X 永久封禁了我分享自己开源库的 15 年账号。
8 分•作者: rifmj•3 个月前
我于7月8日发布了一篇关于我编写的开源库 helix-noise 的帖子——这是一个无散度的3D噪声函数，它返回的是速度而不是值，因此你可以对数百万个粒子进行平流，而无需模拟步骤（MIT协议，演示：https://rifmj.github.io/helix-noise）。
帖子本身没问题。然后我回复了它，附上了GitHub链接和三个标签（#creativecoding #threejs #webgl）。不久之后，我的账号（@dzhumagulov，2011年注册，此前无违规记录）因违反X规则而被永久封禁——被设置为只读模式，无法发帖、点赞或创建新账号。
我通过官方表格进行了申诉：解释说这是我自己的MIT许可项目，没有商业化，只发过一次帖子。但他们的自动化系统发回了拒绝通知，模板中甚至没有提及具体违反了哪条规则——“具体来说：”部分是空的。它只说决定不变，我应该“纠正违规行为”，但没有说明是什么违规行为。
所以目前的情况是：被一个自动化系统封禁，申诉被另一个自动化系统拒绝，而且没有人告诉我我打破了哪条规则。据我所知，“在回复中包含外部链接+标签”似乎被模式匹配为垃圾信息。
我理解在X的规模下进行垃圾信息检测很困难。但分享自己的开源作品，可以说是最正常的开发者行为之一了，而且似乎没有途径可以联系到人工审核员。我很想知道其他人是否遇到过这种情况，以及除了内容病毒式传播之外，是否有什么方法可以解决。
查看原文
On July 8 I posted about an open-source library I wrote — helix-noise, a divergence-free 3D noise function that returns velocity instead of a value, so you can advect millions of particles with no simulation step (MIT, demos: https://rifmj.github.io/helix-noise).<p>The post itself was fine. Then I replied to it with the GitHub link and three hashtags (#creativecoding #threejs #webgl). Shortly after, my account (@dzhumagulov, registered 2011, no prior violations) was permanently suspended for violating the X Rules — moved to read-only, can't post, like, or create new accounts.<p>I appealed through the official form: explained it's my own MIT-licensed project, no monetization, posted once. The rejection came back from their automated system, and the template literally didn't name the violation — the "specifically:" section was empty. It just says the decision stands and I should "remedy the violations," without saying what they are.<p>So the current state: banned by one automated system, appeal denied by another, and no human has told me what rule I broke. As far as I can tell, "external link + hashtags in a reply" simply pattern-matches to spam.<p>I get that spam detection at X's scale is hard. But sharing your own open-source work is about the most normal developer behavior there is, and there's apparently no path to reach a human reviewer. Curious if others have hit this and whether anything short of going viral actually works.