LovisMind

第 00 章

Strang 谈教学：把数学讲成人能想明白的东西

一次访谈里，Strang 讲他为什么不在课上赶进度、为什么愿意在学生面前卡住、以及他如何判断一个细节该讲多深。

Strang on Teaching: Making Mathematics Something a Person Can Think Through

In this interview Strang explains why he refuses to rush, why he is willing to get stuck in front of students, and how he decides how deep a detail should go.

首发于 2019.08.19中 / EN
播客访谈linear-algebrateachingmathematics
语言
字号
目录
篇内目录

这段访谈的位置

这不是一节数学课，而是 18.06SC 的一段访谈：Sarah Hansen 提问，Gilbert Strang 回答他如何教线性代数。把它放在课程最前面读，是因为它解释了后面每一讲的讲法——为什么 Strang 总是先画图、再写符号、最后才证明；为什么他会在黑板上停下来，让全班跟他一起想下一步。

时间轴

时间内容
0:06开场：来自世界各地的学生
0:22「我没有固定套路」：谈教学风格
0:55给学生时间思考
1:57数学有序，但不能急
2:14先写符号，再问含义
2:26让学生自己看出为什么成立
3:0118.06 作为基础课的定位
3:33在课堂上实时解题
3:55为什么喜欢学生
4:19幽默在教学中的作用
4:49把难懂的地方再想一遍
5:35如何感知课堂的接受程度
6:20来自世界各地的反馈
6:58学习数学的建议
7:26结束语

对话脉络

访谈围绕五个问题展开：

  1. 你的教学体系是什么？
  2. 你为什么会在课上「假装困惑」或真的卡住？
  3. 幽默在课堂里起什么作用？
  4. 你怎么决定一个细节讲多深？
  5. 面对几百人、理解程度不一的大班，你怎么和他们建立连接？

Strang 的回答可以概括成一句话：不要替学生把思考做完。

核心主张

没有体系，只有「和你一起想」

被问到教学体系时，Strang 的第一反应是「我没有体系」。他给出的不是方法清单，而是一个姿态：

  • 他喜欢学生，想帮上忙；
  • 关键不是宣布答案，而是和学生一起把问题重新想一遍；
  • 想一遍需要时间，所以不能把证明「嗖」地讲完。

这也解释了他对「假装困惑」传闻的回应：他确实会在关键处停下来，未必是假装，而是因为下一步本来就值得停一下。

符号、意义、原因，要分三次到达

Strang 描述了一个顺序：

  1. 先在黑板上写出符号；
  2. 再问「这是什么意思」；
  3. 最后才问「为什么它是对的」。

如果一上来就要求证明，大多数人还没有建立起「它大概是对的」的直觉。先让直觉成立，证明才有落点。

一句话总结： 教学不是把正确的东西快速说完，而是让听的人有机会自己走到那个正确的地方。

愿意在学生面前卡住

访谈者问他：在几百人面前实时思考、甚至走进死胡同，会不会有风险？Strang 承认会，而且确实发生过。但他认为这恰恰是合理的：

  • 学生本来就会卡住；
  • 老师卡住之后怎么走出来，是一种可迁移的示范；
  • 如果老师永远流畅，学生只会觉得自己不够聪明。

这不是表演性的示弱，而是把解题过程而不是解题结果当作教学内容。

幽默的作用是「让它像人做的事」

Strang 把幽默和「humanize」联系在一起。数学书不是人，但它是人写的；一个双关语（比如「按高斯给定的顺序」）能让学生意识到，眼前这套符号是一个人可以想出来、也可以开玩笑的东西。

幽默在这里不是调味品，而是降低距离感的手段。

细节讲多深：看你是否真的在重新想

他的判断标准很朴素：如果他自己没有在重新思考，那多半是讲得太快了。当他把问题再想一遍，该用哪些词、该走哪几步会自然浮现；反之，照本宣科就会失去和课堂思考的连接。

大班教学的关键也不是覆盖所有人，而是始终意识到听众在哪里，而不是只盯着「A 的逆」这类符号操作。

可行动的结论

对学习者，这段访谈给出三条可操作的提示：

  1. 允许自己停：看不懂时不要跳过，回到「符号—意义—原因」的第一层重来。
  2. 复述过程而不是答案：合上书，把某一步为什么成立讲给自己听；讲不下去的地方就是没懂的地方。
  3. 接受卡住：卡住是学习过程的一部分，关键是练习「从死胡同里退出来」的动作。

对讲授者，则是一个更简单的检查表：

  • 我是不是在替听众思考？
  • 我有没有给「下一步是什么」留出时间？
  • 我讲的是结果，还是到达结果的路？

与课程的关系

后面的每一讲都能看到这些主张的具体形态：第一讲用行图像、列图像、矩阵三种视角讲同一个方程组；消元法把「怎么算」拆成一步步可检查的动作；四个基本子空间把抽象概念落回几何。Strang 在访谈里说的是教学哲学，课堂上呈现的是它的执行版本。

理解检验

1. 为什么 Strang 主张先建立「它大概是对的」的直觉，再给证明？

参考答案：因为证明需要一个落点。访谈里他给出的顺序是「先写符号，再问这是什么意思，最后才问为什么它是对的」。如果一上来就证明，听众手里只有一串变形，既不知道这串变形要把他带到哪里，也无法判断哪一步是关键的。先有「它应该是对的」的直觉，证明才是在解释一个他已经在乎的结论；否则证明只是把符号又排列了一次。另一个实际好处是：有直觉的人能预判下一步该往哪走，也能在计算出错时察觉不对。

