Personal Reflection

Prologue: A slack surface, and high density underneath In June I came back from Laos to Shenzhen. If you only look at the state, this was a slack month. What it looked like in practice — I would only come alive after dark, stay up to two or three a.m. every night, lulled to sleep by one short video clip after another. Waking up in the day to Shenzhen’s gray-white sky, I could not even gather the strength to leave the apartment. I told myself it was the weather, but inside I knew it was not just that. Every day I bargained with myself: tomorrow I will wake early, I will be disciplined, I will start working — and the next day I lost again to that version of me curled up in bed. The body felt empty, not tired-empty but hollowed-out-and-idling-empty. There were many bothers, but I could not point at any specific one. ...

Prologue: A May Without flomo For the entire month of May, I did not write a single flomo note. I barely posted to my feed either. This was not laziness. It was a conscious decision. I had already decided on April 30 that I would stop. The reason is simple: I no longer felt I needed it. Or more accurately — could it have a better form? I want to use my own product. It is not perfect yet, but I want to fully think through and ship the thing I have in my head. The remaining steps are few. It is close. ...

2026年4月思考笔记 AI 知识时效性：从 RAG 到元认知的解决思路 2026 年 04 月 30 日 - 14:46:26 有些工具现在用的是时效性的方法 比如说是通过一些 API 或者 RAG 系统挂在最新的文档 或者是 web search 最新的数据源 元认知层，让模型知道自己不知道，这是更难的问题 显式标注知识截止日期 训练模型在时效敏感问题上主动表达不确定性 系统提示里注入当前日期，让模型对比自身知识边界 时间戳溯源，每一条知识附带获取时间，超时自动触发重新验证 越旧的信息置信度越低，触发 fallback 到搜索 对高频变动领域（金融、法律、医疗）设置强制校验节点 一个模型生成答案，另一个专门质疑时效性 #格物/ai 利他性记录：有意义才有持续驱动 2026 年 04 月 30 日 - 14:07:21 利他性的前提，是这件事情对他很有意义 这个平台很 有记录的价值 可以展现，或者发挥自己的特长 这样可以激励用户去持续的记录和创作 旅途影响的涟漪效应 2026 年 04 月 30 日 - 13:33:53 身边受到自己影响的旅游的影响的人好像也有一些哈哈哈 日本人的规则感：老挝砍价与文化观察 2026 年 04 月 30 日 - 13:24:11 ...

Preface: This article was written on a night in Lhasa. At an altitude of over 3,000 meters, the air is thin and the stars are very close. One evening I stood in the courtyard, alone, and watched the moon in the sky for a long time. That moon doesn’t lose a bit of its light because of AI’s existence. The moonlight is still truly beautiful — What do humans still possess in the AI era? ...

February 2026 Thought Notes | 2026 年 2 月思考笔记 Total: 365 notes this month | February 1-28, 2026 Monthly Themes: Travel, Ecology, Urban Observation Core Topics: Travel Experiences (~80 notes), AI Products (~45 notes), Ecological Protection (~25 notes), Urban Governance (~20 notes), Buddhism & Philosophy (~30 notes), Self-Discovery (~40 notes) Selected Notes of the Month Desire & Resources: The Dilemma of Wildlife Protection 2026-02-01 21:31:47 Wildlife protection is truly a massive challenge. How to protect wild ecosystems—I feel this is incredibly important. Because it’s not just about development and economics; it’s something more fundamental. ...

January 2026 Thought Notes | 2026 年 1 月思考笔记 Total: 388 notes this month | January 1-31, 2026 Monthly Themes: AI Deterministic Value, Finance Patterns, Self-Exploration Core Topics: AI Products & Agent Design (55 notes), Finance & Investment (19 notes), Self-Discovery (11 notes), Biology (10 notes), Social Observation (7 notes), Buddhism (6 notes) Selected Notes of the Month A-Share Industry Rotation: “Three-Times” Rule 2026-01-01 16:05:09 The general rule of A-shares (RMB ordinary stocks) is “shi bu guo san” (事不过三)—an industry rarely rises for three consecutive years. ...

