KNOWLEDGE NOTESKnowledge Notes
Explore curated knowledge, course notes and methods, with their existing content and sources preserved.
This index lists complete English editions. Additional Chinese notes and bilingual updates remain available in the Chinese index.
350knowledge notes2 types of Googlebot (Smartphone primary / Desktop minority) + crawl limits (2MB for HTML / 64MB for PDF / resources counted independently / calculated before decompression) + verification (reverse DNS). All file types Google can index (flat / encoded / media). 6 URL structure best practices + 4 common problems (faceted navigation / parameter abuse / infinite calendar space / relative links). Links must use `<a href>` to be reliably crawled; 5 anchor text guidelines. 3 mobile-first indexing configurations (responsive preferred / dynamic serving / separate URLs) + 13 troubleshooting errors. 9 supported meta tags + 4 unsupported ones (meta-keyword/lang/next-prev/nositelinkssearchbox). 8-element whitelist for valid HTML <head>; iframes/img tags cut off subsequent reading. 5-step crawl troubleshooting + soft 404 fixes + If-Modified-Since 304 optimization + emergency 503/429 crawl reduction. For website testing, do not cloak / use 302 + rel=canonical. AMP overview + Viewer vs Signed Exchange + 3 complete removal paths.
When you need a sitemap (>500 pages / new site / rich media) and when you don’t; how to choose among the 3 formats XML/RSS-mRSS-Atom/text; hard limits of 50MB / 50,000 URLs; sitemap index solution for large sites (500 per site); 4 extension namespaces and combined usage; full table of required and optional fields for image/video/news sitemaps; 4 submission methods; two approaches for cross-site submission; 13 anti-patterns.
Listen: 01:55–06:14. Classifying a cat or dog involves a labeled answer and one decision. Playing a game requires many choices—left, right, or jump—and several paths may win. Reinforcement learning, or RL, improves an action policy using rewards received after acting. It addresses processes with multiple possible choices whose quality becomes clear only later.
RC runs his company on Slock, growing from one person to seven people and forty agents. That describes the contemporary team, not an experimentally optimal ratio. More useful are his responses to duplicated tasks, misplaced context, drifting identity and limited human review capacity.
A synthesis of three perspectives on SEO × LLM positions. Covers Google Search Central (official: how-not-who framework + boundaries of scaled content abuse + the Who/How/Why three questions + AI Overviews / Succeeding in AI Search) / newtype comprehensive guide to llms.txt (two competing format proposals: the robots.txt-like Access Control Model + the Markdown-based Inference Guidance Model / Karpathy's "rewrite software infrastructure for Agents") / newtype knowledge system (wiki.newtype.pro third layer = publishing layer = proprietary "LLM-friendly corpus"). Covers Google's stance on AI content, the current state of the llms.txt proposal, compatibility between E-E-A-T and AI content, and infrastructure for the Agent era.
The episode places Sora’s first demonstrations within a larger question: why can compute and model narratives advance before many mature applications exist? The guests distinguish technical progress, competitive position and application demand.
Product–market fit, or PMF, means that a group of people repeatedly finds a product worth using or paying for. These three interviews show why no single number proves it for every business. An occasional office tool can first demonstrate that individual tasks merit payment. Professional software needs to participate in real delivery. A character entertainment product must explain why people return.
The introduction uses the listings of 智谱 and MiniMax as a three-year milestone, inviting the podcast’s first guest to review AI again. Investment outcomes and market descriptions remain contemporary perspectives. A listing cannot alone validate products, returns or every early hypothesis.
Integration of time management from 3 perspectives + personal OS. Covers yitang Truman's Compulsory Time Management Course (4 parts + 3 leaps + 30+ hypothetical techniques) / newtype Lao Huang (Life OS + three-layer knowledge system + 3 things that do not rely on self-discipline) / Wang Kai (thinking mode by default + absolute rationality + slow thinking). Includes Truman's 3 leaps in time management, Lao Huang's 5 generations of Life OS evolution, and 3 time blocks for deep work
The conversation begins with a contrast: people sound pessimistic while working harder. Perhaps effort has not disappeared; familiar rewards simply arrive more slowly. If so, the task extends beyond changing mood to recognizing feedback cycles, revising evaluation and preparing capabilities.
Qu Kai asks Zhang Hailong, who has more than twenty years of programming experience, to explain the excitement around Cursor. A useful distinction emerges: helping a person operate faster and accepting a task for independent completion are different product promises.
