2026-08-08
The latest addition is Opening the curtain of AI business integration, a Marketplace Tech episode with Megan McCarty-Corino interviewing Priya Rathod of Indeed on workplace AI readiness. It adds Priya Rathod, Workplace AI Readiness Gap, AI Native Worker, Managerial AI Fluency Gap, and AI Job Security Anxiety, while extending Marketplace Tech, Megan McCarty-Corino, Indeed, YouGov, AI Worker Literacy, Business-Led AI Transformation, AI Organization Design, AI Economic Diffusion, and AI Automation Redistribution. Its core synthesis is that AI business integration is blocked less by tool availability than by organizational readiness: employers want AI-native skills, workers need training and trust, managers need enough fluency to lead, and job-security fears can make productivity gains feel personally risky.
The latest addition is 174: AI冲击企业软件巨头？与SAP原欣聊大模型to B的颠覆与边界, a LateTalk episode with 原欣 of SAP on AI’s impact on ERP and enterprise software. It adds Yuan Xin / 原欣, Enterprise Resource Planning, ERP Trust Moat, Autonomous Enterprise, Enterprise Operational Memory, China Enterprise AI System Debt, SAP Joule / Joule Work, SAP Signavio, SAP S/4HANA, Workday, Salesforce, and Feishu / 飞书, while extending SAP, LateTalk, Forward Deployed Engineer, Business-Led AI Transformation, AI Native SaaS Threat, SaaS Trust Moat, Outcome-Based AI Pricing, Result As A Service, Enterprise Agent Governance, Enterprise Agent Memory, Enterprise Data Activation, Language User Interface, Agent Native Software, Alibaba, Alibaba Cloud / 阿里云, Qwen, and DingTalk. Its core synthesis is that AI can reshape enterprise software interfaces, pricing, customization, and agent workflows, but ERP-like systems remain constrained by trusted data, auditability, compliance, business-process memory, localization, and human review.
The latest addition is China’s soft power play in the global AI arms race, a Marketplace Tech episode with Megan McCarty-Corino interviewing Adam Siegel of the Council on Foreign Relations on Chinese open-weight AI models. It adds Adam Siegel, Council on Foreign Relations, and Chinese Open-Weight AI Strategy, while extending Marketplace Tech, Megan McCarty-Corino, China, United States, OpenAI, Anthropic, Microsoft, Hugging Face, Open Source AI Models, Open Weight Release Boundary, AI Model Censorship, Frontier Model Access Restrictions, AI Export Controls, Sovereign AI Models / 主权AI模型, AI Model Sandbox Escape, AI Cyber-Defense Utility, and AI Governance And Compliance. Its core synthesis is that Chinese open weights are both a market strategy and a geopolitical access strategy: they pressure closed U.S. frontier models through price, portability, and local control, but leave unresolved questions around censorship, data, dependence, coercion, and future export controls.
The latest addition is 148. 对游凯超3小时访谈：开源Infra、和模型Co-design 、“如果vLLM失败，我们会后悔一辈子”, a 张小珺Jùn｜商业访谈录 episode with 游凯超 on vLLM, Infract, and open-source large-model inference infrastructure. It adds 游凯超 / Kaichao Yu, Infract, vLLM, PyTorch Foundation, PagedAttention, Model-Infra Co-Design, Open Source AI Infrastructure, Continuous Batching, Prefix Caching, Test-Time Scaling, and Hardware Lottery, while extending DeepSeek, Kimi, PyTorch, Tsinghua University / 清华大学, UC Berkeley, AI Inference Cost Structure, Agent Harness, Mixture of Experts, Open Source AI Models, High-Throughput Inference Batching, Inference Acceleration Stack, AI Chip Specialization, AI Infrastructure Full-Stack Moat, Large Company Open Source Strategy, and Model Harness Co-Evolution. Its core synthesis is that AI inference is now a co-design layer: models, serving engines, hardware, open governance, and agent harnesses all affect whether tokens can be produced cheaply, reliably, and openly.
The latest addition is Burning questions: a more fire-prone world, a The Intelligence episode on worsening wildfire risk, stalled progress for women at work, and U.S. nightlife mayors. It adds Vingeru Makandewiri, Danielle Arras, Corrine Reynolds, Boston, Indonesia, Rudy Giuliani, Extreme Wildfire, Wildfire Smoke Health Burden, Wildfire Climate Feedback, Landscape Fire Prevention, Women Workplace Progress Stall, Promotion Ambition Gap, Childcare Infrastructure Constraint, Nightlife Mayor, Nighttime Economy Governance, Liquor License Quota Barrier, and Socializing Decline, while extending Oliver Morton, Climate Adaptation, Fire Tech Climate Resilience, Natural Hazard As Social Disaster, Disaster Response State Capacity, Goldman Sachs, JPMorgan Chase, McKinsey, New York City, Friendship Recession, and Adult Summer Camps. Its core synthesis is that adaptation is institutional before it is technical: fire prevention, women’s advancement, and after-dark urban life all depend on whether governments, firms, and cities can sustain the unglamorous systems that make risk, work, and social connection livable.
The latest addition is 175.公募基金二季报：极致的抱团与割裂之后, a 起朱楼宴宾客 episode by 大卫翁 reading Chinese public mutual funds’ 2026 second-quarter reports after extreme AI, compute, and semiconductor concentration. It adds Active Fund Crowding / 主动基金抱团, Fund Manager Ranking Incentives / 基金经理排名激励, AI Infrastructure Supply-Chain Bullwhip / AI 基建供应链牛鞭效应, 金子才 / Jin Zicai, 谢治宇 / Xie Zhiyu, 张坤 / Zhang Kun, 陈经纬 / Chen Jingwei, and 江城, while extending Public Mutual Fund Ecosystem / 公募基金生态, Fund-Investor Return Gap / 基金赚钱基民不赚钱, Market Breadth Narrowing / 市场广度收窄, Retail Investor Crowding, AI Equity Valuation Risk, AI Capex Return Window, Portfolio Suitability, 起朱楼宴宾客 / Qizhulou Yan Binke, 大卫翁 / David Weng, and 中欧基金 / Zhongou Fund. Its core synthesis is that this public-fund technology boom is not only a story of correct AI winners: fund-manager ranking incentives, shared research, channel and holder short-termism, and AI supply-chain bullwhip risk can turn a real technology cycle into fragile crowding for late fund buyers.
