Why SCL Eliminates Interpretation
SCL is the deterministic, cryptographic identity protocol for AI run declarations.
SCL eliminates interpretation in representation by enforcing a single canonical form. Identity is derived exclusively from canonical JSON bytes using SHA-256, making it universally verifiable across independent systems.
This makes SCL a deterministic identity system where representation cannot vary.
Why Eliminating Interpretation Is Required
Deterministic identity requires that the same input always produces the same output across independent systems.
If multiple representations of the same declaration are allowed, identity becomes dependent on implementation behavior rather than the declaration itself.
SCL eliminates this by enforcing a single canonical representation.
The Problem: Interpretation Breaks Identity
Without strict constraints, the same declaration can be represented in multiple ways:
- Whitespace differences
- Key ordering differences
- Encoding differences
- Escaping differences
- Serialization differences
Each variation produces different bytes. Different bytes produce different SHA-256 identities.
This makes identity non-deterministic and non-verifiable. Two systems processing the same logical content can produce different identities.
What SCL Eliminates
SCL eliminates interpretation at the representation layer.
- It does not interpret meaning
- It does not evaluate execution
- It does not normalize input
It enforces a single valid representation. Any deviation from the canonical form results in protocol failure.
For any valid declaration:
- Exactly one canonical form exists
- Exactly one byte sequence is produced
- Exactly one SHA-256 identity is derived
How SCL Enforces This
SCL enforces deterministic identity through three strict layers:
1. Deterministic Parsing
Input is processed as raw bytes under a fixed grammar. No normalization, no recovery, no implementation-defined behavior. Invalid input fails deterministically at the first error.
2. Canonical Representation
The parsed structure is serialized into canonical JSON: single-line UTF-8 output, no extraneous whitespace, keys sorted by raw UTF-8 byte order, only defined fields allowed.
Canonical JSON bytes are the sole input to hashing.
3. Identity Derivation
Identity is derived exclusively from canonical JSON bytes using SHA-256. No alternate inputs, no formatting differences, no encoding variation. The same canonical bytes always produce the same identity.
Boundary: This Page Does Not Define Canonical JSON
This page explains why eliminating interpretation is required for deterministic identity. It does not define canonical JSON rules or serialization mechanics.
For canonical representation rules, see canonical JSON specification. For identity derivation mechanics, see SHA-256 identity derivation. For enforcement details, see how SCL enforces deterministic identity.
What This Guarantees
SCL guarantees representation-level determinism.
Given the same valid declaration:
- All compliant implementations produce identical canonical JSON
- All compliant implementations produce identical SHA-256 identity
- No variation is permitted
Identity becomes independent of programming language, runtime environment, and system implementation.
What This Does Not Do
SCL defines identity only.
- It does not make AI execution deterministic
- It does not validate outputs or correctness
- It does not interpret intent or semantics
- It does not control workflows or orchestration
Identity applies only to the declared execution representation.
Why This Matters
Eliminating interpretation enables:
- Cross-system verification
- Reproducible identity across environments
- Reliable audit references
- Stable identity for declared execution
Without eliminating interpretation, identity cannot be trusted.
For the protocol definition, see the definition page. For the complete rules, see the SCL:V1 specification. For what identity covers, see what deterministic identity covers.