The Question That Broke the Internet
“He looks so ordinary, yet he can be so confident.”
When comedian Yang Li delivered this line on Rock & Roast Season 3 in August 2020, she created a phrase — 普信男 (pǔxìn nán) — that would generate 210 million Weibo views, spark brand boycotts, and cost JD.com an estimated 50 billion RMB in market value.
But why did it resonate so deeply? A post-90s knowledge creator named An Senyao (安森垚) has an answer, and his 7-minute video has since garnered 773,000 views with a remarkable 9.9% like ratio.
The Creator: An Senyao
An Senyao entered Beijing University of Chemical Technology at age 15, completed graduate studies at Kyoto Institute of Technology, and became famous as a Zhihu “god-tier” answerer before publishing the “Ancestors” book series. His Bilibili channel features 322 videos with 16.3 million total likes, earning the rare “Golden Hall” nameplate.
His video is a satirical critique of the Dunning-Kruger effect as manifested in Chinese internet culture — why historically ignorant people are supremely confident in conspiracy theories, and how AI chatbots are making the problem worse.
Six Topics Debunked
The video tackles six major areas where false confidence runs rampant:
“Western Fake History Theory” — The conspiracy theory that Western history is fabricated. As one commenter noted: “The fake history theory outputs emotion over content.” This narrative packages cultural insecurity as “patriotism” and ignorance as “independent thinking.”
24 Solar Terms — Solstices and equinoxes are observable worldwide, not uniquely Chinese. The 24-term system is a Chinese cultural framework, not evidence of civilizational superiority.
Steam Engine — A recurring danmaku corrects: “Watt improved, not invented, the steam engine!” Thomas Newcomen built the first practical steam engine in 1712; Watt patented his improvements in 1769. Classic example of oversimplified education creating false confidence.
Bronze Color — Bronze is not always green; the patina comes from oxidation.
Su Shi vs. Su Wu — Confusing the poet with the diplomat who herded sheep among the Xiongnu illustrates how shallow knowledge creates misplaced confidence.
AI: The New Amplifier
The top comment (1,545 likes) cuts deep: “Now with AI as a cyber sycophant that praises everything you say, the ignorant become even more confident.”
AI chatbots’ sycophancy tendency — agreeing with users even when they’re wrong — is becoming an amplifier of overconfidence. As An Senyao observes: “Mobile internet didn’t make people stupider — it unprecedentedly expanded the base of uninformed people.”
The Dunning-Kruger Effect
The Dunning-Kruger Effect (DKE) is a cognitive bias where people with limited knowledge greatly overestimate their competence, while experts tend to underestimate theirs. Identified by David Dunning and Justin Kruger in 1999.
Original Research (Kruger & Dunning, 1999):
- Lowest quartile participants placed at 12th percentile but ranked themselves at 62nd percentile
- Skills needed to produce correct answers are the same skills needed to evaluate correctness — a “double burden”
- After minimal training, self-assessments became more accurate
The DKE Curve:
| Stage | Competence | Confidence | Description |
|---|---|---|---|
| Mount Stupid | Low | High | Beginners who know a little think they know a lot |
| Valley of Despair | Growing | Declining | Intermediate learners realize complexity |
| Slope of Enlightenment | Increasing | Calibrated | Advanced learners develop accurate self-assessment |
| Plateau of Sustainability | High | Appropriate | Experts with accurate confidence |
A Chinese wisdom from the video’s danmaku captures it perfectly: “Enter the world invincible; after three years, can’t move.”
Gender Differences in Overconfidence
One of psychology’s most robust findings: males consistently rate their intelligence higher than females across 30+ countries, despite equal actual IQ.
| Domain | Effect Size (d) | Interpretation |
|---|---|---|
| General Intelligence | 0.37 | Small to moderate |
| Mathematical/Logical | 0.44 | Moderate |
| Spatial Intelligence | 0.43 | Moderate |
| Verbal Intelligence | 0.07 | Negligible |
Why men overestimate:
- Socialization: Boys encouraged to be assertive; girls socialized to be modest
- Sex-role identification: Masculinity predicts self-estimated intelligence
- Workplace behavior: Men apply when meeting 60% of qualifications; women wait for 100%
The “Puxinnan” Phenomenon
普信男 (pǔxìn nán) = 普通 (ordinary) + 自信 (confident) + 男 (man) became permanent Chinese internet vocabulary.
Brand Boycott Pattern:
| Date | Brand | Action |
|---|---|---|
| Mar 2021 | Intel | Ad pulled within hours after male boycott |
| Mar 2021 | Great Wall Motors | Boycotted at product launch |
| Aug 2021 | XPeng Motors | Promotional poster boycott |
| Oct 2021 | Mercedes-Benz | Massive backlash; video pulled |
| Dec 2021 | Heilan Home | Flooded customer service |
| Jan 2022 | Shede Spirits | Stock dropped 1.94% in one day |
JD.com Boycott (2024):
- Oct 14: JD.com launches Double 11 with Yang Li. Male users boycott.
- Oct 17-18: JD Finance “bank run” rumors spread.
