Orchestrated Reality: From Role-Play to Living, Playable Game Worlds -- LLM-Driven World Simulation as a Parameterized-Action POMDP

2026-06-14

AI Summary

Current LLM-driven game worlds struggle to maintain persistent state consistency, hindering the deep integration of autonomous narrative and simulation. This work proposes a playable virtual world framework governed by a single coordinating agent, modeling game state as a normalized JSON entity tree and formalizing it for the first time as a parameterized-action partially observable Markov decision process (POMDP). Structured action execution and state updates are achieved through a Plan-Diff-Validate-Apply pipeline, JSON Schema validation, and a narrative projection observation mechanism. Empirical evaluation across 15 real-world interaction scenarios demonstrates that the framework effectively supports unrestricted player agency while preserving global world consistency.

Abstract

Many games rely on storytelling combined with systems that track levelling, NPC behaviour, and consequence simulation; bridging tightly-authored narrative with deeply-simulated worlds -- most acute in sandbox and open-world settings -- has been prohibitively expensive. LLM-driven worlds open a new path: a single harness can coordinate numerical state, narrative voice, storytelling pacing, and rule logic together. Realising this requires the LLM system to sustain a persistent world (who is where, what has just happened, what is currently true), which today's deployed systems do not: the narrative voice asserts state in free prose without any validated representation, so a fully autonomous game engine remains infeasible. We treat this as an architectural choice, not a limitation of language models, and report work in progress on a framework -- orchestrated reality -- that makes the world a canonical object owned by a singleton orchestration agent analogous to the tabletop-RPG Game Master (GM). We formalise an LLM-driven game world for a human player as a Parameterized-Action POMDP: state is a tree of canonical JSON entities, actions decompose as $a=(k, x_k)$ (a discrete intent kind plus structured JSON parameters), the agent observes only a narrative projection $o=O(s)$ of state, and the transition kernel $F$ is an LLM-driven Plan-Diff-Validate-Apply (PDVA) pipeline that commits schema-validated, content-hashed JSON deltas. We give the formal model, a JSON-state example, a worked single-turn example, and a catalogue of 15 illustrative incidents drawn from a real deployment showing the framework in action. Empirical validation through a planned human player study -- together with multi-NPC concurrent agency and deployment as an RL environment -- is situated as future work.

Problem

• LLM-driven simulation
• persistent world state
• narrative-system integration
• Parameterized-Action POMDP
• autonomous game engine

Innovation

• Orchestrated Reality
• Parameterized-Action POMDP
• LLM-driven simulation
• canonical world state
• Plan-Diff-Validate-Apply