

• Identity_Lantern — the operative fiction these pillars now skin (co-op action-RPG; light-as-territory).
• Identity — the prior fiction (The Awakening Engine, Aether-colony sci-fi) — superseded as operative by DR-048_Lantern_Adoption_Full_Pivot, kept as history/quarry. Skinned via DR-015_The_Awakening_Engine_Fiction_Adoption / DR-009_GameFeel_Identity_FirstBlood (amnesiac Sleeper / Echo / THEM · Aether cyan=ordered / orange=wild Blight).
• Decision records: 07_Sessions/_Decisions/