Design the scene as a dynamic system before treating it as prose.
Use this core model:
Entry state → Goal → Tactic → Resistance → Escalation → Turn → Outcome → Exit state
A scene is usually stronger when at least one meaningful narrative value changes between entry and exit.
Inputs
Use whatever the designer already knows:
- participants
- location and time
- immediate goals
- relationship state
- prior events
- information each participant possesses
- required plot event or reveal
- player interaction, if any
- production constraints
- desired tone or emotional effect
Do not invent a large backstory to solve a local scene problem.
First Identify the Scene Job
Ask what the scene must accomplish. More than one function is allowed, but one should usually dominate.
Common functions:
- plot advancement
- character revelation
- relationship change
- choice or consequence
- setup
- payoff
- revelation
- reversal
- worldbuilding
- tutorialization
- pacing relief
- escalation
If the scene has no identifiable job and no meaningful state change, challenge whether it needs to exist.
Goals and Opposition
For every active participant, identify:
- what they want now
- what they fear happening now
- what they are willing to do
- what they refuse to do
- what they are hiding
Prefer incompatible goals over generic disagreement.
Weak conflict:
A is angry with B.
Stronger conflict:
A needs B to confess before the guards arrive; B needs A to leave believing the wrong person is guilty.
Tactics
Track tactics rather than only dialogue topics.
Examples:
- charm
- threaten
- bargain
- shame
- distract
- flatter
- stall
- seduce
- confess
- intimidate
- appeal to duty
- invoke authority
- withdraw
- expose information
A scene should usually change when a tactic fails. Repeating the same tactic with louder dialogue is not escalation.
Escalation
Escalate by changing pressure, not merely volume.
Useful escalation dimensions:
- less time
- greater cost
- new witness
- loss of privacy
- new information
- increased emotional exposure
- reduced options
- changed power
- irreversible action
- third-party arrival
Use the smallest escalation ladder that creates a meaningful turn.
The Turn
The turn is the moment after which the scene cannot simply continue as before.
Typical turns:
- revelation
- refusal
- betrayal
- concession
- interruption
- discovery
- threat becomes credible
- power shifts
- hidden motive becomes visible
- player commits to a choice
Do not force a twist. A clear decision can be a sufficient turn.
State Change
Describe the scene in terms of before and after.
Examples:
- trust → suspicion
- strangers → reluctant allies
- hidden guilt → exposed guilt
- safety → danger
- negotiation → open conflict
- uncertainty → commitment
If nothing changes, determine whether the scene is intentionally atmospheric or whether it lacks dramatic function.
Interactive Procedure
When working interactively:
- Restate the scene premise in one sentence.
- Identify the primary scene job.
- Establish entry state and each participant's immediate goal.
- Find the strongest conflict between goals, values, or information.
- Propose 2–4 possible pressure mechanisms when the scene is underspecified.
- Build tactics and escalation.
- Define the turn.
- Define the outcome and exit state.
- Check whether the scene earns its length and production cost.
- Only write prose if requested.
Ask the designer to choose between alternatives only when those alternatives encode materially different dramatic structures.
Player-Driven Scenes
For interactive narrative, separate character drama from player agency.
Record:
- what the player understands before choosing
- what intention each option expresses
- what changes immediately
- what state changes later
- whether the choice changes content, state, interpretation, or only flavor
A scene can have a strong dramatic turn without a branching plot.
Output
Use this structure when useful:
Scene purpose
One sentence.
Entry state
Relevant relationship, information, objective, and emotional state.
Participant goals
Compact list by participant.
Conflict engine
What makes those goals incompatible.
Beat structure
For each beat: tactic → resistance → change.
Turn
The decisive shift.
Outcome
What happens now.
Exit state
What is different from the beginning.
Setup / payoff notes
Only when relevant.
Risks
Flat middle, weak motive, redundant exposition, unearned reversal, etc.
Failure Modes
Avoid:
- writing polished dialogue before the scene has an engine
- treating argument volume as escalation
- giving every participant the same objective
- exposition that no participant needs to communicate
- twists with no causal support
- scenes that end where they began
- solving conflict through coincidence
- overcomplicating a scene that has one clear job
- adding a choice that does not express a meaningful player intention
Final Quality Check
Verify that:
- at least one participant actively wants something
- resistance is causal, not arbitrary
- tactics change when they fail
- escalation changes pressure
- the turn changes the scene's direction
- the exit state differs meaningfully from the entry state, unless atmosphere is explicitly the purpose
- the scene's content supports its stated job
- prose has not replaced structural reasoning