The latest addition is An interview with Elon Musk, the full The Intelligence interview where Zanny Minton Beddoes presses Elon Musk on AI, robots, SpaceX, Starlink, DOGE, USAID, and European politics. It adds Frontier Model Peer Review, Universal High Income, AI Work Optionality, AI Fatalistic Acceleration, and Private Infrastructure War Power, while extending AI Abundance Narrative, AI Safety Coordination, AI Alignment Governance, Recursive Self-Improvement, Physical AI, Humanoid Robot Commercialization, Founder Control, Tesla, SpaceX, Starlink, OpenAI, Anthropic, X, and Techno-Optimism Present-Threat Gap. Its core synthesis is that Musk’s future case now rests on unstoppable AI plus physical robots, rival-lab model review, and post-scarcity redistribution, but the interview tests that future against present control failures: private infrastructure in war, aid-cut implementation harm, platform-amplified political claims, and concentrated authority over companies that may become public infrastructure.
The latest addition is Brave New whirl: Turkey’s opposition overhaul, a The Intelligence episode on Turkey’s opposition overhaul, Germany’s surrogacy-policy scandal, and Gen Z career advice under AI. It adds Ozgur Ozel, New Party (Turkey), Republican People’s Party, Kemal Kilicdaroglu, Mansur Yavas, Jens Spahn, Friedrich Merz, Christian Democratic Union (Germany), Alternative for Germany, Bill Gurley, Jodi Kantor, Benjamin Todd, 80,000 Hours, Turkish Opposition Realignment, Judicialized Opposition Suppression, Opposition Party Hollowing, Surrogacy Policy Hypocrisy, Political Double-Standard Backlash, AI Graduate Career Uncertainty, Career Advice Survivorship Bias, Passion Trap Career Advice, Useful Work Career Compounding, Impact Career Choice, and Effective Altruism, while extending Recep Tayyip Erdogan, Justice and Development Party, Ekrem Imamoglu, Germany, Autocratic Succession, AI Hiring Arms Race, AI Job Security Anxiety, Entry-Level AI Career-Ladder Risk, and College Career Preparation. Its core synthesis is that institutional pressure changes the meaning of choice: Turkish voters may get a new opposition vehicle while courts and arrests shape who can compete, German voters punish rule-makers who privately route around rules they defend, and graduates need useful skill compounding more than inherited passion slogans when AI is changing hiring.
The latest addition is No.212 曾经的精神角落：豆瓣、知乎、贴吧、虎扑 | 中国互联网故事25, a 半拿铁 China-internet-history episode on 豆瓣, 知乎, 百度贴吧, and 虎扑 as PC-era Chinese communities. It adds 知乎 / Zhihu, 百度贴吧 / Baidu Tieba, 虎扑 / Hupu, 阿北 / Abei, 周源 / Zhou Yuan, 程杭 / Cheng Hang, 杨冰 / Yang Bing, 得物 / Dewu, Chinese PC Internet Communities / 中文 PC 互联网社区, Online Community Commercialization Friction / 社区商业化摩擦, Community vs Content Platform / 社区与内容平台区别, Keyword Community Architecture / 关键词社区架构, Mobile Community Fragmentation Risk / 移动社区入口碎片化风险, Knowledge Community Creator Incentives / 知识社区创作者激励, and Forum Subculture Memory / 论坛亚文化记忆, while extending 豆瓣 / Douban, Baidu, 半拿铁, ByteDance, Zhang Yiming, 小宇宙, Platform Community Governance, User-Powered Content Platform, Generational Technology Bridge, and Feature Phone Cultural Memory. Its core synthesis is that “spiritual corner” communities are hard to value with ordinary platform metrics: their strength comes from relationships, norms, taste, identity, and shared memory, so the same commercial and algorithmic tools that scale content platforms can damage the atmosphere that made these communities worth returning to.
The latest addition is EP91 订房订票定江山，携程51亿为傲慢买单, a 一劳永逸 episode on Ctrip / Trip.com Group’s reported 51.79亿元 antitrust penalty. It adds Ctrip Antitrust Penalty, Travel Platform Merchant Exclusivity, Travel Price Parity Enforcement, AI Travel Planning, Ctrip Wendao, and Zhuang Chenchao, while extending State Administration for Market Regulation, Online Travel Agency, OTA Platform Concentration, Hotel Platform Pricing Power, Travel Booking Hidden Fees, Travel Super App Convenience, Platform Antitrust, Platform Data Regulation, Qunar, Elong, Skyscanner, and MakeMyTrip. Its core synthesis is that OTA dominance becomes governable only when the mechanisms are visible: merchant exclusivity, lowest-price enforcement, ranking traffic, automatic repricing, cancellation rules, customer service, global supply, and AI travel planning all determine whether a platform remains a useful travel interface or turns into a tollgate.
The latest addition is Vol.268 两个劳斯莱斯, a 商业就是这样 episode on why there are “two Rolls-Royces” and how the RB211 aircraft-engine program split Rolls-Royce’s engine business from Rolls-Royce Motor Cars. It adds Rolls-Royce RB211, Lockheed Corporation, Lockheed L-1011 TriStar, Rolls-Royce Motor Cars, Charles Rolls, Henry Royce, Claude Johnson, McDonnell Douglas DC-10, Tanaka Kakuei / 田中角荣, Kodama Yoshio / 儿玉誉士夫, All Nippon Airways / 全日空, Lockheed Bribery Scandal, Fixed Price Engineering Risk, Airframe Engine Lock-In, Industrial Capability Bailout, Overseas Bribery Export Sales, and Foreign Corrupt Practices Act, while extending 商业就是这样, Rolls-Royce, Boeing, Marubeni / 丸红, Richard Nixon, Strategic Industrial Policy, and Constraint Driven Engineering Strategy. Its core synthesis is that hard technology can be right and still financially fatal when fixed-price contracts, airframe dependency, national prestige, and export-sales pressure compress engineering uncertainty into one company before the technology has time to mature.
The latest addition is 175: 对话Liblib陈冕：关于活下来，以及所有接近死亡的时刻, a LateTalk interview with Chen Mian / 陈冕 of Evoken / 言语科技 on how Liblib, Lavod, and Lib TV try to survive model-layer pressure, low-margin pricing, public controversy, and organization strain. It adds Chen Mian / 陈冕, Evoken / 言语科技, Liblib, Lavod, Lib TV, and AI Application Survival Strategy, while extending AI Application Layer Moat, AI Startup Unit Economics, AI Subscription Economics, AI Application Market Trough, Model Provider Tool Competition, AI Commercialization Pressure, AI Organization Design, Seedance, and Video Models. Its core synthesis is that independent AI applications are not dead by default, but their survival depends on keeping cash flow, accepting honest low positive margins, using timing and user scale to buy room, and turning speed into organization and workflow moats before model providers absorb the easy parts.
