确实值得思考。AI 貌似暂时无法拉动规模需求；油价的持续低位侧面印证了暂时没有补库存的动能，也就无法规模化地拉动信用扩张。存量博弈背景下就是绞肉机行情.
需要思考几个问题。一是全球性的金融投机狂热验证了当前生产力已经很难产生增量，AI只是加速了分配效率，也就是加快财富集中的速度；二是大宗集体大跌不是简单的风险出清，也许预示着底层经济的大衰退，尤其是石油和白银大跌需要警惕，因为第三世界国家好像无法乘着全球化的风起飞了，需求增长会被严重抑制，也就是说如果没富的国家和个人，那也就这样了；三是美股不跌正好说明了全球资本的加速集中，美股是最后的避风港，一旦美股风吹草动，画面太美不可想象；最后，假设美股开始大跌，币圈的底在哪里？
38
结合 <晚点>采访中MiniMax 闫俊杰 关于 DAU 的思考.更好的应用和更多用户，并不会导向更好的模型. 用户规模带来的飞轮效应是惯性的假设. 当然,阿里需要的从来是更多的用户而不是更好的模型.那就和字节拼刺刀吧,这很阿里🤓
Qwen 技术负责人 Junyang Lin 今天在 X 上发了一条简短的告别："me stepping down. bye my beloved qwen."（我退出了。再见，我挚爱的 Qwen。）
以下内容由 Grok 和 Claude 辅助整理：
【1】事情来得很突然
就在昨天，Qwen 团队刚发布了 Qwen3.5 小模型系列，0.8B 到 9B 四个版本，支持视觉理解和推理切换，本地跑只要 7GB 内存，性能吊打好几倍大的模型。社区一片叫好。
结果今天一早，技术负责人就宣布走人了。
更让人意外的是，这不是他自己想走的。Qwen 贡献者 Chen Cheng 第一时间回应：
"leaving wasn't your choice"（离开不是你的选择）
她说昨晚还跟 Junyang 一起发布 Qwen3.5 小模型，无法想象没有他的 Qwen 会是什么样。
【2】不只走了一个人
Junyang Lin 不是唯一离开的。Hyperbolic Labs CTO Yuchen Jin 确认，Qwen 另一位核心负责人 Binyuan Hui 也走了。这两位是那种北京时间凌晨 6 点还在协作上线模型的人。
随后，Qwen3.5、Qwen VL、Qwen Coder 的核心贡献者 Kaixin Li 也发了告别帖：
"Signing off from Alibaba Qwen. Grateful for the chance... Onwards and upwards!"
目前确认离开的至少 3 到 4 人，全是核心骨干。
【3】到底发生了什么？
从各方回应拼出来的图大致是这样的：
阿里云对 Qwen 团队的考核方式变了。Xinyu Yang 批评说，阿里云拿日活（DAU）这类消费级应用的指标来衡量基础模型团队，还用一位 Google Gemini 的非核心成员来替换 Junyang 的位置。一个做基础模型研究的团队，被当成做 App 的来管，创新自然就停了。
You Jiacheng 说得更直接：
"Alibaba-Cloud kicked out Qwen's tech lead"（阿里云把 Qwen 的技术负责人赶走了）
他提到接任者可能是 Hao Zhou，背景偏强化学习方向，社区对他能否带好 Qwen 持怀疑态度。
还有人透露，新领导是绕过 Junyang 直接管理团队的，等于是架空。多位知情人士都指向同一个词：内部政治。
【4】Qwen3 可能就是绝唱了
Panda（Jiaxi Cui）的判断很悲观：Junyang 走后，Qwen3 系列可能是很长一段时间内的最后杰作，团队接下来会转向商业化。
这个判断不是没有道理。Qwen 虽然技术上很强，在 Hugging Face 上下载量巨大，学术界和开发者都在用，但跟字节的豆包比，在国内消费市场的存在感确实弱。阿里云高层可能就是因为这一点，决定换方向换人。
Yuchen Jin 担心的是，Qwen 以后可能不再交付前沿开源模型，转向封闭商业化路线。
【5】开源社区的反应
X 上的讨论几乎一边倒。
Nathan Lambert 称之为"legendary run"（传奇之旅）。Unsloth AI 感谢 Qwen 对开源的贡献。Martin Ziqiao Ma 说 Qwen 一直是前沿实验室和外部社区之间的桥梁，这个桥可能要断了。
