Why is Hacker News like that?
Why is Hacker News like that?
为什么 Hacker News 会变成这样？
Hacker News (aka HN) is a link aggregator where “hackers” gather to discuss technology, politics, and anything which “good hackers would find interesting”. HN is a means by which its host, prominent startup incubator Y Combinator, projects soft power, promotes startups it funds, and feeds people with ideas into its incubation program. Hacker News is one of the most popular forums for discussing technology online today – perhaps the biggest.
Hacker News（简称 HN）是一个链接聚合网站，在这里，“黑客”们聚集在一起讨论技术、政治以及任何“优秀的黑客会感兴趣”的话题。HN 是其主办方——知名创业孵化器 Y Combinator——展示软实力、推广其投资的初创公司，并为孵化项目输送人才的渠道。如今，Hacker News 是在线讨论技术最热门的论坛之一，甚至可能是规模最大的一个。
HN also has a pronounced political bias, which is growing more exaggerated with time. Many of its users have long claimed centrist or “apolitical” views. On a forum created by a startup incubator, it’s no surprise that capitalism is generally taken as given, often with a libertarian angle. Labor is generally de-emphasized (the consensus on HN is almost always anti-union) and openly advocating left economics like communism or socialism is generally frowned upon. Progressives have, on the whole, abandoned the forum.
HN 也有着明显的政治偏见，且这种偏见随着时间的推移愈发夸张。许多用户长期以来声称自己持中间派或“非政治”立场。在一个由创业孵化器创建的论坛上，资本主义被视为理所当然（通常带有自由意志主义色彩）并不令人意外。劳工权益通常被淡化（HN 上的共识几乎总是反工会的），而公开倡导共产主义或社会主义等左翼经济学观点通常会受到排斥。进步派人士总体上已经放弃了这个论坛。
The right wing has not. Yesterday, I saw a post about the new release of the Grok LLM, Elon Musk’s “anti-woke” AI project. Virtually all of the comments are fawning over the release, and debating the finer points of its technical achievements, price, and competitiveness. There are remarks that object to the tool on the basis that its owner, Elon Musk, is a fascist, and that it is designed to amplify hate speech. Almost all of these remarks are “flagged”, moderated by other HN users until they are not visible by default.
右翼则没有。昨天，我看到一篇关于埃隆·马斯克（Elon Musk）的“反觉醒”（anti-woke）AI 项目——Grok 大语言模型发布的消息。几乎所有的评论都在对该发布大加赞赏，并讨论其技术成就、价格和竞争力的细节。也有一些评论反对该工具，理由是其所有者埃隆·马斯克是法西斯主义者，且该工具旨在放大仇恨言论。几乎所有这些评论都被“标记”（flagged），由其他 HN 用户进行审核，直到它们在默认情况下不可见。
Moderation on HN
HN 上的审核机制
Hacker News has two formal moderators, both employed by Y Combinator: Daniel Gackle, aka dang, and Tom Howard, aka tomhow. Scott Bell (sctb) assisted with moderation until 2019. They employ a variety of technical tools to do the job, including:
- Boosting or sinking the rank of posts, or “burying” them entirely
- Deleting posts (entirely) or “killing” posts (marking them as [dead])
- Detaching and/or downranking comment trees
- “Shadowbanning” users
- Overtly banning users
Hacker News 有两名正式的版主，均受雇于 Y Combinator：Daniel Gackle（网名 dang）和 Tom Howard（网名 tomhow）。Scott Bell（sctb）曾协助审核工作直至 2019 年。他们使用多种技术工具来执行审核，包括：
- 提升或降低帖子的排名，或者将其彻底“埋没”
- 删除帖子（彻底删除）或“杀死”帖子（标记为 [dead]）
- 拆分和/或降低评论树的权重
- 对用户进行“影子封禁”（Shadowbanning）
- 公开封禁用户
Moderators occasionally post comments and intervene in discussions, steering things one way or another or explaining their moderation decisions. There are also automated moderation systems, which for example try to detect flamewars with heuristics like the ratios between post score, age, and the number of comments, and apply a downward pressure to the post’s ranking to reduce the visibility of controversial topics. Of course, the topics which are deemed controversial by the automated systems are, more often than not, progressive politics.
版主偶尔会发表评论并介入讨论，引导讨论方向或解释他们的审核决定。此外还有自动化审核系统，例如通过启发式算法（如帖子得分、发布时间和评论数量之间的比例）来检测“口水战”，并对帖子排名施加下行压力，以降低争议性话题的可见度。当然，被自动化系统判定为“争议性”的话题，往往多是进步派政治议题。
Users above a certain (small) amount of “karma” gain the ability to “flag” posts, allowing Hacker News to self-moderate. Posts or comments can be hidden entirely (made “dead”) with a small number of flags, particularly early on, before they gain enough votes to influence the flagging algorithm. This is the mechanism which was used to silence criticism of Grok and Elon Musk in the thread discussed earlier.
当用户的“声望值”（karma）达到一定（较小）数值后，便获得了“标记”帖子的权限，这使得 Hacker News 能够实现自我审核。只需少量标记，帖子或评论就可以被完全隐藏（变为“dead”），尤其是在帖子发布初期，尚未获得足够投票来影响标记算法时。这就是前文讨论的那个帖子中，用来压制对 Grok 和埃隆·马斯克批评意见的机制。
By and large the flagging mechanism is used to silence progressive opinions and articles from being posted on Hacker News. Once a post or comment is flagged, it can be “vouched” for, but this generally does not work in practice due to a relatively straightforward bias: flagged posts are only visible to users who opt-in, are moved to the bottom of threads and hidden from view, and one can flag at any time but cannot vouch for a post proactively. Once a comment or post is gone, it generally stays gone.
总的来说，标记机制被用来压制进步派观点和文章在 Hacker News 上的发布。一旦帖子或评论被标记，它可以通过“担保”（vouched）来恢复，但在实践中这通常不起作用，因为存在一个相当直接的偏见：被标记的帖子仅对选择查看的用户可见，会被移至讨论串底部并被隐藏，而且用户可以随时标记，却不能主动为帖子担保。一旦评论或帖子消失，它通常就再也回不来了。
The moderators tend to view the automated moderation and the flagging mechanism as neutral, and use it to deflect responsibility for moderation decisions onto the tools or the users who use them, i.e. the “community” in the abstract. However, ultimately the moderators are in control of these tools and are aware of their biases and the outcomes they produce.
版主倾向于将自动化审核和标记机制视为中立的，并利用这一点将审核决策的责任推卸给工具或使用这些工具的用户，即抽象意义上的“社区”。然而，归根结底，版主掌控着这些工具，并且清楚它们的偏见以及所产生的结果。