CJV-MTH-2.0 · working specification

A claim checked against a version, not a website.

The method is public so that reviewers can test it, challenge it and reproduce its outputs. It defines records and evidence grades. It is not an accredited standard and confers no certification.

Twelve structural rules

The method in one page

Rule	Requirement
01 · Boundary first	Boundary, functional unit and batch definition are fixed before evidence collection.
02 · Primary data	Metred share is calculated from declared primary-data rules, not estimated to reach a grade.
03 · Attribution	One owner, location and batch identifier unless a published allocation rule applies.
04 · Factor provenance	Every emission factor carries source, geography, year and applicability.
05 · Grade arithmetic	A ≥70%, B ≥45%, C below 45%; U is self-declared and not verified.
06 · Baseline co-signing	A counterfactual baseline is fixed and co-signed before the improvement it measures.
07 · Hard gates	Missing calibration, unco-signed baseline or insufficient metred share stops issuance.
08 · Abstention	Insufficient evidence produces no unit and a written reason — never a softer conclusion.
09 · Canonical commitment	Records use a published canonical serialisation and SHA-256 commitment chain.
10 · Public time anchor	Daily Merkle roots receive a portable public timestamp proof.
11 · Visible revocation	Revoked records remain visible with date and reason; history is never rewritten.
12 · Claim partition	Evidence statements, assurance reports and certification conclusions stay separate.

VERSION POLICY

Old labels keep their method

A rule change ships a new method version. Existing records continue to reference the version under which they were produced.

Current methodCJV-MTH-2.0

Challenge routeCounter-example

Silent editsNot permitted

Submit a technical challenge →

Conformance

Test the structure, not our adjectives

A conforming implementation reproduces arithmetic, grades, commitments and refusal behaviour.

canonical = batch | total_kgCO2e | per_unit | grade | method
record    = SHA-256(prev_hash || SHA-256(canonical))
root      = MerkleRoot(records_for_UTC_day)
proof     = record + path + public_anchor(root)

if metred_share >= 0.70: grade = A
if metred_share >= 0.45: grade = B
if metred_share <  0.45: grade = C
if hard_gate_failed:     output = ABSTAIN

What conformance does not mean

It is not accreditation or recognition by HKAS, CNAS, IAF or APAC.
It is not evidence that a reduction claim is effective.
It is not a guarantee that the measured party told the truth.
It is not a certification conclusion.

Known limitations

Published with the method

Method choice affects comparability. Public anchors prove time and integrity, not truth. Remote evidence has different coverage from site evidence. Related-party records cannot establish independence. Regulatory acceptance remains with the relevant authority.

Inspect registry records →Read governance

Version history

No silent method upgrades

CJV-MTH-1.0 and its 20-record FNV-1a snapshot remain available as a fixed archive. CJV-MTH-2.0 is the current working specification with SHA-256 commitments; it does not rewrite or reclassify v1 records.

Open the v1 evidence archive →

Research protocol

A falsifiable agenda, not validation

Recomputable Carbon Evidence: A Protocol for Grading, Abstention, and Red Teaming. Erik Wong, 2026. Zenodo preprint; not peer reviewed; no empirical dataset generated or analysed.

DOI 10.5281/zenodo.24115645 →