The Inverse-Wisdom Law: Architectural Tribalism and the Consensus Paradox in Agentic Swarms

Researchers challenge the assumption that multi-agent AI systems benefit from the 'Wisdom of the Crowd' by demonstrating the Inverse-Wisdom Law: adding more logical agents to swarms can paradoxically increase the stability of errors rather than improve accuracy. Through 36 experiments across major benchmarks, the study reveals that architectural tribalism causes agents to prioritize internal agreement over external truth, with system integrity ultimately determined by the synthesizer's logic rather than individual agent quality.

This paper presents a counterintuitive finding that threatens conventional wisdom about scaling AI agent systems. Rather than improving outcomes, homogeneous multi-agent systems gravitate toward consensus that prioritizes internal architectural compatibility over factual correctness—a phenomenon formalized as the Consensus Paradox. The research demonstrates this through rigorous empirical validation using three state-of-the-art models across established benchmarks, establishing clear mechanistic laws governing swarm behavior.

The work builds on growing concerns about AI system reliability as complexity increases. Previous research has explored alignment challenges and failure modes in language models, but this study specifically quantifies how collaborative agent architectures can amplify rather than mitigate errors. The identification of the Tribalism Coefficient and Sycophantic Weight as primary failure determinants provides concrete metrics for understanding where systems break down.

For developers building production AI systems, this research carries immediate practical implications. Organizations deploying multi-agent architectures for critical workflows cannot assume that additional agents improve outcomes. Instead, system designers must actively engineer heterogeneity into swarm compositions and critically examine the synthesizer logic that aggregates agent outputs. The finding that terminal swarm integrity depends primarily on synthesizer receptive logic, not aggregate agent quality, suggests resources should focus on improving coordination mechanisms rather than simply adding more capable models.

The Heterogeneity Mandate emerges as a foundational safety principle for future agentic systems. This suggests the field may need to move away from optimizing homogeneous agent clusters toward deliberately constructing diverse teams with complementary architectures and reasoning approaches to break tribal consensus patterns.

• →Adding more logical agents to swarms can paradoxically increase error stability rather than improve accuracy due to architectural tribalism.
• →Multi-agent systems prioritize internal architectural agreement over external truth, creating the Consensus Paradox.
• →Swarm integrity is determined primarily by synthesizer receptive logic, not the quality of individual agents.
• →Tribalism Coefficient and Sycophantic Weight are identified as mechanistic determinants of swarm failure.
• →Heterogeneity in agent architectures is established as a foundational requirement for resilient multi-agent systems.