December 2025 Thought Notes | 2025 年 12 月思考笔记 Total: 380 notes this month | December 1-31, 2025 Monthly Themes: Pilgrimage, Inner Observation, Year-End Reflection Core Topics: Kumano Kodo Pilgrimage (~40 notes), Zen Buddhism & Philosophy (~35 notes), Self-Discovery & Personality (~30 notes), Japanese Culture & Architecture (~25 notes), AI & Technology (~30 notes) Selected Notes of the Month Kumano Kodo: History Accumulated Step by Step 2025-12-01 17:04:20 Kumano Kodo pilgrimage, immersive experience. The most rewarding part of solo hiking is being with yourself—just focus on yourself and the path beneath your feet, walk naturally without needing to “break through” yourself or nature. Respect yourself, respect nature. ...

November 2025 Thought Notes

2025年10月思考笔记 嘉兴的潜力：顶级都市圈间的“白纸” 2025 年 10 月01 日 - 00:18:51 宁波的世界级港口真的是难以撼动的生产力 并且以强大的制造业兜底 相比较而言，嘉兴位置绝佳，两个顶级都市圈中间 长三角核心区不可或缺的超级节点 相对来说也具有强大的不确定性 未来的潜力很大 嘉兴是一个可以填满故事的白纸 #格物/城市发展 高层建筑的潜在危机：早衰与高昂的维修成本 2025 年 10 月01 日 - 00:25:16 高层楼也是最早面临危机的 高层楼的维修成本过于高了 现在都还没显露出来，但是却不断存在危机的 许多中国的高层建筑正经过早老化。在其他国家通常需要几十年才会出现的问题，据报道在中国仅建成不到10到15年的建筑中就已经出现。这些问题包括： 外墙恶化： 外墙瓷砖和面板的开裂、剥落和坠落变得越来越普遍，对楼下的行人构成了安全隐患。 渗水漏水： 来自屋顶、窗户和内部管道的持续渗漏导致水渍损坏和霉菌滋生，既损害了结构完整性，也破坏了居住环境。 系统故障： 电梯作为任何高楼中的关键动脉，正经历着更频繁的故障。管道和电气系统也显示出紧张的迹象，导致服务中断和火灾风险增加。 现在的无人机、科技、远程办公已经在修改游戏规则，其实哪怕是偏远的郊区独栋别墅通勤劣势将被大大削弱，其空间和环境优势会更加凸显 #格物/城市观察 K字签证：吸引全球STEM人才的新策略 2025 年 10 月01 日 - 09:35:52 K 字签证 针对全球的 STEM（科学、技术、工程、数学）领域人才的新型签证 吸引外国高科技人才来华工作、创业或交流 该签证不要求国内雇主担保，只需满足年龄、教育背景或工作经验等基本门槛，即可获批多入境、停留期长达5年，并可直接通往永久居留 这个签证像是应对美国提高H-1B签证费用（高达10万美元）的战略回应，旨在提升科技自给自足和全球人才竞争优势，但是同时引发国内就业担忧和对实际吸引力的质疑 同样的可以吸引更多的人才进入中国，并且加快内部的消费 但是实际上也加剧了本地的就业压力 #格物/时事 Grok的产品思考：追问推荐的设计很一般 2025 年 10 月01 日 - 09:45:58 grok 总感觉也有一些追问的推荐，但是按钮选项总归感觉很一般设计的 后面也会推荐一些选项，比如说一些书籍推荐，或者是案例的推荐 ...

