The confidentiality gap in ledger-based carbon accounting

A thesis page, not a source summary. The claim is mine, made at Martin Pompéry’s instruction (2026-08-17) to give it a page of its own rather than a paragraph on the ledger page. The quotations are verbatim from The Carbon Accounting Landscape Review (ICC, 2026); everything that follows from them is marked ⟨Inference⟩.

The thesis in one line

⟨Inference: the ICC Review states confidentiality as a design requirement of Criterion B, provides no mechanism for it, and then omits it from every subsequent instrument in the document — while the architecture it advocates is the one that makes the problem harder.⟩

What the Review says

Criterion B, Verifiability (§4.2, p. 21) — the full statement:

“Confidentiality and access control are part of verifiability’s design. Because product-level embedded emissions can reveal energy intensity, process efficiency, and cost structure, the system must allow authorised parties to verify the number while limiting unnecessary disclosure of underlying operational data — through permissioned access, aggregation, and data minimisation.”

And, in the same criterion:

“Methods and assumptions are disclosed to relevant parties (e.g., auditors, regulators, those with court orders), while preserving underlying data that is privately owned and competitively sensitive.”

That is the entirety of the treatment. The diagnosis is exact: embedded emissions are a side channel onto energy intensity, process efficiency and cost structure.

Where it then does not appear

⟨Inference: this is an argument from absence, verified by reading each instrument in the document.⟩

Instrument	Confidentiality present?
Table 3 — the eight criteria mapped to regulatory demand (§4.3)	No. Criterion B’s row cites only accredited third-party verification.
§5.2 — contributions and open items, criterion by criterion	No approach is credited or debited on it.
Table 4 — what each approach contributes and what remains open (p. 32)	Not among the open items for any of the six approaches.
§6.2 — missing elements in quantification	Absent.
§6.3 — missing elements in accounting architecture	Absent. Chain of custody, legal transfer, assurance level and transaction-level operability are named; disclosure exposure is not.
§7 — conclusion	Absent.

The requirement is posed once and never treated as unmet. “Permissioned access, aggregation, and data minimisation” is offered in the register of a solution while being a list of desiderata.

Why the ledger sharpens the problem it inherits

⟨Inference throughout this section.⟩

The Review’s own comparison table (Table 1b) is the argument:

• Counting/reporting discloses a period-average, study-scoped figure. Averaging over a period and a product family is itself a disclosure control — it blurs the per-batch variation from which a competitor could reconstruct throughput, yield, downtime, or feedstock switching.
• Accounting records per transaction, aligned to financial periods, “tending towards near-real-time where feasible” (Criterion H), transferred with legal title to the counterparty, and reconciled to entity totals under an audit trail an auditor can follow “end to end” (Criterion B).

So the architecture that best satisfies Conservation (F), Fungibility (E) and Operability (H) is the architecture that maximally exposes what Criterion B says must be protected. Every property that makes the ledger auditable — single recording at source, transfer with title, end-to-end lineage, reconciliation without gaps — is a property that makes the underlying operational data more legible to someone downstream. The Review does not observe this trade-off anywhere.

Two further sharpenings:

• The buyer is not a neutral party. In an exchange model the recipient needs a number. In a ledger model the recipient holds an accountable quantity that must reconcile against its own accounts, which is a stronger claim on the seller’s records. Sellers competing for the same buyer are handing a shared counterparty comparable per-transaction data.
• Primary data makes it worse, and primary data is the goal. The whole incentive direction of the Review — PDS, tightening defaults, “verified primary data as the default” — moves reporting away from industry averages, which are confidential by construction, towards site-specific measured data, which is not.

Why this is not a gotcha

⟨Inference.⟩ The Review is a landscape and gap analysis, not a design document, and it says repeatedly that the framework-development process will resolve the open methodological questions. It is entirely possible the Panel regards confidentiality as an implementation matter for later. The claim here is narrower and, I think, defensible: of the eight criteria, B is the only one whose satisfaction requires a cryptographic or protocol mechanism rather than a rulemaking one, and it is the only one for which the document proposes no mechanism, records no contribution from any of the six approaches, and files no open item. That asymmetry is the finding.

