Skip to content

Graph Propagation & System Contagion

Systems are not isolated.

Failures spread.


A disruption in one node can:

  • cascade
  • amplify
  • propagate

across an entire system.


The Pattern

Graph Propagation models relationships between components.

It evaluates how state changes in one part of the system affect others.


Execution Shape

text
system state + connections

adapter (graph model)

graph@1 (propagation simulation)

(optional) mc@1 (stochastic variation)

ensemble@1 (aggregation)

artifacts + replay

Primitive Composition

  • graph@1 — propagates state across nodes
  • mc@1 (optional) — adds stochastic variation
  • ensemble@1 — aggregates system outcomes

What Gets Computed

Forge does not compute:

  • isolated outcomes

Forge computes:

  • cascade effects
  • propagation paths
  • system-wide impact
  • fragility networks

Output Artifacts

text
propagation_paths
cascade_probability
system_fragility_map
node_influence_score
failure_spread_profile
replay_token

Why It Matters

Many systems fail due to:

  • interconnected dependencies
  • hidden coupling
  • cascading effects

Why Forge

Forge enables:

  • system-level reasoning
  • propagation modeling at scale
  • deterministic replay of cascades