Follow the Incentives: Why Broken Systems Make Sense

🎯 The short version: if you want to know why something turned out the way it did, ignore what the people involved say they want. Look at what they get paid, promoted, protected, or praised for.

Institutions claim to exist for efficiency and fairness. What actually shapes them is where the people who write the rules want the benefits to land. Once you internalize that, a lot of apparently absurd behavior turns out to be perfectly rational.

I. A Venetian Wine Merchant

Imagine you land in 14th-century Venice as the city’s largest wine wholesaler. You have the best Bordeaux in town and a head full of modern business theory. Within three days, you hit a wall.

1. Cutting Prices Turns Out to Be a Crime

You decide to sell more by charging less, so you discount your wine. In theory everyone wins: cheap wine for the city, a more efficient market.

The next morning the wine merchants’ guild raids your shop. The guild master hands you a decree: no wine seller may lower prices on his own initiative, barrel sizes are fixed, and violators forfeit everything they own.

You call it anti-competitive and inefficient. The old man leans in, amused:

Young man, these rules were not written for everyone’s convenience. A few of us paid the Doge three thousand gold florins last month to have them written. We keep our margins, the Doge keeps his gold. The people who can no longer afford wine — who exactly is asking them?

This is the point Douglass North built a career on: institutions are frequently not designed for aggregate efficiency. They look the way they look because whoever can set the rules is steering incentives toward themselves.

2. Why the Constable Doesn’t Want to Catch Thieves

You stop complaining that the rules are stupid and start reading the city through incentives instead.

The constable announces constant crackdowns on night theft and swears he protects the citizenry — yet thieves are everywhere. Then you see the ledger: his salary is fixed, but he keeps thirty percent of every fine he collects from a captured thief.

So he does not want the thieves gone. He needs a steady supply of them to keep the fine revenue flowing.

Once that clicks, you change tactics. You visit him and say: “My lord, I’ve formed a wine escort. I’m not asking you to catch thieves for me. I’ll pay you a monthly security retainer — I only need my barrels to leave the city intact.”

The next day, two armed patrolmen ride beside your wagons, and the road is suddenly clear.

💡 You didn’t change the environment. You redesigned one man’s payoff so that protecting you became his best move. When you can’t change the rules, ask whether you can change the other side’s payoff structure.

3. What the Inspection Checkpoint Is Actually For

Three years later you are the richest merchant in Venice. A young trader newly arrived from the East complains to you: guild officials insist the inspection checkpoints exist to guarantee quality, yet two silver coins to the inspector gets any cargo waved through. Corrupt and wasteful, he says.

You pour him a glass. “Don’t look at what they say they’re doing. Look at what earns them money, power, safety, or standing. The checkpoint was never built to inspect quality. It was built to give the inspector something worth bribing him for.”

II. Three Principles Underneath the Story

Behavior is the shadow of incentives. People don’t act the way you hope; they act the way the system rewards. Slogans (“we think long-term”) are nearly free to produce. The review process (the monthly KPI) is what actually gives orders.
Rules are the residue of a negotiation. An institution is rarely the optimal solution. Often it is simply the tool through which whoever held power maximized their own take.
There is no meaningless inefficiency — only inefficiency that serves someone locally. When a situation looks like everyone is losing, somebody is usually collecting.

🔍 So cutting through a confusing situation takes one question: who is getting money, power, promotion, safety, or status out of this arrangement?

III. What This Is Called in the Literature

The orthodox name is incentive compatibility, and behind it, mechanism design. Different fields emphasize different parts of it.

1. Incentive Compatibility: Private Interest Overlapping Public Interest

Incentive compatibility describes a system in which people pursuing their own interest happen to produce the outcome the rule-maker wanted.

Bill Gates’s philanthropy is a decent example. Once the system made it clear that giving buys reputation, influence, and favorable tax treatment, decisions taken to protect a family’s standing also happened to push global disease control forward. Private and public interest were aligned.

2. Misaligned Incentives: Three Classic Failures

1) Goodhart’s Law: A Metric Dies When You Grade It

When a measure becomes a target, it stops being a good measure.

Set a standard and people will find the cheapest route to the reward, which usually drifts away from whatever you were trying to measure. Google wanted to encourage innovation and started counting new product launches; engineers chasing bonuses and promotions shipped a flood of redundant projects and killed working products, and users paid for it.

2) The Cobra Effect: The Result Inverts the Goal

Colonial administrators in India offered a bounty for every dead cobra. Locals started breeding cobras to collect. When the bounty was cancelled, the breeders released their stock, and there were more cobras than when the program began.

3) Wells Fargo: Roughly 3.5 Million Fake Accounts

To hit account-opening quotas — miss them and you’re fired — branch staff fabricated an estimated 3.5 million unauthorized accounts.

⚠️ The shared lesson: a bad incentive scheme doesn’t merely underperform. It routinely produces the opposite of the stated goal. Before you set a target, ask: what is the laziest way for the most reward-hungry person to hit it?

