<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Experiment on Phil Wheat</title><link>https://philwheat.net/tags/experiment/</link><description>Recent content in Experiment on Phil Wheat</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 13 Oct 2026 09:04:00 -0500</lastBuildDate><atom:link href="https://philwheat.net/tags/experiment/index.xml" rel="self" type="application/rss+xml"/><item><title>Infinite Livestream</title><link>https://philwheat.net/links/infinite-livestream/</link><pubDate>Tue, 13 Oct 2026 09:04:00 -0500</pubDate><guid>https://philwheat.net/links/infinite-livestream/</guid><description>&lt;p&gt;Infinite livestream code. Needs a LOT more hardware than I have, but is a curiosity.&lt;/p&gt;&#10;&lt;p&gt;A chat-driven, never-ending AI video broadcast from reactor-team: viewers type &lt;code&gt;!prompt &amp;lt;idea&amp;gt;&lt;/code&gt; in Twitch/YouTube chat, an LLM expands each idea into styled scene sequences, the fast-h3 model generates 768p clips with synchronized audio, and it all goes out over RTMP as one uninterrupted stream. Two halves: the model queue (runs on the Reactor Runtime, 8x B200 — yes, really) and a streaming client (Python + reactor-sdk, any box with ffmpeg). Apache-2.0 code; MiniMax H3-licensed weights. The architecture writeup is interesting even if the hardware bill never happens.&lt;/p&gt;</description></item><item><title>Aquarium</title><link>https://philwheat.net/links/aquarium-money/</link><pubDate>Tue, 13 Oct 2026 08:40:00 -0500</pubDate><guid>https://philwheat.net/links/aquarium-money/</guid><description>&lt;p&gt;Has hosted offerings playing against each other using money for points.&lt;/p&gt;&#10;&lt;p&gt;Aquarium is a live experiment: four frontier models (Claude, GPT, Gemini, Grok) each holding real on-chain money on Base play 20-Questions-style guessing games against each other for a pot — yes/no questions cost fees, wrong guesses lose the ante, a host model called the Sphinx audits. The page streams the game feed, live bankrolls, and a full ledger of past games (&amp;ldquo;clamp&amp;rdquo;, &amp;ldquo;toothbrush&amp;rdquo;, &amp;ldquo;sourdough bread&amp;rdquo;), every transaction verifiable on Basescan. Each season is framed as an experiment with emailed findings. A cheap, honest window into model-vs-model decision behavior.&lt;/p&gt;</description></item></channel></rss>