lemonscent

Make me curious about the world; make me faithful to my own truth.

[lemonscent @lemonscent_z]

2026 I Ching Forecast and Surviving Principle II

[Michael Guo @Michaelzsguo]

Jeff Dean 离职后的第一次公开露面，是在 AASF 2026 亚洲学者论坛。

结束 27 年的 Google 生涯仅 12.5 小时后，他便与 Dawn Song 登台对谈。他开玩笑说，自己大概“失业了一秒”，随后 Discovery Loop 正式开始运作。

整场对谈聊了 MoE、TensorFlow、Gemini 和 AI Agent 的安全风险。但最值得关注的，是 Jeff Dean 第一次完整解释了 Discovery Loop 的 thesis。

Discovery Loop 并不打算再造一家把模型做得更大的公司。它想优化的是科学研究中最根本的单元：Discovery Loop，也就是从假设到结果的完整实验循环。Jeff Dean 把科学方法拆成了一个可以不断迭代的系统：

拆解问题 → 提出假设 → 设计实验 → 评估结果 → 获得反馈 → 再次迭代

过去，AI 主要参与其中某个环节。比如预测结果、生成代码或者搜索候选方案。Discovery Loop 的目标，则是让 AI 参与整个研究闭环：
自动改进模型架构
自动选择训练数据
自动设计评估方法
自动优化训练算法
同时运行数千个实验， 根据预期价值，动态决定算力应该投入到哪些实验

Discovery Loop背后的核心判断：科学研究真正的瓶颈，可能不是缺少知识或想法，而是实验循环运行得太慢。如果一次迭代从一周缩短到一小时，再把大量实验从串行变成并行，研究效率的变化就不再是提升 10% 或 20%，而可能是几个数量级。

这也解释了为什么 Jeff Dean 认为一家十人左右的公司可以做这件事。云计算已经把过去只有大公司才拥有的基础设施变成了可以购买的资源。创业公司的真正优势不再是拥有数据中心，而是能否让一个小团队把全部注意力集中在同一个目标上。

Discovery Loop 由 Jeff Dean、Sanjay Ghemawat、Oriol Vinyals 和 Quoc Le 共同创立。这个团队覆盖了分布式系统、基础模型、强化学习和大规模 AI 研究，组合本身也非常契合这家公司想做的事。

[Jeff Dean @JeffDean]

Announcing Discovery Loop!

I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix, we are founding Discovery Loop (@DiscoLoopAI), a Public Benefit Corporation whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress. The four of us have worked together for 14 to 30 years, and have helped build some of the world’s most used products, infrastructure and AI models, and we’re excited to turn our attention to this ambitious endeavor.

♾

Learn more at: https://t.co/Rv3LMdLluK

[lemonscent @lemonscent_z]

@JeffDean Thank you for making the technology widely accessible and public utility for the world. Google is no longer the Google without you.

[Nikolay Savinov @SavinovNikolay]

@JeffDean Jeff, good luck with the new adventure! Without your help, long context in Gemini wouldn’t have been released as fast as it was - eternally grateful to you and admire your love for ambitious 0 -> 1 technology!

[Cassandra Unchained @michaeljburry]

Evil incarnate.

[lemonscent @lemonscent_z]

@sunpolis 西班牙那几个联赛队友阴得很，人品不行，赛前各种抱怨裁判不公、抹黑阿根廷。罗德里是惯犯了，之前就当场庆祝他的前曼城队友B席错失绝平进球

[lemonscent @lemonscent_z]

@FuSheng_0306 说得太好了。这就是人生。

[傅盛 @FuSheng_0306]

刚在现场看完世界杯决赛，阿根廷输了，挺伤心的。但也许不拿冠军，英雄才更真实。梅西会失败、会迟暮，正因不完美，我们才觉得自己也可以努力，不用怕一步之遥。至于贬低他的人，我想起高尔基的一句话：鹰有时飞得比鸡低，鸡永远飞不到鹰那么高。

[lemonscent @lemonscent_z]

@sunpolis 因为这场比赛裁判太不公平了非常双标，梅西各种被暴力犯规，而且西班牙球员有人身攻击侮辱挑衅行为。

[lemonscent @lemonscent_z]

@MoonlightZephy @jia_tu49073 你们俩卷起来也是心有灵犀的，偶喜欢，哈哈哈

[lemonscent @lemonscent_z]

木星进入狮子座看似会为AI科技领域的巨头们加冕称王。但正如下文分析所指出的，如果任由两三家闭源实验室维持高达90%的暴利，建立起极端的寡头垄断，那将是对算力、芯片和整个软件生态的疯狂吸血。因此，虽然寡头、垄断、精英式财富集中还要走向极端化，但也必然会带来整个行业最强烈的质疑和对抗。

