跳至主要内容

附錄　資料來源（最終版·中英對照）

Appendix: Source Index (Final · Bilingual)

編輯說明 Editorial Note

第一章教你回到原文，這本書自己也要經得起同一個標準。以下是各章主要事實的出處索引。標注【待補】的項目，出版前需補回完整連結或日期。所有分析與判斷為作者原創立場，文責自負。

Chapter One teaches you to go back to the source. This book must meet the same standard. Below is the source index for key facts in each chapter. Items marked 【TBC】 require full links or dates before publication. All analysis and opinions are the author's original positions; responsibility rests with the author.

第一章　你讀到的新聞，不是原文​

Ch.1: The News You Read Is Not the Original​

  • 黃仁勳於首爾「Korea AI Ecosystem Reception」答記者問：Reuters 現場報導（2026-06-08）【連結待補】 Jensen Huang's response to press at the "Korea AI Ecosystem Reception," Seoul: Reuters on-site report (2026-06-08) 【link TBC】

  • Bloomberg、Seoul Economic Daily、The Standard 交叉比對【連結待補】 Cross-referenced with Bloomberg, Seoul Economic Daily, The Standard 【links TBC】

  • 黃仁勳 All-In Podcast 訪談（2026 年 3 月）【集數待補】 Jensen Huang on All-In Podcast (March 2026) 【episode # TBC】

第二章　遊戲致勝：微軟與世嘉的跨國糾纏​

Ch.2: Game Victory — Microsoft and Sega's Transnational Entanglement​

  • 微軟新聞稿〈Microsoft, Sega Collaborate on Dreamcast〉（1998-05-21） Microsoft press release, "Microsoft, Sega Collaborate on Dreamcast" (1998-05-21)

  • 《Sega Rally 2》Windows CE 版幀數問題：英文維基百科條目 Sega Rally 2 Windows CE frame-rate issues: English Wikipedia entry

  • 大川功與蓋茲的兼容談判：Sam Furukawa 推文（2010）；Kotaku 跟進報導 Okawa Isao–Bill Gates compatibility negotiations: Sam Furukawa tweets (2010); Kotaku follow-up

  • Peter Moore 機場軼事：Moore 本人訪談【出處待補】 Peter Moore airport anecdote: Moore's own interview 【source TBC】

  • DC 遊戲庫 WinCE 開發比例（約 50／600+）【統計來源待補】 Dreamcast library WinCE share (~50 / 600+ titles) 【stat source TBC】

  • 大川功捐款 6.95 億美元、入交昭一郎投資 NVIDIA 500 萬美元、微軟第一代 Xbox 虧損 40 億美元：公開紀錄及《遊戲致勝》 Okawa's $695M donation; Iri Shoichiro's $5M NVIDIA investment; Microsoft's $4B loss on first Xbox: public records and Game Victory

第三章　AI 開始自己寫自己——但誰在付電費？​

Ch.3: AI Is Writing Itself — But Who's Paying the Power Bill?​

  • Anthropic〈When AI builds itself〉（2026 年 6 月） Anthropic, "When AI builds itself" (June 2026)

  • 程世嘉《數位時代》分析【連結待補】 Cheng Shih-chia analysis in Business Next / 數位時代 【link TBC】

  • NVIDIA 資料中心營收 $193.7B／遊戲營收 $16B：NVIDIA 財報【財年口徑待註明】 NVIDIA Data Center revenue $193.7B / Gaming revenue $16B: NVIDIA earnings 【fiscal year convention TBC】

  • OpenAI 預計虧損 $14B、營收 $13B、累計虧損膨脹至 $44B【報導來源待補】 OpenAI projected $14B loss, $13B revenue, cumulative loss ballooning to $44B 【press source TBC】

  • 微軟 2026 年資本支出預計 $190B：公司財報【待補】 Microsoft 2026 estimated capex $190B: company filings 【TBC】

  • Google 資本支出從 $29B 躍至 $75B：公司財報【待補】 Google capex jump from $29B to $75B: company filings 【TBC】

  • GeForce 8800 GTX（2006）/ CUDA Toolkit 1.0（2007）/ AlexNet（2012）：公開產業及學術紀錄 GeForce 8800 GTX (2006) / CUDA Toolkit 1.0 (2007) / AlexNet (2012): public industry and academic records

