The Use of Spies

Core idea

The book closes with what is, structurally, the operational ground of every topic that came before it. Every recommendation in The Art of War depends on accurate knowledge of the enemy. The Use of Spies finally addresses where that knowledge comes from. Sun Tzu's answer is severe: it cannot come from the spirit world, from precedent, or from astrology — it can only come from people.

The topic has three movements:

1. The economic argument. A 100,000-man campaign costs 1,000 gold ingots per day. Yet some rulers begrudge "the few hundred gold ingots needed to confer ranks and emoluments on spies" — and so go to war in complete ignorance of the enemy. "To begrudge the expense is the height of inhumanity."
2. The five kinds of spies. Locals, insiders, double agents, "dead" spies, and live agents — a complete typology of human intelligence, each with its own use.
3. The centrality of the double agent. Of the five, the double agent is the key. He enables all four of the others.

Sun Tzu's argument: "In war, intelligence is of the essence, for it is what the armies depend upon in their every move."

The topic's placement at the end of the book is deliberate. Everything that came before — calculation, position, terrain, manoeuvre, fire — assumed accurate knowledge of the opponent. The Use of Spies is the foundation that the rest of the building rests on.

Why it matters

Intelligence is the cheapest weapon

Sun Tzu's opening argument is a piece of unsentimental cost-benefit reasoning. The campaign costs 1,000 gold ingots per day and disrupts the lives of 700,000 people. A spy network costs "a few hundred ingots" total. The intelligence budget is rounding error against the campaign budget — and yet rulers routinely cut it first. The reasoning: it is invisible until it works. The discipline: spend on it anyway, because the upside dwarfs the cost.

Foresight comes from people, not from systems

The topic's epistemological claim is striking for its era. Sun Tzu explicitly rejects three sources of foresight: the spirit world, past events, and astrology. Intelligence, he says, comes only from "other people — those with the requisite knowledge of the enemy's situation." This is a hard-headed empirical view of strategic forecasting that anticipates modern intelligence practice by two and a half millennia.

The modern equivalent: not "what does the model predict?" but "whom do we know who has direct access to the information?" Strategy benefits more from one well-placed source than from a hundred analyst reports based on public data.

The double agent is the master key

Of the five kinds of spies, the double agent — an enemy spy who reports to your side as well — is the key that unlocks the other four. Why? Because the double agent knows the enemy's network from the inside. Through him, you learn which locals can be recruited, which insiders are accessible, what false information to feed the dead spies, and how to position the live agents. The double agent multiplies the value of every other spy.

The implication: treat the double agent with extreme generosity. Sun Tzu is explicit — "the operative to be treated with particular generosity." A double agent who feels under-rewarded becomes either useless or, worse, a triple agent.

Key takeaways

Mental model

Copy sourceZoom

The five kinds of spies

Local spies

"The enemy's compatriots in our employ." Civilians in the enemy's territory — merchants, farmers, travellers — who know the local conditions and report back. Cheap, low-risk, low-information-density per agent, but vast coverage in aggregate. Modern equivalent: industry analysts, conference attendees, suppliers who serve multiple competitors, customers who shop both sides.

Insiders

"Enemy officials in our employ." People with formal positions inside the enemy organisation — much more dangerous, much more valuable. A single insider can deliver more usable intelligence than a hundred locals. Modern equivalent: former employees with continued contact, partners with deep visibility, board members of allied companies.

Double agents

"Enemy spies who report to our side as well." The enemy sent them to spy on you; you have turned them, so they report to both sides — but the side you've turned them to gets the real information. The double agent is the topic's hero. He knows the enemy's intelligence network, can be used to feed false information back, and reveals the points of access through which all other spies can be placed.

Dead spies

"Pawns in our employ whom we intentionally deceive, so that they relay false information to the enemy." This is the darkest category. The spy himself believes he is reporting real information; he is deliberately fed false information so that, when caught, he convinces the enemy of the false picture. He pays the price — usually with his life — for a deception he did not know he was carrying.

The category exists in modern intelligence too, with various softer forms. Sun Tzu's bluntness about the moral cost is unusual; he does not pretend the dead-spy tactic is anything other than what it is.

Live agents

Spies sent into enemy territory who return with information. Conventional espionage. The lowest-leverage category, but the most reliable — you know what they observed because they came back to tell you.

