Mathine — Evidence In, Trust Out

Layered Field Collapse: Executable Collapse Signatures from Benchmarks to Metaoverfields

A complex mechanical core with glowing lights shattering into pieces amidst smoke and debris.

Layered Field Collapse: Executable Collapse Signatures from Benchmarks to Metaoverfields

Mathine: Layered Collapse Signature Machine
Link: https://doi.org/10.5281/zenodo.18675054 [1]

Operational failures in tool-using agent systems increasingly show up as false closure: outcomes declared “done” under insufficient or mismatched admissible evidence. This paper proposes Layered Field Collapse as a collapse-native framework that makes these failures computable, comparable, and auditable across authority boundaries. [1]

The first contribution is a two-axis stack. On one axis: runtime authority layers (Field → Overfield → Metaoverfield). On the orthogonal axis: a design-time Metafield that governs semantic stability under versioning, so comparability survives contract migration rather than silently drifting. [1]

The second contribution is the Layered Collapse Signature (LCS): a unified diagnostic object computed from receipt-space that integrates (i) worst-subfield gates, (ii) closure integrity under explicit budgets, (iii) cascade dynamics in multi-turn regimes, and (iv) governance lifecycle metrics such as contestation, revocation, and half-life. Collapse becomes something that can be emitted—not merely narrated—by both benchmarks-as-contracts and operational systems. [1]–[3]

Finally, layered collapse is expressed as layer-indexed predicates over LCS, enabling consistent detection of local collapse (field-level), enterprise collapse (overfield-level), and cross-organization reconciliation collapse (metaoverfield-level). The practical shift is direct: from post-hoc incident stories to executable collapse signatures that travel with the governed boundary and remain checkable under declared regimes. [1]

References
[1] R. Figurelli, “Layered Field Collapse: Executable Collapse Signatures from Benchmarks to Metaoverfields”. Zenodo, Feb. 17, 2026. https://doi.org/10.5281/zenodo.18675054
[2] R. Figurelli, “Benchmarks-as-Contracts: A ReceiptBench Spec Template for Regimes and Closure”. Zenodo, Feb. 17, 2026. https://doi.org/10.5281/zenodo.18675035
[3] R. Figurelli, “ReceiptBench for Field Networks: Recursively Composable Governance via Typed Receipts”. Zenodo, Feb. 17, 2026. https://doi.org/10.5281/zenodo.18665376
[4] R. Figurelli, “From Scores to Receipts: Introducing ReceiptBench, a Typed-Receipt Protocol for Governance-Ready Evaluation”. Zenodo, Feb. 16, 2026. https://doi.org/10.5281/zenodo.18661829
[5] R. Figurelli, “Benchmark Convergence As Operational Confirmation Of Large Language Fields (LLFs)”. Zenodo, Feb. 15, 2026. https://doi.org/10.5281/zenodo.18653012

— © 2026 Rogério Figurelli. This article is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0). You are free to share and adapt this material for any purpose, even commercially, provided that appropriate credit is given to the author and the source. To explore more on this and other related topics and books, visit the author’s page (Amazon).