批评声音也很多。Maziyar PANAHI 说希望他们出去建个新实验室，然后被阿里巴巴花 20 亿美元收购回来长个教训。Hugo Pinheiro 问了一个很多人想问的问题：为什么商业决策者总是在事情做得好的时候把它毁掉？
社区里流传最广的一句话是："Qwen is nothing without its people."（Qwen 离开了做事的人，什么都不是。）
【6】他们会去哪？
目前没有确切消息。社区猜测的方向包括：自己创办新实验室、加入 DeepSeek 这类开源友好的团队，甚至有人喊话马斯克的 xAI 和 Hugging Face 赶紧把人抢了。
不管去哪，这些人的能力摆在那里。Qwen 从小模型到大模型、从文本到多模态，能做出这样的成绩，靠的就是这批人。
【最后】
这件事是大公司里开源理想和商业化诉求的老矛盾。Qwen 团队用开源模型攒下了巨大的行业影响力和开发者口碑，但在公司内部的评价体系里，这些可能都不算 KPI。
110
确实.学历一般只在毕业求职前 5 年有用.如果毕业 5-10 年求职的亮点还是"清北",只能说明这 5-10 年白活了.而企业,特别是创业公司,薪资/待遇/福利/稳定性都比不上大厂,还不花时间筛选,扩大筛选范围的话那就不是懒了,是蠢.
今日好文分享：学历歧视成就了Web3
其实我之前创业在招聘的时候是有这方面偏见的，原因很简单：学历虽然不代表一个人的全部能力，但毕竟是一个人适应或学习能力的下限。
学历好的人不一定就真的有能力处理工作中的挑战，但学历不好的人大概率处理的更差…
所以Web3对学历的要求并不是出于歧视， 而源自最基础的筛选…
对于有过CEX经验的人来说，相当于已经有人替你做过这层筛选了，因此才会更容易获得HR的青睐。
毕竟学历和经验，对于大公司来说，有了这些代表你可以快速学习适应，并融入其中成为一个螺丝钉…
但对于创业公司来说，我更倾向于看结果，你有什么样的成果和成就，远比学历来的重要，所以我也雇佣过中专学历的程序员与设计师，后续从结果来讲，效果并不差。
57
我猜她是想说明友商的“断指计划”, 问题是断指之所以能成功是拳头握的不够紧啊.老大哥关心的不应该是蜜糖能不能发胖而是小兄弟冷不冷,吃得够不够饱不是吗😂
真要给我笑死了，也不知道是你们没打过工还是官老爷做久了。
试想一下，假如你是员工，你拿到了更高薪酬的offer，你会把这个offer提交给HR说你看我有这个offer你给我涨工资吧。你会这么干吗？？？
要知道拿到offer只是一个结果，人家拿到offer说明几个月前就已经想离职了并且就已经有动作了。应该搞清楚别人为什么想走，难道就是因为多拿那几块钱工资吗？
这就是典型的管理者的自嗨，压根不去做正确的事，正确的事是什么？是在公司内部主动发现人才，并且给这些人才合理的回报降低他们离职的意愿，用一整套体系留住他们。为什么不去做呢？因为难啊。
出个这种看似合理实则屁用没有的政策多简单啊，怪罪友商多简单啊，PUA员工多简单啊。
73
以其去预测山寨季什么时候来，不如去重新定义下山寨。
在加密货币市场标的物指数级扩容的确定性趋势下，如果山寨季的定义还是普涨行情。简单将现存山寨币市值乘以2倍进行推演（粗略估算），当前市场是否具备支撑如此庞大体量的增量流动性？直白点说，钱从哪来？
#ALTSEASON
3
161
YAN retweeted
Mac Studio 发布时就预计到这一天的到来，即 Undercut 这个对苹果设计具有象征意义但费钱的工艺总将因地位和权力的变化而被干掉。当然凭着苹果的设计能力——就像 Mac Studio 一样——这种转变看上去自然流畅且合情合理，但是 Mac mini 因 Undercut、T 型刀作为 Unibody 设计的代言人自此换面，不由伤感。
7
14
81
22,044
YAN retweeted
GO DIRECT: THE MANIFESTO
I. TRADITIONAL PR IS DEAD.
For too long, founders have yielded control over their narratives to media and middlemen.