Why it matters for SINE

⟨Inference; the underlying facts are on the linked pages.⟩

By the bargaining diagnostic: agreeing the ledger rulebook is a coordination problem — aligned interests, a missing focal point, eight overlapping principle lists and no common architecture. Deciding how much a firm exposes about its own production once the ledger demands transaction-level records is a bargaining problem — opposed interests, private information, and no reason to reveal voluntarily.

The Review is written entirely in the coordination register. The one sentence in it that belongs to the bargaining register is the Criterion B passage, and it is the one that goes nowhere. That is the shape of gap SINE’s stated core description addresses — “makes claims believable to others without those claims having to be disclosed”.

Caveat against over-reading. The wiki’s own record warns that PACT succeeded as coordination work in which “the cryptography was not load-bearing”. A stated requirement is not the same as demand. Whether ledger participants would actually pay for confidentiality-preserving verification, or accept the disclosure and price it into contracts as they do today, is untested and is a question for the SINE response, not an assumption of it. There is also a live objection on record that confidentiality “löst das Eintrittsproblem und beschädigt den Verbreitungsmechanismus” — monitoring that shows nothing may satisfy Ostrom’s fourth design principle technically while destroying its social function (see Ostrom’s commons design principles).

Status ⟨revised 2026-08-17, after two ingests⟩

Checked against SINE’s own research, and it holds — but the gap is narrower and more interesting than first stated. Two papers now in sources/ supply exactly the mechanism the Review does not: Heiss et al. (2023), by a researcher who is part of SINE inside the CCTI programme, and Man & Kleppmann (2026), by CCTI’s Cambridge Co-PI. The joint reading is on Verifiable Carbon Accounting.

The finding of this page is unchanged and is strengthened: the Review poses confidentiality once and treats it as unmet by nobody, while a published, implemented and cost-measured answer had existed for three years. What changes is the shape of the claim SINE should make. Against the eight criteria:

• B (Verifiability) is answered. Proofs to each counterparty that the reported figure is correct, without disclosing the data behind it, at $0.01 per report and sub-second verification.
• C (Comparability) comes with it, unexpectedly — a proof binds a number to a specific audited method, which is a stronger guarantee than a disclosed methodology annex.
• F (Conservation) holds inside a closing constraint — an independently attested aggregate that the reporter does not control. Kleppmann has one, and it is a physical object: an electricity meter.
• G (Completeness) is not answered but reduced ⟨sharpened later the same day, after Martin Pompéry pressed on it⟩. A proof is a statement about a computation; “nothing was omitted” is a statement about the world, unreachable from inside the circuit. It becomes provable only against an aggregate that is independently attested, outside the reporter’s control, and demonstrably covering the boundary. Kleppmann has the first two and concedes he lacks the third — his own proposed repair is to cross-check the meter against the electricity supplier’s bills, i.e. to import a second independent attestation of the same total. Heiss has none and never claims completeness. So G becomes: does such an aggregate exist, and does the report reconcile to it?
• E (Fungibility) is untouched. Nothing here creates a transferable accountable quantity with legal title.

⟨Inference: so the honest position is neither “solved” nor “open”. Supplying a closing constraint across entity boundaries, where no meter exists, is precisely what ICC’s ledger is for — and proving a report against one without opening the books is precisely what the cryptography is for. The two are complements, and neither party has said so. That is the sentence SINE’s response should be built around; see open tasks, item 1.⟩

Addendum ⟨2026-10-01, from the iLEAP specification⟩

A working standard now shows what the gap costs when nobody closes it. The iLEAP Technical Specifications v1.1.0-draft.2 introduce a whole data transaction whose stated purpose is confidentiality — parties “decide to share an aggregated report instead of TCE-level data, due to privacy or competition concerns” (§3.4). This is the first source in the wiki where the confidentiality constraint is not asserted by me from absence, but written into a shipping standard as the reason a feature exists.

It matters for this page in three ways.

