Graph-Structured Congestion with Per-Edge Latency

Develop a full graph-structured extension of the Triadic Cognitive Architecture’s congestion model that incorporates per-edge latency in routing information through tools.

Background

The implemented model treats congestion as a scalar accumulator and, in robustness checks, as a discrete Ornstein–Uhlenbeck drain process. However, real networked tool ecosystems often have graph-structured topologies where per-edge latencies and congestion interact.

The paper notes that a comprehensive graph-structured extension with per-edge latency is not yet implemented and explicitly flags it as future work.