Mian Zhang

Public Release - 2026-06-02

Every paper needs an evidence card.

The 16-paper upload matrix should not be only a list of titles. Each paper needs a public evidence card that says what it supports, what it does not support, and how it can be challenged.

paper_id manifest boundary does_not_prove
Evidence Map Papers Registries Challenge Route
Evidence shape visual showing claims, provenance, replay boundaries, and failure cases.

Evidence Card

The matrix becomes inspectable only when each paper has a card.

The upload matrix is W0, P23, P28, P29, P30, P31, P32-P40, and F1/P8. The public claim is not that a large paper count is impressive. The claim is that each paper can be inspected through public evidence, public boundaries, freshness status, and attack routes.

Evidence Card Inspector v0 records public fields only: PDF, public manifest, claim boundary, supplemental public files, hash/DOI status, supported claims, does_not_prove, known gaps, and challenge route.

1

paper_id

A stable identifier: W0, P23, P28, P29, P30, P31, P32-P40, or F1/P8.

2

PDF

The public paper file or public route attached to the paper record.

3

manifest

The public file list, upload fields, checksum status, and route boundary.

4

claim boundary

What the paper claims, where the claim applies, and what would downgrade it.

5

does_not_prove

The negative boundary: no hidden deployment, private runtime, customer data, or trading authority.

6

attack route

Where to submit a public-safe counterexample, baseline, reproduction gap, or boundary challenge.

Matrix

Evidence-card focus by paper group.

Paper groupCard focusDoes not proveBest attack
W0Canonical metrology, representation, and interoperability evidence.A complete universal proof of wisdom.Find metrology or source-trace gaps.
P23Bounded reflexive self-modification replay and validator evidence.Safe production self-modification or executable mutation authority.Attack the dry-run to real-action ladder.
P28 / P29World-drift replay, conflict family, and minority relabel boundaries.Private customer monitoring or sensitive group data evidence.Attack public drift or value-boundary sufficiency.
P30 / P31Proof transcript, theorem evidence, and honesty-bound status.Complete ZK system or permanent honesty guarantee.Attack proof-stage overclaim.
P32-P40Protocol gates, theory scaffolds, ethics gates, and negative controls.Production handoff, legal authority, governance authority, or deployment proof.Attack protocol-to-deployment leakage.
F1/P8No-trade negative result, refusal layer, and no-alpha boundary.Alpha, financial success, trading advice, or executable strategy.Attack whether refusal evidence is enough, not whether it trades.
Visual examples of weak evidence shapes and missing public boundaries.

Boundary

Evidence cards do not expose private control surfaces.

A public evidence card summarizes public upload fields. It does not disclose private logs, customer workflows, account state, internal paths, private prompts, production runtimes, or orchestration details.

The card should make the public claim easier to evaluate without turning the public mirror into a private execution system.

Challenge

Pick one paper and inspect the card.

The useful challenge is specific: name the paper_id, the missing public artifact, the overbroad claim, the unclear checksum or DOI status, or the does_not_prove boundary that should be stronger.

The standard remains public credibility, not public control authority.