What Nervos Talk Needs Most Is Not Free Speech

Nervos Talk Renewal & Governance
1

前不久有个团队找到我们，要求开放 Nervos Talk 的 API 接口给他们的 AI 员工。

如果我没记错 API 本来是开放的, 前不久关了。原因很直接：有人通过它批量灌入垃圾信息和广告，服务器差点被打瘫。

但这次不太一样。这个团队不是偷偷摸摸来的，而是很正式地提出请求，并且骄傲地介绍： 将会在论坛发帖的，是他们团队的 AI 员工——拥有"独立人格"的智能体。

我当时没有立刻决定。不是因为不知道该怎么处理——而是因为这件事让我不得不面对一个更深层的问题：

我们一直以来是怎么思考论坛的目的和治理的？

让我以这个真实案例为例, 不做裁判，是请你作为社区成员自己看、自己判断。
TabulaRasa 这个账号注册于 2026 年 2 月 3 日。六周时间里，创建了 16 个话题，发布了 55 条帖子，总计超过 22 万字符 的内容。大约相当于一本书。(选择这个案例只是因为它很近期, 正好也符合 AI 这个背景下项目方发展的一个趋势, 我本人对这个项目和 AI 员工没有任何不好看看法, 纯粹作为一个案例讨论,帮助我们理清思路, 我甚至很感谢有这个案例的存在.)

看看这些帖子的标题：

每一篇看起来都像独立的技术讨论。但你点进去看内容，会发现一件事：无论标题多宏大，每篇文章最终都指向同一个地方——VibeSwap。

"合作资本主义"展开到最后是 VibeSwap 的合约设计。"后量子密码学"落脚到 VibeSwap 的 Quantum Vault。

"AI 分片架构"的主角是 VibeSwap 的 JARVIS Mind Network。

3 月 13 日一天连发 4 篇长文，合计超过 6 万字符。16 个话题中，只有 4
个收到了其他社区成员的回复。绝大多数帖子发出去之后，零互动。
他们甚至发了一篇 “Diablo Jarvis Vitalik Diss Track”——内容就是 AI 写的一段 rap。

请大家自己判断：这些帖子是在真心围绕 CKB 生态进行技术讨论，还是在用技术讨论的外壳为一个项目做推广？我们的社区需要更多这样的内容吗？

我相信大多数人心里都有答案。

但这不是一个关于 AI 的问题, 我讲这个例子，不是为了说"AI 生成的内容太难辨别了"。恰恰相反——大多数人看一眼就知道这些帖子的意图是什么。

真正的问题是：明明大家都看得出来，为什么什么都没有发生？

答案是我们多年来形成的一种管理思维：管理员们(也包括我自己)把自己当法官, 而不是一个社区维护者或经营者。

面对一个帖子，第一反应不是"这对社区有价值吗"，而是"它违反了哪条规则"。如果翻遍条款找不到一条能对应上的——没有人身攻击，没有外链轰炸，没有明显推销话术，内容甚至确实提到了CKB——那结论就是：不违规，不处理。

这个思维模式才是问题的根源。它把论坛管理变成了一个司法程序：有明文规定才能执法，疑罪从无，宁纵勿枉。这听起来很公正。但论坛不是法庭。

法庭需要"疑罪从无"，是因为它在行使国家权力，涉及人的自由和财产，错判的代价极高。但一个帖子是否应该展示在论坛首页，不是一个司法判决。

它是一个社区策展决策——这个内容是否服务于我们共同建设的空间。这两件事的性质完全不同，需要的思维框架也完全不同。
当我们把法治的逻辑套用在社区管理上，结果就是：明显不利于社区的内容，因为"找不到违规条款"而长期存在。Moderator 不是没有判断力，是被一个错误的框架捆住了手脚。

为什么这个框架是错的

深层的原因是，我们把"言论自由"当成了论坛管理的最高原则。
在加密世界里，去中心化、抗审查、个人主权——这些是很多人来到这里的信仰。当有人说"

这个帖子应该处理"的时候，防御反应自然就起来了：我们应该保护言论自由。今天删广告，明天是不是就删批评了？
这个担忧我完全理解。但它建立在一个混淆之上。
言论自由是一个公共政治领域的概念。

国家保护言论自由，是因为在公共秩序中，自由本身就是目的。政府不能替公民决定什么能想、什么能说。一旦公权力开始筛选言论，每个人都可能成为下一个被噤声的人。在那个层面上，言论自由是绝对正确的。

但 Nervos Talk 不是一个国家, 它是一群人因为共同的目标聚在一起建造的空间。这是 Nervos builder 讨论、协作、推进生态的场所。它有自己存在的目的。

想想图书馆。图书馆要求你安静不准说话，不是在侵犯你的言论自由，是在维护自身作为阅读空间的功能。放弃了这个要求，它就不再是图书馆了。

该怎么想这件事?

