<?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>Papers on Cephas - Liana Banyan Knowledge Center</title><link>https://cephas.lianabanyan.com/papers/</link><description>Recent content in Papers on Cephas - Liana Banyan Knowledge Center</description><generator>Hugo -- 0.152.2</generator><language>en-us</language><copyright>Ã¢â€Â¬Ã¢Å’Â 2026 Liana Banyan Corporation. All rights reserved.</copyright><lastBuildDate>Fri, 19 Jun 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://cephas.lianabanyan.com/papers/index.xml" rel="self" type="application/rss+xml"/><item><title>The Ten-Fold Compounding: BP087 Empirical Receipt</title><link>https://cephas.lianabanyan.com/papers/10x-compounding-receipt/</link><pubDate>Fri, 19 Jun 2026 00:00:00 +0000</pubDate><guid>https://cephas.lianabanyan.com/papers/10x-compounding-receipt/</guid><description>&lt;h2 id="abstract"&gt;Abstract&lt;/h2&gt;
&lt;p&gt;On June 19th, 2026, a 16-minute build session delivered five parallel build streams, three commits (0d00333, b415a24, 80cd33a), three Edge Functions, and four Firebase deploys while consuming 43% of a 200,000-token context window. The prior baseline (BP063, no cooperative substrate, one build session) consumed 86% of the same window size and still produced errors.&lt;/p&gt;
&lt;p&gt;Normalizing by work per token: the BP087 session delivered approximately 10x the work at approximately half the token cost.&lt;/p&gt;</description></item></channel></rss>