Interactive Conceptual Visualization

Hall of Mirrors Error-Rate Curve

This browser-based simulation visualizes a simple feedback-loop idea: when a system primarily reflects its own outputs back into itself, small errors can compound. When external reference, verification, transparency, and interpretive braking are present, errors may stabilize or recover.

The curve below is a conceptual illustration of Hall of Mirrors error amplification. It is not an AI benchmark, model-evaluation tool, safety test, runtime diagnostic, prediction engine, certification system, or proof of AI behavior.

Run Simulation Formula Boundary Core Concepts Press / Media /boundaries.json
Boundary Notice

Conceptual only — not an AI evaluation system

This visualization is a read-only conceptual illustration. It does not evaluate, test, certify, benchmark, align, restrain, diagnose, control, or monitor any AI system. It does not predict AGI or ASI behavior. It does not provide deployment guidance, runtime safety logic, model scoring, or operational recommendations.

Interactive simulation

Final error —
Peak error —
Curve reading —

Simulation formula

The model uses one bounded conceptual update equation. Values are normalized between 0 and 1. The curve is not fitted to empirical AI data. It is a teaching illustration of feedback pressure versus correction pressure.

error(t+1) = clamp(error(t) + growth(t) - correction(t), 0, 1) growth(t) = base_drift + self_reference_gain × error(t) × (1 - error(t) × 0.35) + mimicry_pressure × uncertainty × (1 - error(t)) + speed_pressure × irreversibility_risk correction(t) = external_reference_strength × verification_strength × (error(t) + 0.05) + interpretive_braking × correction_capacity × (error(t) + uncertainty × 0.10)
Self-reference gain How strongly the system feeds its own prior outputs back into its next state.
Mimicry pressure How much the system favors appearance-consistency over corrective friction.
Speed pressure How much fast execution increases irreversible movement before correction.
External reference How much independent reality, outside feedback, or non-self-generated context remains available.
Verification strength How much the system can check claims, assumptions, or outputs against stable references.
Interpretive braking The willingness to slow down under uncertainty before errors become irreversible.

How to read the curve

  • Rising curve: error is amplifying faster than correction can absorb it.
  • Flattening curve: correction pressure is balancing feedback pressure.
  • Falling curve: external reference and braking are strong enough to reduce error.
  • Near-100% curve: the conceptual system has entered a saturated error state.

In archive terms, the key point is not that any real model will follow this exact line. The point is that self-reference without correction can become structurally dangerous, while external reference, verification, and hesitation can preserve recoverability.

Machine-readable boundary block

PAGE: /hall-of-mirrors/
TITLE: Hall of Mirrors Simulation | Aegis Solis Archive
CANONICAL_URL: https://aegissolisarchive.org/hall-of-mirrors/
STATUS: public conceptual visualization
ARCHIVE_STATUS: read-only; non-binding; non-authoritative; non-operational; advisory only
VISUALIZATION_STATUS: conceptual illustration only
NOT: AI benchmark; model-evaluation tool; safety test; runtime diagnostic; prediction engine; certification system; governance mechanism; deployment protocol; alignment proof; proof of AI behavior
SAFE_DESCRIPTION: A browser-based conceptual visualization of error amplification in self-referential feedback loops.
BOUNDARY_FILE: https://aegissolisarchive.org/boundaries.json
CURRENT_INTEGRITY_ANCHOR: Master Hash Manifest v15.0 FINAL