  • RTX 5090 建議售價與市價【待補】 RTX 5090 MSRP and street price 【TBC】

  • 台灣科技 KOL Facebook 討論串（AI 變現觀察）：匿名化處理 Taiwan tech KOL Facebook thread (AI monetization observations): anonymized

第四章　AI 敘事經濟學：當「取代」變成一門生意​

Ch.4: AI Narrative Economics — When "Replacement" Becomes a Business​

  • Uber COO Andrew Macdonald podcast 訪談【待補】 Uber COO Andrew Macdonald podcast interview 【TBC】

  • Amazon 取消 KiroRank 排行榜；SVP Dave Treadwell 內部發言【報導來源待補】 Amazon cancels KiroRank leaderboard; SVP Dave Treadwell internal remarks 【source TBC】

  • 企業單月 $500M Claude 帳單：Axios【待補】 Enterprise $500M single-month Claude bill: Axios 【TBC】

  • NVIDIA 應用深度學習 VP Bryan Catanzaro 發言【待補】 NVIDIA VP Applied Deep Learning Bryan Catanzaro remarks 【TBC】

  • CloudBees CEO Anuj Kapur 對 Axios 發言【待補】 CloudBees CEO Anuj Kapur to Axios 【TBC】

  • 2026 上半年科技業裁員逾 117,000 人及裁員原因排序：Challenger, Gray & Christmas H1 2026 tech layoffs exceeding 117,000; layoff-reason ranking: Challenger, Gray & Christmas

  • Marc Andreessen 訪談：20VC podcast Marc Andreessen interview: 20VC podcast

  • Sam Altman「AI washing」；Deutsche Bank 分析；Oxford Economics 結論【各待補】 Sam Altman on "AI washing"; Deutsche Bank analysis; Oxford Economics conclusion 【each TBC】

  • Microsoft 自願退休計劃（2026 年 4 月）：年齡+年資 ≥ 70、約 8,750 人、WARN Act 法律框架【待補】 Microsoft voluntary retirement program (April 2026): age + tenure ≥ 70, ~8,750 eligible, WARN Act legal framework 【TBC】

  • Block 裁員 4,000 人與六週後重聘事件：Jack Dorsey 於 Fortune 訪問；當事員工 LinkedIn 貼文【待補】 Block layoff of 4,000 and re-hiring six weeks later: Jack Dorsey in Fortune; affected employees' LinkedIn posts 【TBC】

  • Klarna AI 客服始末：公司聲明及多家媒體報導【待補】 Klarna AI customer service saga: company statements and multiple press reports 【TBC】

  • MIT 研究（2024）：AI 自動化於約 23% 崗位具經濟可行性 MIT study (2024): AI automation economically viable for ~23% of jobs

  • Sony 服務型遊戲計劃（Jim Ryan, 2022）、Concord 下架及多間工作室關閉：各遊戲媒體報導【待補】 Sony live-service games plan (Jim Ryan, 2022), Concord delisting, multiple studio closures: gaming press 【TBC】

  • CNBC 統計：23 家標普 500「AI 裁員」企業股價追蹤（截至 2026-05-15，56% 下跌、平均約 -25%）；中文編譯見 TechNews 科技新報、未來商務（2026-06） CNBC tracking: 23 S&P 500 "AI layoff" companies' stock performance (as of 2026-05-15, 56% declined, avg ~-25%); Chinese coverage in TechNews, FC (2026-06)

  • Gartner 報告：〈Autonomous Business and AI Layoffs May Create Budget Room, but Do Not Deliver Returns〉（含 2028–29 年自動化業務淨增職位預測） Gartner report: "Autonomous Business and AI Layoffs May Create Budget Room, but Do Not Deliver Returns" (incl. 2028–29 net job increase projection)

  • Nike、Salesforce、Fiverr、Cisco 裁員與股價表現：CNBC；《數位時代》2026 年裁員潮整理 Nike, Salesforce, Fiverr, Cisco layoffs and stock performance: CNBC; Business Next 2026 layoff roundup

第五章　資訊封建主義：從免費內容到知識農奴​

Ch.5: Information Feudalism — From Free Content to Knowledge Serfdom​

  • 《第三類法庭》（TVB）【年份待核】 The Third Tribunal (TVB drama) 【year TBC】

  • 《電腦時代》雜誌（香港，1990 年代） Computer Age magazine (Hong Kong, 1990s)

  • Gartner、IDC、McKinsey 報告定價作為知識壁壘案例：一般性行業認知 Gartner, IDC, McKinsey report pricing as knowledge-barrier case study: general industry knowledge

