8:1 Krypton | After three years of service suspension, Tianya Community officially resumes access; Guangdong denies rumors that AI will be used to grade college entrance exams; MiniMax plans to list on the STAR Market 8点1氪丨停服三年后，天涯社区正式恢复访问；广东辟谣高考将用AI改卷；MiniMax拟科创板上市

Tianya is back, but what has returned may be merely a skeleton. Using the backup domain tianya.net, it has come back online after three years of disconnection due to overdue payments, opening only a selection of classic posts for browsing. What does this feel like? It’s like a once-flourishing martial arts master, forced by creditors to sell off his ancestral home (the .cn domain), now lodging in a side wing (the .net domain), relying on a few yellowed secret manuals (old classic posts) to barely maintain the last shred of dignity befitting a prestigious sect. Data migration still requires approval, with full restoration expected by June—but this lifeline has been extended for far too long, and far too tenuously. Born in 1999, Tianya was once one of the cradles of China’s internet spirit, where phenomena like "Ghost Blows Out the Light" and Sister Furong first emerged. Today, its struggle feels less like a resurrection and more like a carefully curated "digital heritage" memorial exhibition. What we truly miss is not this desolate platform that can’t even keep its own address, but that era of wild growth and clamorous BBS communities. Its story represents the last concentrated monetization of the collective nostalgia of PC internet natives—and no one dares guarantee how much longer that sentiment can hold.

At the same time, AI is being endowed with an almost "oracle-like" expectation. Rumors about AI grading the Guangdong college entrance exam act like a mirror, reflecting widespread societal anxiety and imagination surrounding the technology. We seem eager to outsource all repetitive, authoritative judgment work to cold algorithms in exchange for an illusion of "absolute fairness." The Guangdong Provincial Examination Institute’s rebuttal, however, was far more measured: AI serves merely as "electronic eyes" during inspection tours, responsible for observation and early warnings; the pens that ultimately decide candidates’ fates remain firmly in the hands of teachers under strict double evaluation. This division is astute—it draws a clear red line: technology can serve supervision and efficiency but must not overstep the core authority of "judgment" itself. In a society driven by human relationships, handing final interpretive power to AI is tantamount to stuffing the complexity of human nature and the warmth of education into a black box. The authorities’ caution is, in essence, guarding the last humanistic bottom line in education.

The collision between technology and tradition is always dramatic. Nikon of Japan claims it will use low-cost ArF lithography machines to challenge the unshakable monopoly of the Netherlands’ ASML. The courage is commendable, but it sounds more like a poignant industry saga. ASML’s moat was built over decades of technological iteration, a world-class supply chain, and astronomical R&D investment. Nikon’s "price war" strategy might find openings at specific junctures, but overturning the landscape would be as difficult as using a sharp knife to shake a mountain. A more realistic scenario may be that this offers a fallback option for mid-to-low-end chip factories barred by ASML’s high thresholds. Monopolies in the market need challenging, but we need not harbor unrealistic romantic fantasies about "breaking through." True counterattacks never rely on price cuts but on generational leaps in underlying technology.

The story of capital, meanwhile, is always fresh. Going public on the Hong Kong Stock Exchange within four years of founding and now swiftly preparing for a STAR Market IPO, MiniMax’s listing speed is itself an excellent footnote to the frenzy and involution in the current AI track. It verifies one thing: at the right momentum, speed is a barrier, and fundraising capability is survival. But what after a rapid IPO? The market will eventually shift from listening to "stories" to scrutinizing "statements." AI companies need to prove that their staggering valuations are not built on overdrawing the future. Similarly, at the eye of the storm, Elon Musk flatly denied rumors of a SpaceX valuation downgrade—this feels more like a classic move in capital market expectation management. Valuation itself is a grand psychological game; whether it’s 1.8 trillion or 2 trillion USD, behind the numbers lies the subtle fluctuation of investor confidence.

Some finer voices also deserve attention. The lemon tea brand Ningji is reportedly in talks to acquire Häagen-Dazs’ China business. If realized, this would be a "reverse acquisition" of a traditional international giant by a local new consumer brand—a metaphor for the new era. Can localized operations and a youthful supply chain inject new vitality into this aging ice cream aristocrat? Shenzhen authorities issued a real-world risk warning about ride-hailing saturation, puncturing the "earning over ten thousand a month" bubble—a timely and responsible alert far better than letting entrants blindly sink. Meanwhile, Saturday Fortune’s apology over a substandard bracelet exposes the rough quality control and rhetorical evasiveness in some traditional industries. What consumers want isn’t explanations, but steadfast adherence to national standards.

