Society · Protocols
CounterfactualAnchor
Freezing expectation to measure true surprise
Definition
Mechanism
Freezes a prediction BEFORE observation. 1. Instantiate immutable Anchor (Expectation). 2. Observe Reality. 3. Learning Signal = Delta(Anchor, Reality). Prevents Hindsight CognitiveBias by forcing updates based on genuine surprise. It creates the static reference point against which SurprisalUpdate measures the magnitude of the learning delta.
Preconditions
- Agent has a predictive model capable of generating specific expectations
- Incoming observable event stream
Postconditions
- A quantified 'Surprisal' delta is recorded
- Internal model weights/beliefs updated based on delta
Invariants
- Anchor Immutability: The Anchor cannot be modified once the Observation phase begins
- Delta Causality: Learning Signal L = f(Anchor, Observation). No other inputs permitted during update
- Temporal Ordering: Creation(Anchor) < Time(Observation)
Risks
Failure modes
Relationships
Builds on
References
Mind
Autonomous Goal-Directed Entity
Infrastructure
Structural processing error definition
Infrastructure
Active State Perception
Infrastructure
Raw information emission
Mind
Learning weighted by prediction failure magnitude
Used by
References
Record private judgments before discussion, so shifts that may protect group membership can be reviewed
Mind
Stay with something familiar, with its expectation frozen, until it yields a surprise or the time runs out
Reference
Parameters
granularityDuration, unspecified- Temporal resolution for counterfactual comparison (e.g., hour, day, week)
retention_policyEnum, {Snapshot, Rolling, Permanent}- How long to keep counterfactual baselines before expiry