Before the internet, it was by necessity. The way to reach large audiences was through the media, and the way to get media coverage was through professional publicists.
Today, most of the planet is directly reachable by social media or email. There’s no longer a need to go through traditional gatekeepers of information and brokers of reputation — especially as their own credibility has plummeted.
The old PR playbook of relying on third parties with misaligned interests is obsolete.
But while the world has changed, comms norms have not. Still encased in amber are the old habits: prioritizing media over social media, fishing for clicks instead of fostering communities, and avoiding risk by recycling worn-out tactics.
“Corporate communications” itself is now an oxymoron, as nothing meaningful can be communicated by a faceless committee. If press releases read like they were written by a baker’s dozen of middle managers, that’s because they were. Their only discernible purpose seems to be to avoid upsetting anyone and jeopardizing the future job prospects of those middle managers.
The resulting stories are bland and generic, with passion reduced to pablum. Traditional comms is an anachronism.
II. COMMUNICATION IS THE FOUNDER’S JOB.
For a decade, we’ve been told that tech founders are cartoon villains, venture-funded startups are grifts, and new technologies will destroy us all.
Maybe there was a time when founders could just focus on building — they were seen by the media establishment as a curiosity, not a threat to the natural hierarchy who needed to be put in their place. But if that time ever existed, it is now long gone.
You may not be interested in The Discourse, but it is interested in you. And if you bow out, you are forfeiting your license to build a movement and thus build a company.
Building a movement is hard, but it must be done, and it must be done by founders. A founder’s passion, vision, and conviction can’t be simulated by others — least of all the press-release-enjoying middle managers already scouting for their next jobs.
The best spokesperson for any endeavor is not the one who has the most polish, the longest tenure, or the “right” credentials. It’s the person who holds the secret knowledge upon which the enterprise is built, the person who can not only describe the idea but, in the face of inevitable opposition, fight for it and win.
Founders need to take their narrative as seriously as they take the rockets or robots. They would never outsource their product — and when it comes to convincing others to support the mission, the story is the product. Outsourcing comms is as bad as outsourcing code.
As evangelists, founders are irreplaceable.
III. GO DIRECT OR GO HOME.
Going direct to the people who matter is how founders retain control over their narratives and preserve their companies’ uniqueness. Those who are stubborn, unorthodox, and disagreeable should never have their edges filed down for fear of offending entrenched interests.
But going direct doesn’t mean going it alone. It doesn’t mean refusing help or spurning others who can amplify your message. And it certainly doesn’t mean just poasting more.
Going direct means crafting and telling your own story, without being dependent on intermediaries.
Just as founders might have more natural talents at product, management, or engineering, some founders will be naturals at communicating while others have a harder time.
The good news is that going direct and building a movement, while not easy, are skills that can be developed with discipline and time. The bad news is that, unlike with engineering or management, communications failures are immediately public and personally humiliating. It’s not surprising that many are loathe to take on this responsibility.
At the same time, founders willing to pick up that gauntlet will find that it gives them a massive edge in recruiting, fundraising, selling, and shaping the information environment needed for their companies to thrive.
IV. IT’S TIME TO REBUILD THE ROSTRA.
At the center of Rome, as it transitioned from a Republic to an Empire, stood a speaker’s platform from which the city’s leaders would address the public directly.
It was called the Rostra, so named because it stood atop the captured battle rams (or rostrums) of enemy warships. From here, speeches were given that would sway opinion, change regimes, and alter history.
That physical structure has been lost to time, but we now have something much more powerful: a free and open internet with which to build a speaker’s platform of limitless scale. All we need is the will to build it.
The conventional way of communicating has its allure. Outsource your message, let some removed third party go through the motions of getting “impressions,” and spare yourself the risks and discomfort that come with putting your own name on the line.
But that way is incompatible with greatness.
Reject convention — build your own platform, build your own audience, and build your own narrative.
Go direct.
465
1,127
7,249
22,226,564