Data Governance & Editorial Standards

§1 · The MasterBrain: Data Integrity Protocol

GreenCalculus runs a single canonical data layer — the MasterBrain — that supplies every calculator on the platform with emission factors, global warming potential values, and grid intensity figures. The protocol below describes how that data layer is updated, traced, version-stamped, archived, and disclosed.

The 30-Day Update Rule

GreenCalculus commits in writing to updating all affected emission factors and GWP values within 30 calendar days of a major publication event. The qualifying events are: a new IPCC Assessment Report cycle release (e.g., AR7 on publication), the annual DEFRA conversion factor update published each June, and the IEA Global Energy Review annual release.

The most recent test of this rule was the DEFRA 2026 release. DESNZ published the 2026 conversion factors in June 2026; the refresh reached the MasterBrain on 18 June 2026, moving 1,458 rows onto the new edition. The largest single movement was the UK grid electricity factor, which fell from 0.177 to 0.13096 kg CO₂e/kWh — a 26% decrease reflecting real decarbonisation of the 2024 generation mix, and a change large enough that any page still quoting the old figure is materially wrong rather than merely dated. Every affected calculator was version-bumped in the same pass.

The 30-day window is chosen deliberately. Shorter would not allow cell-by-cell verification against the source workbook before deployment. Longer would risk calculators serving stale data during an active reporting cycle, when practitioners are actively citing GreenCalculus values in their own GHG inventories. The window covers the time required to cross-check every changed value against the primary source, update the MasterBrain, propagate version stamps to affected pages, and publish the corresponding changelog entries.

Cell-by-Cell Provenance

Every entry in the MasterBrain carries a machine-readable source object. This is not documentation aspiration — it is the live data structure, embedded against every fuel, every global warming potential value, and every grid intensity figure on the platform. Coverage is total, not typical: all 16,686 canonical rows carry it.

The source object records three things against every value: the publication it was read from (source.id, which resolves in the source registry to publisher, edition, licence and a link to the document), the exact place it was read from — a cell, a table, an annex clause or a page (source.cell_ref), and the date it was retrieved (source.retrieved). Multi-gas rows carry the same fields again on each component, so a CO₂e figure can be decomposed to the individual CO₂, CH₄ and N₂O cells behind it — the AR6 or AR5 basis applied is recorded on the row itself.

The exhibit below is not a transcript. It is read from the live MasterBrain when this page is served, so it cannot fall out of step with the data it describes:

Read live from MasterBrain v2026.203. Verify against the publisher: UK Government GHG Conversion Factors 2026.

A practitioner who wishes to verify that figure can download the DESNZ workbook named above, open the tab and cell it names, and confirm the number directly against the source publication — including the individual gas components, each of which carries its own cell. The same rows are queryable programmatically: GET /wp-json/greencalculus/v1/factors returns the canonical corpus with its source attribution, and GET /wp-json/greencalculus/v1/gwp-values returns the GWP set, so any practitioner, auditor, or downstream developer can check the data layer without reading this page at all.

Version Stamping and Audit Trail

Every calculator, methodology page, and data page carries a version stamp at the top of the article body — for example, MasterBrain v2026.203 · DEFRA 2026 · IPCC AR6 GWP-100 — that links directly to the corresponding entry in the public changelog.

When the MasterBrain version bumps, every affected page receives a Major version bump simultaneously and its _gc_last_reviewed date is updated. This means a practitioner who used a GreenCalculus calculator in Q3 2025 to populate a corporate GHG inventory can return to the public changelog, identify the exact data version that underpinned the result, and reproduce the calculation if challenged during external assurance.

Archival and Reproducibility

Two things are preserved here, and they are not the same thing. The record is permanent: every value change is published in the changelog with the old and new figures stated, and no superseded value is deleted or quietly migrated out of it. The machine-resolvable archive is narrower and newer: it holds point-in-time snapshots that an API call can resolve directly, and it begins at MasterBrain v2026.111.

Every superseded value is retained with its original source reference and the version in which it was retired. The archive is served from the API edge rather than the open corpus endpoint: GET https://api.greencalculus.com/v1/factors/<key>?as_of=<version> resolves a single factor key as it stood at a past MasterBrain version, and requires an API key. The open /wp-json/greencalculus/v1/factors endpoint serves the current corpus only and rejects an as_of parameter outright rather than silently returning today’s value.

Current archive coverage. Every published MasterBrain version from v2026.111 to the live v2026.203 is resolvable — 70 versions, with no gaps. Five had been unreadable (2026.137, 2026.138, 2026.151, 2026.154 and 2026.177) and were restored on 31 August 2026 by rebuilding each corpus from the data layer’s own version history. The last of those records the Ember grid-lifecycle revision, the largest data event of the year.

Two numbers that look like gaps are not. 2026.141 and 2026.176 were never published: the release history runs 140 to 142 and 175 to 177, so those version numbers were allocated and skipped rather than lost. Nothing was archived for them because nothing ever existed at them. We state that rather than counting them as a shortfall we intend to make good — an archive that quietly counts phantom versions as gaps is no more honest than one that hides real ones.

This policy applies equally to corrections (per §4 below) and to scheduled source updates (per the 30-day rule above). Even when a value is corrected because it was wrong, the wrong value is preserved in the archive — because removing it would silently invalidate every report built on it.