1. The diagnosis is confirmed from the field. The Review states that product-level emissions reveal energy intensity, process efficiency and cost structure. iLEAP’s carriers evidently agree, to the point of refusing leg-level disclosure. The side channel is real and priced.
2. The mechanism chosen is the lossiest of the three ICC names. Of “permissioned access, aggregation, and data minimisation”, iLEAP ships aggregation — and documents the cost in its own bold type: once the aggregate is exchanged “it is no longer possible to derive more granular (i.e., TCE-level) data from it, thereby affecting the all data exchanges downstream” (§4.4). One party’s confidentiality is paid for with the granularity of every party below it in the chain. Criterion G (Completeness) and the primary-data push both degrade silently.
3. The cryptographic answer is not reached for, in a document that had every reason to. No selective disclosure, no proofs, no secure computation anywhere in the specification. ⟨Inference: this is the strongest available evidence against the comfortable reading of this page. It is not that standard-setters have failed to notice the problem — iLEAP noticed it and chose a lossy trade anyway. Either the market accepts lossy answers because the cryptographic ones are not legible to it, or the cryptographic ones are not yet cheap or integrable enough to be the default. Both are claims about distribution, not about mechanism — and both are more actionable for SINE than “nobody has solved it”. Note that SINE co-authored this standard, which makes the absence harder to explain as oversight and worth asking Tell Münzing about directly.⟩

Addendum ⟨2026-10-16, Hitachi as a worked instance⟩

Hitachi’s Sustainability Report 2026 is the gap stated by the party suffering from it, in its own words. Scope 3 is 209.18 Mt-CO₂e against 0.46 Mt Scope 1+2 — 99.8 % of the footprint sits outside the company boundary — yet Category 1 is computed “using a calculation method that covers the majority of procurement spending and partially incorporates primary data from procurement partners” (p. 56), i.e. mostly spend-based. The reason Hitachi gives is capability, not unwillingness: SME suppliers “are uncertain about what steps to take, how to schedule these steps, how to apply specialized calculation methods and how to set appropriate targets” (p. 56).

See Hitachi, Ltd. and the Sustainability Report 2026.

And the gap stated outright, by standard-makers, later the same day. The CE Association white paper (March 2026) is the implementation document behind Hitachi’s CVP. Applying CVP across a value network requires each participant’s sales, intermediate-goods costs, COGS and SG&A. Its own study group records the blocker and the remedies:

“The disclosure of numerator and denominator data within VNs is difficult.” → “Data management by financial and other institutions as third parties is also possible.” → “Disclosing numerator and denominator values within a VN can be expected to have an effect in maintaining equity among the stakeholders that comprise the VN.”

⟨Inference: this is the strongest instance the wiki holds, for three reasons. The requirement is stated, not inferred by us. The parties want the disclosure, for fairness, and still cannot make it — so the confidentiality demand cannot be read as mere obstruction. And all four remedies on the table (internal-only use, certification of the final value, year-on-year coarsening, a financial-institution data trustee) are trust arrangements; none makes a figure checkable without showing it. The gap is not an omission in this document. It is a known blocker with no instrument against it, in a standard heading for an ISO 59011 annex.⟩

The primary source names the remedy as a role. The H-AIST research report, §3.3: computing a network-level CVP “requires determining value added and costs for each participating company, but it is not necessarily required to disclose these figures to all participants. A method in which a trusted, impartial third-party organization collects data from individual companies and calculates CVP can enable evaluation without public disclosure to the entire VN. … A third-party body can exist outside the VN, or it can exist within and become part of the VN’s formation.”

⟨Inference: this is the confidentiality gap met with the trust-maximal answer — a collector who sees everything so that no participant has to see anything. It is the same shape as a carbon-data intermediary, and it has the same single point of failure. What makes this instance more tractable than the carbon case is the arithmetic: the network figure differs from the naive aggregate (1.27 vs 0.49 in the authors’ own example) only through the elimination of intra-network purchase costs, and each such cost is a sale on the other side of the same transaction. The parties do not need to reveal their cost bases to each other; they need to agree on figures they both already hold. That is the narrowest version of this gap the wiki has recorded.⟩

⟨Inference, and a boundary on the claim: the research report’s §4.1.1 separately concedes that the CE segmentation “can easily become arbitrary”. That is a different failure — definitional discretion, not confidentiality — and nothing on this page addresses it. Verifiability makes a reported figure faithful to the books; it does not make the classification behind it honest. The two must not be conflated in SINE’s argument.⟩

⟨Correction, 2026-10-16: an earlier version of this addendum said CVP’s numerator is “a fact about objects held by counterparties”. It is not — the inputs are financial-statement items. The gap is real but it is a cost-base-disclosure gap, not a material-data gap.⟩

Addendum ⟨2026-10-16, second entry: the gap stated by a ministry and by a sales division⟩

The Third Open Forum report (H-AIST CE Lab., July 2026) adds two voices that are not standard-writers, which matters because the standard-writers’ version could be read as a drafting problem.