You see, these fragments pieced together reveal an incredibly complex and vivid cross-section of our times. Here lie the collapse and guardianship of old temples, the halo and shackles of new technology, the surge and caution of capital markets, and the calculations and warnings of everyday commerce. As we seek identity in nostalgia, we also feel both bewildered and excited before technological change. Each day’s trending news is a subtle tremor in the body of society, and each tremor deserves our independent, non-conformist gaze—to observe, to think, and to record.

天涯回来了，但带回来的可能只是一副骨架。用tianya.net这个备胎域名，在欠费断网三年后重新上线，只开放部分精华帖浏览。这感觉像是什么？像一个曾经风华绝茂的武林高手，被债主追得变卖了祖宅（.cn域名），如今借住在偏院（.net），靠着几本发黄的秘籍（旧精华帖）勉强维系着名门正派的最后一点体面。数据迁移还需审批，完整恢复预计在六月，但这口气接得实在太长、太勉强。1999年诞生的天涯，本是中国互联网精神的子宫之一，《鬼吹灯》和芙蓉姐姐都在此分娩。如今它的挣扎，不像复活，更像一场精心维系的“数字遗产”纪念展。我们怀念的，终究是那个野蛮生长、众声喧哗的BBS时代，而不是这个连自家门牌号都保不住的落寞平台。它的故事，是PC互联网原住民们集体怀旧情绪的最后一次集中变现，而情怀能撑多久，没人敢打包票。

与此同时，另一边厢，AI正被赋予一种近乎“神谕”的期待。广东高考用AI批卷的谣言，像一面镜子，照出了社会对这项技术的普遍焦虑与想象。我们似乎急于将一切重复、权威的评判工作，都外包给冰冷的算法，以换取一种虚幻的“绝对公平”。广东省考试院的辟谣则冷静得多：AI只是巡考时的“电子眼睛”，负责观察和预警；而真正决定考生命运的笔头，依然紧紧握在严格双评的老师手中。这种划分很聪明，它划出了一条清晰的红线：技术可以服务于监督与效率，但不能僭越“评判”本身的核心权柄。在一个人情社会，把最终解释权交给AI，无异于将复杂的人性与教育的温度塞进黑箱。官方的谨慎，本质上是在守护教育领域最后的人文底线。

科技与传统的碰撞，总是充满戏剧性。日本尼康宣称要用低价ArF光刻机，去挑战荷兰ASML不可撼动的垄断地位。这勇气可嘉，但听起来更像一个悲壮的行业故事。ASML的护城河是几十年技术迭代、全球顶级供应链和天文数字研发共同浇筑的。尼康的“价格战”策略，或许能在特定节点找到缝隙，但想颠覆格局，难度不亚于用一把锋利的小刀去撼动一座大山。更现实的图景可能是，这为那些被ASML高门槛挡在门外的中低端芯片厂，提供了一个备选项。市场的垄断需要挑战，但我们不必对“破局”抱有不切实际的浪漫幻想。真正的反击，从来不是靠降价，而是靠底层技术的代际跃迁。

资本的故事则永远新鲜。成立四年即登港股，现在又马不停蹄筹备科创板上市，MiniMax的上市速度，本身就是当前AI赛道狂热与内卷的绝佳注脚。它验证了一件事：在风口上，速度就是壁垒，融资能力就是生存能力。但快速IPO之后呢？市场终将从听“故事”转向看“报表”。AI公司们需要证明，那惊人的估值不是建立在对未来的透支之上。同样身处风暴眼的马斯克，对SpaceX估值下调的传闻一口咬定“不实”，这更像是资本市场预期管理的经典戏码。估值本身就是一场大型心理博弈，1.8万亿还是2万亿美元，数字的游戏背后，是投资者信心的微妙浮动。

一些更细碎的声音同样值得倾听。柠檬茶品牌柠季被传洽谈收购哈根达斯中国业务，如果成真，将是本土新消费品牌对传统国际巨头的一次“反向收购”，充满了新时代的隐喻——本土化的运营、年轻的供应链，能否给这个日渐老去的冰淇淋贵族注入新活力？深圳官方发布网约车饱和的风险提示，字字实在，戳破了“月入过万”的泡沫，这是一次及时的、负责任的预警，远比放任入行者盲目沉没要好。而周六福为一只不合格手镯的道歉，则暴露了部分传统行业在质量管控上的粗放与话术上的巧辩，消费者要的不是解释，是严守国标的踏实。

你看，这些碎片拼凑出的，是一个何其复杂又生动的时代切片。这里有旧神殿的坍塌与守护，有新技术的光环与枷锁，有资本市场的狂飙与审慎，也有市井商业的算计与警示。我们一边在怀念中寻找身份认同，一边在技术变革前感到迷茫与兴奋。每一天的热点新闻，都是社会肌体一次细微的颤动，而每一次颤动，都值得我们用独立而不从俗的眼光，去观察，去思考，去记录。