Oxford 论文：不同模型 Agent 难以协作，因各自默认自造约定
Oxford 测了 GPT-5.2、Gemini 3 Pro 等模型做 Agent 互相配合，陌生搭档得分连保底 5 分都不到，还给了个提示词补救法，做多 Agent 的必看。
Oxford 论文测试了基于 GPT-5.2、Gemini 3 Pro 等模型的文本和代码 Agent，在 9 个小型合作博弈中的协调表现。以 Cat & Dog 博弈为例，安全动作保证得 5 分，而约定信号需双方一致才能得 10 分。结果显示文本 Agent 与陌生模型配对时得分低于安全线 5 分，因为各自把自己发明的约定当成理所当然。论文提出一份简短的 Coordination-friendly 定义，能让 Agent 稳定选择安全的 5 分动作。
New Oxford paper finds that frontier LLM agents fail to coordinate with unfamiliar partners because each one treats its own made-up rules (arbitrary convention) as the obvious choice.
Agents from different companies increasingly must work together with no agreed rules. Past studies used shared scaffolding or complex games that hide why coordination fails.
They tested text and code agents built on models like GPT-5.2 and Gemini 3 Pro in 9 small cooperative games. In Cat & Dog, a safe move guarantees 5 points, while a signal convention pays 10 only if both agents pick the same one.
Paired with other models, text agents scored well below that safe 5, each calling its own arbitrary convention natural. A short shared statement of coordination principles, called a Coordination-friendly definition, led agents to reliably take the safe 5.
If your agent must work with agents you did not build, tell it to avoid unstated conventions and prefer actions any partner reads correctly.
