Chapter 13: The Use of Spies
From Sun Tzu’s The Art of War (Lionel Giles translation, 1910).
The closing chapter on intelligence. Sun Tzu says victory depends on foreknowledge, and foreknowledge comes from spies — not from omens. He names five classes of agent and demands they be trusted and rewarded.
Key principles
- Foreknowledge wins; it cannot come from superstition or guesswork.
- Use five kinds of agents: native, inward, converted, doomed, and surviving.
- Reward spies well and keep their work secret.
The chapter's argument, in depth
1. Foreknowledge is the whole edge
What Sun Tzu meant. Sun Tzu opens the final chapter by saying victory depends on foreknowledge, and that foreknowledge depends on spies, so that without it you cannot fight to advantage. He adds that this foreknowledge cannot come from omens or from the study of Heaven, but only from human means.
Today. He rules out the two most common sources of false confidence — patterns read from coincidence, and intuition dressed up as insight — and replaces them with deliberate collection. The claim is not that you can know the future, but that you can know the situation you are standing in, and that knowing it is cheap.
Use this when. You are about to commit under uncertainty. Separate what you actually know from what you are predicting, and put a price on the gap between them.
Related lesson: Foreknowledge is the edge.
2. Stinginess on intelligence is cruelty
What Sun Tzu meant. Sun Tzu says a ruler who is stingy with his spies is cruel to his people, because the whole enterprise depends on people being willing to report what they see and knowing they will be treated fairly for it. He names the cost of that economy directly.
Today. Intelligence systems fail on the incentive, not on the tooling. If reporting a bad sign is punished or ignored, the flow of bad news stops and what remains is selection: only the flattering observations arrive. You do not get less information — you get a filtered version of it that confirms you.
Use this when. Your information channels have started to agree with you more than they used to. That is either a sign the world settled or a sign the reporting quietly stopped.
Related lesson: Stinginess on intel is cruelty.
3. Run a spread of sources
What Sun Tzu meant. Sun Tzu names five classes of agent — the local inhabitant, the inward man, the converted enemy, the doomed man, and the surviving man — and says none should be used alone, because a single source can be bought, coerced, or mistaken, while a spread of them cross-checks each other.
Today. The structure survives the vocabulary. A single informant is a point of failure no matter how much you trust them, whereas two or three with different access and different incentives let you distinguish a true signal from one person’s bias. Cross-checking is not distrust; it is what makes any single account worth something.
Use this when. You are about to make a decision on one person’s account. Identify a second source with a genuinely different vantage point before you commit.
Related lesson: Run a spread of sources.
4. Many spies hide the system
What Sun Tzu meant. Sun Tzu says that while the sovereign and the general and the staff are doing their own work, the business of the spies goes on in secret and in disguise, so that the enemy does not know who is watching. He describes a network that is deliberately invisible as a network.
Today. The design principle is deniability of the system rather than secrecy of any single report. The point of many small watchers is that no one person holds the whole picture — including, deliberately, the people running them. Centralized confidence in one observer is the fastest way to be observed.
Use this when. You are building an information-gathering practice inside a larger organization. Decide in advance who knows the full map, because that number ought to be small.
Related lesson: Many spies hide the system.
5. Read the report, not just the words
What Sun Tzu meant. Sun Tzu says the intuitive sagacity of the commander is needed to judge the reports: the skill of the spy lies in the observation, but the skill of the reader lies in judging whether what was observed means what it appears to mean. He warns too against being deceived by an enemy’s apparent state.
Today. A raw observation is not a conclusion. A report says what someone saw, not what it means, and a single signal usually supports several readings. The reader’s job is to hold the alternative explanations open instead of converting the first plausible story into fact.
Use this when. A report arrives that fits a story you already believe. Write down two other explanations for it before you act on any of them, and check which one the evidence actually supports.
Related lesson: Read the report, not just the words.
6. Honor and reward your sources
What Sun Tzu meant. Sun Tzu says the ruler must use the intimate relations with his spies as a source of the greatest advantage to the state, and must reward them so that their trust holds. The obligations run both ways: the source gives honesty, and the ruler gives protection and fairness.
