Skip to content

Black Swan & Resilience

A simulation balancing efficiency against catastrophic tail risks through active inference and the Biological Veto.

The Concept

This simulation models a decentralized network (like a DAO or global market) under constant load. It builds upon Bak's Self-Organized Criticality (SOC), moving the sandpile from a 2D grid to a complex network setting.

Two fundamental opposing forces are explored: 1. Efficiency (Throughput): Maximizing the rate of load processed by the network. 2. Antifragility (Resilience): Surviving fat-tailed outliers (Black Swans) that emerge inevitably in critically poised systems.

Key Metrics & Interventions

  • Spectral Gap (\(\lambda_2\)): Displays the algebraic connectivity of the network. A higher \(\lambda_2\) means better natural dissipation of load, preventing localized clustering.
  • Transfer Entropy Proxy: The system monitors early-warning signs (rising variance, lag-1 autocorrelation) that precede a regime shift.
  • Active Inference Agent: An observer that triggers a Biological Veto. When the agent predicts a Black Swan, it halts the system's throughput temporarily, allowing the network to dissipate accumulated tension before a catastrophic avalanche destroys the topology.

Running

python black_swan_simulation.py