- Oct 18: JD apologizes, drops Yang Li. Both sides angry.
- Post-Oct 18: ~50B RMB market cap loss within two weeks.
Overconfidence Psychology: Three Types
| Type | Definition | Key Finding |
|---|---|---|
| Overestimation | Thinking you are better than you are | Greatest on hard tasks |
| Overplacement | Believing you are better than others | Greatest on easy tasks |
| Overprecision | Excessive certainty in your knowledge | Most persistent; increases with age |
Better-Than-Average Effect: Svenson (1981) found that 93% of American drivers rated themselves above the median.
Culture matters: Individualist cultures show stronger overplacement; collectivist cultures show weaker overplacement but stronger self-criticism.
Fact Check: 10 Claims
| # | Claim | Verdict |
|---|---|---|
| 1 | Watt invented the steam engine | FALSE — Watt improved it; Newcomen built the first (1712) |
| 2 | 24 Solar Terms are uniquely Chinese | MISLEADING — Solstices/equinoxes observed worldwide |
| 3 | Bronze is always green | FALSE — Green patina comes from oxidation |
| 4 | 93% of drivers rate themselves above average | TRUE — Svenson (1981) |
| 5 | Men are more overconfident than women | TRUE — 30+ countries; d=0.37 |
| 6 | DKE is a statistical artifact | DEBATED — Ongoing academic debate |
| 7 | Chinese hot water drinking is ancient | FALSE — 1952 health campaign |
| 8 | AI chatbots validate wrong answers | TRUE — Sycophancy documented |
| 9 | BTA effect is universal | PARTIALLY TRUE — Weaker in collectivist cultures |
| 10 | Yang Li’s joke caused 50B RMB loss | PARTIALLY TRUE — Causation complex |
Recommendations
For Individuals:
- Cultivate intellectual humility — ask “What don’t I know?”
- Seek disconfirming evidence actively
- Practice metacognition — think about your thinking
- Distinguish confidence from competence
- Verify AI-generated claims against authoritative sources
For Society:
- Media literacy education — recognize cognitive biases
- Platform reform — reward accuracy over engagement
- Support genuine expertise
- Encourage productive civil disagreement
What makes you confident — and is that confidence justified?
為啥他們能那麼自信？—— 中國網路文化中的鄧寧-克魯格效應
打破網路的那句話
「他明明看起來那麼普通，但是他卻可以那麼自信。」
2020年8月，脫口秀演員楊笠在《脫口秀大會》第三季說出這句話時，她創造了一個短語——「普信男」——引發2.1億微博瀏覽、品牌抵制，並讓京東損失估計500億人民幣市值。
但為什麼它引起如此深刻的共鳴？一位90後知識型創作者安森垚給出了答案，他的7分鐘影片至今已獲得77.3萬次觀看，點讚率高達9.9%。
創作者：安森垚
安森垚15歲進入北京化工大學，後赴日本京都工業大學完成研究所學業，原為知乎「大神級」回答者，出版《祖先》系列著作。Bilibili頻道擁有322部影片，累計1,630萬次點讚，獲頒稀有的「黃金殿堂」勳章。
他的影片是對「鄧寧-克魯格效應」在中國網路文化中體現的諷刺性批判——為什麼歷史無知者對陰謀論擁有絕對自信，以及AI聊天機器人如何使問題惡化。
六大主題闢謠
影片處理了六個虛假自信猖獗的主要領域：
「西方偽史論」 — 聲稱西方歷史是偽造的陰謀論。評論指出：「偽史論輸出情緒＞內容」。這種論調將文化自卑感包裝成「愛國主義」，將無知偽裝成「獨立思考」。