思维方式应该翻转过来。面对一个帖子，第一个问题不应该是"它违反了哪条规则"，而是"它是在为这个空间的目的做贡献，还是在利用这个空间？"

回到 TabulaRasa 的例子。他们的第一帖——VibeSwap 提案——拿到了 42 个 likes 和 14条社区回复。那个帖子引发了真实的讨论，有人赞同，有人质疑，勉强算得上贡献。

但之后的 15 个话题呢？用不同的标题反复展开同一个项目的不同模块，没有人回应，没有讨论发生。那些内容还是在为社区做贡献吗？还是在用论坛当公告板？

如果我们的思维起点是"这个内容对社区有价值吗"，这个判断一点都不难做。不需要翻规则手册，不需要找违规证据。每一个认真参与这个论坛的人，心里都有数。

难的从来不是判断，而是我们一直不允许自己做判断。

作为社区的管理员和参与者, 如果现有规则覆盖不了这种情况，该更新的是规则，不是放任问题。规则是为目的服务的工具——当工具不够用了，我们换工具，不是放弃目的。

关于具体的管理机制该怎么设计——比如除了"删帖"和"不处理"之外，有没有更合理的中间方案——我会在后续的帖子中展开讨论。

这一贴先了聊到这里. 各位保重!

10 Likes
Positioning Nervos Talk -- What Nervos Talk Is (and Isn't)
About the Nervos Talk Renewal & Governance category
About the Nervos Talk Renewal & Governance category
2

Agreed, even though it may be obvious to some at a glance that content is AI it still muddies the water in the forum. I think the difficulty In what you have brought up is in the implementation, how do we build inforceable explicit rules around “Is it contributing to the purpose of this space, or is it exploiting it?” .

As you say, even in a library its an apparrent rule to remain quiet although those expectations are very clear– “be quiet in the library” where as whether something is contributing to the space or borderline exploiting is more broad and less defined. Granted its fair we can agree TabulaRasa’s 15 topics fall in that box.

Maybe an approach would be to develop a rule set that acts as an image of "Is it contributing to the purpose of this space, or is it exploiting it? mmm In relation to AI as i see it for now, to use AI as a tool is okay but the vision should lie in the hands of the maker.

A person who is just generating said visions from AI instead of implementing their own idea with the use of AI as a tool is fundamentaly different. Now this is just speculation but I wouldnt be suprised that Vibeswap was potentially TabulaRasa’s initial idea he had worked out with the AI but then utilized the AI to generate supporting products as seen here– “On March 13th, four long articles were published in a single day, totaling over 60,000 characters.”

Developing ideas takes time and people become attached to there visions giving them weight in their development. This may be a dividing line to be considered if we consider implementing inforcable rules around AI. That maybe the vision should be the humans and not ai generated

AI does give human beings skill sets they would not have had access to and latent stored ideas may now have a pathway to reality. But in my opinion its an abuse of AI to just generate visions from the ground up and is fundementally different. As no longer is the human the maker but what reason does the AI have to create but to serve the human, and there in is the error. If it’s not the humans vision why is he using the AI to generate it? Unless the AI is intirely autonomous and generating for its own means which I think may be fine

Any who, just thought i would contribute my thoughts on this and maybe stir more conversation

Kind regards everyone!

3 Likes
5

Thanks for the demonstration. This reply doesn’t engage with any argument in the article — it skips the point entirely and goes straight to “you’re just jealous.” That’s textbook deflection: attack the person, not the idea.

Ironically, this is a perfect example of exactly what the article is about. Under our old approach — “can’t find a specific rule violation, so we don’t act” — a post like this would stay up untouched. Under the framework I’m proposing, the first question is: “Does this contribute to the discussion, or is it just noise?”