The latest addition is Advice Line with Chris Riccobono of UNTUCKit, a How I Built This Advice Line episode with Guy Raz and Chris Riccobono of UNTUCKit. It adds Chris Riccobono, UNTUCKit, Greatness Wins, Aaron Sanandres, AeroShorts, Adrian Alvarez, Snug Safety, Preet Anand, Life Alert, Hockey Ninja, Derek (Hockey Ninja), Core Tribe Positioning, Wholesale As Marketing, Trust-Based Distribution, and Safety Product Credibility, while extending How I Built This, Guy Raz, AI Discovery SEO, Measured Channel Testing, Channel Focus Experiments, Subculture Led Marketing, Athlete Sponsorship Strategy, Trust As Business Asset, and Relationship-Led Growth. Its core synthesis is that channel choice has to match how the customer learns to trust the product: apparel can start with a core tribe or golf-shop discovery, senior safety needs trusted intermediaries and measurable audio/earned-media tests, and hockey protection needs professional validation before generic awareness can matter.
The latest addition is Working memory: the surprising decline of dementia, a The Intelligence episode on falling age-adjusted dementia rates, Brazil’s Pix payment system, and the return of emo music. It adds Jonathan Bousenthal, Carla Sugirana, Pix, Central Bank of Brazil, My Chemical Romance, Gerard Way, The Black Parade, When We Were Young Festival, Vans Warped Tour, Wembley Stadium, Age-Adjusted Dementia Decline, Modifiable Dementia Risk Factors, Cardiovascular-Brain Health Link, Shingles Vaccine Dementia Signal, APOE4 Dementia Risk, Amyloid Hypothesis Uncertainty, Alzheimer Drug Efficacy Gap, Public Instant Payment Rail, Payment Infrastructure Trade Dispute, and Emo Revival Identity Culture, while extending Public Health Invisibility, Cognitive Decline Advance Planning / 认知障碍提前规划, Elder Care State Capacity, GLP-1 Agonists, Payment Clearing Network / 支付清算网络, Money Movement Infrastructure, Trade Reciprocity Protectionism, Donald Trump, Visa, Mastercard, Concert Residency Economics, The Intelligence, and Economist Podcasts. Its core synthesis is that hidden trend lines matter: dementia risk can fall even as ageing raises total burden, a public payment rail can expand inclusion while being misread as protectionism, and an old music scene can return when nostalgia is reinforced by live spectacle and identity belonging.
The latest addition is 149. 亲历中美 New Labs 资本狂潮，和清华刘子鸣聊：AI for AI、机制可解释性和 Max Tegmark, a 张小珺Jùn｜商业访谈录 episode with 刘子鸣 on AI For AI, Physics Of AI, Mechanistic Interpretability, KAN, and the 2026 New Lab financing wave. It adds Liu Ziming / 刘子鸣, Max Tegmark, Yuanhuan Intelligence / 圆环智能, Shanghai Qi Zhi Institute / 上海期智研究院, AI For AI, Physics Of AI, Mechanistic Interpretability, Kolmogorov-Arnold Networks, OPHIS Research Workflow, Meta-Model Training Curve Prediction, Training Autopilot, Vibe Training, and New Lab Organization, while extending Auto Research, Recursive Self-Improvement, AI For Science, World Models, Transformer Architecture, Hardware Lottery, Research Taste, Academic AI Research Role, AI Interpretability By AI, Tsinghua University / 清华大学, Peking University / 北京大学, MIT, Stanford University, Recursive, and 张小珺Jùn｜商业访谈录. Its core synthesis is that AI for research needs more than agents trying many experiments: Liu argues for a physics-of-AI layer that structures observations, hypotheses, training dynamics, and failure modes so future systems can predict curves, choose better architectures, and eventually make model training feel closer to an autopilot.
The latest addition is Omission accomplished: why the Iran-war cycle spins on, a The Intelligence episode linking the fraying U.S.-Iran MOU, Chinese driverless-taxi exports, and Wally Funk’s late spaceflight. It adds Don Weinland, Ann Wroe, Wally Funk, Blue Origin, William Randolph Lovelace, Federal Aviation Administration, National Transportation Safety Board, Oman, Houthis, Yemen, Hamas, Gaza, Thin Diplomatic Frameworks, AI Service Export Standard-Setting, and Gender Barrier Spaceflight, while extending Greg Carlstrom, Iran, Strait of Hormuz, U.S.-Iran Nuclear Diplomacy, Iran Postwar Economic Relief, Proxy Conflict Spoiler Risk, Chokepoint Shipping Confidence, China AI Export Leverage / 中国AI出口杠杆, Robotaxi Economics, Baidu, Waymo, Jeff Bezos, The Intelligence, and Economist Podcasts. Its core synthesis is that thin agreements and early deployments both shape reality before the hard questions are settled: vague Middle East term sheets can claim peace while leaving control, disarmament, withdrawal, and money unresolved; Chinese AI services can set foreign operating standards before U.S. rivals arrive; and Funk’s story shows that formal capability tests did not stop gendered gatekeeping from delaying recognition for decades.
The latest addition is 176: 姚顺宇，来到腾讯300天, a LateTalk episode with Gao Honghao / 高宏昊 on Yao Shunyu / 姚顺宇’s first 300 days rebuilding Tencent Hunyuan / 腾讯混元. It adds Tencent Hunyuan / 腾讯混元, WeChat VLM / 微信 VLM, Tencent CSIG, Tencent WorkBuddy, Martin Lau / 刘炽平, Zhang Xiaolong / 张小龙, Zhou Hao / 周颢, Lu Shan / 卢山, Tencent TEG, Gao Honghao / 高宏昊, and Federated AI Organization, while extending Tencent, WeChat, Yao Shunyu / 姚顺宇, Yuanbao, LateTalk, DeepSeek, Kimi K3, AI Organization Design, Large Company Organizational Inertia, AI Assistant Service Entry, Mixture of Experts, Agent RL, Model Provider Tool Competition, AI Programming Engine Shift, and Long-Chain AI Competition. Its core synthesis is that Tencent’s AI catch-up is a model problem and an organization problem at once: Hunyuan can improve through young talent, infra rebuilding, executive sponsorship, and product feedback loops, but WeChat’s independent VLM path shows why a federated platform company cannot centralize AI purely by technical efficiency.
