The Art of War, Chapter 13: 用間 (The Use of Spies)

Chapter 12 ended on fire attacks and the discipline of acting only when the gain is clear. Chapter 13 closes the book by asking where the information behind all those decisions comes from. Its answer is people who actually know, not omens, analogies or guesses, and it argues that paying for that knowledge is cheap compared with the cost of a campaign fought blind.

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

13.1

原文

凡興師十萬，出征千里，百姓之費，公家之奉，日費千金，內外騷動，怠于道路，不得操事者，七十萬家。相守數年，以爭一日之勝，而愛爵祿百金，不知敵之情者，不仁之至也，非人之將也，非主之佐也，非勝之主也。

中文解說

凡是出動十萬大軍、遠征千里，百姓的負擔和官府的供給，每天要耗費千金。國內外一片騷動，人們疲於奔波在道路上，無法安心耕作的人家多達七十萬戶。雙方相持好幾年，只為爭一天的勝利，卻吝惜爵位俸祿和百金的賞賜，不肯花錢去了解敵情，這是極不仁慈的做法。這樣的人不配做軍隊的統帥，不配做君主的輔佐，也不是能取得勝利的君主。

English

Raising an army of a hundred thousand and marching it a thousand li costs the people and the state a thousand pieces of gold a day. There is unrest at home and abroad, people are worn out on the roads, and seven hundred thousand households cannot work their land. Armies then face each other for years to win a single day. Yet some rulers will not spend a little on rank or gold to learn the enemy's condition. Sun Tzu calls this the height of inhumanity. Such a person is no leader of men, no adviser to a ruler, and no master of victory.

The figures illustrate scale, as in Chapter 2. The argument is about proportion: a small outlay on information set against an enormous outlay on the campaign.

13.2

原文

故明君賢將，所以動而勝人，成功出於眾者，先知也。先知者，不可取於鬼神，不可象於事，不可驗於度，必取於人，知敵之情者也。

中文解說

所以英明的君主和賢能的將領，一出動就能戰勝敵人，成功超出一般人，靠的是事先掌握情況。這種事先的了解，不能靠祈求鬼神得到，不能靠拿相似的事情去類比，也不能靠天象度數去驗證，一定要從人那裡取得，也就是從了解敵情的人那裡取得。

English

The reason enlightened rulers and capable generals win when they move, and achieve more than ordinary people, is foreknowledge. Foreknowledge cannot be obtained from spirits or gods. It cannot be got by analogy with past events, and it cannot be checked against calculations of the heavens. It has to come from people who know the enemy's situation.

This is the core of the chapter. Three sources are ruled out: omens, analogy and astronomical calculation. Some readers take 象於事 to mean reasoning by resemblance to similar cases. That is the reading used here.

13.3

原文

故用間有五：有鄉間，有內間，有反間，有死間，有生間。五間俱起，莫知其道，是謂「神紀」，人君之寶也。

中文解說

使用間諜有五種：有鄉間，有內間，有反間，有死間，有生間。五種間諜同時運用，敵人無從知道其中的門道，這叫做神妙的法則，是君主的法寶。

English

There are five kinds of spy: local, inside, converted, doomed and surviving. When all five operate together, no one can trace the method. This is called the divine skein, and it is a ruler's treasure.

The point is the combination. Each kind has a different vantage point, and together they make the picture hard for the other side to see through or disrupt.

13.4

原文

鄉間者，因其鄉人而用之；內間者，因其官人而用之；反間者，因其敵間而用之；死間者，為誑事於外，令吾間知之，而傳於敵間也；生間者，反報也。

中文解說

鄉間，是利用敵方的同鄉百姓；內間，是利用敵方的官吏；反間，是利用敵方派來的間諜；死間，是在外面製造假情報，讓我方的間諜知道並傳給敵方的間諜；生間，是能活著回來報告情況的人。

English

A local spy is recruited from the enemy's own countrymen. An inside spy is recruited from among the enemy's officials. A converted spy is an enemy agent turned to your use. A doomed spy is given false information to pass on, and carries it to the enemy's agents. A surviving spy comes back and reports.

The definition of the doomed spy is terse and some commentators read it differently, but the usual reading is that he carries deliberate misinformation and is expected to be lost when it is found out.

