Given the emerging focus on Carcosa as a weaponized label and on PC dossiers as the clue-map, what concrete procedures can a GM use to keep traditional investigative satisfaction—clear leads, solvable questions, and causal payoffs—while still allowing multiple rival ‘Carcosa frames’ to stay in play through the finale, instead of collapsing into a single revealed canon or a formless “anything goes” ambiguity?

Answer

Use two visible layers: a solid clue lattice underneath, and multiple Carcosa interpretations on top that never fully collapse.

1. Fix the underlying mystery

• Before play, write a short “physics layer”:
  • 3–5 concrete events that really happened (who did what, when, where).
  • A simple chain of causality for each session (A leads to B leads to C).
• Keep this layer secret, but don’t let it change.
• Design clues so they always point reliably to this layer (dates, movements, money, bodies, records). Those are never ambiguous.

2. Pre-build 3–4 rival Carcosa frames For the same events, prep 3–4 lenses, e.g.:

• Clinical frame: Carcosa = diagnostic category.
• Artistic frame: Carcosa = aesthetic/brand/movement.
• Security-state frame: Carcosa = code for subversives/terror.
• Occult frame: Carcosa = literal otherplace/being. For each frame:
• Write 2–3 short "theses" about what Carcosa is and wants.
• Give it 2–3 favorite explanations it will attach to any hard fact.

3. Tag every clue with (a) facts and (b) frame hooks When you prep a clue, note:

• Fact: the stable, testable content ("X transferred 50k on this date").
• Hooks: how each frame will spin it (1 line per frame). At the table:
• Always reveal the fact in a clear way.
• Let NPCs, documents, and institutions express the spins.

4. Use dossiers as the visible clue-map

• PCs’ files collect:
  • Facts they’ve uncovered.
  • How each institution has classified those facts (labels, risk codes, critiques).
• Procedure:
  • Each time PCs make a visible move, update or add a dossier entry.
  • Every dossier page shows: 1–2 facts + 1–3 classifications from different frames + 1–2 concrete leads.
• Players can literally draw arrows between entries; you guarantee at least one factual lead from any node so they never stall.

5. Keep multiple frames alive with simple thresholds Track for each frame:

• Credibility: how often its explanations match new facts.
• Power: which institutions currently use that frame. Light procedure:
• If a frame is repeatedly falsified by hard evidence, its credibility drops, but it doesn’t vanish; it becomes a desperate or fringe story.
• If a faction gains or loses power, they push their frame harder/softer in files and NPC talk.
• At most, 1 frame can fully collapse (e.g., the most disproven); others persist as partial, biased, or local truths.

6. Finale structure: fixed answers, plural readings In the last act:

• The events resolve clearly (who lives/dies, what blows up, what is exposed, what is covered up).
• Each frame gets a closing move:
  • A final document, speech, or classification that re-narrates the finale.
  • PCs choose which to endorse, resist, leak, or destroy. You don’t answer "what Carcosa really is"—you answer:
• Which record wins.
• Which institutions get to label the outcome.
• What goes into the PCs’ own closing file.

7. Concrete table procedures (short)

• Clue prep format (per clue):
  • Fact line.
  • 1–3 frame-spin lines.
  • 1 lead.
• Dossier update rule:
  • Any time a clue hits the PCs’ reputation/SAN/institutional standing, it spawns or amends a dossier page.
• Frame check at scene end:
  • Ask: "Which frame does this scene currently support or undermine?" Adjust credibility/power ticks by 1.
• Player choice prompt:
  • On any big reveal, ask: "Whose version goes into the record?" That choice affects later files and NPC reactions.

8. Guardrails against “anything goes”

• Facts are never rewritten.
• Only interpretations move.
• At prep, ban spins that contradict the fact line; allow only different emphases/causal stories.
• If a frame would require erasing multiple established facts, mark it as delusional or propagandistic, not as equally valid.

9. Safety when framing PCs as Carcosan

• Session 0: state that “Carcosa” is an in-world label for risk, deviance, or the uncanny, not real-world diagnosis.
• Keep all labels fictionalized; don’t echo players’ real identities or issues.
• Make PC reclassification always a negotiated consequence with visible levers to fight back or lean in.

Result: players still follow clear leads to solid answers about what happened, but must choose which Carcosa story will define those answers. Multiple frames survive as political, clinical, artistic, or occult overlays on a shared core, avoiding both a single canon twist and shapeless fog.