An in-depth Chinese-English typography reference based on Google Fonts' "Knowledge" column, which breaks down typefaces from "looks good by feel" into actionable judgment that PMs can follow when selecting type and setting type. It covers eight main sections: ① Type anatomy and vertical metrics (baseline / x-height / ascenders & descenders / em square · why baselines shift when you change typefaces). ② Type classification (genre: serif / sans serif / slab serif / monospaced / script / display · historical classification · seven Chinese type categories: Song, Kai, Hei, etc. · the three-layer Skeleton-Flesh-Skin type matrix). ③ Type selection methodology (4 major checklists: fit for purpose + design completeness + reliable files + usage license · emotional considerations · selection by project scale / text length / font size · optical sizes · weight and style · avoid fake bold and fake italic). ④ Type pairing (4 reasons why you need a second typeface · family metaphor: brothers / cousins / distant relatives · serif paired with sans serif is the safest choice · differences should be clear but not too distant). ⑤ Legibility and accessibility (legibility vs readability · lookalike characters Il1 / bdpq mirroring · open counters · WCAG contrast ratios 4.5:1 / 3:1 / AAA 7:1 · body text ≥16px · 8% of men have color vision deficiency · Lexend / Atkinson Hyperlegible). ⑥ Fully retained practical typography rules of thumb (line height 115–150% · line length 45–75 characters / 15–20 for Chinese · tracking: tighten for large sizes, loosen for small sizes · kerning almost never needs manual adjustment · widows and orphans · reverse text weight · techniques for breaking the rules). ⑦ CJK and multilingual script (full-width monospacing · no baseline, relies on visual center · no true italic · Chinese punctuation · Japanese three-script system + Tsume). ⑧ A quick-reference glossary of 30+ terms. Target audience: non-technical PMs, with in-line explanations of jargon, and Why / When to use / Counterexamples for each section. Division of labor with nngroup-visual-design-principles (an overview of visual design principles, with only one section on typography) and refactoring-ui (UI rules for non-designers, covering type scale / line length): this piece is a dedicated in-depth typography reference, with more detailed numerical values and glyph feature descriptions.
Integration of unit models and economies of scale from 3 perspectives. yitang Truman (6 levels of unit models + 15 classic ways to mess up + top 10 unit models + L6 economies of scale, the most complex diagram from a Yitang class in 2021 + Excel template) / Wang Huiwen Tsinghua Product Class (scale effect = gravity of the business world + three ABC shapes + network effect + high-frequency beating low-frequency + STP+4P+PMF) / Wang Kai ("good assets" hundred-billion-yuan monopoly / multi-technology cost accounting / do not challenge giants on their main battlefield). Covers selection of the top 10 unit models, SKU model vs order model vs store model, and the 50,000-car iteration case of a car-sharing unicorn
A "panoramic map of AI interaction paradigms" for AI product PMs, reorganizing the will-design-notes design resource section into 7 paradigms: ① Conversational (linear Q&A, the ChatGPT paradigm, the starting point of the AI revolution) ② AI in old UI (attaching AI to familiar interfaces such as Word/Notion/Figma/Linear, including 6 sub-types: natural language search / referencing to build trust / context-bound preset buttons / flexible batch document extraction / infinite canvas / non-linear conversation) ③ Agentive UX (AI works silently in the background, including 4 sub-types: you complete me intelligent completion / you surprise me proactive prediction / you exist invisible learning / you assist me returnable control) ④ Human in the loop (highlighted text / highlighted visuals / collaborate with me / real-time generation while thinking, AI acts as a partner not a domineering boss) ⑤ Hyper personalization (aiOS conversational operating system / AI as a product standalone hardware Rabbit R1 · Humane Pin) ⑥ Multi-modal (6 natural inputs: shoot / write / see / listen / draw / move, replacing precise prompts) ⑦ Trending (the interface evolution history from CLI→GUI→conversational→intent-driven + 6 major macro trends). Each paradigm includes Why it works + real product cases + design key points + when to use / counterexamples. Target readers = non-technical PMs, with in-line explanations of jargon and real-life analogies. It is complementary to [[shape-of-ai-patterns]] (pattern library, finer-grained reusable components) and [[nngroup-ai-ux-design]] (deep dive into conversational interfaces + empirical research); this article is at the "paradigm classification skeleton" level, and overlapping content is cross-referenced rather than rewritten.
