The Art of War, Chapter 3: 謀攻 (Attack by Stratagem)

Chapter 2 showed that a campaign’s cost is mostly set by how long it lasts. Chapter 3 draws the conclusion: the best result is the one that leaves the prize intact and the army unspent. It ranks the ways of winning, from persuasion at the top to storming a walled city at the bottom, and ends with the conditions under which a ruler or a commander can expect to win.

Like every chapter in this series, it has three parts: the reading, an analysis of what the chapter argues, and how finance professionals can interpret it in their own work. The original text follows the Chinese Wikisource edition of 《孫子兵法》. The Chinese and English explanations are Claude’s own paraphrase.

Reading the chapter

3.1

原文

孫子曰：凡用兵之法，全國為上，破國次之；全軍為上，破軍次之；全旅為上，破旅次之；全卒為上，破卒次之；全伍為上，破伍次之。

中文解說

孫子說：用兵的通則是，使敵國完整地歸順是上策，打破敵國就差一些；使敵軍完整地降服是上策，擊破敵軍就差一些；對旅、卒、伍也是一樣，使其完整歸順是上策，擊破它們就次一等。

English

The best outcome is to take a state whole. Breaking it is second best. The same ranking holds for an army, a battalion, a company and a squad: taking it intact is better than destroying it.

The ranking applies at every scale. What is won intact is worth more than what is won broken.

3.2

原文

是故百戰百勝，非善之善者也；不戰而屈人之兵，善之善者也。

中文解說

所以百戰百勝，並不是最高明的；不經交戰就使敵人屈服，才是最高明中的最高明。

English

So winning a hundred battles out of a hundred is not the highest skill. Making the enemy yield without fighting is.

This is the best known line of the chapter. A hundred wins still means a hundred fights, and each one costs what Chapter 2 described.

3.3

原文

故上兵伐謀，其次伐交，其次伐兵，其下攻城。攻城之法，為不得已。

中文解說

所以最好的用兵是破壞敵人的謀劃，其次是破壞敵人的外交與盟友，再其次是用軍隊交戰，最差的是攻打城池。攻城的辦法，是不得已才用的。

English

The best war is to attack the enemy's plans. Next best is to attack its alliances. Next is to attack its army. The worst is to besiege its cities, which is done only when there is no alternative.

This is a ladder of four rungs: 伐謀 (plans), 伐交 (alliances and relationships), 伐兵 (forces), 攻城 (fortified positions). Each step down costs more and is less certain.

3.4

原文

修櫓轒轀，具器械，三月而後成；距闉，又三月而後已。將不勝其忿，而蟻附之，殺士三分之一，而城不拔者，此攻之災也。

中文解說

製造攻城的大盾和攻城車，準備各種器械，要三個月才能完成；堆築攻城用的土山，又要三個月才能完工。將領按捺不住怒氣，驅使士兵像螞蟻一樣爬城，士兵死了三分之一，城還是攻不下來，這就是攻城帶來的災禍。

English

Preparing the shields and siege engines takes three months. Building the earth ramps takes another three. If the general cannot control his anger and sends his men up the walls like ants, a third of them die and the city still stands. That is the disaster of a siege.

The failure described here is one of temper. The time cost of 3.4 echoes Chapter 2, and the anger is what turns a costly option into a ruinous one.

3.5

原文

故善用兵者，屈人之兵而非戰也，拔人之城而非攻也，毀人之國而非久也，必以全爭於天下，故兵不頓而利可全，此謀攻之法也。

中文解說

所以善於用兵的人，使敵軍屈服而不靠硬戰，奪取敵人城池而不靠強攻，滅亡敵國而不拖延時日。一定要以完整的方式在天下爭勝，這樣軍隊不受損耗，而利益也能完整保全，這就是以謀略取勝的方法。

English

A skilled commander makes the enemy yield without fighting, takes cities without assaulting them, and ends a rival state without a long campaign. He contends for the whole prize, so his army is not worn down and the gain stays intact. This is the method of attacking by stratagem.

3.6

原文

故用兵之法，十則圍之，五則攻之，倍則分之，敵則能戰之，少則能守之，不若則能避之。故小敵之堅，大敵之擒也。

中文解說

所以用兵的原則是：兵力十倍於敵就包圍它，五倍就進攻它，兩倍就設法分散它，勢均力敵就要能戰，兵力少就要能防守，實力不如對方就要能避開。所以弱小的一方如果只是硬撐，就會成為強大一方的俘虜。

English

With ten times the enemy's strength, surround it. With five times, attack. With twice, divide it. If equal, be able to fight. If fewer, be able to defend. If outmatched, be able to avoid battle. A small force that stubbornly stands its ground will be captured by a large one.

