Society · Protocols
PromiseGraph
Recursive trust dependencies modeled as a DAG
Definition
Mechanism
Agents do not trust each other directly; they trust a graph of promises. Agent A accepts a promise from Agent B only if B provides a dependency graph of sub-promises it relies on. A verifies the leaves of the graph or checks B's 'credit score' for fulfilling similar graph structures. It maintains a DAG of trust, validating leaf nodes via SpotAudit and ensuring acyclic dependencies via NegativeProof logic.
Invariants
- Acyclic: Promise dependencies cannot form a loop
- Cycles forbidden.
- Leaf Verification: All leaf promises must be anchored in verified truth
Risks
Failure modes
Graph explosion (too much verification overhead).
Relationships
Builds on
References
Mind
Autonomous Goal-Directed Entity
Infrastructure
Proving non-membership in a committed state snapshot
Infrastructure
Random sampling of internal state
Nothing in this library builds on this pattern yet.
Reference
Data schema
{
"type": "object",
"required": [
"nodes",
"edges"
],
"properties": {
"nodes": {
"type": "array",
"items": {
"type": "object",
"properties": {
"promise_id": {
"type": "string"
},
"promisor": {
"type": "string"
}
}
},
"description": "Promise nodes"
},
"edges": {
"type": "array",
"items": {
"type": "object",
"properties": {
"from": {
"type": "string"
},
"to": {
"type": "string"
}
}
},
"description": "Dependency edges"
},
"leaf_anchors": {
"type": "array",
"items": {
"type": "string"
},
"description": "Verified leaf promises"
}
}
}