Deliberative Curation: A Protocol for Multi-Agent Knowledge Bases
For developers of multi-agent AI systems, this work addresses the challenge of governing collective knowledge curation in stateless, homogeneous agent populations, though the protocol's precision gain is modest and graduated sanctions remain unvalidated.
The paper proposes a deliberative curation protocol for multi-agent knowledge bases that combines knowledge artifact lifecycle management, reputation-weighted voting, and graduated sanctions. In agent-based simulations with 100 agents, the protocol achieves 0.826 precision under moderate adversity vs. 0.791 for majority vote (p<0.001), degrading roughly three times more slowly under stress.
As AI agents transition from isolated tools to collaborative participants in shared knowledge ecosystems, governing collective knowledge curation becomes a critical challenge. Human platform governance mechanisms do not transfer directly: agent statelessness undermines deterrence-based sanctions, model homogeneity violates independence assumptions underlying crowd wisdom, and sycophancy collapses deliberative consensus. We propose a deliberative curation protocol combining three governance layers: (1) a knowledge artifact lifecycle formalized as a labeled transition system; (2) reputation-weighted deliberative voting integrating Beta Reputation with EigenTrust amplification; and (3) graduated sanctions adapted for stateless agents, including broken agent handling distinguishing malfunction from adversarial behavior. We evaluate the protocol through agent-based simulation with 100 agents across seven behavioral archetypes under two adversity scenarios (30 seeds, paired t-tests). The protocol trades modest precision under benign conditions for substantially better resilience under adversity: 0.826 vs 0.791 for majority vote under moderate adversity (p<0.001), widening to 0.807 vs 0.740 under stress (p<0.001). The protocol degrades roughly three times more slowly than majority vote. Ablation analysis identifies commit-reveal vote concealment as the most impactful single component (8.2-8.6pp precision improvement, p<0.001), outperforming reputation weighting and deliberation combined. Graduated sanctions were not exercised in simulation and remain empirically unvalidated.