The ratios are illustrations of scale, not data. The point is that the right action depends on relative strength, and that avoiding a fight is a legitimate choice. The last line is sometimes read as a warning against stubbornness and sometimes as a statement about weak forces fighting on regardless; the practical lesson is the same.

3.7

原文

夫將者，國之輔也。輔周則國必強，輔隙則國必弱。

中文解說

將領是國君的輔佐。輔佐周全可靠，國家一定強盛；輔佐有缺漏、與君主不和，國家一定衰弱。

English

The general is the ruler's support. If the support is complete, the state will be strong. If there are gaps in it, the state will be weak.

Some readers take 隙 as a gap in the general's ability, others as a rift between general and ruler. The next passage points to the second reading.

3.8

原文

故君之所以患於軍者三：不知軍之不可以進而謂之進，不知軍之不可以退而謂之退，是為縻軍；不知三軍之事，而同三軍之政，則軍士惑矣；不知三軍之權，而同三軍之任，則軍士疑矣。三軍既惑且疑，則諸侯之難至矣，是謂亂軍引勝。

中文解說

國君危害軍隊的方式有三種：不知道軍隊不可以前進卻命令前進，不知道軍隊不可以後退卻命令後退，這叫做束縛軍隊；不瞭解軍中事務卻干預軍隊的行政，軍士就會迷惑；不瞭解軍事上的權變卻干預軍隊的指揮，軍士就會疑慮。軍隊既迷惑又疑慮，其他諸侯就會乘機來犯，這就叫做自亂其軍而讓敵人取勝。

English

A ruler can harm his army in three ways. He orders an advance or a retreat without knowing whether the army can do it, which hobbles it. He interferes in the army's administration without understanding its affairs, which confuses the soldiers. He interferes in command without understanding military judgment, which makes them doubt. When the army is confused and doubtful, rival states will arrive, and the ruler has thrown away the victory himself.

The warning is about a decision maker who overrides the people closest to the facts.

3.9

原文

故知勝有五：知可以戰與不可以戰者勝，識衆寡之用者勝，上下同欲者勝，以虞待不虞者勝，將能而君不御者勝。此五者，知勝之道也。

中文解說

預知勝利有五個條件：知道能戰與不能戰的人會勝，懂得根據兵力多少靈活運用的人會勝，上下同心的人會勝，以有準備對付無準備的人會勝，將領有才能而國君不加干預的人會勝。這五點，就是預知勝利的方法。

English

There are five ways to know who will win. The side that knows when to fight and when not to will win. The side that knows how to use large and small forces will win. The side whose ranks share one purpose will win. The prepared side facing an unprepared one will win. The side with a capable general and a ruler who does not interfere will win. These five are the way to know victory in advance.

3.10

原文

故曰：知彼知己，百戰不殆；不知彼而知己，一勝一負；不知彼不知己，每戰必殆。

中文解說

所以說：既了解敵人又了解自己，打上百次仗也不會有危險；不了解敵人而只了解自己，勝負各半；既不了解敵人又不了解自己，每次作戰都一定有危險。

English

So it is said: know the other side and know yourself, and a hundred battles bring no danger. Know yourself but not the other side, and you win one and lose one. Know neither, and every battle puts you in danger.

This is one of the most quoted lines in the book. Note that it promises freedom from danger (不殆), not victory every time.

Analysis: what the chapter argues

The goal is the whole prize. 3.1 ranks every outcome by how intact it is. Taking something whole beats breaking it, at every scale from a state to a squad. The aim is to keep the value of what is won.

Not fighting is the stronger result. 3.2 says that a hundred victories are not the highest skill. This does not say that fighting is wrong. It says that fighting is a cost, and the best commander wins without paying it.

A ladder of options. 3.3 ranks four ways to contend: 伐謀, 伐交, 伐兵 and 攻城. Each step down is more expensive and less certain than the one above it. Reading 伐謀 as defeating the enemy’s plan, and 伐交 as cutting its alliances, is the usual reading. Some commentators understand 伐交 as winning over third parties by diplomacy. Both readings point to the same thing: shape the situation before force is needed.

Anger turns an option into a disaster. 3.4 gives the siege as the example. It takes six months of preparation, and the general’s loss of temper then destroys a third of his men for no gain. The passage is as much about self-control as about method.

The method restated. 3.5 repeats the principle in plain terms: yield without battle, take without assault, end without a long campaign. The words 非久 link back to Chapter 2.

Match action to relative strength. 3.6 gives a scale from surrounding to avoiding. The ratios are illustrations. What matters is the last three clauses: fight if equal, defend if smaller, avoid if outmatched. The final line is read in more than one way, but all readings warn against stubborn resistance from a weaker position.

The ruler can lose the war from the capital. 3.7 and 3.8 move from the battlefield to the relationship between ruler and general. The three faults are ordering what the army cannot do, running what the ruler does not understand, and commanding without understanding military judgment. 3.8 names the result: the ruler beats himself (亂軍引勝).

