Absurd Behavior Is Usually About Incentives

🎯 The one-line version: when you are trying to understand strange behavior, look at what the person said next to what they are actually rewarded for. Incentives are good at generating explanations, but they do not replace evidence.

An institution has stated goals, but it is also shaped by resources, execution capacity and the interests of the people taking part. Working out who benefits and who pays helps us find problems. It does not mean that someone who benefits caused the problem.

1. A story about a Venetian wine merchant

The following is a fictional story written to explain incentives. It is not Venetian history. Imagine you have traveled back to 14th century Venice and become the largest wine wholesaler in the city. You are about to cut prices to move stock, and you discover that the guild rules forbid it.

1. Cutting prices is against the law

You plan to sell thin margins at high volume and discount the wine. On paper, everyone in the city gets better wine for less and market efficiency improves.

The next day, the wine merchants’ association brings people and raids your shop. The guild master throws a decree at you: in this city, no wine merchant may cut prices on their own. Barrel sizes must be uniform. Violators forfeit all property.

You call it an anti-competitive rule that destroys efficiency. The old guild master snickers and leans over:

Young man, this rule was not written for everyone’s convenience. Last month a few of us large merchants gave the governor three thousand gold coins to get it. We keep our margins, the governor gets his coins, and as for the common folk who cannot afford wine, who cares?

This fictional story borrows one angle from institutional economics: rules change what participants choose and what they get. North studied how institutions affect long-run economic performance, but the guild, the fines and the dialogue in this story are not conclusions from his historical research. For background, see the 1993 Nobel Prize in Economic Sciences press release.

2. Why the constable does not want to catch thieves

You stop complaining that the rules are absurd and start looking at the city through the lens of “seeing the incentives.”

The constable declares every day that he is cracking down hard on night theft and protecting the citizens, yet thieves stay rampant. Going through the accounts, you find the reason: the constable’s salary is fixed, but he can take thirty percent of every fine collected from a caught thief.

So he does not want to eliminate thieves. He needs a steady supply of them in the city so the fine cuts keep coming.

Once you see that, you change your approach and tell the constable: “Sir, I am forming a wine guard. I am not asking you to catch thieves for me. I will pay you a monthly consulting fee for security, as long as my barrels leave the city safely.”

The next day, two fully armed patrol soldiers are riding alongside your wine caravan, and the whole route is clear.

💡 You did not change anything about the environment. You redesigned the constable’s incentives so that he voluntarily became your shield. When you cannot change the rules, first ask whether you can change what the other side gets paid for.

3. What the inspection checkpoint is really for

Three years later you are the richest merchant in Venice. A young merchant newly arrived from the East complains: the guild officials keep saying the inspection checkpoint exists to guarantee quality, yet slipping two silver coins to the inspector gets immediate clearance. Terribly inefficient and corrupt, he says.

You hand him a glass. “Do not look at what they claim to be doing. Look at what they get money, power, safety or status for. That checkpoint never existed to inspect quality. It exists to give the inspector room to extort.”

2. Three things underneath the story

1. Behavior is shaped by incentives. Alongside the public slogans, look at what is actually measured, rewarded and punished. Personal values, ability and information shape the choice too.
2. Rules are often the result of bargaining, power relations and historical accumulation, and they do not necessarily reach the efficient outcome. To judge how a specific rule formed, you still have to look at how it was made.
3. Inefficiency may serve someone’s local interest, and it may also come from technical limits, coordination costs or missing information. List several explanations first, then compare them with evidence.

🔍 So seeing through a complicated phenomenon comes down to one question: Who, because of this rule or behavior, received money, power, promotion, safety or status?

3. What academics call this

Institutional, incentive and information problems can be analyzed through mechanism design, institutional economics and public choice. These fields are related, but you cannot call every odd behavior “incentive compatible.”

1. Incentive compatibility: how rules make truth-telling the profitable choice

Incentive compatibility has a stricter technical meaning. In a direct mechanism, if truthfully reporting private information is the participant’s optimal strategy, that mechanism is incentive compatible. It does not cover every case where private interest happens to line up with the public interest. For the definition, see the Nobel Prize advanced information on mechanism design.

For example, a procurement rule may let suppliers report their costs. The designer has to consider whether a supplier profits by inflating the number. That is exactly the kind of problem mechanism design handles. And when you analyze charitable giving, you cannot infer the donor’s tax position or private motives just from the donation.

2. Incentive misalignment: three classic counterexamples

1) Goodhart’s law: once a metric is a target, it stops working

When a measure becomes a target, it ceases to be a good measure.

Metrics can be gamed, and they sometimes need to be used alongside quality checks. Say a team is rewarded only on “number of projects shipped.” Members may then prefer to split work into many small projects and neglect maintenance. This is a hypothetical example, not a factual claim about how Google assesses its staff internally.

2) The cobra effect: the result comes out exactly opposite to the goal

The familiar “cobra bounty” story describes a mechanism where paying per dead snake encourages people to raise snakes, and where removing the bounty can produce snake release instead. It is used here as an illustrative story about incentive misalignment, not as proof of causation built on unverified historical detail.