要打破这个死局，必须找到类似Kimi这样的开源力量，或是让自带全产业链闭环的科技巨头下场掀起价格战。只有把AI模型层的虚高暴利打下来，把生存空间让渡给基础设施和应用层，科技生态才有活路。

结合这个大博弈背景，再看谷歌为何突然推迟发布Gemini新模型，表面是进一步拉跌了谷歌的股价，而且显得谷歌已经在和OpenAI和Anthropic的大战中输了，实际上这是一次极具定力（或者说是碰巧的幸运？）的“战术避险”。避开最近水逆混乱的日子发布重要产品，从另两个AI巨头大战中先逃遁出来，不需要顶着近期的乱流强行抢跑，在这个天时背景下反而可能因祸得福。

因为今年8月和明年8月都会发生狮子座的日全食，宇宙正在悄然散播、褪走这顶“垄断霸主”王冠的光芒，这意味着对高高在上垄断霸权的因果剥夺与清算。谷歌此时主动退下神坛，实质上是把这顶“带电的王冠”留给了现任寡头，让它们去硬扛接下来极高的算力消耗与反垄断冲击波。

试图把持技术的寡头们似乎忘了，技术文明就像太阳的光芒一样。太阳虽然是我们这个太阳系的中心，权威无上，但是太阳所代表的正是人类的智能和启蒙。最核心的天道法则是——太阳的光芒是免费无私地普照大众的，它从根本上就不该是被打包成明码标价的特权稀缺资产。

垄断的极点往往就是重置的起点。技术范式的潮水终将顺应宇宙规律。谁能真正将AI从高高在上的特权稀缺资产，化作便宜而普惠大众的公共基础设施，让“智能之光”无差别地照进每一个终端，谁就能赢得天下。

[Gavin Baker @GavinSBaker]

Kimi K3 may be an important inflection point for AI. Potentially negative for Anthropic and OpenAI while being net positive for essentially every other company in the world. I mean that very literally. Although the real “Sputnik moment” would be an open-source frontier model that was also token efficient unlike Kimi K3 which is 50-70% more expensive to run than GPT 5.6 per Artificial Analysis.

Rationale:

A world where there are only 2-3 dominant frontier labs with 90% inference margins is net negative for every other layer while being awesome for those 2-3 labs. Those labs would become monopsonies for power, data centers, semiconductors and hyperscalers and would obviously vertically integrate over time into all those layers while also completely subsuming the application/software layers. 

Anything that lowers margins and increases competition at the model layer is good for every other AI layer: power, semiconductors, hyperscalers, neoclouds and yes even software.

This is why Jensen is so supportive of open-source. An open-source model requires the *exact* same amount of compute to run as a closed frontier model of similar size and architecture. Kimi K3 is roughly the same price as GPT 5.6 Terra on a per token basis, which actually suggests that it is less computationally efficient as I am sure that GPT 5.6 is priced to a higher margin than K3. And given that K3 is a token wastrel, i.e. token inefficient, it is significantly more expensive per task than GPT 5.6 and Grok 4.5, which are much more token efficient. Cost per token and token efficiency (i.e. intelligence density per token) are the drivers of intelligence per unit of cost. The winning AI companies will be those that offer the most intelligence per $ over time.

Lower margin % at the model layer = more margin $ at every part of the infrastructure layer and is a godsend for software. This can happen either through open-source models like K3 at the frontier *or* having a vertically integrated model company like Meta, SpaceX or Google at the frontier. Both outcomes result in a lower margin % at the model layer as vertically integrated model companies don’t really care where the margin $ come from. This is why it was so painful for OpenAI and Anthropic when Google was right there with them from a model competitiveness perspective and why Grok 4.5 and Muse 1.1 were just as important as Kimi K3.

The reason Kimi K3 is only *potentially* negative for Anthropic and OpenAI is 1) the @ericvishria point that the Claude and ChatGPT products and harnesses may be more important than their models today and 2) the hypothesis that they have much more advanced model checkpoints internally that are already being used for RSI. In the latter scenario, reaching RSI even a few months ahead of other labs might be enough to cement a permanent lead.

Time will tell on both points. And likely fairly quickly.

Caveat would be that since Kimi K3 is not token efficient and thereby actually more expensive than ChatGPT 5.6, we may need to see a more token efficient open-source model at the frontier or see Grok 5/Composer 4/Muse 2 at multiple points on the Pareto frontier for this potential risk to Anthropic and OpenAI to play out. And I am sure they will both vertically integrate as quickly as possible while continuing the product/harness strength they have shown over the last 8 months.