2. 「在学生面前卡住」对学习者有什么具体价值？

参考答案：它把教学内容从「结果」换成了「过程」。学生自己本来就会卡住，老师当场卡住再走出来，示范的正是可迁移的那部分能力——如何判断走进了死胡同、如何退回上一个可靠的位置、如何换一个视角。反过来，如果老师永远流畅，学生会把「想不出来」误当成自己不够聪明，而不是当成解题的正常阶段。

3. 如果让你用一句话判断自己是否真的理解了一个定理，你会怎么判断？

参考答案：合上笔记，能否不问「结论是什么」，而把「每一步为什么必须这样走」重新推一遍。换句话说：能复述过程而不是复述答案。讲不下去的那一步，就是还没懂的地方——这正是访谈里说的「符号—意义—原因」三层里最后没有打通的那一层。

Where this interview sits

This is not a mathematics lecture. It is an interview from 18.06SC: Sarah Hansen asks, Gilbert Strang answers, and the subject is how he teaches linear algebra. It belongs at the front of the course because it explains the style of every lecture that follows — why Strang always draws the picture before writing the symbols, and why he stops at the board to let the room find the next step with him.

Timeline

TimeTopic
0:06Opening: students around the world
0:22"I have no system": on teaching style
0:55Giving students time to think
1:57Mathematics is orderly, but you cannot rush
2:14First the symbols, then the meaning
2:26Let students see why, rather than telling them
3:01The role of 18.06 as a basic course
3:33Solving problems live in class
3:55Why he likes students
4:19The role of humour
4:49Working a hard idea through again
5:35Sensing how the class is following
6:20Feedback from students worldwide
6:58Advice on learning mathematics
7:26Closing

The conversation's shape

Five questions organize the discussion:

  1. What is your teaching system?
  2. Why do you sometimes feign confusion, or genuinely get stuck?
  3. What role does humor play?
  4. How do you decide when to go into detail?
  5. How do you connect with a large lecture hall full of people at different levels?

His answers reduce to one instruction: do not do the thinking for the student.

The central claims

No system, just thinking alongside them

Asked for his system, Strang says he has none. What he offers instead is a stance:

  • he likes students and wants to help;
  • the key is to think the question through again with them, not to announce the answer;
  • thinking takes time, so a proof cannot be zipped through.

This also explains the "feigned confusion" rumor. He does pause at critical points, and the pause is not a performance: the next step is genuinely worth a moment.

Symbols, meaning, reason — in three passes

Strang describes an order:

  1. put the symbols on the board;
  2. ask what they mean;
  3. only then ask why the statement is true.

Demanding a proof first assumes an intuition that most listeners have not built yet. Let the intuition stand up, and the proof has somewhere to land.

One-sentence summary: Teaching is not saying the correct thing quickly; it is giving the listener a chance to arrive at the correct place themselves.

Willing to get stuck in public

The interviewer asks whether thinking live in front of hundreds of people is risky. He admits it is, and that he has lost the thread or hit dead ends. He thinks that is fine:

  • students will hit dead ends anyway;
  • watching a teacher climb out of one is a transferable lesson;
  • a teacher who is always fluent only makes students feel inadequate.

The point is to make the process of solving, not the result, the content of the class.

Humor humanizes the subject

Strang ties humor to humanizing mathematics. A book is not a person, but a person wrote it; a pun — "keep things in their Gauss-given order" — signals that the symbols in front of you were produced by someone who could also make a joke.

Humor is not decoration here. It reduces distance.

How deep to go: are you actually rethinking it?

His criterion is simple: if he is not thinking it through himself, he is probably going too fast. When he re-derives the step, the right words and the right sequence appear on their own; reciting from memory breaks the connection with the room.

Connecting with a large class does not mean reaching every person. It means staying conscious of where the audience is, rather than staring at an operation like "A inverse".

Actionable conclusions

For a learner, three practices follow:

  1. Allow the pause. When a step does not land, go back to the first layer — symbols, then meaning, then reason.
  2. Recite the process, not the answer. Close the book and explain to yourself why a step holds. Where the explanation breaks is where the understanding is missing.
  3. Accept being stuck. Being stuck is part of learning; the skill to practice is the move that gets you out of the corner.

For a teacher, a short checklist:

  • Am I doing the audience's thinking for them?
  • Did I leave time for "what is the next step?"
  • Am I teaching the result, or the road to it?

How it relates to the course

Every later lecture shows these principles in action: Lecture 1 reads one system through the row picture, the column picture, and the matrix; elimination turns "how to compute" into a sequence of checkable moves; the four subspaces bring abstraction back to geometry. The interview states the philosophy; the lectures are its execution.

Check your understanding

1. Why does Strang build the intuition that a statement is true before offering a proof?

Reference answer: A proof needs somewhere to land. Strang's stated order is: write the symbols, ask what they mean, and only then ask why the statement is true. Without the first two steps the listener has a string of manipulations with no destination and no way to tell which step matters. Intuition also does practical work: someone who expects the result can anticipate the next move and notice when a computation has gone wrong.

2. What specific value does watching a teacher get stuck have for a learner?

Reference answer: It moves the lesson from the result to the process. Students get stuck anyway; a teacher who gets stuck and then climbs out demonstrates the transferable skill — recognizing a dead end, backing up to the last reliable point, and changing viewpoint. A teacher who is always fluent instead teaches students to read difficulty as a sign that they are not smart enough.

3. In one sentence, how would you tell whether you truly understand a theorem?

Reference answer: Close the notes and re-derive why each step is forced, rather than reciting what the conclusion says. You understand it when you can reconstruct the path; the step where the reconstruction stalls is the part you have not understood — precisely the missing layer in the symbols–meaning–reason sequence.

来源