METI, naming the cause and prescribing no mechanism. Junichiro Mimaki, Director of the Resource Efficiency and Circular Economy Division:

“Data is a strategic asset for companies. However, advancing the CE requires a shift in mindset by recognizing the value of collecting and sharing information that is useful to other stages of the CE loop. Achieving this will require deeper trust and stronger collaborative relationships among companies.”

⟨Inference: the first sentence is the correct diagnosis — the data is withheld because withholding it is rational — and the second asks the parties to stop acting on that rationality. This is the clearest instance in the wiki of the pattern the page describes: the obstacle identified precisely, the remedy left as an appeal to disposition. A construction that does not require the mindset to shift is the thing nobody in the room proposes.⟩

Hitachi’s commercial side, stating the commercial consequence. Hiroko Shinkai, Green Digital Transformation Division:

“Unless information on the quality, provenance, quantity, pricing, logistics, etc. of recycled materials can be effectively shared and integrated, CE businesses will not be commercially viable.”

“Even when companies wish to use recycled materials, they often face challenges related to inconsistent quality, supply volume and pricing, while data exchange across companies remains underdeveloped.”

and the sufficiency criterion:

“What matters is not the sheer number of data fields, but whether the information required by the next player in the value chain can be transmitted seamlessly and without interruption.”

⟨Inference: that last line is the most useful sentence to borrow. It frames the problem as what the next party needs, arriving intact rather than as disclosure — which is precisely the frame under which selective disclosure and verifiable computation are improvements rather than compromises. It is also the only place in the whole Hitachi corpus where the requirement is stated by someone with revenue responsibility.⟩

Where the gap is cheaper to close than on the carbon side. The second of H-AIST’s two proposal halves, residual-value grading, names the economic quantity directly: substantiating quality and reliability information “reduces uncertainty-related costs” and expands demand for higher-grade recycled material. ⟨Inference: the beneficiary of the verification is the party who performs it — a sorter who can substantiate grade sells higher. Compare the carbon case, where a verified figure’s benefit usually lands on a customer or a regulator and the cost lands upstream. Self-financing verification is the rarer and easier sale.⟩

Addendum ⟨2026-10-26, the same gap outside carbon accounting entirely⟩

Li et al., Agentic Communication Networks (2026) is a networking white paper with no connection to emissions data, and it states this page’s problem as its central research question. §7.2, set as a display line in the original:

“What is the minimum network-relevant information required to support an agentic task?”

With the framing: “An agent may possess extensive private context, reasoning history, memory, goals, and application state. Most of this information should never be required by the network. ACN therefore needs to identify a narrow and stable boundary between application semantics and network-relevant service semantics.” And in §1: “a rigorous agentic network architecture must preserve the separation between private content and network-visible service expressions.”

A coordinating layer that must see enough to act and not enough to expose — the same shape as Criterion B. ⟨Inference: three differences are worth holding.

First, it is better stated than in the ICC Review. The Review names confidentiality once and then omits it from every subsequent instrument; this paper keeps returning to it and makes it the headline question of a section. It also states the right design rule — privacy “must therefore be considered when defining every ACN information element, not added as a separate protection mechanism after the architecture is established” (§7.4).

Second, the answer is the same non-answer. The mechanisms offered are authorisation, policy enforcement and audit logging (§5.5): expose less, to fewer parties, under rules. No cryptography. That is the ICC Review’s “permissioned access, aggregation, and data minimisation” in different vocabulary.

