介绍语义距离
This week Moss and Geoffrey Huntley (creator of the Ralph loop) wrote about the importance of automated feedback as back pressure for agents. It’s the idea that the controls you build around and for an agent determines how far it can run autonomously. If you want to deep dive in the topic, you should check this session by Dex and Vaibhav.
本周，Moss 和 Geoffrey Huntley（Ralph loop 的创造者）撰写了关于自动化反馈作为智能体反压的重要性。其核心理念是，你围绕智能体并为其构建的控制机制，决定了它能自主运行多远。如果你想深入了解这个话题，应该看看 Dex 和 Vaibhav 的这场分享。
This post focuses on how to build a metric to control back pressure; we will frame it with two basic questions:
本文重点介绍如何构建一个指标来控制 back pressure；我们将通过两个基本问题来展开探讨：
- The When question: When should a human step in? The later, the better.
- “何时”问题：人类何时应该介入？越晚越好。
- The What question: What should the human look at? The less, the better.
- What 问题：人类应该看什么？越少越好。
AI’s ultimate goal is to scale to levels unimaginable before. Each time you reduce the gap of what and when, you should naturally go faster due to the increase of time. Hence what and when will naturally go up again. It’s an optimization game.
AI 的最终目标是扩展到以前无法想象的水平。每次你缩小 what 和 when 之间的差距时，由于时间的增加，你自然应该走得更快。因此 what 和 when 自然会再次上升。这是一场优化游戏。
Each time you reduce the gap of what and when, you go faster — then both naturally go up again.
每次你缩小 what 和 when 的差距时，你的速度就会加快——然后两者自然都会再次上升。
Now that we are aligned that back pressure is the key to control what and when, we might want to bisect how the feedback is structured. This is, understanding the different levels of back pressure we can differentiate, which ones should be automated and which ones should be part of the human loop. I decided to call it semantic distance.
既然我们已经达成共识，back pressure 是控制内容和时机的关键，我们可能想要二分查找反馈是如何构建的。也就是说，理解我们可以区分的不同层级的 back pressure，哪些应该自动化，哪些应该成为人类循环的一部分。我决定将其称为 semantic distance。
Semantic Distance: the different levels of backpressure measures for your coding agent.
语义距离：为你的编码代理提供的不同级别的反压措施。
It’s a simple scale that describes the different types of backpressure measures you can set to your coding agent to be able to reduce what an...