3) Wells Fargo: roughly 3.5 million possibly unauthorized accounts

The US Consumer Financial Protection Bureau stated in its September 8, 2016 announcement that employees opened accounts without customer authorization in order to hit sales targets and collect bonuses. Wells Fargo reported in its August 31, 2017 expanded review announcement that the review covered January 2009 through September 2016 and identified roughly 3.5 million “potentially unauthorized” accounts. That category may include some accounts that were in fact authorized. The figure is a screening result and does not equal a count of individually confirmed unauthorized accounts.

⚠️ The shared lesson: a badly designed incentive scheme does not merely “perform a bit worse.” It often produces results exactly opposite to the goal. Before you set a metric, ask: what is the cheapest way for the person who most wants the reward to hit it?

3. Two deeper angles

1) Public choice theory: analyzing the institution and the individual together

Buchanan’s public choice work connects individual choice with the political and economic rules of decision making, and analyzes what participants gain and what constrains them inside an institution. That helps you test policy explanations, but it does not prove that every official is only chasing private gain. See the 1986 Nobel Prize in Economic Sciences press release.

2) Munger: change the incentives, not the ideas

When investors discuss decision bias, they often remind people to look at what rewards are producing. This article borrows that angle. Specific cases still need to be checked against records, and a famous person’s opinion is not a substitute for evidence.

The practical summary of this article: ask what reward, punishment and constraint a behavior brings, then check whether the explanation fits the record. That is my own synthesis, not a quote from Munger.

There is a plainer version of this in investigative journalism: follow the money. The phrase comes from the description of the Watergate break-in in the film All the President’s Men. When every side tells a different story and the fog is thick, trace where the money and the power went, and the truth often clears up immediately.

4. Turning this insight into a reflex

Here is a three-step method you can apply directly. When you hit anything odd, absurd or baffling, take it apart in this order.

1. Step one: separate “what is said” from “what is measured”

The biggest fog in human society is noble language wrapped around a cold mechanism. So start with a two-column split:

Two small exercises:

• A company says it “encourages innovation.” Ask: who pays when innovation fails? Do people who simply follow procedure still get their performance rating?
• A policy says it will “streamline the process.” Ask: after streamlining, does the original approving department get bigger or smaller in headcount and power?

2. Step two: six elements for tracing where the benefit flows

Treat these six as analysis probes and see who had resources redistributed because of it:

1. Money: where the cash flow comes in, and whose pocket it finally lands in.
2. Power: who gained decision rights, approval authority or a veto.
3. Name: who took the moral high ground and the industry reputation.
4. Position: who got promoted, or whose scope of management and headcount grew.
5. Safety: when something goes wrong, who gets the risk and the cost dumped on them, and who gets the backstop.
6. Attention: who gained attention, exposure and the right to explain.

These six resources help you raise questions, but they are not the whole of a person’s motives. Once you have found where the resources flow, you still have to check ability, information and institutional constraints, so you do not read correlation as causation.

3. Step three: ask the reverse question, “who is hurt?”

Run Cui Bono (who benefits?) backwards, and three steps are enough:

1. Assume this bad phenomenon disappears completely and everything becomes highly efficient.
2. List who would be worse off in that new world.
3. Judge whether the injured party has enough power to stop the change from happening.

A hypothetical example: an agency keeps paper approvals forever. You can ask whether the department’s budget depends on that process, and you should also check data compatibility, security requirements and legal retention rules. A cumbersome process on its own does not prove that staff are deliberately obstructing digitization.

5. Analysis exercise: a helmet rule and its enforcement, what to verify first

First separate vehicle categories from local jurisdiction. Article 51 of the Road Traffic Safety Law explicitly requires motorcycle riders and passengers to wear a safety helmet as prescribed. The specific requirement for electric bicycles depends on the applicable local rules. You cannot merge the two vehicle types into one nationwide fine schedule. See the official legal text.

Observing one absence of an on-site check does not mean enforcement does not exist, and it does not directly imply anything about the management department’s motives. What follows is a set of verification questions, not an investigation into enforcement in any particular place.

1. Enforcement cost: what records you need

You can look at vehicle counts, accident risk, police deployment, persuasion and penalty records, and compare across different times and road segments. Resource constraints are one possible explanation and need record support.

1) Headcount and road segments

First ask which people, which road segments and how many days the survey covered. A short personal observation cannot represent an entire city.

2) Fines and management goals

Establish the applicable clause, the penalty standards and the enforcement procedure before discussing cost. The benefit of traffic safety management cannot be measured by fine revenue alone.

3) Safety outcomes and alternative explanations

Check whether there was publicity, road inspections or other measures. Put enforcement capacity, risk priority and reward mechanisms on the table as separate explanations, rather than deciding in advance that nobody is managing it.

💡 When you analyze incentives, list the hypotheses and the evidence gaps first. Who might benefit is a starting point for an investigation, and it cannot prove the real motive on its own.

