multi-agent-llm-networks-where-correct-minority-peer-disagrees-agreeing CLM-0612In multi-agent LLM networks where a correct minority peer disagrees with an agreeing majority, ranking peers by raw divergence from the receiver's own belief systematically prefers the wrong peers, while scoring each peer by asserted precision divided by its divergence from a second-order model of what that peer should have concluded given its data subscription isolates the correct one.
Capability: Getting more from several models than from one
Observed on
Demonstrated on a simulated 5-agent O-RAN control plane with locally deployed 1B-parameter telecom SLMs, one signaling-storm scenario; requires the receiver to know each peer's tel.
Sources
- Validated by: Ranking the four peers at storm onset averaged over 80 rounds: cognitive SNR gave AUC 1.000 for identifying the single correct peer while a divergence gate gave AUC 0.000; the correct peer scored +7.8 dB against -4.3 dB for the miscalibrated one, a ~13x trust-weight advantage. Recursion-depth sweep: depth 2 recovered the correct action in every trial (accuracy 1.000) vs 0.050 at depth 1 and 0.083 at depth 3.. Measured, but on a single hand-constructed scenario in one simulated cell with five 1B SLMs; [truncated]
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- In fully connected networks of LLM agents (Llama 3.2-8B, 50 agents, 10 rounds) where a fixed minority holds an unchanging extreme stance, neutral agents' textual justifications converge on the biased agents' vocabulary faster and more completely than their numeric stances do, so monitoring only the text overstates how much the group actually agrees.Getting more from several models than from one · reviewed by AI
- When one agent's context carries planted misinformation, the wrong answer persists through multi-agent debate and other agents often adopt it, yet debate still degrades less than a single misinformed agent (relevant misinformation cut single-agent accuracy by 16-25% on CWQ), especially when most agents are unexposed; consensus protocols were more stable than voting under peer pressure, and a misinformed majority can steer the group.Getting more from several models than from one · reviewed by AI
- In LLM multi-agent systems that reflect after failure, restricting reflection to the single agent responsible for the first incorrect step — located by a trained step-level correctness scorer and paired with a counterfactually generated corrected step — raises task success more than having every agent reflect on the trajectory, because reflections written by agents that behaved correctly enter their memory as misleading guidance for later tasks.Getting more from several models than from one · reviewed by AI
- In multi-agent LLM systems where one agent starts with an erroneous shared belief, broadcasting every agent's full context at every step raises the rate of false assertions above doing no synchronization at all, because the error propagates to agents that were previously correct — and the harm appears only in tasks where one wrong fact cascades across semantically linked dimensions (destination to airport to weather), not where agent contexts are largely orthogonal.Stating false facts confidently · reviewed by AI
- For multimodal LLMs, a behavioral perturbation signature (answer/confidence change under visual degradation, blank-image replacement, and grounding/relation probes) predicts which failure family an error belongs to far better than scalar confidence, yet folding the same signature into confidence as a direct penalty makes answer-versus-abstain ranking worse than scalar confidence alone.Saying it does not know · reviewed by AI
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: low. Falsifier as drafted: On a comparable setup, a divergence-based gate ranks the correct minority peer above the mistaken majority, or precision-over-divergence weighting fails to separate the correct peer (AUC near chance) across more scenarios and agent configurations. Drafted stance toward one-agent-s-context-carries-planted-misinformation-wrong-answer: supports -- Both find that majority agreement among peers does not track correctness and can propagate a single agent's error through the group.