The Ringelmann Effect in Multi-Agent LLM Systems — a Scaling Law for Effective Team Size

arXiv·medium signal

Accepted at ICML 2026, this paper documents the 'Ringelmann effect' in multi-agent LLM systems: adding more agents yields diminishing (and eventually negative) returns on effective team productivity, and proposes a scaling law for the optimal team size. It's a useful counterweight to the 'just add more agents' instinct, giving builders a principled reason to cap fan-out. Directly applicable to anyone tuning agent-count in orchestration pipelines.