3. Two Deeper Lenses

1) Public Choice Theory: Don’t Budget for Virtue

James Buchanan’s central claim: politicians and civil servants are ordinary people, so don’t expect them to serve the public interest out of nobility. Politician, executive, or philanthropist, their decisions inside a given system are largely self-interest maximization — votes, authority, budget, reputation.

2) Munger: Change the Incentive, Not the Opinion

Charlie Munger called it “the power of incentives,” and argued most people badly underestimate how completely incentives govern behavior — strong enough that people unconsciously rationalize their own bad conduct.

Never ask the baker whether his bread is good. Look at how much he makes selling it.
If you want to change someone’s behavior, change their incentives — don’t try to change their mind. — Charlie Munger

Investigative journalism has a plainer version: follow the money. The phrase comes from the film All the President’s Men, about Watergate. When every party tells a different story, trace where the money and power flow, and the picture usually resolves fast.

IV. Turning This Into an Instinct

Here is a three-step routine. Any time something looks perverse, absurd, or simply unexplainable, run it through in order.

1. Step One: Separate What Is Said From What Is Measured

The great fog of social life is noble language wrapped around cold machinery. So split the thing in two:

What to split out	What to look for
What they say	Slogans, culture decks, press statements. Anything phrased as mission, long-term thinking, or the common good gets discounted on sight
What they measure	KPIs, commission rates, voting rules, allocation of authority. Who gets punished for contradicting the slogan? Who gets paid for following the machine?

Two quick exercises:

A company says it rewards innovation. Ask: who carries the blame when an experiment fails, and can someone who simply executes the plan still score well on review?
A policy says it simplifies a process. Ask: after simplification, does the approving department’s headcount and authority grow or shrink?

2. Step Two: Trace Six Kinds of Flow

Treat these as probes and check who ends up with a bigger share because of the arrangement:

Money: where cash enters, and whose pocket it lands in.
Authority: who gains decision rights, sign-off, or a veto.
Reputation: who occupies the moral high ground and the professional credit.
Position: who gets promoted, or ends up managing a larger scope.
Safety: when it goes wrong, who absorbs the risk and who is indemnified.
Attention: who captures visibility and the right to define the narrative.

Behavior inevitably migrates toward wherever these six pool up. Find the flow and the absurdity starts to look reasonable.

3. Step Three: Ask Who Would Lose

Run cui bono in reverse. Three moves:

Assume the bad situation vanishes and everything becomes maximally efficient.
List who is worse off in that world.
Ask whether those losers hold enough power to block the change.

An example: why do bureaucratic paper procedures at hospitals, universities, and large agencies survive untouched for a decade? Digitize them and approval takes a second — at which point the office whose job is to stamp and verify has no reason to exist, and its managers lose headcount and budget. For them, keeping it cumbersome is the optimal strategy.

V. Case Study: Helmet Laws Nobody Enforces

Most places with e-bikes, mopeds, or shared scooters have a helmet rule on the books, and on a normal day almost nobody enforces it. Run this through the framework and it resolves quickly.

On the surface it looks like a traffic-management problem: the law exists, so why the weak enforcement? Traced through the incentive chain, it is really a contest between enforcement cost and enforcement payoff.

1. Enforcement Costs Far More Than It Returns

For a local police force, every stop carries a full cost.

1) The cost side: too many riders, impossible to cover

Two-wheelers are numerous, fast-moving, and spread across every side street. Achieving car-level compliance would mean flooding the streets with officers to stop riders, write tickets, and argue with them.

2) The payoff side: minimal, sometimes negative

Helmet fines are typically small, and in many jurisdictions the default is a warning. That never covers the cost of the stop, the paperwork, and any appeal. Meanwhile the riders are largely commuters, delivery workers, and parents on the school run — a population where aggressive ticketing generates friction fast.

3) The risk side: the officer absorbs it

If hard enforcement produces a viral confrontation or a complaint, the local commander is the one who takes the hit. In most performance systems, “an incident happened” is far more damaging than “nothing was done.”

💡 In incentive terms: strict enforcement equals exhausting work, high cost, and a real chance of getting blamed. Looking the other way equals conserved manpower and a quiet beat. The choice is obvious.

2. The Real Function of the Statute Is Liability Transfer

Many people assume a law exists solely so that everyone complies. Back to North’s view of institutions: some rules exist to give administrators liability transfer (risk offloading).

If the law requires helmets: when a rider dies of a head injury, responsibility is unambiguous — the rider was not wearing the legally required helmet and bears primary responsibility. The agency has, legally speaking, discharged its duty to inform and regulate.
If the law says nothing: the public and the family ask directly why no safety standard was set and no risk was flagged. Administrators are exposed to both press and formal liability.

Mapped to safety in the six flows above: the core incentive for writing a paper rule is moving risk from the institution to the individual. Once the law is printed, most of the institutional objective is already met. Whether anyone enforces it on the street is a separate cost question.

