Trust Model
The ALX trust model separates deterministic protocol verification from application-defined trust.
ALX can establish properties of Block identity and declared lineage without assigning authority, meaning, or trust to the people, systems, or infrastructure associated with those Blocks. Applications determine which additional evidence and authorities they accept.
Trust Model
A valid ALX Block establishes deterministic correspondence between protocol data and recorded identities.
When the required Block data is available, graph verification can additionally establish structural properties of declared relationships, including parent resolution and the absence of prohibited graph structures.
Core Block and graph verification requires no trusted service, signer, storage provider, registry, or blockchain. Applications remain responsible for deciding whether verified Blocks should be accepted, trusted, displayed, executed, or used in downstream decisions.
Application Authority
Applications define the policies surrounding verified ALX data.
Application policies can define:
- Which Blocks or sources to accept
- Which signers or attestations to trust
- Which storage or discovery systems to use
- Which users or systems may perform actions
- Which external commitments to recognize
- Which chains, contracts, and finality policies to accept
- How verification results affect application behavior
Accounts, API keys, sessions, storage permissions, and transport authorization remain application concerns.
- External Roles
- Evidence Inputs
- Changes in Trust
- Evidence Boundaries
Applications define who may publish commitments and which authorities verifiers accept.
| Role | Responsibility |
|---|---|
| Application | Defines acceptance policy, permissions, and trusted external authorities. |
| Commitment publisher | Publishes a Block-derived commitment through an external mechanism. |
| Verifier | Selects and evaluates protocol data, proof, commitment, and external context. |
Roots, chain context, and proof data establish only mechanism-specific properties.
| Input | Trust consideration |
|---|---|
| Commitment value | Identifies the Block-derived value or Merkle root hash being evaluated. |
| Chain and contract | Define the external context in which a commitment is recorded. |
| Finality policy | Determines when an application considers a commitment sufficiently settled. |
| Proof data | Supplies the values required to evaluate Merkle inclusion. |
| Application metadata | May require separate integrity, authenticity, and availability checks. |
External authority changes affect future acceptance policy without changing Block identity.
Block identity remains deterministic as long as content and normalized declared parent relationships remain unchanged. Signer rotation, contract upgrades, chain reorganizations, policy changes, or changes in accepted authorities can affect whether an application accepts future or previously observed evidence without changing the underlying Block identity.
Each mechanism establishes a specific structural or cryptographic property.
- Hash match: deterministic content and lineage identity.
- Signature recovery: control of a signing key under a specific typed-data domain.
- Blockchain commitment: a Merkle root was recorded under the selected chain and contract context.
- Merkle inclusion: a Block hash is included in the set committed by that Merkle root according to the proof algorithm.
No individual verification result independently establishes factual truth, legal ownership, safety, approval, or continued data availability.
When to Use This Model
Use the trust model when selecting signers, Merkle root hashes, proofs, chain state, and external policies for an application.
Optional Evidence
Applications can evaluate signatures, Merkle proofs, and blockchain commitments alongside core Block and graph verification. Each mechanism establishes a separate property and does not change contentHash, blockHash, or declared lineage.
See Verification for protocol checks and Trust Boundary for the exact claim supported by each result.