13.5

原文

故三軍之事，莫親於間，賞莫厚於間，事莫密於間，非聖智不能用間，非仁義不能使間，非微妙不能得間之實。微哉！微哉！無所不用間也。間事未發而先聞者，間與所告者皆死。

中文解說

所以在全軍的事務中，沒有比間諜更親近的，賞賜沒有比間諜更優厚的，事情沒有比間諜工作更機密的。沒有超凡的智慧不能使用間諜，沒有仁慈和道義不能驅使間諜，沒有細膩精妙的心思不能分辨間諜帶回的情報是否真實。微妙啊，微妙啊！沒有哪個地方不能用間諜。間諜的事還沒進行就被洩露出去，那麼間諜和聽他報告的人都要處死。

English

Of all in the army, no one should be closer to the commander than the spy. No one should be rewarded more generously, and no matter should be kept more secret. Using spies takes wisdom, handling them takes decency and fairness, and judging what they bring back takes subtlety. It is a delicate business, and there is no place where spies cannot be used. If news of a spy's mission leaks before it begins, the spy and the person who was told are both put to death.

The last sentence is harsh and shows how seriously secrecy was taken. The sentence before it is the more useful one: the hard part is judging whether the report is true.

13.6

原文

凡軍之所欲擊，城之所欲攻，人之所欲殺，必先知其守將、左右、謁者、門者、舍人之姓名，令吾間必索知之。必索敵人之間來間我者，因而利之，導而舍之，故反間可得而用也；因是而知之，故鄉間、內間可得而使也；因是而知之，故死間為誑事，可使告敵；因是而知之，故生間可使如期。五間之事，主必知之，知之必在於反間，故反間不可不厚也。

中文解說

凡是要攻擊的軍隊、要攻打的城池、要殺掉的人，一定要先了解他的守將、左右親信、負責傳達的人、守門的人和門客的姓名，要命令我方的間諜一定去查清楚。一定要搜查出敵方派來刺探我方的間諜，用利益收買他們，加以引導後放回，這樣反間就可以為我所用。靠反間得到的情況，鄉間和內間就可以被利用；靠這些情況，死間才能傳播假情報給敵人；靠這些情況，生間才能按預定時間回來報告。五種間諜的事，君主必須掌握，而掌握的關鍵在反間，所以對反間不可不給予優厚的待遇。

English

Before attacking an army, besieging a city or killing a person, first learn the names of the commander, his attendants, the gatekeepers, the door guards and his staff. Have your spies find these out. Look for enemy spies who have come to spy on you. Tempt them with gain, guide them, and send them back, and a converted spy can be used. Through that, local and inside spies can be recruited. Through that, the doomed spy can carry false news to the enemy. Through that, the surviving spy can return on schedule. The ruler must understand all five, and that understanding depends on the converted spy, so the converted spy must be treated generously.

This passage gives a chain of dependency and a practical rule: know the specific people around the target, and do not stint on those who are your best source. The converted spy is the weakest part of any finance reading, because it involves turning someone against their own side.

13.7

原文

昔殷之興也，伊摯在夏；周之興也，呂牙在殷。故明君賢將，能以上智為間者，必成大功。此兵之要，三軍之所恃而動也。

中文解說

從前殷朝興起時，伊摯曾在夏朝任職；周朝興起時，呂牙曾在殷朝任職。所以英明的君主和賢能的將領，能夠用最聰明的人做間諜，必定成就大功。這是用兵的關鍵，全軍都要依靠它來行動。

English

When the Yin dynasty rose, Yi Zhi had served under the Xia. When the Zhou rose, Lü Ya had served under the Yin. So an enlightened ruler and capable general who can use the most intelligent people as spies will achieve great things. This is the essence of warfare, and the whole army depends on it in order to move.

Some readers take Yi Zhi and Lü Ya as examples of spies in the narrow sense. Others read them as wise men who knew the old regime from the inside and so could advise the new one. The text itself does not settle it.

Analysis: what the chapter argues

Ignorance is the expensive choice. 13.1 sets the daily and social cost of a campaign against a small payment for information. Refusing the small payment is called the height of inhumanity, because the cost of the error falls on others. The numbers are scale illustrations, not data.

Foreknowledge must come from people. 13.2 rules out three sources: spirits, analogy and calculation. The positive claim is that what you need to know sits in the heads of people who are close to the situation. This is a statement about method more than about spying.

