Skip to content

Black Swan & Resilience

A toy network-load simulation with an early-warning pause policy.

The Concept

This simulation models load accumulation and redistribution on a selected network. It is inspired by sandpile models; it is not calibrated to a DAO, market, or infrastructure network.

Two quantities are compared in this implementation: 1. Efficiency (Throughput): Maximizing the rate of load processed by the network. 2. Toy resilience: avoiding large modeled cascades under the selected load rule.

The code does not establish a universal efficiency–resilience trade-off, antifragility, or inevitable power-law tails.

Key Metrics & Interventions

  • Spectral gap (\(\lambda_2\)): Reports algebraic connectivity. Any relation to load dissipation is a property to test under this code's redistribution rule, not a general graph-theoretic identity.
  • Early-warning statistics: Monitors variance and lag-1 autocorrelation. These are not transfer entropy and are not universal precursors.
  • Pause policy: Temporarily halts throughput when the selected warning crosses a threshold. The comparison should report false alarms, missed cascades, and total throughput as well as avoided losses.

The current artifact is an exploratory simulation, not evidence that the policy works in a real system.

Running

python black_swan_simulation.py