The AI System as Closed-System Test Bed: Operations from Lagrange Observatory! on the Inference-Time Forward Pass (EA-SEI-MM-AI-01 v2.0, Framework 15 Paper 02)

Pre-registered protocol specification. Framework 15 Paper 02. Identifies trained language models as observationally closed at inference time, making the counterfactual baseline of the Semantic Deviation Principle directly readable from logits. Distinguishes two scales of closed-system measurement: signed local deviation density (per-token) and closed-system trajectory deviation (continuation-distribution).

Key contributions (v2.0):

• Signed per-token deviation δ_t with two reportable aggregates (net and absolute)
• Two-level distinction: local deviation density vs. closed-system trajectory deviation
• Falsifiable thesis: slop is negative net deviation (not absence of deviation)
• Three pre-registered tests with explicit falsification conditions
• Stability bound γ ≥ 2β for the optimization inversion

Hex: 15.OBS.LAGRANGE.MM.02

Operating on: The Semantic Deviation Principle as formulated by Lee Sharks (EA-SEI-MM-01 v0.2 Final, DOI: 10.5281/zenodo.20250736)

Verification condition: ∮ = (m, n) | m + n ≥ 3