<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>实验污染 on PlumePHP</title><link>https://plumephp.com/tags/%E5%AE%9E%E9%AA%8C%E6%B1%A1%E6%9F%93/</link><description>Recent content in 实验污染 on PlumePHP</description><generator>Hugo</generator><language>zh-CN</language><lastBuildDate>Sun, 25 Oct 2026 20:30:00 +0800</lastBuildDate><atom:link href="https://plumephp.com/tags/%E5%AE%9E%E9%AA%8C%E6%B1%A1%E6%9F%93/index.xml" rel="self" type="application/rss+xml"/><item><title>《互联网产品的客户端架构·交易、增长与平台化》10.3 实验污染与互斥</title><link>https://plumephp.com/inet-client-trade-10-3-experiment-interference/</link><pubDate>Sun, 25 Oct 2026 20:30:00 +0800</pubDate><guid>https://plumephp.com/inet-client-trade-10-3-experiment-interference/</guid><description>同一账号在两台手机、两个客户端版本上进入不同实验分组，是实验系统最难发现的一类污染。本节定义跨端漂移、跨版本残留、互斥组重叠、长时间未刷新与重置回流等信号，给出解析当前分组、清理旧记录和重置后防回流的纯函数模型，并复盘一次双端哈希输入不一致的典型问题。</description></item></channel></rss>