<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel><title>Game Anatomy</title><link>https://gameanatomy.net</link><description>Evidence-led experiments on AI game generation, testing, and the verification harnesses that make results inspectable.</description><item><title>One Build Failed. Then It Passed. Then It Failed Again.</title><link>https://gameanatomy.net/posts/one-build-failed-passed-failed-again/</link><guid>https://gameanatomy.net/posts/one-build-failed-passed-failed-again/</guid><description>One Space Invaders route received primitive FAIL, engineered mechanical PASS, and human-terminal practical FAIL—and all three verdicts were honest.</description><pubDate>Mon, 13 Jul 2026 00:00:00 +0000</pubDate></item><item><title>The Mechanic Was Real. Human Review Found It Too Subtle.</title><link>https://gameanatomy.net/posts/mechanic-real-human-review-too-subtle/</link><guid>https://gameanatomy.net/posts/mechanic-real-human-review-too-subtle/</guid><description>Command Cells passed every named machine gate, but one operator review found the mechanic too subtle and set a 30-second magnitude bar for the next search.</description><pubDate>Mon, 13 Jul 2026 00:00:00 +0000</pubDate></item><item><title>The Accepted Build Took Three Model Roles.</title><link>https://gameanatomy.net/posts/accepted-build-took-three-model-roles/</link><guid>https://gameanatomy.net/posts/accepted-build-took-three-model-roles/</guid><description>The accepted 03.2B Space Invaders base combined a cleaner GPT seed, a Flash presentation donor, and a Pro executor under machine and human verification.</description><pubDate>Mon, 13 Jul 2026 00:00:00 +0000</pubDate></item><item><title>The Untestable Game Won</title><link>https://gameanatomy.net/posts/the-untestable-game-won/</link><guid>https://gameanatomy.net/posts/the-untestable-game-won/</guid><description>In a Breakout experiment, mechanically testable builds passed while uninstrumented grafts won the human review.</description><pubDate>Sat, 11 Jul 2026 00:00:00 +0000</pubDate></item><item><title>All Three Breakouts Passed. None Could Restart.</title><link>https://gameanatomy.net/posts/all-three-breakouts-passed-none-could-restart/</link><guid>https://gameanatomy.net/posts/all-three-breakouts-passed-none-could-restart/</guid><description>All three engineered Breakouts passed the mechanical driver, but none provided the player a complete restart loop.</description><pubDate>Fri, 10 Jul 2026 00:00:00 +0000</pubDate></item><item><title>My Lie Detector Lied</title><link>https://gameanatomy.net/posts/my-lie-detector-lied/</link><guid>https://gameanatomy.net/posts/my-lie-detector-lied/</guid><description>A deterministic Pong test harness falsely failed correct AI-generated games twice, showing why falsifiers need their own autopsy.</description><pubDate>Wed, 08 Jul 2026 00:00:00 +0000</pubDate></item><item><title>Fun Is Not a Spec</title><link>https://gameanatomy.net/posts/fun-is-not-a-spec/</link><guid>https://gameanatomy.net/posts/fun-is-not-a-spec/</guid><description>An AI Pong specification can enforce correctness, but this experiment shows why player feel still needs separate human judgment.</description><pubDate>Wed, 08 Jul 2026 00:00:00 +0000</pubDate></item><item><title>Three Pongs From One Line</title><link>https://gameanatomy.net/posts/three-pongs-from-one-line/</link><guid>https://gameanatomy.net/posts/three-pongs-from-one-line/</guid><description>Three AI systems turned one vague Pong prompt into different games, then an explicit test contract made the results verifiable.</description><pubDate>Tue, 07 Jul 2026 00:00:00 +0000</pubDate></item></channel></rss>