  • Claude Skill 架構分析文章被七個以上 Facebook 群組壓下事件：作者親歷，匿名化處理 Author's Claude Skill architecture analysis suppressed by 7+ Facebook groups: first-hand experience, anonymized

第六章　別人的學費——Ring 0 那顆地雷​

Ch.6: Someone Else's Tuition — The Ring 0 Landmine​

  • NVIDIA RTX Spark / Grace CPU / Blackwell GPU 規格：NVIDIA 公開發佈資料 NVIDIA RTX Spark / Grace CPU / Blackwell GPU specs: NVIDIA public announcements

  • Windows on ARM 三批白老鼠歷史（Tegra 3, 2012 / Qualcomm 2017–2023 / Snapdragon X Elite 2024–2025）：公開產品歷史 Windows on ARM three-batch history (Tegra 3 2012 / Qualcomm 2017–2023 / Snapdragon X Elite 2024–2025): public product history

  • 微軟 Windows RT 虧損 $900M：微軟財報 Microsoft Windows RT $900M write-down: Microsoft filings

  • Prism x86 模擬層 Ring 3 / Ring 0 結構性限制：技術文件 Prism x86 emulation layer Ring 3 / Ring 0 structural limitations: technical documentation

  • Intel BIOS 韌體生態 40 年歷史：產業公開知識 Intel BIOS firmware ecosystem 40-year history: public industry knowledge

第七章　學習的扭曲：焦慮是最好的商品​

Ch.7: The Distortion of Learning — Anxiety Is the Best Product​

  • 專業技能半衰期壓縮（10–15 年→5 年以下；軟體/AI 領域 2.5 年或 18 個月）：World Economic Forum《Future of Jobs Report》；Skillable, 2025；Morson Edge, 2025 Professional skill half-life compression (10–15 yrs → <5 yrs; software/AI fields 2.5 yrs or 18 months): WEF Future of Jobs Report; Skillable, 2025; Morson Edge, 2025

  • 70-20-10 學習模型：美國勞工統計局（BLS）；Center for Creative Leadership 70-20-10 learning model: U.S. Bureau of Labor Statistics (BLS); Center for Creative Leadership

  • 只有 11% 員工將正式培訓應用到工作：Association for Talent Development（ATD）, 2020 Only 11% of employees apply formal training to work: Association for Talent Development (ATD), 2020

  • 半數企業管理者缺乏支持推動員工發展；45% 員工缺乏使用培訓項目的指引：D2L / LinkedIn, 2025 Half of companies say managers lack support for employee development; 45% say employees lack guidance on using training programs: D2L / LinkedIn, 2025

  • Cisco 裁員砍傳統路由部門，CCIE 認證人員被裁；「哪些認證剛剛被釋放到公開市場」：KORE1 分析（2026） Cisco layoffs cut traditional routing division; CCIE-certified staff laid off; "which certifications just hit the open market": KORE1 analysis (2026)

  • 《小朋友齊打交 2》（Little Fighter 2）用戶規模與收費嘗試【待補】 Little Fighter 2 user base and monetization attempts 【TBC】

  • 創作者贊助平台付費轉化率 1–3%【數據來源待補】 Creator patronage platform paid conversion rate 1–3% 【data source TBC】

第八章　文筆不是內容，程式碼不是資產​

Ch.8: Writing Is Not Content, Code Is Not an Asset​

  • 舊系統維護佔企業 IT 總支出 60–80%：Gartner、Forrester、Deloitte 產業基準 Legacy system maintenance at 60–80% of enterprise IT spend: Gartner, Forrester, Deloitte industry benchmarks

  • 美國聯邦政府 IT 預算逾 $105B，其中 79% 用於運維和維護：美國政府問責署（GAO）, 2025 U.S. federal IT budget >$105B, 79% allocated to O&M: U.S. Government Accountability Office (GAO), 2025

  • 維護成本年增 18–25%；70% 組織難以找到具備多重技能的大型機人才：Kyndryl / BMC, 2025 Maintenance costs rising 18–25% annually; 70% of organizations struggle to find multi-skilled mainframe talent: Kyndryl / BMC, 2025