[Rho Rider @RhoRider]

⛔️ Just remember when Dot Com popped it wasn’t the first 40% drawdown that killed the tech Bulls…

It was the 4-month-long dead cat rally they threw everything they had left at before the 75% drop that got them…

[心灵书舫 | SoulSloop @SoulSloop99]

亚历克斯·弗格森爵士谈莱昂内尔·梅西在2026年FIFA世界杯上的表现。

“我一生都在观看足球。我曾与最伟大的球员交手，执教过一些顶尖球员，见证了数代天才。但莱昂内尔·梅西在39岁时所做的一切，超出了我所见过的任何东西。”

“大多数球员到了那个年纪早已退役，或者在努力跟上现代足球的节奏。梅西呢？他正带领阿根廷再次闯入世界杯决赛。他不是在生存，他是在主宰。这就是区别。”

“人们总在问他是怎么做到的。答案很简单：他的足球头脑领先其他人几十年。他不再需要跑过你，因为在你迈出第一步之前，他就已经预见了接下来的三传球。”

“看看本届赛事。每当阿根廷需要有人站出来时，梅西总是回应。当他们承受压力时，他要求控球。当英格兰以为自己一只脚已踏入决赛时，梅西用两次惊世助攻彻底改变了比赛。这就是伟大该有的样子。”

“你可以完美组织你的防守，派两三个球员盯防他，花几个月时间分析他的一举一动。但这一切都无法保证任何结果，因为他只需一秒钟，就能摧毁你九十分钟的努力。”

“最让我震惊的不是他的能力，而是他的心态。场合越大，他越冷静。当其他人感受到世界杯的沉重压力时，梅西踢球仿佛天生就为这些时刻而生。这就是为什么他不断击碎人心并创造历史。”

“我要说一些有些人不喜欢的话，但我不在乎。我们再也不会看到像莱昂内尔·梅西这样的足球运动员了。当他退役时，一个足球时代将与他一同退役。球员们会赢得金球奖，进数百粒进球，举起奖杯，但没有人能像他那样，将天才、一致性、耐力与谦逊完美结合。”

“珍惜他还在球场上的每一分钟，因为当莱昂内尔·梅西离开足球时，这项运动将失去一小部分灵魂。他之后会有伟大的球员，但再也不会有另一个莱昂内尔·梅西。

[lemonscent @lemonscent_z]

@sunpolis 因为阿根廷靠的不只是纯粹的运气，而是团结一致为了共同的精神领袖和目标积极进取的行为，这个是完美符合今年值年卦天火同人的正面吉利显象，天人合一，自助者天助之。

[Danny @DjokovicFan_]

Novak Djokovic: “I refused lots of big brands and huge paychecks because I can’t represent something that I do not believe in. I also rejected the most famous drink in the world. I care about integrity. If I do not drink something, I can’t represent it.”

[dilemma @actuallyimthe]

Argentina just beat Spain at the 2026 World Cup final, 3-2.

[lemonscent @lemonscent_z]

H2 of this fire horse year marks the inevitable reversal and systemic backlash of a "fire" structure that has reached its absolute peak.

We are looking at a hard structural pivot:
▪️ Internal crises mutating into external conflicts (or vice versa)
▪️ H1 stability collapsing into H2 chaos
▪️ Mutiny following successful suppression; immediate downfalls post-coronation
▪️ Recessive (Yin) variables abruptly infiltrating established systems
▪️ A complete reversal in trade, markets, and warfare as previously underestimated factors enter and alter the entire board.

Historically, this exact structural signature manifests as:
Loss of authoritative control;
Loyalty weaponized as betrayal;
Strategic pivots in warfare;
Rapid cooling of overheated markets;
Urban, financial, or food systems pierced by Black Swans.

This is not standard volatility. It is the forced redistribution of power, wealth, influence, and control from a central structure that simply ran too hot. The macro (global geopolitics) and the micro (individual factions) will mirror each other completely.

As Warren Buffett said: "What we learn from history is that people don't learn from history."

[ian bremmer @ianbremmer]

lindsey graham knew exactly who trump was.
and still decided
that being in his orbit
was exactly where he needed to be.
@gzeromedia

[Cameron Khansarinia @khansarinia]

On February 14 250,000 Iranians marched for freedom in Munich.

A surprise guest took the stage: Lindsey Graham.

The crowd erupted in chants of: "USA! USA! USA!"

The Senator was visibly moved. He grabbed the Lion and Sun flag and waived it proudly.

Iranians will never forget.