A Million Agents Is a Distributed Systems Problem
What happened
InstaCloud 的 CTO 把代理（Agent）大规模部署看作分布式系统问题，而不是单纯的模型能力问题。他引用了 Google Research 的一项研究，测试了 180 种代理配置后发现：多加代理在能并行的任务上最多提升了 80.9% 的表现，但在需要严格按顺序推理的任务上，表现反而下降了 39% 到 70%。不加协调的代理会把错误放大 17...
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pagePickA Million Agents Is a Distributed Systems Problem
InstaCloud's CTO frames agent scaling as a distributed systems problem. Google Research's 180-config study found multi-agent setups boosted parallel tasks by 80.9% but hurt sequential reasoning by 39–70%. Uncoordinated agents amplified errors 17.2×; an orchestrator cut that to 4.4×. In ACL 2026's Silo-Bench, teams of 2–100 agents talked a lot but reasoned poorly, with zero success on the hardest tasks at 50 agents. The takeaway: persist state, not the agent—schedule agents like processes and recover them like nodes.