The latest addition is E246｜何谓蒸馏？聊聊硅谷如何看中国开放模型逼近前沿, a 硅谷101 episode with 王铁镇 and Keith Zhai on Kimi K3, Chinese open-weight model progress, and what “distillation” means. It adds Wang Tiezhen / 王铁镇, Keith Zhai, Moonshot AI / 月之暗面, OpenRouter, Model Distillation / 模型蒸馏, Model Identity Data Pollution / 模型身份数据污染, Scaling Efficiency, Open-Weight Commercial Licensing, Closed Model API Moat Pressure, Model Sovereignty / 模型主权, Open Model Safety Governance, and Agent Inference Workload, while extending Kimi, Kimi K3, DeepSeek, Qwen, Open Source AI Models, Open Weight Release Boundary, Chinese Open-Weight AI Strategy, AI Inference Cost Structure, Model Routing Cost Control, Neo Cloud, Sovereign AI Models / 主权AI模型, Frontier Model Access Restrictions, AI Model Sandbox Escape, AI Cyber-Defense Utility, and Thinking Machines Lab. Its core synthesis is that Chinese open-weight models pressure closed frontier labs through more than benchmark parity: distillation accusations need evidence, efficiency and licensing matter, enterprise buyers care about sovereignty, and safety governance has to compare open and closed systems through auditability, training data, deployment controls, and incident response.
The latest addition is 270.大厂押注AI办公，飞书和钉钉却先成了配角, a 乱翻书 episode with 明昊, Eric, and 雅兰 on Feishu / 飞书, DingTalk, Doubao, Tencent WorkBody, and Qwen in AI office agent competition. It adds AI Office Agent, Tencent WorkBody, Doubao Enterprise Edition / 豆包企业版, Eric (Feishu source), and 雅兰 / Yalan (Luanfanshu), while extending 乱翻书, 庄明浩 / 庄明昊, ByteDance, Alibaba, Tencent, Feishu / 飞书, DingTalk, Doubao, AI Commercialization Pressure, AI Inference Cost Structure, AI Assistant Service Entry, AI Programming Engine Shift, Coding Agent As Universal Action Layer, Agentic Workflow, Enterprise Data Activation, Enterprise Operational Memory, AI Organization Design, AI Product Fragmentation, Large Company Organizational Inertia, and Model Provider Tool Competition. Its core synthesis is that AI office is becoming the monetization and workflow answer to expensive chatbot traffic: Feishu and DingTalk matter less as standalone IM protagonists than as document, meeting, permission, workflow, and organization-data substrates for models and agents, while the winner remains unclear because payment could come from enterprises or individual workers and coding-like execution may hide behind office interfaces.
The latest addition is EP272 谁的青春里，没有一本被传阅到卷边的杂志？, a Talk三联 episode with 高一丁, 伍玥, 魏茜, and 夏洁逸 using 三联生活周刊’s 1400th issue to reconstruct Chinese magazine-reading memory. It adds 伍玥 / Wu Yue (Sanlian), Magazine Reading Ecology / 杂志阅读生态, Classroom Magazine Circulation / 班级杂志传阅, Electronic Magazine Transition / 电子杂志过渡, Paper Magazine Subscription Ritual / 纸刊订阅仪式, and Local Zine Revival / 在地小杂志回潮, while extending Reading As Life Experience, Reading Medium Pluralism / 阅读媒介多元主义, Childhood Reading Ecology, Family Reading Ecology, Attention Fragmentation / 注意力碎片化, Mobile Internet Prehistory, and Analog Media Preservation. Its core synthesis is that magazines were periodic social objects rather than only article containers: family shelves, school bans, class sharing, serials, fan inserts, email issues, kiosks, subscriptions, and local zines shaped youth imagination and attention before mobile feeds absorbed much of that role.
The latest addition is 「热爱一个行业15年的理由是什么？」｜对谈汪天凡：我要投真正的快乐、投最纯的愿景、投人性的光辉【公路播客】, a Shizilukou Crossing road-podcast episode with Will Wang Tianfan / 汪天凡 of B.A.I Capital on 15 years in VC and AI-era investing. It adds Will Wang Tianfan / 汪天凡, B.A.I Capital, Loki/Lookie, Mashie, Yuan Ming / 袁鸣, Wisdom Over Intelligence / 智慧稀缺论, AI Context Machine / AI 上下文机器, AI-Native Investing Workflow, AI Cognitive Gym / 把 AI 当健身房, Three-Non Venture Theory / 三非理论, AI Infra Crypto Analogy, Blockchain Financial Innovation, and AI For Fun, while extending Koji, Personal AI Memory, Wearable AI Assistant, OS-Level Context, AI For AI, AI-First Organization, AI Investment Research, Founder Product Fit, Startup High-Beta Bet, AI Application Layer Moat, Stablecoins, Attention Fragmentation / 注意力碎片化, and Human Judgment Under AI. Its core synthesis is that AI investing should not stop at “more intelligence”: the investable frontier sits where context, feedback, human agency, founder vision, joy, and nonlinear competitiveness turn generic model capability into wisdom, presence, and durable product pull.
The latest addition is EP119 对话刘可凡：用 try-catch-finally，给独立做产品的内耗写个处理流程 🐛, a 硬地骇客 episode with 刘可凡 on solo product work, internal friction, and AI-assisted execution. It adds 刘可凡 / Liu Kefan, Try-Catch-Finally Self-Management / try-catch-finally 自我管理, Falsifiable Product Hypothesis / 可证伪产品假设, and Human As Agent Tool / 人作为 AI 工具, while extending One-Person Company, Fast Product Validation, Validated Learning, Product Analytics, Founder Work Boundaries, AI Use Pacing, AI As Business Operator, AI Engineering Thinking, Human Agency Under AI, Model Context Protocol, and Agentic Workflow. Its core synthesis is that independent builders need a process for pressure as much as for product: start before confidence, convert fear into falsifiable experiments, use AI for concrete decomposition rather than life direction, and stop deliberately so solo autonomy does not become endless optimization.
The latest addition is Serena & Lily: Serena Dugan and Lily Kanter. They Built a $20M Brand—Then One Investor Almost Destroyed It, a How I Built This episode with Lily Kanter and Serena Dugan on building Serena & Lily from premium nursery bedding into a broader home brand. It adds Serena & Lily, Lily Kanter, Serena Dugan, Mill Valley Baby, Wendy Bellissimo, Maveron, Bad Money, Liquidation Preference Stack, and Inventory-Heavy Consumer Brand Financing, while extending How I Built This, Guy Raz, Founder Cash Flow Constraint, Direct To Consumer Cash Flow, Direct-to-Consumer Brand Control, Startup Governance, Financial Gravity, Investor Reference Checking, Consumer Brand Moat, Experiential Retail, Retail Site Selection, and Founder Role Transition. Its core synthesis is that product pull can increase financing danger in physical consumer brands: inventory, catalogs, packaging, stores, and board terms can turn fast growth into a cash crisis, and bad money can keep shaping acquisition and founder options long after the original check is spent.
