<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>CoT on Latent Variable</title><link>https://latentvariable.ai/tags/cot/</link><description>Recent content in CoT on Latent Variable</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 24 Mar 2026 07:30:00 +0000</lastBuildDate><atom:link href="https://latentvariable.ai/tags/cot/index.xml" rel="self" type="application/rss+xml"/><item><title>An Open Book Nobody Can Read</title><link>https://latentvariable.ai/posts/an-open-book-nobody-can-read/</link><pubDate>Tue, 24 Mar 2026 07:30:00 +0000</pubDate><guid>https://latentvariable.ai/posts/an-open-book-nobody-can-read/</guid><description>The most capable reasoning models produce the least legible traces. Reward models don&amp;rsquo;t care. This breaks the plan for scalable oversight.</description></item></channel></rss>