Today. Sources are relationships with a renewal date. If the value only flows one way, the source stops taking risks for you and the flow stops — usually without announcement, which is why intelligence quietly degrades long before anyone notices the change.
Use this when. You are relying on people outside your organization to keep you informed. Ask what you have done recently that keeps their cooperation worth the risk it carries.
Related lesson: Honor and reward your sources.
The intelligence audit
Six questions to run before you commit to a decision made under uncertainty. The final chapter is short, but the discipline it describes takes longer than reading it does.
- Truth. What do I actually know, as opposed to what am I inferring? Write the two lists separately and do not mix them.
- Sources. How many independent sources back my most important belief, and do they see the situation from different angles?
- Motivation. What does each source gain from telling me this, and what do they stand to lose by telling me the truth?
- Coverage. Who inside the other side sees what I need to know, and which vantage point am I failing to reach?
- Read. What are the two other explanations for the report I am about to act on?
- Cost. What am I paying for this intelligence, and what is the cost of the ignorance it replaces?
Common misreadings
- “Sun Tzu endorses spying on people, full stop.” The chapter is about obtaining foreknowledge in the context of war, and it is unusually explicit about how that information must be handled — rewards, protection, and the ruler’s own obligations. Read in a modern setting it is a chapter about the ethics and economics of knowing, not a license for surveillance.
- “The five classes of spy describe five kinds of undercover agent.” They are five sources with different relationships to the enemy — local inhabitants, people already inside, turned enemies, and the two desperate categories — not five tradecraft techniques. The classification runs on access and incentive, which is exactly why a spread of them cross-checks.
- “Sun Tzu rules out omens, but he leaves room for astrology.” He rules out divination as a source of foreknowledge, not as an activity. The chapter’s claim is epistemological: omens cannot tell you what the enemy is about to do, whereas people can. That is a statement about where knowledge comes from, not a prohibition on ritual.
- “If you cannot get intelligence, you should not act at all.” He says that without foreknowledge you cannot fight to advantage, but his own advice is to create foreknowledge by building the means to obtain it, and to act on the preparation you do have. The instruction is to go and find out, not to freeze until someone hands you certainty.
- “Reward the spies and everything is fine.” He also warns that the system must be hidden and that the ruler’s judgment is the final filter. Compensation without appraisal produces confident, well-paid reports that nobody checks. The reward keeps the flow honest; the reading keeps it true. You need both.
The forecast that only confirmed the plan
Imagine a product team that has spent a quarter building a scheduling feature for clinics. Two weeks before launch, a founder asks a well-connected contact for a read on the market. The contact is generous, encouraging, and uninformative — they describe the sector as promising and mention that several teams “are exploring this space.” The founder records this as validation. But the sentence contains no fact about any specific clinic, and no second source was consulted. The chapter’s method, applied honestly, would have made the gap visible: what do we know, as opposed to what are we hoping? A second source with a different vantage point — a clinic manager rather than a consultant — would probably have been far less encouraging. The launch decision would likely have been the same, but taken with an accurate picture of the uncertainty instead of a flattering one. That is the whole argument of the final chapter: foreknowledge is not a feeling that things will work, it is a specific and checkable account of what you know and what you are guessing.
Quotations from this chapter
Drawn from our sourced lesson database — each links to its permanent page with the full interpretation.
In the field
Real campaigns where this chapter was tested — including where the record shows it failing.
The doctrine is demonstrated on the attacking side and broken on the defending side, where a better plan lost to a better guess about the commander’s reputation.
The text’s intelligence was gathered and delivered and then discarded, which is the one thing the chapter’s machinery exists to prevent.
Modern applications
- Pay for real market intelligence instead of trusting hunches.
- Build trusted inside sources; protect them and compensate them fairly.
Related strategy topics
Frequently asked questions
Why does The Art of War end with spies?
Because Sun Tzu treats victory as a product of foreknowledge. The final chapter argues you cannot win without knowing the enemy’s true state, and that knowledge must be earned through reliable intelligence, not omens.