2. Accident liability: a helmet alone cannot decide it

Whether a helmet was worn cannot, by itself, determine primary responsibility in a traffic accident, and it cannot prove the management department has discharged all of its statutory duties. Specific responsibility depends on applicable law, accident evidence and a formal determination.

• Check the cause of the accident, each party’s behavior and the formal liability determination, instead of writing “violation” straight into “bears primary responsibility.”
• Separate the safety obligation, the enforcement duty and accident liability. These three need separate verification.

When discussing how responsibility is divided, name the clause and the facts you are relying on. Do not substitute the phrase “separation of duties” for legal analysis.

3. Concentrated inspections: first establish whether one happened

If concentrated inspections appear somewhere, compare the public notices, enforcement records and risk changes before and after, and only then discuss the relationship to assessment.

• Look for published traffic safety arrangements, with their applicable dates and scope.
• Look for whether safety risk and enforcement resources changed, rather than only hunting for material that fits your earlier judgment.

Only when the timeline and the records line up is there reason to assess the effect on assessment criteria further. One concentrated inspection is not evidence that an official is chasing promotion.

4. Write the question as a checkable claim

🎯 A reusable set of questions: what did I observe? What else could explain it? What records would tell these explanations apart?

6. Hypothetical reasoning: what could drive a company’s AI spending

What follows discusses the choices of a fictional company, as practice for incentive analysis. It does not represent any real company’s finances or executive motives. Analyzing a real company requires annual reports, capital expenditure, compensation disclosures and business data.

Put “trace the incentives” on it and two invisible saddles show up, matching money and position from the six elements above: the share price and talent.

1. The first saddle: how the market values the company

Say a large tech company’s main business growth slows this year. Ads are not selling and user growth has plateaued. Now it has two possible stories.

1) How cost cutting and growth expectations can affect valuation

The first story is “we cut costs.” That may improve profit and cash flow. How the market reacts still depends on durability, business outlook and other information, so you cannot promise the share price will only rise a little.

The second story is “we are investing in AI infrastructure.” Investors may revise growth expectations, and may equally worry about cost and return. Telling a growth story does not guarantee a higher valuation.

💡 In a simplified valuation model, both profit expectations and the valuation multiple can move. This model helps you break the problem apart. It does not predict any company’s share price.

2) Which choices stock incentives can influence

In this hypothetical, executive pay includes stock incentives. A real company’s compensation structure and assessment items need to be checked against that year’s public disclosures, and you cannot generalize.

You can put forward a hypothesis like this: if the reward weighted short-term valuation more heavily, the person in charge might lean toward emphasizing the growth story. Testing it means comparing the actual assessment metrics, the public decisions and the long-run business results. You cannot infer what an executive is thinking.

2. The second saddle: the movement of top talent

A large tech company is not only competing for customers. It is competing for engineers and scientists.

1) What conditions talent may care about

For some technical staff, compute, projects, research opportunities and career development may matter. Salary, work environment and personal preference may also affect whether they stay.

2) What risks cutting frontier projects can create

Say the company cuts the budget for frontier projects. Some employees may worry about growth opportunities, while others may prefer to put their energy into stable business. Whether talent actually leaves needs hiring and departure records to support it.

In this hypothetical, you can list several risks that need testing:

1. Are some employees leaving because project opportunities shrank?
2. Is the movement happening at team level, or is it one or two individuals’ choice?
3. Has the outside candidate’s view of the projects and the business outlook changed?
4. Has recruiting become harder, and can you rule out pay and market conditions as causes?

⚠️ Keeping a research team may be worth something in retention and in experimental opportunities, but whether the scale of spending is reasonable still requires comparing cost, business goals and the alternatives.

3. Testing both hypotheses together

Growth expectations and talent opportunities are two possible explanations. You should also compare customer demand, competitive pressure, technical capability and cash flow constraints. You cannot conclude from “who benefits” that the company had no other options.

🎯 The same questions carry over to the next hype cycle: What does following the crowd get you? What does not following lose? Who ends up paying the bill?

7. Three questions worth asking repeatedly in daily life

✅ 1. If I were the person involved, under the rules as they stand, what would I do to get the most benefit and carry the least risk?

2. What is this rule punishing? What is it permitting?

3. What is the rule maker’s own core KPI?

Once you habitually read the news, company decisions and personal relationships this way, the world stops being chaos and randomness and becomes as readable as code: the logic of code is syntax, the logic of the world is incentives.

8. Critical thinking exercise: separate facts, explanations and evidence

Say you read this sentence: “This approval process is slow, so the department is deliberately holding on to power.” Split it into three layers. How long the process takes on average is a checkable fact. “Deliberately holding on to power” is an explanation of motive. Approval logs, rule changes and resource allocation are the evidence you need to supply. You also have to test other explanations such as safety requirements, understaffing and information asymmetry.

That is one workable step in critical thinking: write the claim, note the evidence and the gaps, then ask what new record would change your mind. You can use the fact-checking prompt to organize the material. For going further into the source of a story, read Solving Information Asymmetry and Follow Where Real Power Goes.