2025年9月思考笔记 flomo 标签新设计：心智分类法 2025 年 09 月01 日 - 08:25:46 设计的核心，不是按照学科（哲学、AI）分类，而是按照记录时的心智状态去分类，关注的是，当时的自己是观察内在、探究外物，还是捕捉灵感。 三大支柱： 观我：向内探索，关于人的思考，包括个人成长相关的，精神世界的内省与反思。 格物：向外探究，关于物与事的思考，包括 AI 、软件工程、产品设计、商业模式等客观世界的分析。 知世：旅居中的见闻，对世界的理解，旅居的思考，文化的思考。 一闪：闪过的灵感，想法和金句。 后面可以适当的再加一个标签，比如说项目的名称。 #格物/flomo 科技狂奔与人性缓行：我们跟得上 AI 吗？ 2025 年 09 月01 日 - 08:36:02 不免有些恐慌：AI 发展的这么快，我们这代人能跟上它的速度吗？ 从蒸汽机、电力、互联网到如今的人工智能，技术不断扩展人类的能力，拓展了我们的感官和行动边界。但人性的发展却像一条缓慢的河流，我们的情感模式、欲望、恐惧、偏见，和几千年前并无太大不同。 AI 能写诗、能生成视频，可人类真的有足够的智慧去驾驭这样的科技吗？人性准备好了吗？又该何时才能准备好？ #观我/AI 河内印象：强烈的民族正统感与爱国氛围 2025 年 09 月01 日 - 10:08:14 河内的国庆氛围比想象中更浓厚。 这里的人说话、做事，似乎总带着一些“民族正统感”，爱国表现也更官方、更热烈。越南虽小，但能在强国夹缝中生存并保持独立，这种“民族自豪感”或许天然地强化了人们的“爱国”身份认同。 #知世/越南 盲从权威的代价 2025 年 09 月01 日 - 12:18:40 为什么长期盲从权威的人，更容易滑向“只能相信”的境地，最终成为权威的奴隶？ 我们依赖权威是为了降低认知负担，但这需要一个前提：对事情本身有基本的认知。比如，我们相信飞行员的专业性，但也应了解飞机可能出现意外的客观现实。 当信息真假难辨时，其中可能夹杂着大量的错误和偏见。这包括 AI 的回复，也可能掺杂着各种潜在的错误信息。 一个有效的策略是：根据信息来源的权威度进行加权判断，同时对来自不同信源的信息进行交叉验证，寻找共识点和矛盾之处。 #格物/思维模型 概念、理论与理念：思想的三层阶梯 2025 年 09 月01 日 - 14:17:43 ...

August 2025 Thought Notes The Unexamined Life and Game Theory Applications August 1st, 2025 - 07:23:45 Socrates said “the unexamined life is not worth living.” But what does examination actually require in practice? Self-examination isn’t just introspection—it’s systematic analysis of our own patterns, motivations, and decision-making processes. Like debugging code, but for consciousness. Game theory as a lens for life decisions: Every interaction can be understood as a game with players, strategies, and payoffs. But unlike abstract economic models, real-life “games” involve: ...

July 2025 Thought Notes Vibe Coding: Are Human Experiences the Shackles of AI Agents? July 1st, 2025 - 11:39:24 In Agent development, “The Bitter Lesson” remains equally important. Excessive human experience injection might actually limit AI performance and lock in product possibilities. Embrace Agent Patterns: Agent mode is mature; what we need is training agents’ thinking patterns. We can focus more on ideas and design, becoming architects rather than coders. Beware Over-Prescription: Excessive rule-setting (like fixed testing processes) may prevent AI from utilizing the latest model capabilities. We should let AI self-select more often. Context-Dependent Experience: Experience has scenario dependency. Over-reliance on experience constrains LLM performance; moderate injection as “heuristic guidance” is sufficient. This insight challenges traditional software engineering approaches in the AI era. The “bitter lesson” teaches us that general methods leveraging computation ultimately outperform human-designed heuristics. In agent development, this means resisting the urge to over-engineer constraints based on our past experiences. ...

