Why Major essay 是補充文書裡最容易寫成空話的一篇。「我從小喜歡科學」「我想用科技改變世界」——這些句子寫完等於沒寫。本文用 3 幕結構教你怎麼把 250 字寫到 adcom 信服。
Why Major Essay 怎麼寫不變空話：3 幕結構與具體性法則
發布於 2026 年 5 月 14 日
每年 10 月，我和學生開 Why Major essay 會議時，第一個問題永遠是：「為什麼你要念這個 major？」
70% 學生的第一句答案是：「因為我喜歡 / 我有興趣 / 我覺得有意義。」
這就是 Why Major essay 寫不好的根源。「喜歡」「興趣」「意義」這三個字，是 essay 的死刑判決。Adcom 看到這幾個字會直接跳過——因為 50,000 個申請者寫的都一樣。
Why Major（也叫 Intended Major Essay、Academic Interest Essay）是美國頂校補充文書裡最高頻、也最難寫的題目。Stanford、MIT、UPenn、Cornell、Northwestern、CMU、UMich、UT Austin、Duke——幾乎每個 Top 30 都會問。本文用我 15 年帶過 600+ 位學生的實戰經驗，拆解這篇 250-650 字怎麼寫得讓 adcom 信你真的懂這個 major。
一、Why Major 是補充文書裡最難的一篇
很多家長覺得 Why Major 比 Common App PS 簡單——「PS 要寫人生故事，Why Major 只要說『我為什麼念 CS』，難度應該低吧？」
錯。Why Major 比 PS 難 3 倍。原因：
維度
Common App PS
Why Major
字數
650 字
通常 100-650 字（更短）
自由度
你選題材
題目限定
知識門檻
不需要懂 major
必須懂該領域
重複性
個人故事各有特色
主題雷同，比細節
評分人
一般 adcom
可能由該系教授看
最後一點最關鍵——Why CS essay 在很多 Top 30 學校是讓 CS 系教授看的。教授看一段「I love coding because it lets me create things」會直接翻白眼，因為他知道你根本沒寫過認真的程式。
二、為什麼「I love science because Einstein」永遠不行
這幾年我看過幾百個 Why Major 的爛開頭，可以歸納成 5 種：
爛開頭類型
範例
為什麼爛
名人崇拜型
"Ever since I read Einstein's biography..."
名人不是你，你也不是 Einstein
童年回憶型
"As a child, I was always curious about..."
5 萬人都這樣寫
抽象抒情型
"Science is the language of the universe..."
給狗看狗都搖頭
救世主型
"I want to use technology to change the world..."
TED Talk 句型，無內容
履歷重複型
"I've taken AP Bio, AP Chem, AP Physics..."
你的成績單已經有了
正確的 Why Major 開頭應該是「一個具體的時刻 / 觀察 / 失敗」。看下面對照：
CS Major：壞 vs 好
壞：
Ever since I started coding in middle school, I have been passionate about computer science. Programming allows me to solve problems creatively.
好：
The first time I wrote for i in range(10): and watched it print "Hello" ten times, I didn't think much of it. Two years later, when my recursive Fibonacci hit Python's default recursion limit at 1,000, I finally understood what a stack was.
I have always been fascinated by the human body and how it works. I want to become a doctor to help people.
好：
My grandmother's HbA1c was 9.2 the morning her cardiologist mentioned "metabolic memory." I went home and read three PubMed papers on hyperglycemic memory in vascular endothelium. None of them used the word "fascinating."
Economics fascinates me because it explains human behavior and helps us understand the world.
好：
My father's bubble tea shop raised prices by NT$5 in 2022. Sales dropped 23%. I spent that summer trying to figure out whether we had crossed our demand elasticity threshold, then learned the word for it.
"I was 9, my Lego bridge collapsed under a textbook. I added a triangle. It held three."
"I always loved reading"
"I read 'Norwegian Wood' in 8th grade. I didn't understand why Watanabe wouldn't choose Midori until I was 16."
"I was fascinated by stars"
"The first time I saw the Andromeda galaxy through my dad's old telescope, I realized I was looking at light that had left 2.5 million years ago."
注意：好的 Origin 都有 具體年齡 + 具體物件 + 具體理解。三者缺一不可。
五、第二幕 Exploration Evidence：證明你深入了
Exploration 是最技術的一幕。它的功能：讓教授相信你不只是「喜歡」，而是「鑽進去」。
可用素材：
證據類型
強度
範例
自學/自做專案
最強
"I built a Flask app to scrape Taipei rental prices..."
競賽 / 研究
強
"My ISEF project on titration sensitivity used..."
課堂 deep dive
中
"In AP Bio I asked Ms. Lin why we treated meiosis as discrete steps..."
閱讀 / 課外書
中
"I read Hofstadter's GEB the summer before junior year and..."
線上課程 (MOOC)
弱
"I completed Andrew Ng's ML course on Coursera..."
興趣 / 喜好
無
"I love reading about physics"
規則：給的 evidence 越「私人」、越「無人逼你做」，越有信服力。
例如：
弱 Exploration：
I have taken AP Calculus BC, AP Statistics, and AP Computer Science A. I also participated in math club for three years.