The economics

The topic's opening cost-benefit reasoning is one of the cleanest pieces of economic argument in the book:

| Line item | Cost | | ----------------------------------------------- | ------------------------------------ | | 100,000-man campaign, 1,000-league deployment | 1,000 gold ingots/day | | 700,000 people displaced from their normal work | Indirect, but real | | Full intelligence network with five spy types | A few hundred gold ingots, total |

Sun Tzu's point: the intelligence budget is approximately 0.01% of the campaign budget — and the rulers who skimp on intelligence are the ones who lose the campaign. A modern equivalent: large companies spending hundreds of millions on a market entry while their competitive intelligence team has two analysts and a $50,000 budget.

The names rule

A small but striking line in the topic: it is always necessary to know in advance the family and personal names of the enemy commander's retainers, counsellors, gate officers, and sentries. Not their roles — their names.

The rule is more strategic than it appears. To deploy a spy effectively you need to know whom to bribe, whom to flatter, whom to threaten, whom to feed false information. None of this works against an abstract "the enemy general's adjutant." It works against "Lieutenant Han, whose wife's brother served under us at the previous campaign and who is in debt to the moneylender at the eastern gate." Specificity is the precondition for leverage.

The discipline around the weapon

Like Attack by Fire's fire attack, intelligence is a weapon with strict preconditions:

• Only the most perceptive ruler understands how to employ spies.
• Only the most humane and just commander knows how to place them in the field. (Humane because spies require trust, often built across years; just because they need reliable rewards.)
• Only the most subtle interrogator can get the whole truth out of them.
• Premature disclosure means death for the spy and everyone he confided in.

The discipline runs through the topic: what cannot be done with maximum secrecy should not be done at all. A leaky intelligence operation is worse than no intelligence operation, because it betrays your own people and warns the opponent.

Practical application

Invest seriously in competitive intelligence

For any organisation operating in a contested market, the intelligence budget should be a non-negotiable line item, sized not by what it costs but by what it prevents. A well-placed source who tells you a competitor's pricing change three weeks early saves you a quarter of bad decisions. A relationship with a former employee of a rival is worth more than ten analyst reports.

Map the network, not just the players

Sun Tzu's "names rule" applies directly. Don't track competitors at the company level; track them at the individual level. Who is on the team? Who reports to whom? Who is unhappy? Who is overdue for a promotion? Who has a tense relationship with the CEO? This kind of map is what makes intelligence actionable.

Cultivate the modern double agent

The corporate equivalent of the double agent is the high-trust insider relationship — the senior engineer at a rival who has lunch with your CTO once a quarter, the former PM who left for a competitor and stayed friendly, the board member who serves on multiple boards. These people are not "spies" in the Cold War sense; they are humans with continued relationships across an organisational boundary. The relationships are built over years, with care and reciprocity, and become decisive when contested moves happen.

Treat your information sources with extreme generosity

Sun Tzu's instruction — treat the double agent with particular generosity — generalises. The people who give you the most leveraged information are not the highest-paid in your organisation, but they are the highest-leveraged. Make sure they know it. The conversation that prevents a strategic error is worth ten times the bonus that recognises it.

Example

A growth-stage company is preparing to enter a new market where two incumbents already operate. The CEO's instinct is to hire an analyst, buy an industry report, and develop a "competitive teardown" deck. This is The Use of Spies's third-best option — buying the public version of information everyone has.

The The Use of Spies reading is different. The company spends three months on relationship intelligence. A former VP of Sales from Incumbent A is invited to consult for a small project. A current senior engineer at Incumbent B is the brother of the founder's college roommate; they meet for coffee. A board member of an ally company sits on the board of Incumbent A's main partner; an introduction is arranged. None of this is illegal; none of it is unethical; all of it is operating in the network rather than reading the report.

By the time the company is ready to enter, it knows: which incumbent is about to lose a key engineer (and can be poached this quarter), which one is over-extended on the partner side and cannot defend against a direct integration play, what the typical sales cycle pricing actually is (not the published number), and which of the major customers is open to switching if a credible alternative arrives. The market entry that follows looks brilliant from the outside; the reality is that it was foreseen — The Use of Spies's central word — because the network supplied the foresight the public report could not.

Related lessons