Third, this makes three. The Review states the requirement and supplies no mechanism; iLEAP answers it by aggregation; ACN answers it by authorisation and audit. Three communities, three domains, no contact with one another, the same gap — and in none of the three is the cryptographic instrument rejected on the merits, because in none of the three is it considered. ⟨That is an argument about demand formation rather than about technology, and it is now the sharpest open question this page carries. See open tasks.⟩⟩

Addendum ⟨2026-10-27, the counterexample — a body that does reach for cryptography⟩

⟨From OECD, Data Spaces between Co-ordination Challenges, Financial Viability and Public Policy (2026).⟩

Every entry above shares a shape: a community states the confidentiality requirement, and then does not consider the cryptographic instruments that answer it. Open task 19 asked for the counterexample — “a standards or networking document that does consider and reject the instrument” — as the thing that would discriminate between the three candidate readings (not known / known but not deployable / known and not wanted). The OECD report is it, and the answer it gives is none of the three cleanly.

It knows the instruments. Box 3.2 names four categories of privacy-enhancing technology and what each is for: federated learning (models travel to the data), secure multi-party computation (joint computation without revealing inputs), trusted execution environments including confidential computing and “Data Control Rooms” (enclaves plus policy enforcement and audit), and synthetic data. This is the same inventory as the CCTI stack, minus ZK and zkVMs.

It does not reject them. Support for PETs is one of the report’s eight policy considerations (§5.5), alongside usage control, with the formulation “Neither alone is sufficient”. It records the G7 Industry, Digital and Technology Ministerial Declaration of December 2025 committing to “advanc[ing] DFFT through the promotion of privacy-enhancing technologies (PETs)” and naming the OECD Global Repository on PETs.

What it does instead is bound the claim. Three statements, and they should be read together:

1. In §3.2, PETs appear as an alternative mechanism that falls short: they “address the confidentiality dimension” but “do not address the institutional co-ordination problem of who governs, who pays and how collective value is allocated”.
2. In Box 3.2, PETs appear as a complement that makes the contractual mechanism enforceable rather than merely promised — the connector negotiates not only who may access data and for what purpose but “through which PET the data must be processed”, so that “the provider no longer needs to trust the recipient’s compliance because the technology structurally limits what can be done with the data”.
3. In the same box, the barrier to adoption is named and it is not ignorance: “additional computational costs, specialised expertise requirements and the need for co-ordination among participants on which technologies to deploy” (Auñón et al., 2024) — explicitly “the same collective-action challenges identified throughout this section”.

What this does to the three candidate readings

Reading	Verdict after the OECD
Not known — a reach deficit	Falsified at the policy level. A committee-declassified OECD report and a G7 ministerial declaration both name PETs explicitly. The instruments are in the vocabulary of the institutions that set the agenda.
Known but not deployable — engineering maturity	Partially supported, but the obstacles named are cost, expertise and agreement on which technology, not capability. The third is a governance obstacle wearing engineering clothes.
Known and not wanted — incentive deficit	Not supported in this form. The OECD wants them. What it does not do is treat them as the answer.

⟨Inference, and this is the substantive update: the reason the carbon and networking communities in the entries above do not reach for cryptography may be neither ignorance nor refusal but sequencing. The OECD’s position is that confidentiality is the second problem. The first is who governs, who pays, and how collective value is allocated — and a community that has not solved that has nothing for a PET to be deployed into. On that reading, the four communities recorded on this page stating the confidentiality requirement and stopping are not failing to see the instrument; they are stuck one layer below it. Which is uncomfortable for a cryptography-first pitch and comfortable for SINE’s actual position, since Data Commons is a governance-first argument with cryptography as its enforcement layer, not the other way round.

The corollary is a sharpening of what we should be selling: not “here is a confidentiality instrument you have overlooked” but “here is the governance arrangement, and it is enforceable rather than promised, because the contract specifies the PET”. That is the OECD’s own construction, and it is closer to our stack than anything previously recorded on this page.⟩

One limit the report states that we have not

“Once data leave the provider’s infrastructure, enforcement depends on the data user’s compliance” — said of the ODRL/DSP/connector mechanism, and the reason provenance and audit logs rise from 25 % to 48 % of initiatives between all stages and operational stage while usage-policy blocks fall from 35 % to 28 %. The field is visibly migrating from prevention to “deterrence through accountability”. ⟨Inference: that migration is the market gap, stated in adoption data rather than in argument. The instruments on the CCTI stack are the only ones that do not require the migration.⟩

Addendum ⟨2026-11-06, the gap in plastics — and the first named mechanism in production⟩

Source: POLYPROBLEM 06 — Der Circularity Code (Röchling Stiftung / Wider Sense, December 2023).

