The Constitutional Document Nobody Has Written

— The Absence

There is a document that governs how electricity flows across the United States. It is called the Federal Power Act, and it is not interesting to read. It establishes the Federal Energy Regulatory Commission, defines its authority, constrains what it can do, and specifies who it answers to. It is bureaucratic, precise, and largely invisible to anyone who is not a utility lawyer. It is also the reason the lights work.

There is a document that governs how the internet is coordinated. Several, in fact — the IETF's RFC process, the ICANN bylaws, the root zone management agreement. They specify who can add a top-level domain, how protocol standards are set, and what no single government or corporation is permitted to do with the naming system. They are the reason the internet did not become a series of incompatible national intranets controlled by telecoms.

There is a document that governs how European capital markets are structured. It is called the Markets in Financial Instruments Directive, and with its accompanying regulation it specifies who may operate a trading venue, on what terms, with what transparency obligations, answerable to which authority. It is dense, contested, and imperfect. It is also the reason a share can be bought in Milan and sold in Amsterdam under one set of rules.

Now ask: where is the equivalent document for the infrastructure layer of financial market settlement? Not the rules for what can be traded. Not the disclosure requirements for issuers. The constitutional document for the infrastructure itself — the layer that determines who controls finality, on what terms, accountable to whom. It does not exist. And that absence, it turns out, is the most important fact about the current system.

I — Who Benefits From the Absence

The standard explanation for why the constitutional document has not been written is complexity. Settlement infrastructure is technically intricate, legally multijurisdictional, and operationally interdependent. This explanation is not wrong. It is incomplete. Complexity explains why the document is difficult to write. It does not explain why nobody is seriously trying to write it. For that, you need to ask a different question: who benefits from the current governance vacuum?

Four beneficiaries of the vacuum

• Incumbent CSDs. Euroclear, Clearstream, and DTCC operate as regulated monopolies. Their fee structures, float income, and data revenues are products of market power that a constitutional framework with cost-recovery constraints and open access criteria would directly threaten. A document that specified access criteria constitutionally would end the commercial negotiation through which incumbents extract rents from new entrants. The absence of that document preserves a revenue model worth billions annually.
• Large dealer banks. The largest participants in clearing and settlement systems sit on the governance boards of those systems. They do not merely use the infrastructure — they govern it. A constitutional framework that insulated the governing body from participants would end this arrangement. The conflict of interest is structural: the banks most capable of articulating what a fair constitutional framework would look like are precisely the banks who would lose the most from its implementation.
• Dominant exchanges. Exchanges that have vertically integrated clearing and settlement extract rents from the integration. A constitutional framework that separated infrastructure governance from participant governance, and mandated open access to clearing, would unbundle a business model built on vertical integration. The absence of the framework protects the integration.
• Regulators themselves. A constitutional framework for settlement infrastructure would require regulators to define, precisely and publicly, the limits of their own authority. Many regulators prefer the current arrangement — broad supervisory discretion, informal guidance, bilateral negotiation — precisely because it preserves their flexibility. A constitutional document that constrained the infrastructure operator would, necessarily, also constrain the regulator. That is not universally welcome.

The governance vacuum is not an oversight. It is a product — maintained deliberately, by the people who profit from its maintenance, in plain sight.

II — Six Clauses Without Which It Is Not a Constitution

Assume the political moment arrives. A crisis makes the cost of absence undeniable. A regulator with sufficient mandate decides to write the document. What would it need to contain? Not every provision. But six clauses without which the document is not a constitution — it is a press release.

Clause I — The Mandate

The founding failure of every attempted infrastructure governance reform is mandate ambiguity. The authority must govern the infrastructure layer — settlement finality, access to the settlement rail, the finality protocol — and explicitly not govern the application layer above it. The mandate boundary must be written into the founding document with precision, because every incumbent will attempt to use the infrastructure mandate to influence outcomes at the application layer. The document must make this structurally impossible, not merely inadvisable.