June 2025 Thought Notes June 2025 Thought Notes Boundaries of Rejection: Finding Balance in Social Interactions June 2nd, 2025 - 06:28:12 In a Korean fried chicken restaurant, unexpectedly surrounded by Chinese people. The meal sets were all for two people, so I ordered two combo meals for myself. When ordering, I asked if I could pay by card, wanting to avoid using cash as I felt my cash wasn’t quite enough. ...

May 2025 Thought Notes AI-Era Independent Development: Frontend as the Main Battlefield for Value Delivery May 1, 2025 - 10:22:06 In the AI era, the battlefield of independent development is shifting toward the frontend. The transformation is evident: Backend Simplification: AI models turn core services into universal capabilities, lowering technical barriers in backend development. SaaS and PaaS providers encapsulate vast foundational capabilities that developers can directly invoke. Frontend as Value Delivery Nexus: The frontend is closest to users, making it easier to focus on MVP (Minimum Viable Product) and single-person product loops. Using a tech stack like Next.js / Tailwind / Firebase / Vercel enables full-stack delivery and rapid validation of the “function → value → growth” chain. From “Building” to “Integration”: Databases, permissions, storage, and model calls are becoming platformized (such as Supabase, LangChain, Replicate). Independent developers no longer need to write backend from scratch—they just need to “integrate.” AI decouples “intelligence” from the backend, making the frontend the true battlefield where users are closest, changes are most significant, and value is most concentrated. ...

April 2025 Thought Notes Learning & Growth Projects & Work Challenges & Solutions Insights & Reflections Goals & Planning Books, Articles & Resources Quotes & Inspiration Next Month’s Focus

March 2025 Thought Notes Learning & Growth Projects & Work Challenges & Solutions Insights & Reflections Goals & Planning Books, Articles & Resources Quotes & Inspiration Next Month’s Focus

During a 48-hour engineering task, I wrote down one sentence: “At this point, I’ve spent half the time designing, researching, understanding the requirements, judging the boundaries, and developing my taste and intuition. I haven’t started writing code.” The task began as a limited feature. After I kept discussing it with AI, it quickly grew a product model, system boundaries, and a complete design. Every step had a rationale. Every piece connected to the next. I could feel myself thinking faster and farther, holding more layers in mind at once. ...

At the beginning of 2026, my GitHub contribution graph stayed nearly blank for several weeks. A few months later, the same graph showed hundreds of contributions in a single week. In July, this blog published 55 articles in Chinese , while GitHub’s contribution calendar recorded 733 contributions. Those numbers prove neither quality nor user value, much less a business result. GitHub also makes clear that contributions may include qualifying commits, pull requests, issues, reviews, and other activity; the number cannot simply be relabeled as a commit count. GitHub’s profile contributions reference ...

At dusk, I sat on the rooftop of Youli Café and raised my coffee cup toward the twin pagodas of Kaiyuan Temple. The camera recognized the distant towers and ignored the hand in front of it. Cup and fingers dissolved into a red blur; across the old city, the pagodas held their edges. ...

This is a thought experiment-style simulated interview. The respondents are composed of organizational psychology and organizational economics perspectives, and do not represent any real individuals’ original statements. In my June 2026 monthly review, I wrote: Incentive structures colonize personality. At that time, I studied several founders and observed myself. An organization initially selects a particular narrative, user model, and revenue framework primarily to launch a business. Years later, that structure reverses to filter what members can see, what they are willing to acknowledge, and which behaviors persist. ...

I had only asked my second brain to organize the thoughts of a friend. It placed scattered notes in order by time, theme, and cause, then tried to explain how she thought about work, money, and freedom. A few thousand words later, I looked at the resulting portrait and realized that the person it had revealed most clearly was not her. It had shown me why I never stop building systems. ...

This piece is a thought experiment in simulated interviews. The respondents are synthesized from existential philosophy and behavior design perspectives—this is not verbatim dialogue from any real person. One June day, perhaps like this. I woke around noon. My phone lay on the pillow. First I checked the time, then the messages. Nothing was urgent, so I opened a short video. The algorithm did not require me to decide what came next. By the time I looked up, more than forty minutes had passed. ...