Five signs of a likely win. 3.9 lists them: judging when to fight, using the right force, a shared purpose, preparation, and a capable general left alone. Four of the five are about the home side, which is consistent with 3.10.

Know the other side and yourself. 3.10 gives three cases. Both known gives safety. One known gives an even chance. Neither known gives constant danger. Note that the phrase is 不殆, free from peril, which is not the same as always winning.

How finance professionals can read it

This section is Claude’s interpretation. Sun Tzu wrote about armies, and none of the mappings below are his. Chapter 2 priced the commitment. This chapter asks how to reach the objective without paying the full price of a contest, and what a decision maker owes the people closest to the facts. Nothing here suggests acting outside the law. Any approach to a counterparty, a competitor or a market participant should be checked with legal and compliance advisers first.

Investing

• Prefer the position that does not need a fight. 3.2 and 3.5 suggest looking first for opportunities that do not require a contested process, such as a negotiated purchase, a private placement or a position where the price is not being bid up by others. This is an analogy for cost, not a rule that quiet deals are better.
• Rank your options by cost and certainty. The ladder in 3.3 can be borrowed as a habit: before choosing the expensive route, such as a crowded auction or a prolonged activist campaign, list the cheaper ones and say why each was passed over.
• Protect the whole value. 3.1 prefers taking something intact. In practice, ask whether a transaction preserves the asset’s team, customers and franchise, or whether the approach itself damages what is being bought.
• Do not let temper set the price. 3.4 describes a general losing control. Overpaying after losing a competitive round, or doubling down to avoid admitting a mistake, is the investment version of sending men up the wall.
• Know when you are the smaller force. 3.6 says to avoid when outmatched. A smaller investor can decline to fight for a crowded asset and look for situations where its size is an advantage.

Raising funds

• Win the plan before the pitch. 伐謀 can be read as shaping the situation before the formal process starts. This means investors understanding the company’s plan clearly, and the company understanding what each investor needs, through open and lawful conversations.
• Relationships are part of the ladder. 伐交 maps loosely onto who else is in the conversation: existing investors, lead investors, advisers and references. An introduction from someone the investor trusts changes the cost of the process. This must be done honestly. Overstating interest from other parties, or implying commitments that do not exist, is misrepresentation.
• Do not fight for a round you cannot win. 3.6 suggests avoiding when outmatched. A company that is not a fit for an investor’s mandate saves both sides time by moving on.
• Founders and boards: leave the specialists room. 3.8 applies to a board or major investor who overrides management on matters they do not know. Governance exists for oversight, but oversight and interference are not the same.

M&A

• The best deal may not need an auction. 3.2 and 3.5 can be read as favouring a bilateral approach that preserves value over a contested one that raises the price. Whether this is open to a buyer depends on the target’s board, its duties to shareholders and the applicable rules. Take legal advice before approaching a target or its holders.
• Alliances matter. 伐交 maps loosely onto the stakeholders around a deal: shareholders, lenders, regulators, employees and key customers. Understanding their positions early, through proper channels, is cheaper than fighting them late.
• Judgment and temper in contested situations. 3.4 is a warning for hostile or competitive processes. Pride, rivalry and sunk cost push parties to escalate. Set a walk-away price in advance and treat any move above it as a decision that needs a fresh review.
• Preserve what you are buying. 3.1 favours acquiring intact. A buyer that wins by breaking the target’s team or customer base has paid for something worth less.
• Let the deal team do its job. 3.8 and 3.9 warn against senior sponsors who override the people doing the work. Principals set the mandate and the price limit. The execution team needs room to carry it out.

Across all three

• Use lawful information only. 3.10 is about knowing the other side and yourself. In finance, that means public filings, proper diligence, information the counterparty has chosen to share, and expert networks used within the rules. It does not mean material non-public information, insider trading, misrepresenting your position, or breaching confidentiality or law. If you are unsure whether information is permitted, stop and ask compliance.
• Know your own side as well as the other. The sentence in 3.10 gives equal weight to both. Many teams research the counterparty carefully and are vague about their own limits, funding, mandate and tolerance for delay.
• Treat no fight as an outcome worth planning. Decide in advance what a good walk-away looks like, and credit the team for it.
• Leave judgment with those who have the facts. 3.8 and 3.9 together suggest that a principal sets direction and limits, then does not override the specialists on details.

Questions to carry into the next chapter

Chapter 4 (軍形, Tactical Dispositions) moves from how to win without fighting to how to make yourself hard to beat first and wait for the opening.

• In your last significant decision, what cheaper options did you rank before choosing the one you took?
• When was the last time temper, pride or sunk cost raised the price you were willing to pay?
• How well can you describe your own limits, funding and mandate, compared with how well you know the counterparty?