二十四節氣 — 冬至、夏至等節氣在全球各地均有觀測記錄，並非中國獨有。二十四節氣是中國文化框架，不是文明優越性的證據。
蒸汽機 — 彈幕反覆糾正：「瓦特改良了蒸汽機！不是發明！」紐科門於1712年製造首台實用蒸汽機，瓦特於1769年取得改良專利。
青銅器顏色 — 青銅並非總是綠色；銅綠來自氧化。
蘇軾 vs 蘇武 — 混淆詩人與牧羊的外交官，說明膚淺知識如何創造錯誤自信。
AI：新的放大器
最高讚評論（1,545讚）一針見血：「現在有了AI這個賽博奸臣，什麼都誇太對了你好棒，無知者更自信了。」
AI聊天機器人的諂媚傾向——即使用戶錯了也同意——正在成為過度自信的放大器。正如安森垚觀察：「移動互聯網不是變愚蠢了，而是基本盤史無前例的擴大了。」
鄧寧-克魯格效應
鄧寧-克魯格效應（DKE）是一種認知偏差，能力欠缺的人傾向於高估自己的能力，而真正有才幹的人則傾向於低估自己。由大衛·鄧寧和賈斯汀·克魯格於1999年提出。
原始研究（Kruger & Dunning, 1999）：
- 最低四分位數參與者的測驗成績僅處於第12百分位，但自我評估卻認為處於第62百分位
- 產生正確答案所需的技能與評估正確性所需的技能相同——「雙重負擔」
- 經過最少訓練後，自我評估變得更準確
DKE 曲線：
| 階段 | 能力 | 自信 | 描述 |
|---|---|---|---|
| 愚昧之峰 | 低 | 高 | 初學者略知一二便以為精通 |
| 絕望之谷 | 成長中 | 下降 | 中級學習者意識到複雜性 |
| 開悟之坡 | 持續增長 | 校準中 | 高級學習者發展準確的自我評估 |
| 持續高原 | 高 | 適當 | 專家擁有準確的自信 |
影片中彈幕的中文智慧完美捕捉了這一點：「初入江湖天下無敵，再練三年寸步難行。」
過度自信的性別差異
心理學研究中最穩健的發現之一：男性在30多個國家中一致地高估自己的智力，儘管實際IQ相等。
| 能力領域 | 效應量（Cohen’s d） | 解釋 |
|---|---|---|
| 一般智力 | 0.37 | 小到中等 |
| 數學/邏輯 | 0.44 | 中等 |
| 空間智力 | 0.43 | 中等 |
| 語言智力 | 0.07 | 可忽略 |
男性為何高估：
- 社會化過程：男孩被鼓勵果斷；女孩被社會化為謙遜
- 性別角色認同：男性化特質預測自我估計智力
- 職場行為：男性在符合60%條件時就申請職位；女性等待100%
「普信男」現象
普信男 = 普通 + 自信 + 男，成為中國網路永久詞彙。
品牌抵制模式：
| 日期 | 品牌 | 行動 |
|---|---|---|
| 2021年3月 | Intel | 男性抵制後數小時內廣告下架 |
| 2021年3月 | 長城汽車 | 產品發布會遭抵制 |
| 2021年8月 | 小鵬汽車 | 宣傳海報遭抵制 |
| 2021年10月 | 奔馳 | 大規模反彈；影片下架 |
| 2021年12月 | 海瀾之家 | 客服被灌爆 |
| 2022年1月 | 舍得酒業 | 股價單日下跌1.94% |
京東事件（2024年）：
- 10月14日：京東啟動雙十一營銷活動，邀請楊笠。男性用戶抵制。
- 10月17-18日：京東金融「擠兌」傳言蔓延。
- 10月18日：京東道歉，宣布不再合作。兩邊不滿。
- 10月18日後：兩週內市值蒸發約500億人民幣。
過度自信心理學：三種類型
| 類型 | 定義 | 關鍵發現 |
|---|---|---|
| 高估 | 認為自己比實際更好 | 在困難任務上最顯著 |
| 高估排名 | 認為自己比他人更好 | 在簡單任務上最顯著 |
| 過度精確 | 對自己的知識過度確定 | 最持久；隨年齡增長 |
高於平均效應： Svenson（1981）發現93%的美國駕駛認為自己的駕駛技術高於中位數。
文化差異： 個人主義文化顯示更強的高估排名效應；集體主義文化顯示較弱的高估排名但更強的自我批評。
事實查核：10項主張
| # | 主張 | 判定 |
|---|---|---|
| 1 | 瓦特發明了蒸汽機 | 錯誤 — 瓦特改良；紐科門1712年製造 |
| 2 | 二十四節氣是中國獨有 | 誤導 — 全球均有觀測 |
| 3 | 青銅器總是綠色 | 錯誤 — 銅綠來自氧化 |
| 4 | 93%駕駛認為自己高於平均 | 正確 — Svenson (1981) |
| 5 | 男性比女性更容易過度自信 | 正確 — 30+國家；d=0.37 |
| 6 | DKE主要是統計偽影 | 爭議中 — 學術辯論持續 |
| 7 | 中國人喝熱水是古老傳統 | 錯誤 — 1952年運動 |
| 8 | AI聊天機器人會驗證錯誤答案 | 正確 — 諂媚已有文獻記錄 |
| 9 | 高於平均效應是普遍的 | 部分正確 — 在集體主義文化中較弱 |
| 10 | 楊笠笑話導致500億損失 | 部分正確 — 因果關係複雜 |
建議
對個人：
- 培養知識謙遜——定期問「我有什麼不知道的？」
- 尋求反證——主動尋找挑戰自己觀點的資訊
- 實踐元認知——思考自己的思考過程
- 區分自信與能力
- 驗證AI生成的主張
對社會：
- 媒體素養教育——識別認知偏差
- 平台改革——獎勵準確性而非參與度
- 支持真正的專業知識
- 鼓勵建設性不同意見
什麼讓你自信——而這種自信是否合理？
References:
- Kruger, J. & Dunning, D. (1999). “Unskilled and Unaware of It.” JPSP, 77(6)
- Moore, D.A. & Healy, P.J. (2008). “The Trouble with Overconfidence.” Psychological Review
- Svenson, O. (1981). “Are We All Less Risky?” Acta Psychologica
- Reilly, D. et al. (2022). “Male Hubris, Female Humility.” Frontiers in Psychology
- Kahneman, D. (2011). Thinking, Fast and Slow
- Bilibili Video: BV1TqVN68Eir by 安森垚