  • 全球約 2,200 億行 COBOL 仍在運行；處理 95% ATM 交易、80% 當面信用卡交易、每日清算逾 $3T；43% 銀行系統建於 COBOL：World Economic Forum；Open Mainframe Project, 2025 ~220 billion lines of COBOL still running globally; handles 95% of ATM transactions, 80% of in-person card transactions, >$3T daily settlements; 43% of banking systems built on COBOL: WEF; Open Mainframe Project, 2025

  • 大型機退出專案逾 70% 預計無法達成預期效益：Gartner, 2026 >70% of mainframe exit projects expected to fall short of target benefits: Gartner, 2026

  • 42% 曾採用微服務的組織正在整合回較大部署單元：CNCF 調查, 2025 42% of organizations that adopted microservices are consolidating back to larger deployment units: CNCF survey, 2025

  • 60% 團隊後悔在中小型應用中使用微服務；單體架構平均節省 25% 成本：Gartner 60% of teams regret using microservices for small-to-mid apps; monoliths save 25% cost on average: Gartner

  • Amazon Prime Video 從微服務回退成單體架構，基礎設施成本降低 90%：Amazon 技術部落格 Amazon Prime Video migrated monitoring from microservices back to monolith, cutting infra costs 90%: Amazon tech blog

  • Segment 從 150+ 微服務回退到單一應用；Service mesh 採用率從 18% 降至 8%：業界報導 Segment reverted from 150+ microservices to a single app; service mesh adoption dropped from 18% to 8%: industry reports

第九章　企業文化的死結​

Ch.9: The Dead Knot of Corporate Culture​

  • 年齡歧視實證數據：結構性論證【具體來源待確認】 Age discrimination empirical data: structural argument 【specific source TBC】

  • 組織判斷力的劣幣驅逐良幣（Gresham's Law 類比）：作者原創分析框架 Gresham's Law applied to organizational judgment: author's original analytical framework

第十章　當生產接近免費，人還需要什麼？​

Ch.10: When Production Approaches Free, What Do People Still Need?​

  • 暢銷書生成器（AI 生成的叔本華安慰文）：作者實測生成 Bestseller generator (AI-generated Schopenhauer consolation text): author's live test

  • World Model / Memory / Agent / Self Reflection 研究方向：AI 研究界公開文獻（一般性引用） World Model / Memory / Agent / Self Reflection research directions: open AI research literature (general reference)

第十一章　刀很利，但你要斬什麼？​

Ch.11: The Blade Is Sharp — But What Are You Cutting?​

  • 資深作家回顧華文寫作代際變遷、AI 重度用戶 $200→$60 降級記錄：社交媒體貼文，匿名化處理 Senior author's generational survey of Chinese-language writing; heavy AI user downgrade from $200→$60: social media posts, anonymized

第十二章　得意忘象——言、象、意​

Ch.12: Grasp the Meaning, Forget the Image — Word, Image, Meaning​

  • 言象意框架源自中國古典思想：《周易》繫辭傳；王弼《周易略例·明象》 The Word/Image/Meaning framework from classical Chinese thought: Xici Zhuan (Commentary on the Appended Phrases, I Ching); Wang Bi, Zhouyi Lüeli: Mingxiang

  • 《道德經》/ 太極圖：古典文獻 Tao Te Ching / Taiji diagram: classical texts

第十三章　認知模型、約束條件、演化能力​

Ch.13: Cognitive Models, Constraints, Evolutionary Capacity​

  • 結構性論證為主；案例引用自本書前述章節（第二章微軟世嘉案例、第六章 Ring 0 案例） Primarily structural argument; case studies drawn from earlier chapters (Ch.2 Microsoft-Sega, Ch.6 Ring 0)

  • 《道德經》「大道無形」：古典文獻 Tao Te Ching, "the great Way has no form": classical text

第十四章　仁者無敵為何失效，又為何仍然重要​

Ch.14: Why "The Benevolent Are Invincible" Fails — And Why It Still Matters​

  • 《論語》《孟子》《孫子兵法》《道德經》：古典文獻 Analerta, Mencius, Art of War, Tao Te Ching: classical texts

  • 黃仁勳韓國綜藝節目 humanity 發言 / NVIDIA「光速執行」管理哲學【出處待補】 Jensen Huang on Korean variety show re: humanity / NVIDIA "light-speed execution" management philosophy 【source TBC】

  • 世嘉案例（回扣第二章） Sega case study (cross-reference Ch.2)

