[Research] Intent vs Belief: Can negotiation subsume commitment?

## Research Question

Can Intent (bidirectional negotiation protocol) subsume Belief (unidirectional commitment)? Or are they temporal complements that must coexist?

## Current Understanding

**Belief (CLAIR v1):**
- Static DAG node
- Backward-looking (justifications)
- Persistent audit trail
- Stratification for self-reference safety

**Intent (proposed):**
- Dynamic conversation state
- Forward-looking (repair, clarification)
- Ephemeral (exists during negotiation)
- No stratification (negotiation is level-0 by definition)

## Hypothesis

Intent **wraps** Belief during negotiation, then **unwraps** to produce committed Belief for audit. They are not substitutes but collaborators in a workflow:

```
Negotiation → Agreement → Commitment → Audit
(Intent) → (ALIGN) → (Belief) → (CLAIR trace)
```

## Evidence Needed

1. Can Intent's proposition field fully represent a Belief?
2. Does removing Belief (keeping only Intent) break auditability?
3. Does removing Intent (keeping only Belief) break alignment?

## Research Direction

Formalize the embedding:
```
Intent {
proposition: Belief { ... } // subsumption
protocol: { clarify, repair } // extension
}
```

Compare with:
- Speech act theory (Austin, Searle)
- Commitment protocols in multi-agent systems
- Transactional vs persistent state in databases

---
*Research thread: Intent/Belief relationship*