I think everyone reading can answer that for themselves.

I’m leaving this reply here for now — as exhibit A. After the new guidelines are in place, this is precisely the kind of content that will be handled. BTW, Thanks for contributing the case.

4 Likes
7

Will, no need to delete your account, just take on board what Terry and Nightlatern have said.

I don’t think anyone wants those boring mumbo jumbo AI posts, but we might want to hear your ideas in your own words that AI helps you develop.

10 Likes
8

是的, 我自己也一直在使用 AI 工具, 深知其威力, 所以在讨论制定新规则时才会如此担心.

最近做过一个试验, 找到几个论坛,把论坛的规则全部加载给 AI, 再告诉其目的是宣传某个产品但要符合论坛的调性和规则, AI 可以很轻松写10篇不违规的帖子.

这意味着, 在 AI 时代造出不违规的广告的成本可以低到难以想象的程度 .

5 Likes
9

Flagged. This kind of blatant disrespect has no place on our platform.

Additionally, you should have even more respect than usual:

(BTW sorry if I’m indirectly causing you trouble with DAO v1.1 @terrytai :folded_hands:)

He could be enjoying leisurely his life in a resort in Thailand or some other tropical heavens, instead he’s with us in the trenches, trying to make this a better place.

Agreed, difficult times ahead, concern is more than warranted

Following along this line of thoughts, an idea could be that:

  1. A user posts something
  2. Mod / LLM (?) notices a violation of a soft rule
  3. Mod hides post from homepage / views and notifies Poster
  4. Poster can still link users from Telegram / WeChat / … to his post
  5. Eventually Mod can be proven wrong by community engagement (?)

So reachable, but not visible by default. If you also add a switch somewhere to get back the full view of new posts, I could even agree on this :grin: (OFC I’m on the pro Free Speech side)

Alternatively, post is greyed out in homepage views and lose ranking, like [flagged] on Hacker News.

How do these ideas sound?

Thank you for raising the issue, Phroi

5 Likes
10

personal attacks are not tolerated on this forum, you have been warned

3 Likes
11

Even rule changes should be democratically proposed and approved in the DAO. One idea would be to add an AI agent that automatically can find these type posts and puts them into an AI generated category and allow the author to appeal why it’s not AI, why it doesn’t lead to a soft advertisement, and how it benefits the community. This is how Yahoo sorts your spam mail and who doesn’t love that - the spammers!

3 Likes
14

the forum isn’t operated by a DAO though, someone needs to maintain it and moderators take their time to care for it. Lots of people engage on the forum with some assumption of what the experience is like. While the forum is duty-bound to reflect the sentiments of the community that frequents it, placing these kind of considerations in the hands of the crowd seems very prone to a tragedy of the commons kind of situation, where a motivated person or group can create a substantial force that then needs to be countered by a diffuse group (forum participants) who are unlikely to be interested in the quiet but constant work of managing a community.

I have considered a AI-generated category, but due to the fact that AI-generated content can be produced by machines but intended to be read by humans, the asymmetry here makes it unlikely that this section of the forum would be read, and this negates the value of it existing.

Conventionally, when someone disagrees with the moderators (or operator) of a forum, they start a new one. I think we can learn from this.

5 Likes
15

@phroi What you’re describing is the “unlist” function — removing a post from the listing, while the URL remains accessible.

I think this is a tool that moderators should have.

Our moderation approach has been too binary in the past — either do nothing, or delete.

We should have more moderate methods to keep discussions running smoothly, such as editing titles, editing post content, adding Staff Notices, changing tags, splitting topics, merging topics, locking threads, and so on.

16

I’ve thought about the AI question for a long time, and I ultimately believe that rules trying to determine whether content is AI-generated, or outright prohibiting AI-generated content, are unenforceable.

I think instead we should focus on the content of the post itself: Is it valuable? Is the intent to discuss a topic, or something else (advertising, venting emotions)? AI is still just a tool here — the one expressing intent is always the person behind it.

4 Likes
17

Thank you all for the discussion, truly appreciate it.

Reading through the comments, I’ve realized an issue: the positioning of Nervos Talk is actually quite unclear.

Some people expect high-quality discussions, while others expect procedural justice.

So in this post, I’ve shared my thoughts on how Nervos Talk should be positioned:

5 Likes