✦   ✦   ✦

breaking news

News, read through The Primal Race
← 全部评论 · all commentary

把尺子换成AI,能力没变New Ruler, Same Capability

科技 结构层 · 文化层 404 Media ↗ 2026-10-31 § 链接
新闻摘要Proton CEO Andy Yen 在内部强力推动 AI,称质疑者为「loser thinking」,并设置邮箱让员工举报阻碍 AI 的人。公司被曝将重新面试员工以决定去留,引发内部恐惧。Proton 发言人未否认报道,承认部分员工感到焦虑。
The newsProton CEO Andy Yen is aggressively pushing AI internally, calling critics 'loser thinking' and setting up an email for employees to report AI obstacles. The company reportedly will re-interview staff to decide who stays, causing fear. A Proton spokesperson did not dispute the reporting and acknowledged some employees are anxious.
换尺子不换能力,暴力还在原地
Swap the ruler, keep the capability: the violence stays put.

Proton 的 CEO Andy Yen 在内部 Slack 里骂一个要求先测几天代码再上线的安全工程师:“那是狗屁,是失败者的思维。当然可以又快又安全。”这句话本身就是全部的证据:一个做加密邮箱的公司,把“先测再上线”称作 loser thinking。

先看清这里换的是什么。不是能力,是评价方式。快和安全的取舍是真实存在的,pentest 需要时间,这是生物墙一样的东西,不是态度问题。Yen 的解法是宣布这道墙不存在,然后把说墙存在的人改叫失败者。这和芯片做不出来就宣布一条新定律是同一个动作:现实不配合,就重新定义现实。区别只在于,τ law 至少还印在 PPT 上,这里连 PPT 都省了,直接是“only losers think like that”。

暴力三角在这里算得很清楚。Potential 是“又快又安全”的产品,Actual 是跳过测试直接上线的代码,差额落在谁身上?落在用户身上,落在那些以为 Proton Mail 的卖点是“谨慎保护隐私”的人身上。这个差额不流血,所以不会被认作暴力,它是结构层的:一个安全产品的风险被系统性地转移给了不在场的人。而让这个转移合法化的,是文化层的一句话——“快就是强,慢就是弱”。

这句话不是中性的管理风格。它是男性中心叙事最标准的模板:把谨慎、细致、手工打磨标记为 feminine 的、可耻的、需要被 smash 的东西,把速度、粗暴、ruthless 标记为强。Yen 说要“ruthlessly smash everything that stands in the way of AI velocity”,smash 的对象是谁?是那个在 Slack 里小心翼翼提建议的工程师。元暴力在这里不需要论证,它自己报了名。

再看共谋的结构。公司装了软件统计员工生成了多少 AI 代码,设了一个邮箱专门收集“阻碍 AI 采用”的人和事,宣布今年是最后一年招工程师,以后只招会用 AI 造东西的人,然后重新面试员工,看他们够不够拥抱 AI。这一整套下来,还有哪个员工会举手说“这段代码需要测三天”?留下来的人不是被说服了,是被筛选了。报道里说员工“hesitant to even mildly critique Yen’s views”,这不是文化,这是筛选机制的成品。

有意思的是,这套“velocity”叙事自己漏了底。内部文件说 Proton 因为“insufficient velocity”达不到年度经常性收入目标,同比增长“远低于市场认为的好”,按所有重要指标“我们是一家表现不佳的公司”。也就是说,AI psychosis 的燃料不是技术突破,是收入焦虑。一个前 CERN 粒子物理学家,用“失败者思维”这种词去堵住安全工程师的嘴,为的是财报。

Proton 的发言人说“我们有时太直白,或者解释得太慢”,没有否认任何一条报道。这句话翻译过来是:我们知道自己在干什么,只是没想到会被写出来。

还有一件事报道没写:那个在 Slack 里被公开骂成 loser 的工程师,后来怎么样了。报道里没有名字,没有后续,没有他是否还在 Proton。这个缺席本身就是材料——在这套叙事里,提出风险的人不配有名字,只配有一个标签。

“又快又安全”这句话里,快是真的,安全是承诺。承诺不需要 pentest,只需要一个邮箱地址,专门用来 smash 提出问题的人。

Andy Yen, CEO of Proton, told an employee in the company's security Slack channel that wanting a few days to pentest new code before pushing it live was "bullshit and loser thinking." "Of course it's possible to be fast and secure," he wrote. That sentence is the whole story: an encrypted email company calling "test before shipping" loser thinking.

Look at what actually got swapped here. Not capability — the evaluation method. The tradeoff between speed and security is real; pentesting takes time. That is a biological-wall kind of fact, not an attitude problem. Yen's solution is to declare the wall doesn't exist, then relabel everyone who says it does as a loser. This is the same move as announcing a new law of physics when you can't shrink the chip: reality won't cooperate, so redefine reality. The only difference is that the τ law at least made it onto a slide. Here it's just "only losers think like that."

Run the Violence Triangle. Potential: a product that is both fast and secure. Actual: code shipped without testing. Who eats the difference? Users — the people who believed Proton Mail's selling point was careful privacy protection. The gap doesn't bleed, so it won't be recognized as violence. It's structural: risk from a security product is systematically transferred to people who aren't in the room. What legalizes the transfer is a cultural sentence — fast is strong, slow is weak.

That sentence is not a neutral management style. It is the standard template of the masculine-centric narrative: mark caution, care, and handcrafting as feminine and shameful, as something to be smashed; mark speed, bluntness, and ruthlessness as strength. Yen wants to "ruthlessly smash everything that stands in the way of AI velocity." Smash whom? The engineer who raised a mild objection in Slack. Meta-violence doesn't need an argument here. It signed its name.

Now the co-conspirator structure. The company installed software tracking how much AI code workers generate. It set up an email address for reporting anything blocking AI adoption, so leadership can smash it. It announced this is the last year it will hire engineers at all. Then it re-interviewed staff about their AI use, with the clear implication that the interviews decide who gets fired. After that, which employee raises a hand to say this code needs three days of testing? The people who stay weren't persuaded. They were filtered. The report says employees are hesitant to even mildly critique Yen's views. That isn't culture. That's the finished product of a filter.

The velocity narrative also leaks its own fuel. An internal document says Proton is behind its annual recurring revenue target due to "insufficient velocity," that year-over-year growth is far from what the market considers good, and that by all metrics that matter, "we're an underperforming company." So the AI psychosis runs on revenue anxiety, not technical breakthrough. A former CERN particle physicist calls a security engineer a loser so the numbers look better.

Proton's spokesperson did not dispute any of the reporting. They said the company has sometimes been "too blunt or too slow in explaining ourselves." Translated: we knew what we were doing; we just didn't expect it written down.

One thing the report doesn't say: what happened to the engineer called a loser in front of everyone. No name, no follow-up, no word on whether he still works there. That absence is material. In this narrative, the person who names the risk doesn't get a name. He gets a label.

In "fast and secure," fast is real and secure is a promise. Promises don't need pentests. They need an inbox for smashing the people who ask questions.