AI systems2026Measurement framework

Order Effects and Non-Classical Contextuality in Language Models

A measurement framework treating premise-order sensitivity in language-model outputs as a non-classical, post-selection-free signal, triangulated across three independent tests. It reports a structural property of response statistics, not a claim about physical substrate. Prepared for open publication to establish priority.

PROGRAMThe Tell
CURRENT POSTUREOpen publication candidate
DISCLOSUREPublic briefing
01 / PUBLIC ABSTRACT

ORIENTATION, NOT OVERCLAIM.

A measurement framework treating premise-order sensitivity in language-model outputs as a non-classical, post-selection-free signal, triangulated across three independent tests. It reports a structural property of response statistics, not a claim about physical substrate. Prepared for open publication to establish priority.

02 / CURRENT POSTURE

The public surface contains an orientation briefing. The full manuscript and supporting package remain subject to revision and release control.

03 / EVIDENCE BOUNDARY

This page does not imply peer review, acceptance, experimental replication, or public availability of every supporting asset. The canonical register distinguishes those states explicitly.

04 / RELEASE GATE

Authorship, source verification, claims-to-evidence mapping, reproducibility, and patent-disclosure review must be cleared before any final external release.

PUBLIC APERTURE

PUBLIC BRIEFING

The public surface contains an orientation briefing. The full manuscript and supporting package remain subject to revision and release control.

NEXT RESEARCH FILEGY-R-026Training Collapse as Parametric Resonance: A Floquet–Mathieu Reframe