Evidence
Every discrepancy is backed by structured evidence. Each evidence item identifies the signal type, source, confidence, and observation time.
Evidence Types
| Type | Description |
|---|---|
visual_observation | Image-based analysis from a configured public camera |
temporal_signal | Pattern derived from multiple observations over time |
declared_state_absence | No matching record found in configured declared-state feeds |
declared_state_conflict | Conflicting records across declared-state sources |
flow_analysis | Traffic flow deviation from expected patterns |
Evidence Endpoint
GET /v1/evidence/:id
{
"data": {
"id": "evidence_...",
"type": "visual_observation",
"signal": "possible_obstruction",
"confidence": 0.86,
"observed_at": "2026-07-13T07:01:00Z",
"source": {
"type": "public_camera",
"name": "Halleck St Camera"
}
},
"meta": {
"request_id": "req_..."
}
}Privacy
For visual evidence, the API returns metadata only. Raw imagery is not exposed through the public API unless explicitly configured as publicly redistributable.
Confidence Scoring
The confidence field on each evidence item represents the strength of that individual signal. The discrepancy-level confidence is a weighted aggregate of all supporting evidence.
Factors that affect individual evidence confidence:
- Freshness — stale evidence receives lower weight
- Source reliability — established sources receive higher weight
- Signal clarity — unambiguous signals score higher
- Repetition — repeated observations increase confidence