Everything above was built from carbon-accounting instruments. This report shows the same gap blocking a different data layer — plastics production data — two years earlier, and names it as the first-order reason firms withhold. Heino Claussen-Markefka (ProData GmbH / R-Cycle), p. 7:

Unternehmen zögern damit, ihre Produktionsdaten zu übermitteln – schon allein, weil damit die Frage verbunden ist, wer diese Daten speichern und auswerten darf und wie dabei Betriebsgeheimnisse gewahrt werden können.

Note the construction: three questions in one sentence — custody, permitted use, and trade-secret protection — and “schon allein” places them ahead of cost, effort and interoperability, which the same report treats at length elsewhere. This is the pattern the ICC’s Criterion B states and then drops.

What is new: a mechanism, named, with a supplier and running projects

Christoph Kugler (SKZ), pp. 10–11, describes cross-company machine learning on production data, needed because firms will not experiment on live lines:

Damit Produzent A keinen Einblick in die Daten des Produzenten B erhält, werden dazu spezielle Verfahren wie das föderierte Lernen eingesetzt.

He names the supplier — the Hamburg firm Katulu — and reports model projects already running, then draws the architectural conclusion: “Eine übergeordnete, also unternehmensübergreifende, digitale Infrastruktur ist etwas, das wir immer wieder propagieren.” Asked whether the remaining obstacle is cultural rather than technological, he answers “Absolut.”

⟨Inference: this is the first source in the wiki where a practitioner names a deployed confidentiality-preserving mechanism in a non-carbon domain. It should not be filed as an answer to the gap above, because it answers a different question. Federated learning lets parties train a shared model without pooling inputs. The carbon instruments need something else: a way to transmit a claim — a PCF, a recyclate share — that a counterparty can verify without seeing the inputs that produced it. ZK proofs and MPC address that; federated learning does not. The right reading is that the field has solved the easier of the two problems and is using that solution as evidence the harder one is tractable.⟩

A second observation worth keeping: the attested cases of confidentiality blocking data flow now span carbon accounting (ICC, iLEAP, Hitachi), circular-economy indicator work (CE Association / CVP), plastics production data (this source) — and are absent exactly where the reporting entities are states with nothing to hide (GEO BON / Kim et al.). That distribution is itself evidence that the constraint is competitive rather than technical.

Addendum ⟨2026-11-06, the gap read at the level of the wire format⟩

From the Technical Specifications for PCF Data Exchange v2.2.0 (WBCSD/PACT, April 2024) — the protocol itself, now in the wiki. Every earlier instance of this gap was read off a strategy paper, a methodology or a report. This one is read off the bytes.

The complete security inventory of the protocol: HTTPS mandatory, OAuth2 Client Credentials, Bearer tokens with expiry, optional OpenID Connect discovery, and validation of the source field on an inbound event. That is all of it. A ProductFootprint travels as plain JSON. companyName is a free-text string; companyIds are URNs the sender asserts about itself; the assurance block is a set of claims about an audit with nothing attached a recipient can check. The words cryptography, encryption, signature, proof, zero-knowledge and multi-party computation do not occur in 2,421 lines.

Confidentiality is access control, and the access control is advisory. Hosts “SHOULD implement an access management system and only return the product footprints for which the data owner granted access”. §7.2.1 then places authentication and access management, credentials management and logging entirely out of scope — they belong to “solutions” built on top. Who may see what, on what evidence, with what audit trail, is therefore not an interoperable property of the network.

Non-disclosure is a specified outcome. §4.1.2: when a recipient requests a PCF it may not see, “the data owner decides on making the PCF available or not” and the host answers AccessDenied; when the PCF does not exist, “the decision making and decision making protocol for this case is up to the discretion of each data owner”. The protocol handles the confidentiality problem by delegating it, per party, per request, with no aggregate record of how often data was withheld.