Dual GWP Basis Disclosure

GreenCalculus operates two concurrent GWP bases transparently — a practice rare among public environmental calculators. Both are declared in every calculator’s audit trail and every data export. Neither is hidden, and neither is treated as a silent default.

Practitioners choose the basis appropriate to their reporting framework. Both bases are published in full at our IPCC AR6 GWP reference table and applied within our published DEFRA emission factors.

§2 · Editorial and Quality Control Standards

GreenCalculus content is produced and reviewed against a three-layer model. Each layer has a defined role, a defined output, and a defined point of accountability.

Policy on AI Assistance

GreenCalculus is explicit about where AI tools are used and where they are not. AI tools may assist with formatting HTML, scaffolding code structure, and identifying candidate internal links between pages. All methodological decisions and all numerical outputs are hard-coded by Jeremiah Say from primary source documents.

Review Depth by Content Type

Different content types are subject to different review depths. The bar chart below shows the relative depth applied to each surface, followed by the full review matrix.

§3 · Independence and Conflict of Interest

No organisation can pay to have specific emission factors, preferred methodologies, or favourable assumptions featured on GreenCalculus. The factors used on this platform are those published by the relevant Tier-1 regulatory body — full stop.

GreenCalculus editorial policy

Independence

GreenCalculus is an independent technical resource. It is not a target-setting body — it does not set emissions reduction targets and does not validate corporate pledges, and is therefore distinct from organisations such as the Science Based Targets initiative. It is not a verification or certification body — it does not issue certificates, does not audit third-party emissions claims, and does not assert assurance opinions. It is not a consultancy — it does not provide bespoke paid advice. Its sole function is the impartial, accurate implementation of existing global standards into accessible calculation tools.

No pay-to-rank

No organisation can pay to have specific emission factors, preferred methodologies, or favourable assumptions featured on GreenCalculus. The factors used on this platform are those published by the relevant Tier-1 regulatory body — DEFRA for the United Kingdom, US EPA for the United States, IEA for grid intensity, IPCC for global warming potentials. No commercial relationship influences the data layer or the methodology documentation. If GreenCalculus ever receives sponsorship or advertising revenue from a party with an interest in any specific emission factor, that relationship is disclosed separately in standard site disclosures and does not affect editorial or data decisions.

Revenue model transparency

GreenCalculus is supported by display advertising and is building toward premium API access for enterprise users who require programmatic access to the MasterBrain data layer. Neither revenue stream creates editorial conflicts. Advertisers do not influence content selection, content ordering, or the values published in the data layer. API consumers query the same data layer that all public users see — there is no premium-tier divergence in the underlying numbers.

§4 · Correction and Feedback Policy

The Errata Process

When an error is identified — whether through internal review, an external practitioner challenge, or a MasterBrain audit — the following six steps are taken without exception. No silent fixes; no exceptions.

The changelog entry is written before the corrected page is published. Any practitioner who subscribed to the changelog RSS feed at the time of the error will see the correction in their feed.

Worked example: the v2025.2 CH₄ correction

Every calculator that referenced the affected fields received a Major version bump, and the v2025.1 snapshot — including the incorrect 25.0 value — remains resolvable through the archive endpoint above for any practitioner reproducing a pre-correction figure. See the corresponding changelog entry for the full attribution.

Technical Feedback Channel

A useful submission includes the URL of the affected page, the specific value or methodology step in question, and the source document being cited as authority for the proposed correction. Challenges are reviewed by Jeremiah Say. For general site enquiries unrelated to data or methodology, see the standard contact page.

§5 · Compliance Framework Alignment

GreenCalculus’s internal governance is modelled on principles drawn from two internationally recognised verification frameworks: ISO 14064-3 (the standard for verification and validation of GHG statements) and the GHG Protocol Corporate Standard. The principles below are not aspirational. They describe how the data layer, the editorial process, and the correction policy already operate today.

ISO 14064-3 Principles

GHG Protocol Principles

§6 · Citing GreenCalculus

Every published calculator and methodology page exposes the live MasterBrain version stamp at the bottom of the page. Cite that stamp in any disclosure (CDP, SECR, CSRD/ESRS, internal audit) so the reader can reproduce the exact calculation.

The homepage provides three ready-to-copy formats:

• ISO 690 — EU compliance default, recommended for CSRD/ESRS reports.
• APA 7 — corporate sustainability academia.
• Vancouver — life-cycle assessment and science journals.

If your house style requires a different format, three principles let you build it:

1. Always include the MasterBrain version (e.g. v2025.55). The version stamp pins the exact data layer the calculation ran against, even after subsequent updates.
2. Always include the access date. Combined with the version stamp, this gives a reproducible audit trail for any output.
3. Always include the URL — root-relative or absolute. Each calculator and methodology page is a stable permalink.

Errata and corrections to past versions are documented in the public changelog and follow our Correction and Feedback Policy above — superseded factors remain queryable indefinitely so historical reports stay reproducible without needing to rerun any calculation.

§7 · Current Data Version Status

The table below shows the edition of each Tier-1 source currently loaded in the MasterBrain, and the date that edition was retrieved from its publisher. Retrieval dates are recorded in the data itself — each is the source.retrieved field on the rows drawn from that publication, not an editorial assertion about them. The review column states a cadence rather than a date, because a fixed date on this table becomes an accusation the moment it passes.

§8 · Frequently Asked Questions