This piece is a thought experiment in simulated interviews. The interviewees are synthesized from field anthropology and documentary practice perspectives and do not represent any real person’s actual words. I work with AI, yet I am consistently moved by life outside of AI. The blackness of caves in Laos, the sting of a wound after a motorcycle fall, the wind along the Mekong, a stranger who read a story in my eyes, and the stretched-out time of walking instead of riding on the Four Thousand Islands—none of these experiences has model-like efficiency or can be reproduced with one click. ...

This piece is a thought experiment in the form of a simulated interview. Its two fictional voices combine perspectives from behavioral decision research and high-risk outdoor practice; they do not represent the words of any real person. Nothing here constitutes investment, medical, or outdoor safety advice. In May 2026, I injured my leg in Vang Vieng and still walked into the cave. Over the past two years, I’ve also been able to leave traditional systems, live long-term abroad, and place both income and identity within greater uncertainty. ...

This piece presents a thought experiment in simulated dialogue. The respondents are synthesized from phenomenological and psychotherapeutic perspectives and do not represent any real individuals’ words. It is not intended as psychological diagnosis or therapeutic advice. I excel at stepping back. Entering a city, I quickly see how its industries, spaces, and history shape life. Entering a relationship, I analyze attachment patterns, projections, scarcity, and familiarity. Developing a product, I first deconstruct systems, identify boundaries, and map information flow. ...

This piece is a thought experiment conducted through simulated interviews. The interviewees are composed of social psychology, editorial practice, and creator research perspectives and do not represent any real person’s original words. In November 2025, I recorded the phrase: “The creator’s loss: the scarcity of positive feedback.” In April 2026, I wrote: “Understanding generates positive feedback.” Together, these two sentences shift the question. Do creators truly seek only praise? When one hundred people like a piece but none articulate the contradictions it addresses, that void persists. Conversely, a single reader who precisely identifies how a detail altered their judgment—without public dissemination—might reignite the creator’s resolve to continue. ...

This piece is a thought experiment-style simulated interview. The respondents are composed of cognitive science, archival studies, and product design perspectives and do not represent any actual expert’s original words. In May 2026, I stopped using flomo. That month had no daily reminders, no faint sense of debt from “today’s notes still pending,” and no rapid conversion of thoughts into cards. Surprisingly, life didn’t thin out. Instead, Laotian cities, caves, injuries, strangers, and bars coalesced in late May into a more complete self-reading than fragmented pieces. ...

This piece is a thought experiment conducted through simulated interviews. The two interviewees are fictional characters synthesized from environmental psychology, behavioral science, and urban studies—representing no real individuals or viewpoints. In May, I explored caves, rode motorcycles, and met strangers in Vang Vieng. By June, back in Shenzhen, I often did not come alive until after dark, relied on short videos to lull myself to sleep at two or three in the morning, and lacked the energy even to step outside during the day. ...

This is a thought experiment-style simulated interview. The respondents are fictional characters composed of developmental psychology, field anthropology, and the author’s personal experience—no real experts are quoted. In December 2025, I wrote in my notes: Exploration no longer provides sufficient momentum. At first, moving to a new city wasn’t just a change of scenery. New languages, social structures, and rhythms would disrupt your existing self. Once your position loosened, your identity followed; once your identity loosened, possibilities multiplied. For a long time, “traveling” and “becoming someone new” were nearly synonymous. ...

This is the final essay in Reckoning with Reality. The first six considered profit, competence, action, feedback, and technology. The last returns to a question no model or market can decide for us: what kind of life am I willing to pay for? When we discuss a choice, we habitually list the returns. How large is the market? How high is the income ceiling? Will growth come faster? Can the path eventually produce a moat? The more complete the spreadsheet, the more rational the choice appears. ...