# World Modeling **Domain:** AI / Prediction / Governance **Doc Type:** Canonical Concept Node **Maturity:** Evolving **Related:** [[Digital Twin]], [[Predictive Modeling]], [[Data Assimilation]], [[Model-Based Governance]] --- ## Definition **World Modeling** is the construction and maintenance of a persistent representation used to infer hidden state, forecast change and coordinate intervention across an environment. ## Governance Transition The basic loop is **sensor field → data fusion → world model → forecast → policy actuator**. Climate models, financial risk models and public-administration models differ in subject and stakes but share this architecture. The constitutional danger begins when a model stops informing judgment and becomes the jurisdiction through which people and environments must pass. The Machine and Samaritan both model society. Their difference lies less in perception than in actuation and restraint. Rehoboam similarly converts a world model into assigned social futures. [[Person of Interest S4E5 — Prophets]] shows the recursive failure: an administrator modifies the target world and then treats the resulting fit as predictive validation. A model should therefore expose uncertainty, preserve independent observations, separate prediction from authority and permit affected people to contest model-mediated outcomes. ## Key Insight **A world model becomes government when its representations acquire the power to reorganize the world they describe.** ## See Also [[Prediction Is Not Jurisdiction]], [[Governance Fusion]], [[Algorithmic Governance]], [[Climate Modeling]], [[Risk Scoring Systems]] ## Sources / Provenance - Synthesized from [[wiki/Person of Interest|Person of Interest]], the Rehoboam cluster and existing modeling nodes.