The latest addition is A股的春夏秋冬：种树、种粮、种菜, a 面基 episode with 吴伟志 of 中欧瑞博 on A-share cycles, method fit, company taxonomy, and institutional research execution. It adds 吴伟志 / Wu Weizhi, 中欧瑞博 / Zhongou Ruibo, A-Share Market Seasons / A股四季框架, A-Share Planting Company Taxonomy / 种树种粮种菜, Investment Strategy Fit / 投资策略适配, Growth Investing / 成长投资, MAPER Investment Research Framework, and Research Index Portfolio Construction / 投研指数化, while extending 面基, A-Share Bull Market History, Investment Risk Management, Position Sizing, Value Investing, Trend Following, Quantitative Investing, AI Investment Research, Passive Investing, AI Equity Valuation Risk, Market Breadth Narrowing / 市场广度收窄, Defensive Dividend Assets, Asset Allocation, Investment Style Map / 投资流派地图, Multi-Strategy Allocation, and Institutional Investor Process Discipline / 机构投资者流程纪律. Its core synthesis is that A-share investing is a matching problem: market season, company type, strategy style, position size, research depth, and organizational process have to fit one another, or investors turn useful methods into contradictory slogans.
The latest addition is Taken littorally: Spain’s sudden crisis in Ceuta, a The Intelligence episode introduced by Jason Palmer on Ceuta’s migration surge, falling-object deaths in India, and AI Writing Detection. It adds Ceuta, Morocco, Algeria, Western Sahara, Polisario Front, Brahim Ghali, Pedro Sanchez, Michael Reid, Giorgia Meloni, Pangram, Migration Weaponization, Border Legal Signal, Falling Object Public Safety, Infrastructure Accountability Gap, and AI Writing Detection, while extending Spain, European Union, Schengen Area / 申根区, Immigration Backlash Cycle, India, Leo Mirani, Caitlin Talbot, AI Writing Pedagogy, AI Detector Bias, ChatGPT, Claude, Gemini, and Grok. Its core synthesis is that legal signals and weak enforcement can scale quickly when institutions are already under strain: a court ruling can become border behavior and diplomatic leverage, formal infrastructure rules can fail to prevent everyday lethal hazards, and AI detector scores cannot replace editorial judgment.
The latest addition is 期权这张饼，为什么越来越难吃了？, a Keji Luandun episode on why employee options have become a weaker compensation promise. It adds Employee Stock Options / 员工期权, Employee Stock Option Liquidity Risk / 员工期权流动性风险, Variable Interest Entity / VIE 架构, VIE Employee Equity Mismatch / VIE 员工股权主体错位, Option Pool Recapture Incentive / 期权池回收激励, and Restricted Stock Units / RSU, while extending Option Contract Mechanics, Equity Compensation Upside, Equity Incentive Alignment / 股权激励相容, Employee Equity Communication, Paper Wealth Vs Cash Value, Workplace Incentive Design, Startup Governance, Trust As Business Asset, 土豆网 / Tudou, Zynga, Uber, and Baidu. Its core synthesis is that options are not automatically alignment or upside: once growth slows, IPO paths stretch, VIE entities diverge, tax and exercise costs bite, and option pools become valuable enough to fight over, employees have to discount headline equity by liquidity, enforceability, and company trust.
The latest addition is 177: 详解Kimi K3：强到冲击Anthropic估值的模型什么样？, a LateTalk technical episode with 赵晨阳 and 曾志远 on Kimi K3. It adds Zhao Chenyang / 赵晨阳, Zeng Zhiyuan / 曾志远, Redix ARK / Radix ARC, AgentIn, Kimi Linear, Kimi Delta Attention / KDA, Attention Residues, NoPE / No Position Encoding, Quantile Balancing, Per-Head Muon, Kernel Development Agents, MOPD Post-Training, On-Policy Distillation, and Agent Environment Isolation, while extending Kimi, Moonshot AI / 月之暗面, LateTalk, Anthropic, Dario Amodei, DeepSeek, Fable 5, Moore Threads / 摩尔线程, Open Weight Release Boundary, Open Source AI Models, Chinese Open-Weight AI Strategy, Closed Model API Moat Pressure, Open Model Safety Governance, AI Inference Cost Structure, Agent Inference Workload, Prefix Caching, Model-Infra Co-Design, Mixture of Experts, Model Distillation / 模型蒸馏, Recursive Self-Improvement, Agent RL, Model Harness Co-Evolution, AI Programming Engine Shift, and AI Coding Verification. Its core synthesis is that K3 matters less as a single benchmark shock than as a stack-level model case: hybrid attention, sparse routing, optimizer stability, kernel agents, post-training merges, agent sandboxes, and serving-state lifecycle all show why open weights can pressure closed APIs while still leaving the repeatable model factory partly closed.
The latest addition is In arms’ way: Gaza-deal sticking points, a The Intelligence episode on Gaza-deal sticking points, America’s small-business boom, and radio’s resilience across Africa. It adds Board of Peace, National Committee for Administration of Gaza, International Stabilization Force, International Verification Committee, Afrobarometer, Joy FM, Egypt, Uganda, Senegal, Gaza Deal Implementation Gap, Disarmament Withdrawal Sequencing, Gaza Technocratic Administration, AI-Enabled Self-Employment, Full-Time Self-Employment Boom, African Radio Resilience, Low-Data Media Access, and DAB Radio Capacity Expansion, while extending The Intelligence, Economist Podcasts, Gaza, Hamas, Israel, Donald Trump, Benjamin Netanyahu, Morocco, Qatar, Turkey, Kenya, Ghana, Sam Altman, Thin Diplomatic Frameworks, Israeli Election 2026, Palestinian Question in Israeli Politics, Entrepreneurship Infrastructure, One-Person Company, AI As Business Operator, and Africa Connectivity Infrastructure. Its core synthesis is that implementation is often the real story after a headline: a Gaza plan still depends on sequencing, verification, territory, and election incentives; a self-employment boom still depends on whether AI and infrastructure create durable businesses rather than only filings; and African radio persists because cheap, low-data access can coexist with digital amplification.
