Information Selection for Intervention Before Irreversible Deadlines: Exact finite frontiers for fixed and adaptive observation policies
Year: 2026
Type: preprint
Abstract: Theoretical preprint, version C2; NOT PEER REVIEWED. A finite worst-case analysis of selecting fixed or adaptive information channels before irreversible action deadlines. Includes exact conditional feasibility proofs, bounded conflicting-state witnesses, a latest-start adaptive recursion and reproducible deterministic computations. The information-delay trade-off, diagnosis, shielding, covering and dynamic programming have established foundations; no priority claim over those principles is made. All computation data are SYNTHETIC_CONSTRUCTED: human participants 0; field records 0. Results include 4,394 base-grid configurations, 35,152 subset evaluations, 2,744 direct-policy checks, 7,168 heterogeneous edge cases, 65,536 adaptive Boolean crosschecks and a 1,372-configuration adaptive construction. No empirical human or deployment safety result is claimed. Substantive OpenAI ChatGPT assistance in formulation, proofs, source discovery, code, figures and English drafting is disclosed; Michael chu has reviewed and approved C2 and accepts responsibility. No funding or competing interests. Article PDF, documentation and constructed data: CC BY 4.0; source code reproduce.py: MIT, as scoped in the companion archive. No unrelated manuscript under review is included.
Author: chu, Michael
Language: English
Topic: Game Theory and Applications
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7236627392
