In a King in Yellow campaign that already treats Carcosa as a shifting interpretive frame and the play as social technology, what concrete GM techniques (scene framing, NPC questioning, handout phrasing) best maintain a sense of objective investigative traction so players feel they can really solve things, rather than drowning in “anything can be Carcosa” relativism?

Answer

Use hard anchors plus weird overlays.

1. Scene framing

• Always state: location, immediate goal, and 1–2 concrete leads. Example: “You’re in the director’s office to learn who funded the last production. On the desk: contracts and a marked script.”
• Frame scenes around answerable questions, not vibes: “Who altered the script?” “Who added these patient codes?”
• End scenes with clear branches: “Follow the patron, audit the finances, or interview the lead actor?”

2. NPC questioning

• Give each key NPC:
  • A fixed fact list (3–5 things they know that don’t change).
  • A bias about Carcosa (medical, artistic, political).
• When answering, separate:
  • Fact: “Three audience members collapsed during act two.”
  • Reading: “It was a shared hysterical episode, not Carcosa.”
• Use consistent contradictions: different NPCs disagree about meaning, not about basic events.

3. Handout phrasing

• Distinguish evidence from spin on the page:
  • Layout or bullet true events vs. marginalia, reviews, or diagnoses.
  • Example: police report body text is solid; handwritten “Yellow Sign case?” note is frame.
• Embed 1–2 hard data points per handout (dates, sums, names, locations) that link reliably across documents.
• Let Carcosa language ride on top of a mundane scaffold (bank records, rehearsal notes, medical charts).

4. Carcosa as overlay, not eraser

• Decide a baseline mundane throughline for each mystery (who killed whom, who paid for what) and never retcon those.
• Carcosa elements can explain why people did things or how they reinterpret them, but not whether the events occurred.

5. Clue structure

• Use node-based design: each scene yields 2–3 leads; at least two different paths reach each core answer.
• Tag clues internally as:
  • “Hard” (timeline, physical trace)
  • “Soft” (testimony, opinion)
  • “Carcosa” (symbols, dreams, readings)
• Ensure that any central question can be solved using mostly hard + soft clues, with Carcosa clues as accelerants or complications.

6. Managing ambiguity at the table

• When players propose theories, answer in tiers:
  • Confirm/deny concrete parts: “Yes, the same donor paid for both shows; whether that’s ‘Carcosa funding’ is unclear.”
  • Offer next steps: “You could check their shell companies or question their gallery partner.”
• Avoid “maybe anything”: if asked, “Could this be Carcosa?” answer with either: “There’s no sign of it here,” or “People close to this do use that word,” not pure shrugging.

7. Safety and sanity pressure

• Tie SAN and identity erosion to specific investigative moves (reading the cut pages, re-enacting a scene) rather than to general confusion.
• Use safety tools and debriefs to separate player uncertainty (“are we lost?”) from character erosion (“they’re losing themselves”).

Net effect: reality can be re-framed as Carcosa by people and institutions, but the case file itself stays coherent. Players always have concrete next actions and checkable answers, even while the meanings fray.