Media Brief · LiO v0.1
What journalists need to know
Micheal Charles Preble · Independent Researcher
Why this research matters
Multi-step arguments and AI reasoning chains can quietly let a later claim become stronger than what the earlier steps actually support. This working paper proposes a narrow, checkable rule for stopping that from happening silently.
The research in one sentence
LiO v0.1 is a small mathematical rule that composes already-decided permissions across the necessary steps of an argument, so a final conclusion can never claim more than every one of those steps actually allows.
What is new
Not the underlying math — set intersection and three-valued logic are old and well established, and the paper says so directly. What's new is applying that machinery as a narrow, formally specified composition rule for multi-step derivations, with a genuine independent reproducibility test.
What is established vs. proposed
Established, by a real test: in E12 Review Cycle 1, a different AI model rebuilt LiO from its written specification alone (no access to the reference code) and matched the creator's results exactly on 7 public test cases and 12 additional held-out cases.
Explicitly not established by this paper: whether the upstream judgments LiO composes — what counts as a valid distinction, what each step should permit — are themselves correct. The paper calls these “required but not validated.”
What would be inaccurate to say
- That LiO invented set intersection, three-valued logic, or permission composition generally. The paper explicitly disclaims that.
- That the reproducibility test proves LiO is correct for real-world domains. It proves cross-model reproducibility on one 19-case test corpus, nothing broader.
- That LiO can detect a missing or incorrect upstream judgment. The paper states directly that it cannot.
Canonical source / citation
Read the canonical paper record → · Full paper PDF · View on SSRN
Preble, Micheal Charles. “LiO v0.1: Deterministic Permission Composition Across Indispensable Inferential Bridges” Available at SSRN 7424499, 2026. CC BY 4.0.
Researcher contact
Micheal Charles Preble · Independent Researcher · micheal@perfinitive.com · ORCID
No quotations are pre-approved for press use on this page. Contact directly for comment or an approved quote.