The test: could a new entrant read the mandate and know, without asking anyone, whether a specific decision falls within the authority's scope?

Clause II — Access Criteria

Access to the settlement infrastructure must be governed by published, objective criteria — regulatory authorisation, technical standards, capital requirements — that any participant meeting them can satisfy as of right, without commercial negotiation with the infrastructure operator or existing participants. The access criteria must be set and enforced by the constitutional authority, not by the operator, and certainly not by a board on which existing participants sit. Administration by the operator is commercial negotiation by another name. The constitutional document must specify who administers the criteria, what the appeals process is, and what timelines apply — because delay in access administration is functionally equivalent to access denial.

The test: could a new entrant connect to the infrastructure within a defined period, without the consent of any existing participant?

Clause III — Fee Governance

Settlement infrastructure governed as a public utility should price at cost recovery — not profit maximisation. The document must define the cost-recovery principle, specify what counts as a recoverable cost, establish an independent audit function, and set maximum fee review periods. The politically explosive sub-question is float income: incumbent CSDs earn significant revenue from the cash and securities balances that participants hold overnight. A cost-recovery framework must specify whether float income counts as revenue against which fees are offset — or whether it is simply extracted. The document must resolve it constitutionally, not leave it to administrative discretion.

The test: could an independent auditor determine whether fees comply with the cost-recovery principle without asking the operator for its interpretation?

Clause IV — Conflict-of-Interest Architecture

The governing body of the infrastructure authority must not include, in any voting capacity, representatives of entities that participate in the infrastructure. This sounds obvious. It is violated by every existing settlement governance structure. Euroclear's board includes representatives of its user-shareholders. DTCC is governed by its participants. (Euroclear; DTCC, governance disclosures) The constitutional document must specify who appoints the governing body, what qualifications are required, what cooling-off periods apply to former industry participants, and — critically — what the removal process is. An insulated governing body that can be removed by governments under political pressure is not insulated. The insulation must extend to the appointment and removal process, not just the day-to-day governance.

The test: could a decision that reduced incumbent revenue and benefited new entrants be made by the governing body without any participant being able to block, veto, or commercially retaliate against it?

Clause V — AI Accountability

This clause does not exist in any current infrastructure governance document. It needs to. AI will operate at every layer of future settlement infrastructure — risk monitoring, compliance attestation, collateral optimisation, anomaly detection. When those systems make consequential decisions that cause harm, the accountability chain must be specified in advance. The document must define what decisions may be made by automated systems without human review, what decisions require human authorisation, what the escalation process is when an automated decision is contested, and who bears liability for automated decision errors. The accountability gap in AI-operated infrastructure is not a future problem — it is a present one, and a constitutional document written today that does not address it is already obsolete.

The test: if an AI system operated by the infrastructure authority caused a participant material harm, could that participant identify, within 24 hours, the human or institution accountable for the decision?

Clause VI — Enforcement

This is the clause that determines whether the framework is real. A constitutional document with no credible enforcement mechanism is not a constitution — it is an aspiration. The enforcement clause must specify sanctions for participant violations that are proportionate, automatic where possible, and not subject to override by the participant or their home regulator. It must address the specific failure mode of large-participant violation — where the participant is systemically important enough that enforcement creates its own systemic risk. The document must specify a resolution mechanism: graduated sanctions, pre-positioned liquidity backstops, mandatory wind-down procedures — whatever is required to make enforcement credible against the largest participant in the system.

The test: if the largest participant in the settlement system violated Clause II tomorrow, could the authority enforce a sanction without triggering a systemic event?

III — Why It Has Never Been Attempted

The six clauses above are not exotic. They are the standard provisions of any credible regulatory constitution. (CPMI–IOSCO, 2012) They are not technically novel. They are politically impossible — for four compounding reasons.

