Most current designs assume investigative horror needs at least one relatively fixed anchor of ‘mundane reality’ for contrast; what if instead you generalize The King in Yellow campaign as a contest between multiple equally unstable but competing realities (institutional, artistic, Carcosan, player-authored), and how would that change scenario goals, victory conditions, and safety tools when there is no expectation of restoring a baseline—only choosing which reality gets to solidify around the investigators by the end?

Answer

Treat the campaign as a reality contest: the point is not to restore normality but to choose which unstable reality hardens around the group.

1. Reality tracks

• Define 3–4 “realities” as tracks:
  • Institutional (law, medicine, bureaucracy)
  • Artistic (bohemia, salons, the play)
  • Carcosan (signs, dreams, layered scenes)
  • Player-authored (PC reports, theories, flashbacks)
• Each has a simple 0–5 track. Scenes and choices advance one or more tracks.

2. Scenario goals

• Per scenario, set:
  • Primary contest: which 1–2 tracks this case pushes.
  • Local stakes: who/what flips if that reality wins here (a clinic, a troupe, a city block).
• Scene framing: every clue is tagged to a reality; following or believing it bumps that track.
• PCs can try to “cross-interpret” clues (read an institutional clue in artistic terms, etc.) to push a different track.

3. Campaign “victory conditions”

• At campaign end, the highest track “sets” the fiction:
  • Institutional win: the world stays legible but rewritten as policy, diagnoses, sealed archives.
  • Artistic win: decadence, masks, and performance norms define truth; life becomes theater.
  • Carcosan win: layered reality, unstable time, KiY textures everywhere.
  • Player-authored win: the PCs’ dossiers and theories are canon; the world conforms to their record.
• PCs don’t restore baseline; they steer which lens becomes default.

4. How play feels different

• Clues are about alignment as much as facts.
• “Solving the case” = choosing which reality’s explanation sticks.
• Sanity pressure comes from being pulled between tracks and watching earlier anchors convert (e.g., a neutral hospital becomes an open Carcosan front or an art-world myth factory).

5. Safety tools for no-baseline play

• Session 0: name the tracks, their tones, and non-goals (e.g., no real-world diagnostic labels).
• Use clear IC/OOC boundaries for “what’s real at the table” vs “what’s real in-fiction now.”
• Periodic reality check-ins: quick debriefs where players restate the current highest track and how that feels; adjust lines/veils if needed.
• Offer opt-out from specific track pressures (e.g., a player can say “less institutional-gaslighting horror” and you bias toward artistic/Carcosan in their scenes).

6. Simple procedure sketch

• Each scenario:
  • Pick 2–3 key NPCs and give each a favored reality.
  • Give each major clue a default reality tag and an alternate one.
  • When PCs believe, endorse, or act on a tagged interpretation, tick that track.
• Every 2–3 sessions, show an “alignment event”: something mundane re-skinned by the leading reality (courtroom as theater, gallery as triage ward, etc.).

7. Endgame choice

• In the last arc, present 2–3 concrete actions that would spike different tracks (publish a report, stage a performance, sign institutional orders, accept Carcosa entirely).
• The “ending” is which they back, not whether they avert the horror.

This keeps ambiguity but replaces the lost-baseline goal (“fix the world”) with a legible steering goal (“which reality are we crowning?”), and it demands stronger, ongoing table consent because the fiction never snaps back to normal.