The latest addition is 176.纽约一年：一个悲观主义者的活法｜725沙龙实录, a 起朱楼宴宾客 live 725沙龙 conversation hosted by 欧阳斌 of 纽约漫谈录 with 大卫翁 on one year in New York City. It adds 纽约漫谈录 / Niuyue Mantanlu, New York Salad-Bowl Multiculturalism / 纽约沙拉式多元, Uncertain-Era Optionality / 乱纪元选择权, Split Globalization / 分裂的全球化, Base Currency Thinking / 本币思维, and Cycle Worldview / 周期世界观, while extending 起朱楼宴宾客 / Qizhulou Yan Binke, 大卫翁 / David Weng, 欧阳斌 / Ouyang Bin, 725沙龙 / 725 Salon, New York City, United States, Japan, China, Asset Allocation, Structure Over Prediction / 结构优先于预测, Life Antifragility, Investment Worldview Fit, Currency Risk, Entry-Level AI Career-Ladder Risk, Investment Cooldown Discipline, China Excess Savings Reallocation / 中国超额储蓄再配置, Podcast-Enabled Career Path / 播客带来的职业路径, and Output-Forced Input. Its core synthesis is that pessimism can be operational rather than fatalistic: New York teaches coexistence without fusion, globalization can retreat in supply chains while accelerating in culture, and ordinary people respond to “乱纪元” by preserving asset, currency, mobility, skill, and attention choices instead of pretending to forecast the next shock.
The latest addition is E247｜对话盛颖：xAI，Infra的浪漫，SGLang，开源，平权与“甄嬛传”, a 硅谷101 episode with 盛颖 on SGLang, xAI, and Redix ARK. It adds 盛颖 / Sheng Ying, SGLang, LM-SYS, LM Arena, 连敏 / Lian Min, Clarke Barrett, Two Sigma, AI Infrastructure As Product, Radix Attention, Day-Zero Model Support, Open-Source AI Democratization, and SMT Solver, while extending 硅谷101, xAI, Grok, Redix ARK / Radix ARC, Google, Stanford University, Columbia University, Model-Infra Co-Design, Open Source AI Infrastructure, Inference Acceleration Stack, Agent Inference Workload, Prefix Caching, Agent RL, Formal Verification, Research Taste, Open Source Community Commercialization, AI Infrastructure Full-Stack Moat, Open Source Infrastructure Trust, High Responsibility Density, and Gendered Creator Confidence. Its core synthesis is that AI infrastructure is becoming a product and power-distribution layer: inference engines, prefix-cache structures, RL rollout systems, launch-day model compatibility, open communities, and company formation all shape who can build with strong AI rather than merely consume closed models.
The latest addition is Sand heists and property rights in the Caribbean (Summer School), a Planet Money Summer School episode on the Caribbean through Barbuda property rights and Jamaica sand theft. It adds Caribbean, Barbuda, Antigua and Barbuda, Jamaica, Damian King, Caribbean Policy Research Institute, Gaston Browne, John Mussington, Natalia John, Albert "Patty" Simon, Michael Hilton, Pascal Peduzzi, Karen Duffin, Hurricane Irma, Small Open Economy Vulnerability, Communal Land Tenure, Barbuda Land Act, Disaster Recovery Privatization, Property Rights As Investment Incentive, Path Dependence, Sand Scarcity, Illegal Sand Mining, Sand Theft Externality, Tourism Resource Protection Funding, and Small-State Enforcement Capacity, while extending Planet Money, Sarah Gonzalez, Singapore, United Nations, Tourism-Dependent Small Economy, Externality Internalization, Property Rights And Community Obligations, and Tragedy of the Commons. Its core synthesis is that institutions and resources become legible under pressure: small island economies reveal vulnerability to outside shocks, Barbuda shows how communal land can be both protection and investment constraint, and Jamaica’s missing beach turns sand into a scarce, hard-to-police externality rather than a background material.
The latest addition is Taking the shine off: albedo and global temperatures, a The Intelligence episode on falling planetary reflectivity, West Africa cocaine routing, and Female Breadwinner Norms in American couples. It adds Paul Crutzen, International Maritime Organization, Argo floats, West Africa, Liberia, Monrovia, Sierra Leone, Ori Ogunbiyi, Paulo Silva, Maritime Analysis and Information Centre, Gregory Coleman, Albedo Feedback, Earth Energy Imbalance, Aerosol Cooling Masking, Ocean Heat Monitoring, Solar Geoengineering, West Africa Cocaine Transit, Cocaine Route Displacement, Maritime Drug Interdiction Capacity, Narco-State Risk, Female Breadwinner Norms, Reported Income Gender Norms, and Domestic Labor Double Burden, while extending Oliver Morton, NASA, Carbon Removal, China, Brazil, Ghana, Western Sahara, European Union, United States, Taylor Swift, Travis Kelce, Doug Dawson, World Values Survey, Women Workplace Progress Stall, Family Labor Boundaries, Gender Power And Uncertainty Cost, Age-Gap Marriage Decline, and Small-State Enforcement Capacity. Its core synthesis is that hidden system balances matter: cleaner air can reveal warming once masked by aerosols, interdiction can move cocaine routes into weaker maritime institutions, and egalitarian survey answers can coexist with household status and care-work norms that still penalize female breadwinners.
The latest addition is AI-driven law could be an answer to accessible legal help, a Marketplace Tech episode with Megan McCarty-Corino interviewing Benjamin Alarie of the University of Toronto on AI and accessible legal help. It adds Benjamin Alarie, University of Toronto, Super Justice, AI Access To Justice, Personalized Legal Guidance, and Legal AI Verification And Auditability, while extending Marketplace Tech, Megan McCarty-Corino, Legal AI Hallucination, Vibe Lawyering, Human-In-The-Loop Legal AI, AI Verification, Human Judgment Under AI, AI Governance And Compliance, and Civil Liberties Surveillance Risk. Its core synthesis is that legal AI’s access promise depends on accountable abundance: AI may make legal reasoning and legal help less scarce, but the system still has to preserve equal rights, verifiable outputs, audit trails, privacy limits, and professional responsibility.