第十五章　熱情的重量——追求所愛的賬單​

Ch.15: The Weight of Passion — The Bill for Pursuing What You Love​

  • 論證與個人陳述為主；三種人分類、熱情三種重量比喻為作者原創分析 Primarily argument and personal narrative; three-person taxonomy and three-weight metaphor are the author's original analysis

第十六章　被偷走的梯子——努力為何換不到回報​

Ch.16: The Stolen Ladder — Why Hard Work No Longer Pays Off​

  • 學歷通脹 / 多勞多得崩塌 / 生存空間壓縮：結構性論證【具體統計數據來源待確認】 Credential inflation / collapse of the work-reward equation / shrinking survival space: structural argument 【specific statistical sources TBC】

  • 魯冰花案例：台灣電影引用 Lupinus (魯冰花) case: Taiwanese film reference

第十七章　最笨的路（上）——一道無解的數學題​

Ch.17: The Dumbest Road (Part 1) — An Unsolvable Math Problem​

  • 創作者贊助平台轉化率 ~2%：Okoa, 2024 Creator patronage conversion rate ~2%: Okoa, 2024

  • 《小朋友齊打交 2》（Little Fighter 2）案例：同第七章 Little Fighter 2 case: same as Ch.7

  • 電影盜版案例：一般性觀察 Film piracy case: general observation

第十八章　最笨的路（下）——為什麼仍然要走​

Ch.18: The Dumbest Road (Part 2) — Why Walk It Anyway​

  • 美國教科書價格 1977–2015 漲幅 1,041%：美國勞工統計局（BLS） U.S. textbook prices 1977–2015, increase of 1,041%: U.S. Bureau of Labor Statistics (BLS)

  • OECD 2024 報告：最富有 10% 家庭擁有超過 52% 家庭財富；美國為 79% OECD 2024 report: wealthiest 10% of households hold >52% of total household wealth; 79% in the U.S.

  • OECD 2025 財富分佈報告：即使相對不平等輕微上升，絕對差距在總體財富增長下急劇擴大——「這些擴大的絕對差距對社會流動性構成了更大的障礙」 OECD 2025 wealth distribution report: even as relative inequality rose modestly, absolute gaps widened sharply amid overall wealth growth — "these widening absolute gaps pose greater barriers to social mobility"

  • 鐘型→M型/K型社會退化：結構性論證，基於上述 OECD 數據 Bell-curve → M-shaped / K-shaped societal regression: structural argument based on above OECD data

終章　送得到未來的書​

Closing Chapter: A Book That Reaches the Future​

  • 回引全書數據（Gartner、CNBC、OECD 等）：無新增外部來源 Cross-references book-wide data (Gartner, CNBC, OECD, etc.): no new external sources

  • 易經第六十四卦「未濟」（火水未濟）：古典文獻 I Ching Hexagram 64 "Wei Ji" (Before Completion / Fire over Water): classical text

後記　關於你正在做的事​

Afterword: About What You Are Doing​

  • 個人陳述，無外部數據引用 Personal statement; no external data cited

延伸閱讀 Further Reading​

  • 《遊戲致勝：從像素到 AI，娛樂如何暗中重塑全球科技霸權》（全十七章） Game Victory: From Pixels to AI — How Entertainment Secretly Reshaped Global Tech Hegemony (17 chapters)

  • 《微軟與蘋果的 AI 生態暗戰：全景式產業研究》（全八章） Microsoft vs. Apple: The Hidden AI Ecosystem War — A Panoramic Industry Study (8 chapters)

  • 《遊戲之星：微軟與世嘉的跨國糾纏》 Game Star: Microsoft and Sega's Transnational Entanglement

  • 長篇小說《鏡界》 Novel: Mirror Realm (鏡界)

待處理事項摘要 Outstanding Items Summary​

類別 Category數量 Count說明 Description
【連結待補】Link TBC~15已知來源但需補回 URL Known sources needing URLs
【出處待補】Source TBC~8需確認具體出處 Specific sources need confirmation
【年份/集數待補】Date/Ep TBC~3需核實年份或集數 Years or episode numbers to verify
【統計來源待補】Stat source TBC~4數據需追溯原始出處 Data needing original source traced

v4 附錄覆蓋範圍 v4 Appendix Coverage：第一至七章（加上第六章/終章備註） 本最終版新增覆蓋 This Final Version Adds：第八至第十八章 + 終章

© 星忘塵 Nebula Walker · 創象引擎 MYTHOGEN ENGINE · 2026