<?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>Writing on Matthew Norgren</title><link>https://norgren.org/writing/</link><description>Recent content in Writing on Matthew Norgren</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Wed, 06 May 2026 09:00:00 -0400</lastBuildDate><atom:link href="https://norgren.org/writing/index.xml" rel="self" type="application/rss+xml"/><item><title>What systems can't fix</title><link>https://norgren.org/writing/what-systems-cant-fix/</link><pubDate>Wed, 06 May 2026 09:00:00 -0400</pubDate><guid>https://norgren.org/writing/what-systems-cant-fix/</guid><description>Building systems that make the right thing automatic is high-leverage. But systems shape behavior at the margins. They don&amp;#39;t fix a person who&amp;#39;s checked out, and learning that was the hardest lesson of leading engineers.</description></item><item><title>It looks like it works</title><link>https://norgren.org/writing/it-looks-like-it-works/</link><pubDate>Mon, 12 Jan 2026 18:45:00 -0400</pubDate><guid>https://norgren.org/writing/it-looks-like-it-works/</guid><description>I&amp;#39;m all in on AI for building software. But it makes it easy to trick yourself: the demo that dazzles, the plan that&amp;#39;s polished and quietly wrong. Judgment is the scarce thing now.</description></item></channel></rss>