Validate Edge

Validate a causal edge based on observed outcomes.

When a prediction is made and later verified/falsified, update edge strength:

• Success: Increase strength (edge was correct)
• Failure: Decrease strength (edge was wrong)

Uses Bayesian averaging to accumulate evidence:

new_strength = (old_strength * count + adjustment) / (count + 1)

Also tracks validation_count and invalidation_count for meta-analysis:

• Edges with many validations are reliable
• Edges with mixed validation are uncertain
• Completely invalidated edges can be removed

Why This Matters

Causal edges are hypotheses. This endpoint turns them into tested knowledge:

• Observing outcomes validates predictions
• Strength gradually approaches ground truth
• Enables continuous improvement via experience

This is how Ninai learns from the world: discover → predict → observe → validate.

Args: request: Edge ID, success flag, and strength adjustment current_user: Authenticated user db: Database session

Returns: Updated edge with new strength and validation counts

Example: ``` Edge: "user_context → model_accuracy" (strength 0.6) Observed: Accuracy improved 23% when context increased

Request:
{
    "edge_id": "edge_123",
    "success": true,
    "strength_adjustment": 0.25
}

Returns:
{
    "id": "edge_123",
    "strength": 0.725,
    "validation_count": 5,
    "invalidation_count": 1,
    "last_validated_at": "2026-03-02T10:30:00Z"
}
```