The NEMS Framework for Physicists: No External Model Selection as a Criterion of Foundational Completeness

This paper introduces the No External Model Selection (NEMS) framework to a physics audience. We explain, in physical rather than model-theoretic language, what NEMS requires, why it matters, and what it implies for programs that rely on external choosers (measures, vacuum selectors, observer maps) to extract determinate predictions. The central result is a trichotomy: any candidate fundamental theory is either record-level categorical, internally selecting, or not fundamental in the NEMS sense. We summarize the Externality Reduction bridge (every "outside dependency" is selection in disguise), the semantic visibility closure (the semantics cannot hide outside the records), and the results of three illustrative audits applied to multiverse cosmology, the string landscape, and computational-universe programs. The core theorem spine is machine-checked in Lean 4 (v2.0.0, 8051 jobs, zero sorry, zero custom axioms). The framework does not adjudicate which theory is correct; it provides a precise, auditable criterion for when a theory qualifies as a complete foundational explanation. Trust boundary. This is a physics-facing survey; precise axioms and the classification theorem appear in Papers 1–2, with pinned exports in nems-lean . See .