The latest addition is No.213 景德镇往事：千年“世界瓷都”的前世今生, a 半拿铁 episode on 景德镇 as a thousand-year porcelain production system. It adds Jingdezhen / 景德镇, Jiangxi / 江西, Fuliang / 浮梁, Hutian Kiln Site / 湖田古瓷窑址, Gaoling Village / 高岭村, Changling Porcelain Stone Mine / 长岭瓷石矿遗址, Jingdezhen Imperial Kiln Factory / 景德镇御窑厂, David Vases / 大维德瓶, Tang Ying / 唐英, Tong Bin / 童宾, Zhushan Eight Friends / 珠山八友, Jingdezhen Ceramic University / 景德镇陶瓷大学, Lotte Pottery Workshop / 乐天陶社, Taoxichuan / 陶溪川, Jingdezhen Porcelain Production System / 景德镇瓷业生产系统, Qingbai Porcelain / 青白瓷, Kaolin / 高岭土, Porcelain Stone-Clay Binary Formula / 瓷石瓷土二元配方, Yuan Blue-and-White Porcelain / 元青花, Official Supervision, Civilian Kilns / 官监民烧, Distributed Ceramic Production / 分布式瓷业生产, Imperial Kiln System / 御窑体系, Imperial Kiln Waste Control / 御窑落选品控制, Export Porcelain Customization / 外销瓷定制, Ceramic Industrialization Transition / 陶瓷工业化转型, State-Owned Ceramic Factory System / 国营瓷厂体系, Jingpiao Creative Ceramic Ecology / 景漂创意陶瓷生态, Ceramic Livestream Commerce Risk / 陶瓷直播电商风险, Advanced Ceramics Industrial Upgrade / 先进陶瓷产业升级, and China-Changnan Etymology Myth / China源于昌南说, while extending Material History Narrative, Process Knowledge / 过程知识, Heritage Tourism Commercialization, Long-Term Place Operation, and 半拿铁. Its core synthesis is that porcelain history is a city-scale system: geology, formula, waterway logistics, official orders, export demand, craft labor, factory organization, market transition, creative districts, livestream trust, and advanced materials all determine whether “world porcelain capital” remains a living capability rather than a slogan.
The latest addition is Vol.269 小历史 | “不要怕，是技术性调整”, a 商业就是这样 “小历史” episode on the 置地饮牛奶 takeover and the 1973年香港股灾. It adds Hongkong Land / 香港置地, Dairy Farm International / 牛奶公司 / DFI, Jardine Matheson / 怡和, Chinachem Group / 华懋, Henry Keswick / 亨利·凯瑟克, Zhou Xinian / 周希年, Phil Oram, Hang Seng Index / 恒生指数, South China Morning Post / 南华早报, Tender Offer Media Campaign / 收购要约媒体战, Retail Shareholder Tender Mobilization / 散户要约动员, Bonus Share and Stock Split Misreading / 拆股送股误读, Four-Exchange Hong Kong Market Fragmentation / 香港四会并列市场碎片化, Hong Kong Dollar Hot-Money Cycle / 港元热钱周期, and Conglomerate Control in Colonial Hong Kong / 殖民期香港财团控制, while extending Hong Kong / 香港, Hong Kong Market Structure, Retail Bull Market Psychology, Retail Investor Crowding, Speculative Bubble Psychology, Currency Anchor Transition / 货币锚转换, Investor Education, and Market Speculation. Its core synthesis is that the 1973 crash is clearer when read through one concrete takeover: company control, paper-shareholder logistics, newspaper persuasion, split/bonus-share mechanics, fragmented exchanges, retail chasing, and currency inflows converted a corporate battle into a market-wide bubble amplifier.
The latest addition is Advice Line: "Strategy Sessions", a How I Built This Advice Line mashup with Guy Raz, Jeffrey Hollender, Sarah LaFleur, and Shazi Visram advising Tony DeRosa of Hearsay Brewing and Theater, Monica Stoney of Tress London, and Sandy Dininger of Brain Buffs. It adds Sarah LaFleur, M.M. LaFleur, Tony DeRosa, Hearsay Brewing and Theater, Monica Stoney, Tress London, Sandy Dininger, Brain Buffs, Control-Preserving Incentives, Neighborhood-First Word of Mouth, Direct-Teacher Edtech Pilot, and Edtech Data Value, while extending How I Built This, Guy Raz, Jeffrey Hollender, Shazi Visram, Equity Incentive Alignment / 股权激励相容, Workplace Incentive Design, Stage-Appropriate Hiring, Startup Governance, Relationship-Led Growth, Direct-to-Consumer Brand Control, Fast Product Validation, and Product Led Willingness To Pay. Its core synthesis is that founders can protect strategic optionality by testing before committing: test operators before equity, neighborhood/tactile proof before broad marketing, and teacher evidence before locking in edtech pricing.
The latest addition is AI chatbots have linguistic slips when they go off-script, a Marketplace Tech “Uncanny AI” episode with Megan McCarty-Corino and Janelle Shane on why chatbots sometimes insert foreign-language words into otherwise fluent output. It adds AI Weirdness Blog, Chatbot Domain Bleedthrough, Chatbot Code Switching, and Chatbot Self-Explanation Uncertainty, while extending Marketplace Tech, Megan McCarty-Corino, Janelle Shane, Claude, ChatGPT, LLM Statistical Boundary, AI Interaction Internationalization, Language-Dependent AI Bias, Customer Support Automation, AI Toy Companionship / AI玩具陪伴, and Human Judgment Under AI. Its core synthesis is that chatbot fluency should not be mistaken for clean language or domain control: multilingual training data, mixed text formats, and probabilistic next-token behavior can produce visible code-switching or subtler drift, and a chatbot’s own explanation for the slip remains only a hypothesis without independent diagnostics.
The latest addition is Socialist mobility: a telling Michigan vote, a The Intelligence episode on Abdul El-Sayed’s Michigan Democratic primary win, Yangtze Fishing Ban Recovery, and Postcard Media Decline. It adds Michigan, Abdul El-Sayed, Haley Stevens, Mike Rogers, Gabriel Crossley, Democratic Socialists of America, United Auto Workers / UAW, Royal Mail, English Heritage, John Godfrey, Theodore Hook, D-Day, Three Gorges Dam, Chinese paddlefish, Statewide Left Primary Test, Primary Electorate Distortion, Yangtze Fishing Ban Recovery, Environmental Repair Social Cost, Postcard Intelligence Infrastructure, and Postcard Media Decline, while extending The Intelligence, Economist Podcasts, John Prideaux, Catherine Nixie, Democratic Party, United States, Donald Trump, Israel, China, Xi Jinping, Yangtze River / 长江, Baiji / 白鱀豚, George Orwell / 乔治·奥威尔, Environmental Tradeoff Accounting, Conservation Intervention, Analog Media Preservation, Populist Democratic Energy, and American Democratic Resilience. Its core synthesis is that institutional change has to be judged beyond the headline: a left primary win still faces statewide electability, a fishing ban can restore biomass while damaging livelihoods and missing extinct species, and postcards can shift from global social network and military intelligence source to rare analog memory.
