协议压缩改变谁付费：跨组织LLM智能体通信中的双边成本
Protocol Compression Changes Which Party Pays: Bilateral Cost in Cross-Organization LLM Agent Communication
浏览论文内容
中文总结 AI 辅助
研究跨组织LLM智能体通信中协议压缩的双边成本，发现压缩放大成本离散度且运行时协商虽成协议但谈判失败，多数对话陷入僵局，成本节省源于提前终止而非压缩本身。
中文摘要 AI 辅助
跨组织通信的智能体之间交换的长消息按令牌计费。因此，一种更短的表示法看似是一种节省，除了达成使用它的协议外无需任何成本。近期研究报告称这种节省是有条件的。压缩表示法反而可能使总令牌数比JSON基线增加8%至11%，因为解析失败会迫使额外的模型调用。这是针对单一付费方测量的。在两个组织之间，任何一方都无法在另一端安装解码器，且各自按自己的分词器、价格和缓存状态付费。我们测量了双方。一项预注册的令牌级研究覆盖了六个供应商的198个内容匹配项对，共2,376个原生使用单元。随后，我们在一个具有精确枚举可行集的双边采购谈判中叠加了英文基线、运行时模式协商后压缩以及注入模式压缩。该叠加覆盖了3个模型对的1,215单元网格中的1,053个完整对话，以及协商条件的405个对话重跑。压缩将跨供应商成本离散度放大了1.078倍，95%置信区间为[1.066, 1.091]，并且有两个供应商对颠倒了哪个端点更便宜。运行时协商作为协议成功，但作为谈判失败。双方在135个主要对话中的121个中达成了模式协议，但没有一个是我们会提供的模式。他们在这些对话中仅解决了9个任务，并在106个中陷入僵局。协商会话平均为10.8轮，而对比为17.6轮，成本为英文总成本的52%，因为会话更早结束，而不是因为握手被偿还。盈亏平衡点范围从20轮到70轮，低端仅在条件性核算下出现，且所有这些都高于观察到的每个英文会话。在一个跨供应商对上，双方都保留了约一半的成本。在另一个对上，接收方在高缓存命中率下支付更多。
英文摘要
Agents that talk across organizations exchange long messages billed by the token. A shorter notation therefore looks like a saving that costs nothing but an agreement to use it. Recent work reports the saving is conditional. Compressed notation can instead raise total tokens by 8% to 11% over a JSON baseline, when parsing failures force extra model calls. That is measured for one payer. Between two organizations neither side can install a decoder at the other end, and each pays under its own tokenizer, price, and cache state. We measure both sides. A preregistered token-level study covered 198 content-matched item pairs across six vendors, for 2,376 native-usage cells. We then overlay an English baseline, runtime schema negotiation followed by compression, and injected-schema compression on a two-party procurement bargain with an exactly enumerated feasible set. The overlay covers 1,053 completed dialogues of a 1,215-cell grid across 3 model pairs, plus a 405-dialogue rerun of the negotiated condition. Compression amplifies cross-vendor cost dispersion by a factor of 1.078, with a 95% CI of [1.066, 1.091], and two vendor pairs reverse which endpoint is cheaper. Runtime negotiation succeeds as a protocol and fails as a bargain. The parties agree a schema in 121 of 135 headline dialogues, none of them the schema we would have supplied. They settle the task in only 9 of those dialogues, and they reach impasse in 106 of them. The negotiated sessions average 10.8 turns against 17.6, and cost 52% of the English total because sessions end sooner, not because the handshake is repaid. Break-even horizons run from 20 to 70 turns, the low end only under the conditional accounting, and all of them lie above every observed English session. On one cross-vendor pair both parties keep about half their cost. On the other the receiving party pays more at a high cache-hit rate.
发表机构
- Redrob
机构由 AI 辅助整理，请以论文原文为准。