Combining sources is the strength. 13.3 and 13.4 list five types, each with a different vantage point. The value is in operating them together, since a single source can be deceived or lost. The exact sense of 死間 and of 象於事 is read differently by different commentators, and the readings here are the common ones.

Relationship and reward are part of the method. 13.5 says the informant should be close, well rewarded and kept in confidence. It also says that handling such people takes decency and fairness, and that judging their reports takes subtlety. The weak point is not collecting information, it is deciding whether it is true.

Specific names beat general impressions. 13.6 asks for the actual names of the people around the target before acting. The passage then builds a chain in which each source depends on the one before. Its stress on the converted spy shows that the most valuable information comes from those who already stand between the two sides.

Capable people are worth enlisting. 13.7 gives two historical cases of wise men who had served the previous regime. How literally they count as spies is disputed. The consistent message is that the best information holders are intelligent people, and that the decision to move rests on what they report.

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.

This chapter is about espionage, and the finance reading is limited to lawful information gathering and professional posture. That means public filings, management meetings, expert calls conducted within the rules, customer and channel checks that respect confidentiality, and data rooms opened under an NDA. It does not mean material non-public information, insider trading, misrepresentation, bribery, inducing anyone to breach confidentiality, or any breach of law or professional duty. The double agent and the doomed spy have no finance equivalent and are not mapped here. Compliance and legal teams decide what is permitted in any given situation.

Investing

• Weigh the cost of research against the size of the position. 13.1 argues that skimping on information is the expensive choice. A thin research budget on a large position is the same mistake in small form.
• Prefer people who know to people who infer. 13.2 rules out omens and analogy. For an investment, that suggests giving less weight to narrative and to resemblance to past cases, and more to filings, management, customers, suppliers and licensed expert networks used within the rules.
• Use several independent sources. The five types in 13.3 suggest combining views from different positions: the company, the market, the industry, former staff through proper channels, and your own analysis. Agreement across independent sources means more than volume from one.
• Test every source for reliability. 13.5 puts the hard part in judging whether the report is true. Ask what each source gains from saying what it says.
• Know the specific people. 13.6 asks for names. Know who runs the business, who signs off and who can overrule them.

Raising funds

• Learn the investor before the pitch. Public information on a fund’s mandate, past investments, check size and decision process is lawful and useful. Approach people who can speak for the fund, not a back channel.
• Take references and feedback seriously. Ask for the names of the people involved in the decision, and ask passed investors honestly why they passed. Treat the answers as data, and check them against what other investors say.
• Reward the people who help. 13.5 says to treat informants generously. For a raise, the equivalent is giving advisers, introducers and early supporters honest credit, clear terms and prompt follow-up, never improper inducements.
• Be careful what you share. If investors receive confidential material, keep to the confidentiality terms and to the securities rules on what can be said and to whom.

M&A

• Treat diligence as the chapter’s information spend. The cost of a mistaken acquisition is very large. The cost of thorough diligence is small by comparison. Skimping is the 13.1 error.
• Use the data room, management sessions and advisers. These are the lawful channels. Use them to find the real decision makers, the key customers, the people the business depends on and the points where the numbers rest on judgment.
• Run customer and channel checks inside the rules. These can be done with consent and under the terms agreed by the parties. Do not approach the target’s staff or customers in ways that breach confidentiality or the exclusivity terms.
• Read the seller’s own materials critically. 13.5 warns that the hard part is judging what is reported. A management presentation is a source, not the truth, and each claim needs a second source.
• Do not induce a breach. Do not offer anything to an employee, adviser or counterparty in return for confidential information. This is bribery or an inducement to breach duty, and the transaction can be unwound by it.

Across all three

• Make the information budget explicit. Decide in advance what the decision is worth and how much lawful research it justifies.
• Separate what is known from what is inferred. Mark each key fact by its source, and mark what rests on analogy or narrative.
• Stay inside the lines. If a piece of information arrives and it is unclear whether it is public, stop and ask compliance before acting on it.

Questions to carry forward

• On your current largest decision, what have you spent on finding out, and is that proportionate to what is at stake?
• Which of your key beliefs rest on analogy or narrative rather than on someone who actually knows?
• Who are the specific people you would need to understand before committing, and do you know their names and incentives?