Four reasons

• The jurisdictional problem. No single regulator has the mandate to write it. Securities regulation is national. Settlement infrastructure is global. Writing it requires multilateral agreement that no existing body has the mandate to negotiate.
• The incumbent capture problem. The bodies most capable of drafting the document are the ones most threatened by it. The working groups that produce settlement reform proposals are staffed by people seconded from incumbent institutions. The document cannot be written by the people in the room — and the people in the room control who is in the room.
• The complexity alibi. Settlement infrastructure is genuinely complex. This complexity is weaponised as a permanent deferral strategy. Every reform proposal generates a call for further study, a pilot programme, a sandbox, a consultation paper. These are not steps toward a constitutional framework — they are substitutes for one.
• The giving-something-up problem. Constitutional documents require someone to give something up. Access criteria mean incumbents give up the commercial negotiation. Fee governance means operators give up rent extraction. Conflict-of-interest architecture means large participants give up board seats. The people who would lose something are the people at the drafting table.

IV — The Blockchain Test, Scored

Every serious blockchain settlement proposal is an implicit constitutional document. Score the main proposals against the six clauses:

Scores against six clauses

• Permissioned bank consortium chain — 2 / 6. Fails Clause IV (conflict of interest — banks govern a system they use) and Clause II (access criteria set by participants). Passes Clause I only partially — mandate is clear but too narrow.
• Public chain (Ethereum) — 2.5 / 6. Passes Clause II (open access) and partially Clause III (fee governance via gas market). Fails Clause V (no AI accountability framework), Clause VI (no enforcement against large validators), and Clause IV (validator concentration without accountability).
• DAO-governed settlement protocol — 1.5 / 6. Fails Clause VI catastrophically — token-weighted governance has no credible enforcement mechanism against a large token holder. Fails Clause IV — token concentration creates conflicts structurally identical to board capture.
• Wholesale CBDC settlement rail — 3.5 / 6. Strongest candidate. Passes Clause I (clear mandate), Clause III (central bank cost-recovery), Clause VI (central bank enforcement credibility). Fails Clause IV if governed by a single central bank with political exposure. Clause V (AI accountability) unaddressed in any current design. Clause II depends entirely on implementation.

The wholesale CBDC model comes closest — but 3.5 out of 6 is not a constitution. It is a starting point. The clauses it fails — conflict of interest at the central bank level, AI accountability, and access criteria — are precisely the ones that will determine whether the resulting infrastructure serves the system or the incumbent operators of it.

— The One Precedent That Worked

In 1930, the major central banks created the Bank for International Settlements — a multilateral institution with a specific mandate, insulated from national governments by treaty, governed by its member central banks rather than by any single state. It was created because a crisis made the cost of the governance vacuum undeniable, and because a political moment existed in which major powers had sufficient shared interest to give something up. (Toniolo, 2005)

The BIS is not a perfect model for settlement infrastructure governance. Its governance reflects the power structure of 1930 more than that of 2026. But it demonstrates the essential point: a multilateral infrastructure governance body with constitutional constraints can be created, can survive for nearly a century, and can perform genuine functions that no national body could perform alone.

What a financial market infrastructure equivalent would need to do differently: broader membership from the start. Explicit access criteria. A fee governance framework with independent audit. And — the critical innovation — a pre-specified AI accountability framework, because the BIS equivalent for settlement infrastructure will operate AI systems at every layer, and the accountability gap cannot be left to future administrative guidance.

None of this is technically difficult to design. Constitutional scholars, regulatory economists, and infrastructure lawyers could draft the document in months. The knowledge exists. The precedents exist. The technology exists. What does not exist is the political moment. Constitutional documents for infrastructure are almost always written in the aftermath of a crisis that made the cost of absence undeniable. The crisis that will precipitate the constitutional moment for settlement infrastructure has not yet arrived. The time to draft it is now, when there is no crisis and no urgency and therefore no incumbent with sufficient interest to prevent a serious attempt.

The vacancy at the centre of financial market infrastructure governance is not a gap waiting to be filled by technology. It is a decision waiting to be made by people with the authority and the will to make it. AI will not write the document. Blockchain will not write it. What is missing is not the pen — it is the political will to pick it up before the building is already on fire.