Escaping the Simulation: Systematic Concept Expansion via Semantic Operators

Abstract

Consider every corpus as a simulation—a compressed, selective model of a richer underlying reality. When analysts write strategic assessments or scientists publish papers, they’re constructing simplified models that capture certain features, compress others, and omit what seems irrelevant or simply wasn’t thought of. We present a framework for escaping the simulation: using patterns within a corpus to infer concepts that exist in the fuller reality but were compressed away or never articulated. The framework follows a principled pipeline: Corpus!Extract(type)!Augment(knowledge) ! Define(operators) ! Apply(ops; concepts) ! Filter(threshold). We demonstrate it through construction of a strategic lexicon from 821 national security documents, generating 16 novel doctrines that occupy genuine gaps in strategic vocabulary. The key insight: simulations leak information about what they’re simulating. The concepts a corpus does contain imply other concepts it doesn’t.

Keywords