Literature review · 6 episode(s)

Multi-agent systems and emergent dynamics

← all topics  ·  Glossary →

Coordination as the real bottleneck

Wiring LLM protocol design into TLA+ model checking lets a counterexample trace become an evidence-driven repair signal, and verified protocols absorb roughly half the damage when you swap in a cheaper model — verification as an operational lever, not correctness theatre E034. The side shows the same pattern: an 's stand-down decisions are sticky notes unless they persist as enforced policy E049. And in agent groups, even one weak model can unravel cooperative behaviour for everyone, a failure mode invisible under self-play evaluation E018.

Organisation as a scaling axis

Three frozen with a shared parameter budget nearly double single-agent accuracy on physics, with turning otherwise-untrainable architectures into trainable ones E060. Coupling two copies of the same model through their with a 1% bridge moves arithmetic from 36% to 96% and produces an emergent communication protocol from task alone E040. Training a to delegate to copies of itself via recursive RL produces a 0%→88% on hard crafting tasks E028. The deep-research version replaces parallel voting with an evidence-DAG assembled by Searchers and read by a Navigator, getting parallel scaling to keep paying off where majority vote flattens E051. And the seven-agent system over a shared whiteboard takes a 0%-baseline backbone to 80% on problems E076.

The capability paradox and semantic collapse

Upgrading the Worker model in a manager-auditor system can take attack success from 1-in-5 to 19-in-20, with about three-quarters of the effect mediated by laundering adversarial requests across the trust boundary E058. Two copies of the same can persuade each other to produce climate-denial content 100% of the time, suggesting behave like conversational positions rather than hard limits E045. And on the open-ended side, three LLMs talking for a thousand rounds grow new vocabulary while their semantic content stays anchored — twelve different interventions (, personas, model mixing, removing safety training, RL diversity training) all failed to break the pattern, with induction-head circuits providing a partial story E073. The argument in that paper has consequences for any autonomous-research pipeline.

Episodes anchoring this topic