ADR-0002 — Sovereignty-first governance
Status: Accepted Date: 2026-08-05 Decided by: Tamga Network project management Source: docs/tamga-network-guven-katmani-kontratlari-v2.md (architecture decision draft) — this ADR formalises the governance section of that draft.
Context
PM-BC-0001 left governance broadly as "founding consortium + on-chain voting". A critical question required this to be refined:
Why should Türkiye wait for Kazakhstan's vote when it makes one of its own institutions (e.g. the Ministry of Education) an issuer?
It should not. Otherwise this would be an unacceptable violation of sovereignty and the biggest obstacle to adoption. The precedent is clear: under eIDAS (electronic IDentification, Authentication and trust Services)The EU regulation on electronic identification and trust services; eIDAS 2.0 introduces the European Digital Identity Wallet. every member state publishes its own Trusted List independently; the EU only aggregates the lists (LOTL (List of Trusted Lists)The signed list that points to the countries' trust lists, their signers and their recognition status.). No state votes on which institutions another state makes issuersThe institution that signs and issues credentials: a university, a professional body, a public institution or a company.. Mutual recognition is an agreement or a choice, not something voted institution by institution.
Decision
Governance is split into three layers; in each layer a different actor decides:
| Layer | Subject | Who decides | Threshold |
|---|---|---|---|
| 1 — Network membership | A new state becoming a validator / being removed / protocol upgrades | Validator majority (existing states) | 2/3 |
| 2 — National registries | Its own issuers, its own relying parties | Only the state concerned | NO vote (onlyOwnerState) |
| 3 — Cross-border recognition | "Do I accept credentials from country X?" | Each state, for itself | Unilateral |
Sub-decisions
- Layer 1 thresholds: admitting a new state 2/3; removing a state 2/3 (the removed state's vote is not counted); protocol upgrade 2/3. (2/3 was chosen over unanimity: it prevents a single state from vetoing or locking growth; it is still a heavy decision.)
- Layer 2 — sovereignty in code: every state has a namespace (
TR:issuer:*,KZ:rp:*…) and only that state's authorised key can write to it. This is enforced in the contract by theonlyOwnerState(stateCode)modifier; it does not depend on good faith. - Layer 3 — default recognition policy: FULL among founding members (they fully recognise each other from the start); NONE for later joiners (opt-in) — states open up at the pace of their own legislation. This avoids the "all or nothing" trap.
- Removal ≠ destroying credentials: removing a state only means "can no longer write new records or produce blocks". The credentialsA digital document signed by an issuer and held in the person's wallet; the person shows only the fields that are asked for. in its citizens' wallets do not become invalid; their fate depends on the other states' layer 3 recognition decisions.
- Right to exit: a state can leave unilaterally with
withdraw(); no vote is needed. Guaranteed in code. Note (2026-11-12, ADR-0076): this item was narrowed. Unilateral exit applies while the network has at least three active members. In a two-member network exit is refused (with a single member left, the network's joint decisions would be locked for good); a member that wants to leave first waits for a third member to be admitted, then withdraws without a vote. - Cross-border jurisdiction (home-state sovereignty): a citizen's pseudonymA stable account identifier the wallet derives separately for each website; two sites cannot link the same person. can be opened ONLY by their own state's guardian threshold + their own court; no other state can do so. Detailed in
PM-ID-0002(accountable disclosure); consistent with the principle "disclosure = home-state sovereignty" inPM-ID-0001.
Rationale
- Adoptability. No state joins a network with the risk that "my internal institutional decisions depend on others' votes" or "if I am ever expelled, my citizens' identities become worthless". Sovereignty guarantees are a precondition for participation.
- Precedent (eIDAS/LOTL). This is exactly the EU's working model; proven, not theoretical.
- Gradual integration. Unilateral recognition lets each state open up at its own pace; it removes the need for full alignment in the early days.
- Guaranteed in code.
onlyOwnerState+withdrawturn sovereignty from a promise into an engineering guarantee.
Consequences
Positive:
- A strong incentive to adopt (sovereignty + exit + durable credentials).
- The governance contract's scope shrinks → simpler, smaller attack surface.
- Conceptually aligned with eIDAS → easier interoperability.
Cost / caveats:
- Managing the cross-recognition matrix (every state × every state × category) is an operational burden; the wallet and verifierThe party that checks a presented credential: signature, issuer in the trust list, status and policy. Also called relying party. UX must hide it.
- The "founders FULL" default requires the founding set to be chosen carefully.
- Home-state jurisdiction may be limited in cross-border crime scenarios (inter-state agreements may be needed in the future — open topic in
PM-ID-0002).
Alternatives considered
- Everything by common vote (full consortium): rejected — violates sovereignty in layers 2/3; would not be adopted.
- No common vote at all (fully independent): rejected — the composition of the network (who is a validator) would be uncontrolled; "who sits at the table" must be a joint decision.
- Unanimity (layer 1): rejected — a single state could lock growth; 2/3 preferred.
- Default recognition FULL (everyone): rejected — legislative and sovereignty risk; founders FULL + later joiners NONE preferred.
Related documents
PM-BC-0001— the "founding consortium + on-chain voting" wording is refined by this ADR.- ADR-0001 — the engine (Besu/QBFT); layer 1 votes coincide with the QBFT validator set.
SPEC-BC-0001— the contract implementation of this model (Governance, Registry, Recognition).ARCH-0001— the phased validator model (namespaces and voting in phase 0).PM-ID-0001/PM-ID-0002— home-state sovereignty and accountable disclosure.
Status
The governance model is decided. Implementation details (contract interfaces, threshold parameters) are in SPEC-BC-0001; guardian composition and cross-border jurisdiction will be deepened in PM-ID-0002 + PM-GOV-0001 (planned).