Narrative sim audit: world layer outpaces hero layer

When auditing an emergent-simulation game engine for playability gaps, separate the aggregate world layer (economy, politics, populations, ecology, factions) from the individual hero layer (a protagonist). The world layer is usually far more complete. The recurring gaps that block actual story playthroughs: individual-scale combat/encounter resolution (engine only resolves conflict at faction scale via power-margins + seize-fractions), autonomous LLM-driven NPC and antagonist minds (the GM/director may be LLM while NPCs stay deterministic + narrated), and discovery-driven progression vs a pre-authored rank ladder. Also: grep the source tree for stat/quest/character modules before trusting a roadmap doc — code often has more than the roadmap admits.