Human-Centric System Architecture: Decentralized DecisionMaking to Eliminate Architectural Bottlenecks

Abstract

Modern systems face limitations imposed by centralized control. These limits lead to single points of failure, uneven information flow, and slow decisions. I present a multi-layer mathematical model for human-centric, decentralized systems. Our model offers quantitative tools to identify and reduce bottlenecks by distributing decisionmaking. The framework introduces core metrics: Bottleneck Index, Decentralization Degree, Decision Efficiency Function, Collective Intelligence Score, and Resilience Index. A four-layer architecture—Strategic Human Decision, Decentralized Coordination, Autonomous Agent, and Technical Infrastructure—is described. We validate the approach using thematic analysis and simulation across organizational, healthcare, and autonomous settings. Ablation studies show that modular design, self-organization, and adaptability reduce bottlenecks. The system improves CIS by 41.3% over centralized systems. This framework guides engineers and leaders to build resilient sociotechnical systems