⟨Inference, and the reason this instance is the strongest one on the page: the Tech Specs are an interoperability profile ⟨Per Martin Pompéry, 2026-11-06⟩ — their entire method is selecting other people’s standards and constraining them. OData, CloudEvents, OAuth2, OIDC, RFC 8288, URN, UN CPC, ISO 3166 were all picked up and profiled. JSON Web Signature was available on identical terms and was not. So the gap here cannot be explained by unfamiliarity with the instruments, by cost, or by the difficulty of inventing something: the document shows, on every other page, exactly how cheap adoption would have been. It is a selection decision, dated April 2024, by a body with SINE as its named technology partner and pathfinder.sine.dev as the host in its normative examples.⟩

⟨A second observation for the SINE argument: §6’s lifecycle model is a full provenance graph — immutable id, version, precedingPfIds, Deprecated status, validity windows — made entirely of self-assertions. The slots a proof would occupy already exist and already carry claims. Adding verifiability to PACT is not a redesign of the data model; it is changing what the existing fields are worth. See verifiable carbon accounting.⟩

Cross-references

• The Carbon Accounting Landscape Review (ICC, 2026) — the source
• iLEAP · iLEAP Technical Specifications (2026) — aggregation chosen over cryptography, in a SINE co-authored standard
• Ledger-based carbon accounting — the architecture
• Decision-grade carbon-data criteria — Criterion B in full
• Coordination failure vs. bargaining failure — the diagnostic used
• Li et al., Agentic Communication Networks (2026) — the same gap reached from networking ⟨2026-10-26⟩
• Verifiable Carbon Accounting — the mechanism, jointly from both papers
• Heiss et al. (2023) · Man & Kleppmann (2026)
• Cryptographic trust infrastructure (CCTI stack) — SINE’s instrument inventory
• Commercial clean-room vendors — who else sells this
• Data Commons — Security as one of the three values
• Tech Specs for PCF Data Exchange v2.2 (2024) · Pathfinder Network — the gap at the level of the wire format ⟨2026-11-06⟩
• Circular Value-added Productivity (CVP) · Value networks and the ISO 59000 series · Hitachi, Ltd. — the gap as a live instance, stated by the standard’s own authors
• Residual-value grading — the same gap on goods rather than accounts, with the willingness-to-pay on the right side
• Third Open Forum report (July 2026) — METI and Hitachi’s Green DX division stating it
• OECD, Data Spaces (2026) — the counterexample: a body that names PETs and bounds them ⟨2026-10-27⟩ · Data space boundary conditions

Sources

• sources/whitepapers/2026-08-17-icc-carbon-accounting-landscape-review/2026-08-17-icc-carbon-accounting-landscape-review.md (§4.2 criterion B; §4.3; §5.2; §6.2; §6.3; §7 — the last four cited for what they do not contain)
• sources/whitepapers/2026-08-17-icc-carbon-accounting-landscape-review/tables-recovered-from-pdf.md (Tables 1b, 3, 4)
• sources/standards/2026-10-01-ileap-technical-specifications-v1.1.0-draft.2.md (§3.4, §4.4 — the first addendum)
• sources/hitachi/2026-10-15-hitachi-sustainability-report-2026.pdf, pp. 36, 56 (the second addendum)
• sources/standards/2026-10-16-ce-association-social-implementation-cvp/tables-recovered-from-pdf.md (Table 1 — the second addendum)
• sources/hitachi/2026-04-13-h-aist-cvp-research-report.pdf, §§3.2–3.3, 4.1.1
• sources/hitachi/2026-07-30-h-aist-ce-lab-third-open-forum.md (Mimaki and Shinkai, panel topics 1–2 — the third addendum)
• sources/research/2026-10-27-oecd-data-spaces-co-ordination-viability-public-policy/2026-10-27-oecd-data-spaces-co-ordination-viability-public-policy.md (Box 3.2, §3.2, §3.3.3, §5.5 — the fourth addendum)
• sources/standards/2026-10-29-pact-data-exchange-protocol-v2.2.md (§4.1.2, §5.2, §5.4, §6, §7.2.1, §7.3, §7.6, §7.7, §7.8.4 — the 2026-11-06 addendum)