Quell story progression: design

design · draft · Narrative and writing · edited 2026-11-06

Early sketch, started from a conversation; expect it to change. The belief model it builds on is NPC stances: data model.

The loop

Each run follows the story graph to the eye in the sky, which asks two questions: “Has she forgiven you?” and “Have you forgiven yourself?” Whatever the player answers, the eye says “Then you must try again,” and the player is sent back to the beginning with their beliefs and knowledge intact.

flowchart LR
  START["beginning"] --> GRAPH["story graph<br/>(gated by beliefs)"]
  GRAPH --> EYE{"the eye:<br/>has she forgiven you?<br/>have you forgiven yourself?"}
  EYE -->|"yes or no"| AGAIN["'then you must try again'"]
  AGAIN -->|"Stances and Inventory kept"| START
  EYE -.->|"the questions no longer hold"| OUT["the eye disappears:<br/>letting go"]

The goal is to get out of the guilt cycle: for the eye to disappear from sight. There is no perfect outcome, because none exists. The player can’t change what was done, heal the wounds, or fix the world. The way out is to let go, much as the true ending of Sovereign Tower is to destroy the tower and let the people decide their own fates, not to make war or peace.

Letting go is aporia

Answering yes or no is still holding a stance (Affirm or Deny), so either answer keeps the player in the loop. The way out is when the questions no longer have a grip: the stances they ask about are Suspended.

Suspension can’t be chosen. It comes only through argument, by elenchus or equipollence, once the player has committed to enough beliefs for them to collide. So:

• it comes late without being scripted as late;
• a player who never commits to anything stays in the loop, because a missing stance isn’t a suspended one. This is aporia, the release of dukkha, not passivity;
• arguing is required. Winning arguments isn’t the point, but there’s no way out without them.

The ladder

“Held, then lost” is Wittgenstein’s ladder: “My propositions serve as elucidations in the following way: anyone who understands me eventually recognizes them as nonsensical, when he has used them — as steps — to climb up beyond them. (He must, so to speak, throw away the ladder after he has climbed up it.)” (Tractatus 6.54). The player’s beliefs are the rungs. They have to be climbed: held, argued for, set against each other. Only then can they be thrown away. A player who never climbs has nothing to throw away.

The Buddha’s raft makes the same point: the teaching is a raft for crossing the river, not something to carry on your head once you’ve crossed (MN 22, Alagaddūpama Sutta).

Wrong questions

“Has she forgiven you?” and “Have you forgiven yourself?” are wrong questions. They aren’t hard to answer; they’re the wrong thing to be asking. Each is an instance of clinging (upādāna): to her verdict, to a self that judges and is judged, to guilt as a debt that could be settled. Clinging is what produces dukkha, so the questions don’t describe the suffering; they are it. That’s why every answer leads back to the start: answering accepts the question.

The Buddha set some questions aside in the same way. In the parable of the poisoned arrow (MN 63, Cūḷamālukya Sutta), a wounded man who won’t let the arrow be removed until he knows who shot it dies with his questions answered by no one. MN 2 (Sabbāsava Sutta) lists questions about the self (“Did I exist in the past? What am I?”) as unwise attention that ends in a thicket of views.

In the belief model, a wrong question rests on statements it takes for granted (its presuppositions). Answering it affirms them; the way out is suspending them. The eye’s questions are only the surface. What has to loosen is what they presuppose.

The story graph

Within a run, the story is a directed graph read left to right. Branches converge again, so it’s a DAG, not a tree. Nodes are events and story beats; edges are gated by conditions on beliefs:

• mostly the player’s Stances, which are the record of their change of mind and carry over between runs;
• NPC stances only locally: what an NPC does, says or is willing to talk about. They may not need to touch the story graph at all.

The graph resets each run; the beliefs don’t. A later run can open paths an earlier one couldn’t, because the player now holds (or no longer holds) something.

Decisions are offered, not argued

Debates settle what is the case and how good or bad it is (F and V statements); they never settle what to do. Decisions are separate: when someone asks the player what to do, the player picks from a few authored choices built from the conversation so far, as in Sovereign Tower. The player can’t make one up. Otherwise the game would need an open-ended space of behaviour on top of the space of beliefs.

NPCs mostly decide for themselves based on their own beliefs. Asking the player directly should be rare, reserved for NPCs who trust the player’s judgement.

See also

• NPC stances: data model — stances, Suspended, and elenchus and equipollence as the routes to aporia.
• Debate — the debate mechanic: reading, arguing, carrying over between runs.
• Quell integration — the player’s loop in Quell.
• Value-based argumentation — why an NPC who ranks values differently may never be moved, and whether losing that way is satisfying.

What links here: NPC stances: data model