3. Why Enforcement Suddenly Gets Strict

You have probably noticed the pattern: a place with zero enforcement for months suddenly has officers stopping every rider for two weeks. Nobody’s conscience improved. An external incentive temporarily changed.

Funded enforcement waves: national road-safety programs — the US “Click It or Ticket” mobilizations are the textbook case — attach grant money and reporting requirements to concentrated enforcement periods. Departments must show activity to keep the funding.
Compliance targets and campaigns: a regional safety plan sets a measurable helmet-use or citation target for the quarter, and it lands in someone’s performance review.

The moment “check helmets” is wired to a commander’s budget and position, resources appear regardless of cost. When the reporting window closes, the incentive disappears and enforcement reverts to baseline.

4. Three Behaviors, Side by Side

Behavior	Stated reason	Underlying incentive	Resulting action
No routine enforcement	”We favor education over penalties”	Costly, low return, easy to get blamed	Selective enforcement
The law stays on the books	”To protect lives”	Completes legal liability transfer	Rule lives on paper
Sudden crackdown	”Special safety initiative”	Grant money and quarterly targets tied to promotion	Short burst, then back to normal

🎯 Next time you see a rule on the books that nobody enforces, ask one question: what does the enforcer gain by enforcing it, and what does it cost them?

VI. Case Study: Why Every Big Tech Company Has to Do AI

On the surface, the pile-in looks like a technology call: large models are the next platform, so everyone goes. But conviction about technology cannot explain why every company keeps raising spend while losing money on it.

Apply the incentive lens and two invisible harnesses appear, matching money and position in the six flows: the share price and the talent market.

1. First Harness: How Capital Markets Price You

Assume a large company’s core business is decelerating this year — ad spend is flat, user growth has topped out. It has two things it can say.

1) Cost cuts don’t move the multiple; a story does

Option one: “We took a lot of cost out this year.” That’s a one-time margin improvement, priced at the current multiple, worth a small bump at best — and investors readily read it as “this company has nothing left to invest in, so it’s trimming.”

Option two: “We are investing heavily in AI and building the infrastructure for the next decade.” That is a growth narrative, and the market stops pricing current profit and starts pricing the size of the imagined opportunity.

💡 The two are priced through different channels: cost cuts move the numerator (earnings); the AI story moves the multiple. At large-company scale, a small move in the multiple swamps whatever the savings were worth.

2) Executive pay is wired directly to the stock

Cash salary is a small slice of a big-tech CEO’s compensation. The bulk is stock and options, and long-term awards are frequently tied to share-price or market-cap milestones.

So “was the technical call correct” appears nowhere on the scorecard. Only the stock does. Even a CEO who privately believes current AI spending is a bubble has a payoff structure that forbids saying so publicly. Follow the narrative and any error is a shared industry error. Break from it and being right is forgotten, while being wrong is entirely personal.

2. Second Harness: Where Top Talent Goes

These companies are not only competing for customers. They are competing for engineers and researchers.

1) Elite researchers chase compute and projects, not salary

People at that level are not short of money. What they care about is access to large-scale compute, participation in frontier work, and output with real influence — the things that set their standing in the field three years from now.

2) A company that says “we’re sitting this out” gets drained

Suppose a large company announces it will skip large models and stick to its existing business. To its best technical staff, that announces a hard ceiling on compute budget, project ambition, and personal market value.

An ugly chain reaction follows:

The strongest people leave first, because they have the most options.
They take teams with them, and departures become structural.
Outsiders start reading the company as technically out of the game, and new graduates stop applying.
Even if leadership reverses course two years later, hiring no longer works. Money can turn on a dime; a reputation in the talent market cannot.

⚠️ Which is why many companies would rather fund an expensive model team with no near-term return: they aren’t only buying models, they’re buying the right to say they’re still at the table. Leaving the table costs far more to undo.

3. Stack the Two Harnesses and There Is No Choice Left

Option	Valuation	Top talent	The executive personally
Go all-in on AI	Gets a growth narrative, multiple holds up	Can hire and retain	Equity appreciates; if it’s wrong, the whole industry was wrong
Sit out, focus on cost	Labeled no-growth, multiple compressed	Core talent leaves and won’t come back	Equity shrinks, and the blame is theirs alone

Both rows point the same direction. Big tech’s AI spending isn’t purely a shared belief that general intelligence is imminent. Under the current incentives, the cost of not doing it falls on one executive, while the cost of doing it badly is spread across the industry.

🎯 The same question transfers directly to the next hype cycle: what does a follower gain, what does a holdout lose, and who ultimately pays for it?

VII. Three Questions Worth Asking Constantly

✅ 1. If I were the person inside this system, what would maximize my gain and minimize my exposure under the current rules?

What does this system punish, and what does it quietly tolerate?

What is the rule-maker’s own core metric?

Once you read news, corporate decisions, and personal dynamics this way, the world stops looking chaotic and starts reading like source code: code runs on syntax; the world runs on incentives.