Highlights from Will's Design Notes *UX Metrics and ROI* (based on the NN/g research report of the same name + a 5-edition case library). It focuses on the three things PMs need most: ① Metric system checklist — what each UX metric (task completion rate / time on task / success rate / conversion rate / retention / CLV / bounce rate…) looks like, how to calculate it, when to use it, and what its limitations are; how to interpret standardized questionnaires SUS (System Usability Scale, full score 100), NPS (Net Promoter Score = %Promoters − %Detractors), and CSAT; the five dimensions of Google's HEART framework (Happiness / Engagement / Adoption / Retention / Task success) as a scaffold for metric selection; what questions each of the four major data sources (surveys / quantitative usability testing / analytics tools / customer service and marketing data) answers. ② UX→ROI conversion methodology — improvement score formula (positive metric = post-change ÷ pre-change − 1; negative metric = pre-change ÷ post-change − 1), three-step method (select KPI → convert UX metrics to KPI by unit → report responsibly), how to convert to monetary value for four types of business models (e-commerce by sales volume / content sites by PV-based advertising / intranets by person-day wages / sites with clear target actions by conversion × average order value), retaining all exact figures including health insurance site saving 1.5 million yuan/year, HR tool saving $6,112 per month / 60% of work hours, financial subscription trial increasing monthly revenue by $20,000 from 76 to 187, and intranet ROI reaching 8/20/50 times the cost for companies with 1,000/10,000/100,000 employees; also presents Athenahealth's chained correlation model of "usability → satisfaction → willingness to recommend → retention and paid referrals", teaching PMs to connect soft metrics to revenue KPIs using statistical correlation. ③ Expectation management and reporting talking points — the "narrowing improvement" pattern that UX redesigns deliver an average 75% lift (95% confidence interval 8%~104%), and that this figure shrank from 247% to 75% from 2006 to 2020 (p=0.017), broken down by metric category (effectiveness 131% / efficiency 104% / adoption & retention 93% / engagement 87% / satisfaction 19%); 5 iron rules for reporting to bosses (estimates are not financial reports / compare against R&D costs / tell stories instead of piling tables / be transparent about assumptions / extend the timeline to 2–5 years) plus counterexamples. It complements *nngroup-analytics-metrics*: that piece covers NN/g's "analytics and measurement perspective" (vanity metrics / confidence intervals / statistical significance / sampling / triangulation / four-step ROI principle), while this piece is a "complete metric system checklist + improvement score math + multi-business-model ROI conversion recipes + empirical baseline magnitude data + reporting talking points"; overlapping content (four-step ROI / HEART / confidence intervals) is only cross-referenced and not rewritten. Target audience: product managers.
Integration of Vibe Coding and independent development from 5 perspectives. Karpathy (Software 3.0 + definition of Vibe Coding + Iron Man Suit + Agentic Decade) / newtype Lao Huang (6+ practical non-technical independent products + 3 core mindsets on Unit Economics + Vibe Coding cleansing prophecy) / Xiaohongshu vibecoding (three core mindsets + 0-80 vs 80-100 make-or-break line + Ci Le Ge Ci interview + Astro Kid's 400 million financing) / Wang Kai (product manager transition + AI Coding changing collaboration workflows + disruptive nature of data analysis) / **zara Zhang Zala (liberal arts background + Build small + Opinionated products + Build first learn later + Coding ≠ Shipping 70/30 + 4 core mindsets for working with models + Frontend Slides 16k+ stars)**. Covers definition of Vibe Coding, cleansing prophecy, 0-80 vs 80-100 make-or-break line, platformization of Vibe Gaming, and cross-over for liberal arts graduates.
Once AI writes code, can small games be consumed in a scrolling video-like feed? Zhang’s team explored that direction, then chose deeper creation tools and play among friends. Ordinary generated material may offer strangers limited value, while friendship supplies a shared reason to care.
A large collection of AIGC industry reports amassed by Wang Kai in 2023, totaling over 20 PDFs. Coverage includes the Sparks of AGI paper + CAICT/JD AIGC white papers + industry research from Huaxi Securities/Qiming Venture Partners/Aifenxi/Colliers/Minute/ZhongAn Insurance/Fudan University and others + early lectures (Fu Sheng/Li Piji) + Azure OpenAI documentation + domestic and international large model evaluations. This is a collection of the industry's immediate reactions in the 6–12 months after the ChatGPT moment (2022-11).
Aschenbrenner's 165-page Situational Awareness manifesto + How People Use ChatGPT (OpenAI) + Why LLMs Hallucinate (OpenAI) + Generative AI Seniority-Biased Change + large-scale Reddit persuasion experiment + OpenAI Practical Guide to Building Agents + Google Prompt Engineering (Lee Boonstra) + Looking at DAO 2025 (Weijin Research) + DeepSeek primer + AGI introduction. 10 documents that together answer five core questions: "What is AGI / How soon will it arrive / How to use it / Who will be impacted / How to build Agents"
Overseas promotion practice accumulated from 226 posts: platform-specific SOPs for Reddit / Twitter / YouTube / TikTok / HackerNews / ProductHunt / Dev.to, overseas phone number/bank card/payment configuration, multi-Agent multi-browser multi-account matrix