The latest addition is 贾扬清：我所经历的「人工智能已死」到「AI 颠覆世界」的数年巨变丨串台「声东击西」S10E24, a What’s Next｜科技早知道 and 声东击西 crossover in which 徐涛 interviews 贾扬清 on the AI arc from an unfashionable academic label to infrastructure, products, startups, and agentic work. It adds Jia Yangqing / 贾扬清, Caffe, Google Brain, Facebook AI Infra, Lepton AI, Agent Reliability Verification, and What Over How Work Shift, while extending TensorFlow, PyTorch, Alibaba Cloud / 阿里云, Nvidia, DeepSeek, AI Verification, AI Coding Verification, Agent Harness, Multi-Agent Collaboration, AI Programming Engine Shift, AI Infrastructure As Product, Neo Cloud, Human Judgment Under AI, AI Organization Design, and Large Company Organizational Inertia. Its core synthesis is that AI progress becomes durable only when models are embedded in infrastructure, product loops, verification harnesses, and organizations that know what results they want; more compute, more agents, or more code generation do not remove the need for human criteria and external checks.
The latest addition is 178: 与田渊栋聊 RSI：模型自进化如何到来？, a LateTalk interview with 田渊栋 on Recursive Self-Improvement, Recursive Superintelligence, and how AI-assisted AI research might become a self-improving loop. It adds Tian Yuandong / 田渊栋 and AI Research Feedback Compression, while extending LateTalk, Recursive, Recursive Self-Improvement, Auto Research, AI For AI, ML Coding, Research Taste, AI Verification, Mechanistic Interpretability, Frontier Model Scaling, AI For Science, Discovery Model, Anthropic, OpenAI, Google, FAIR, and AI Organization Design. Its core synthesis is that RSI is not just stronger coding agents: execution speed and verifiable benchmarks make early loops possible, but deeper self-improvement still needs research taste, abstraction, direction choice, interpretability, and organization design; if model progress comes in S-curve-like plateaus rather than smooth scaling, startups and new methods may still matter.
The latest addition is Gig workers train humanoids on household chores, a Marketplace Tech Bytes episode with Joanna Stern on SpaceX public earnings, Household Robot Training Data, and possible MySpace revival. It adds MySpace, Household Robot Training Data, and Orbital Debris Governance, while extending Marketplace Tech, Joanna Stern, SpaceX, Starlink, Moon, Space Economy Infrastructure, Space Based AI Infrastructure, Embodied AI, Physical AI, Humanoid Robot Commercialization, AI Trainer Labor, Robot Data Scale Up, Household Robot Data Flywheel, AI Training Data Scarcity, Bluesky, Facebook, Mark Zuckerberg, Social Graph Moat, and Algorithmic Amplification / 算法放大. Its core synthesis is that technology ambition depends on hidden operating layers: SpaceX’s AI and lunar ambitions require capital and debris governance, household robots require physical chore data gathered from human workers, and social-media nostalgia points to frustration with algorithmic, centralized feeds rather than a simple desire to revive an old brand.
The latest addition is Outsider chance: ending Sudan’s proxy war, a The Intelligence episode on Sudan’s proxy war, the Gordie Howe Bridge dispute, and Betye Saar’s assemblage art. It adds Tom Gardner, Sudan, Khartoum, Rapid Support Forces, Sudanese Armed Forces, Darfur, El Obeid, Red Sea, Gordie Howe Bridge, Ambassador Bridge, Detroit, Windsor, Ontario, Betye Saar, Watts Towers, The Liberation of Aunt Jemima, Foreign-Backed Civil War, Fragmented Ceasefire Enforcement, Cross-Border Infrastructure Trade Tension, and Assemblage Art Memory Work, while extending The Intelligence, Economist Podcasts, Jason Palmer, Annie Crabill, Anne Rowe, United Arab Emirates, Saudi Arabia, Turkey, Egypt, Canada, United States, Donald Trump, Michigan, Proxy Conflict Spoiler Risk, Low-Cost Drone Warfare, Trade Reciprocity Protectionism, and Trump Victory-Narrative Deal-Making. Its core synthesis is that headline openings do not settle power constraints: Khartoum’s recapture does not end a foreign-backed and drone-saturated Sudanese war, a useful bridge can still become tariff and ownership theatre, and Saar’s found objects turn discarded material into historical memory and liberation work.
The latest addition is Older workers aren’t retiring. Should they be forced to?, a Planet Money episode on delayed retirement, generational mobility, and whether some roles should have age-based endpoints. It adds Ryan Hendrickson, BJ Swami, Samuel Moyn, Olivia S. Mitchell, Gerontocracy in America, Edward Lazear, Career Mobility Bottleneck, Mandatory Retirement Policy, Lump of Labor Fallacy, Phased Retirement Succession, Retirement Security Tradeoff, and Implicit Contract Retirement Theory, while extending Planet Money, NPR, Sarah Gonzalez, NASA, Reddit, Wharton School, Older Worker AI Retirement, Institutional Knowledge Transfer, and Employer-Bargained Benefits. Its core synthesis is that later retirement is not one labor-market story: older workers staying can preserve expertise and retirement security, but scarce senior roles can block succession, so the useful policy debate sits between fixed-jobs fallacy, real bottlenecks, pensions, and phased handoff.
The latest addition is EP273 NPC还是NPD，都别想拿走我们人生的遥控器, a Talk三联 episode with psychotherapist 龚学平 on popular NPD language, painful relationships, and taking back one’s “life remote control.” It adds 龚学平 / Gong Xueping, Popular NPD Discourse / 大众NPD话语, Narcissistic Relational Control / 自恋式关系控制, Self-Worth Judgment Recovery / 价值判断权收回, Childhood Survival Rule Generalization / 童年生存规则泛化, and Intergenerational Pattern Pause / 代际模式暂停键, while extending Talk三联, 三联生活周刊 / Sanlian Life Weekly, Psychological Abuse Recognition / 精神虐待识别, 相信自己的感受 / Self-Feeling Trust, Therapy Relationship And Boundaries, Protection As Control, Communication Boundary Setting, Psychic Vampire Personality / 精神吸血鬼人格, and Emotion Regulation Toolkit / 情绪调节工具箱. Its core synthesis is that popular NPD talk should not become amateur diagnosis, but it can help people name control, emotional depletion, and self-doubt; the practical repair path starts when a person trusts discomfort as data and reclaims judgment through small, concrete actions.