szl-holdings/szl-papers

SZL Holdings academic corpus — preprints, thesis lineage, bounty problems, prior-art disclosures. Doctrine v11 LOCKED 749/14/163.

stars

112

commits

TeX

primary language

Sep 7, 2026

updated

a-11-oy.com

academic-corpus

agentic-ai

compliance

doctrine-v11

formal-verification

governance

governed-ai

lambda-conjecture-1

preprints

prior-art

research

szl-holdings

Browse cluster: Policy-as-Code and Kubernetes Governance →

README

SZL Papers

The published research behind SZL's governance math — preprints, thesis lineage, open bounty problems, and prior-art disclosures.

What this is

Every claim SZL makes about AI governance traces to a versioned, machine-checked source. This repository is that paper trail: the preprints, thesis versions, and open problems that underpin the proof backbone.

The repository preserves 26 thesis source versions (v1 through v26). The public Zenodo concept DOI 10.5281/zenodo.19944926 currently resolves to the published v21 record (10.5281/zenodo.20490218); v22–v26 are repository source versions and must not be described as separately DOI-published until matching Zenodo records exist.

Directory	What it is
`preprints/puriq/`	PURIQ preprint — formal specification of the Λ trust aggregator
`thesis/ouroboros/`	Ouroboros thesis — the full governance substrate, 26 versions
`bounty/`	Open mathematical challenges — characterize uniqueness under weaker conditions
`prior-art/`	Prior-art disclosures for IP protection

The thesis in brief

SZL's core thesis: AI governance should produce cryptographic receipts — not dashboards — and the trust math behind those receipts should be machine-checked in a proof kernel, not claimed in marketing copy.

Latest version: v26 — Governed Post-Determinism (GPD), Locked-Eight Edition

Key milestones:

8 formulas in the locked-proven registry at proof snapshot 5cfaf9a3b55fd75e8a2a1197502a93dd8ca68be3: receipt replay, DAG acyclicity, FIFO ordering, ledger conservation, bounded coupling, Reed–Solomon recovery, entropy budget, append-only monotonicity. The distinct frozen Doctrine-v11 count baseline is c7c0ba17 (749/14/163).
~185 theorems machine-checked across Waves 11–23 (CI-green, not in the locked 8).
Λ uniqueness: Conditional uniqueness is proven axiom-free (Theorem U). Conjecture 1 is machine-checked false and disproved as stated; its counterexample is published.
Open bounty: Characterize or prove uniqueness under weaker conditions. See bounty/.

Full proof library and commit history: szl-holdings/lutar-lean

Honest proof status

Tier	Count	What it means
Published locked registry	8 — F1, F4, F7, F11, F12, F18, F19, F22	Proof terms checked at snapshot `5cfaf9a3…`; registry count is exact. Release audit must capture per-declaration `#print axioms` output.
Machine-checked (experimental)	~185	CI-green across Waves 11–23. Real work, honest stage. Not in the locked count.
Open questions	2	Weaker-condition Λ uniqueness and Khipu BFT safety (Conjecture 2). Conjecture 1 itself is disproved as stated.

Cite this work

@misc{szl2026governed,
  title  = {Governed Post-Determinism: A Proof-Backed Substrate for Consequential AI},
  author = {Lutar, Stephen P.},
  year   = {2026},
  doi    = {10.5281/zenodo.19944926},
  url    = {https://doi.org/10.5281/zenodo.19944926}
}

SZL Holdings · Apache-2.0 code · CC BY 4.0 papers · Concept DOI 10.5281/zenodo.19944926

_{No production ATO claimed · repository-scoped SLSA L1 · Λ = Conjecture 1 (advisory, not a theorem) — the original unconditional uniqueness statement is disproved as stated (machine-checked counterexample published; conditional Theorem U holds) · Not affiliated with Defense Unicorns}