強 Exploration：
The summer after sophomore year, I tried to solve the Monty Hall problem on paper. My intuition kept saying the probability was 1/2. I wrote a Python Monte Carlo simulation to prove myself wrong. After 100,000 iterations, the door-switching probability was 0.6671. I finally understood Bayesian updating—but only because I refused to believe it.
"At MIT, I want to take 6.006 Introduction to Algorithms, then UROP with Prof. Erik Demaine on computational geometry."
"I want to study Bio at Duke to become a doctor."
"Duke's interdisciplinary Trinity Genome Sciences major + DUSON's clinical exposure starting sophomore year would let me bridge wet-lab to bedside earlier than most BS-BA programs."
"Economics at Penn will help me succeed in business."
"Wharton's M&T dual-degree, combined with Penn's Behavioral Economics minor under Prof. Maurice Schweitzer, is the only U.S. undergraduate path to the negotiation research I want to do."
規則：必須提到至少 1 個 specific course / professor / program。沒提的話，這篇 Why Major 跟其他 30 所學校的 Why Major 完全可以互換——adcom 會立刻看出。
七、「No Major」與「Specific Major」的差別
不是所有 Why Major 都長一樣。看學校：
學校類型
文書類型
寫法
Brown PLME（無 major 制）
"Why do you want a liberal arts education?"
寫跨領域的好奇心，不強調單一 major
UPenn Wharton
"Why business / why Wharton?"
寫極具體的商業興趣（不能是 generic finance）
MIT EECS
"Why engineering?"
寫技術好奇心，可以略不挑系
CMU SCS（直接申 CS）
"Why CS / Why CMU SCS?"
寫到演算法 / 系統 / AI 哪個次領域
Stanford undeclared
"Intellectual vitality"
寫思考方式，不一定要 major
最容易寫爛的是 UPenn Wharton 與 CMU SCS——這兩所要求極度具體的職涯方向。
Wharton 真實開頭示範
壞：
I want to study at Wharton because business has always interested me and I want to make an impact in finance.
好：
The first time I saw a "cap table" was when my father, a 38-year founder of a precision tooling firm, sold equity to a Korean investor. I didn't understand why the post-money valuation was 1.4× the pre-money. I spent two months reading "Venture Deals" and learned the difference between a 1× non-participating preference and a fully diluted basis.
❌ "I love coding because I can build anything I imagine"
❌ "Steve Jobs / Elon Musk inspired me"
❌ "AI will change the world"
✅ 寫具體技術問題（race condition、memory leak、ML overfit）
✅ 寫一個 buggy code 的下午
Bio / Pre-med 陷阱
❌ "I want to help people"
❌ "Watching my grandfather suffer made me want to be a doctor"
❌ "Volunteering at the hospital changed me"
✅ 寫具體 biology mechanism（不只是「我喜歡細胞」）
✅ 寫一個你自己提出的問題
Econ 陷阱
❌ "Economics explains everything"
❌ "I want to work in investment banking"
❌ "I want to fight poverty in developing countries"
✅ 寫具體市場觀察（不只是「我讀 Freakonomics」）
✅ 寫一次你的 prediction 對了 / 錯了
Humanities 陷阱
❌ "Books changed my life"
❌ "I want to be a writer / lawyer / professor"
❌ 引用一句名人格言
✅ 寫一本你真的吵架過的書
✅ 寫一個你重新讀後看法變了的文本
九、CS Major 完整範例段落（450 字）
組合一下完整結構。以下是我幫一位學生改完後的 final draft 開頭兩幕。
[第一幕 Origin]At 2:47 AM, my Python script for calculating π crashed after 11,238 digits. The error said OverflowError. I had just learned what a float was three weeks earlier. That night, I read the IEEE 754 standard from start to finish, fell asleep on my keyboard, and woke up with the keys "G" and "H" imprinted on my forehead. >[第二幕 Exploration]The next summer, I rewrote the script using decimal.Decimal. Then I learned that wasn't fast enough for 100,000 digits, so I switched to mpmath with arbitrary precision. I joined a Discord server for π-day enthusiasts and got into a three-hour argument with a Belgian retiree about whether the Chudnovsky algorithm converges faster than Ramanujan's series. I read both papers. I built a Streamlit dashboard comparing them. The Belgian was right. >By junior year, I was less interested in π and more interested in why certain algorithms converge faster than others. I took Stanford's CS161 (Algorithms) on YouTube and started reading "Introduction to the Theory of Computation" by Sipser. I didn't understand chapter 7. I am still trying. >[第三幕 Forward]At Carnegie Mellon's SCS, I want to take 15-251 Great Ideas in Theoretical Computer Science with Prof. Klaus Sutner, then pursue research on algorithmic information theory under Prof. Bernhard Haeupler's group. CMU is the only U.S. undergraduate program where I can start theoretical CS coursework before sophomore year—and where the floor for "I don't understand chapter 7" is high enough that I'll find friends who don't either.
「I love physics」沒人信。「I spent three weeks trying to derive Lagrangian mechanics from Newton's laws and gave up on day 19 when I couldn't see why kinetic energy was T = ½mv²」——adcom 立刻信。
