How Anything Gets Built

Victor asked: how do humans keep things in order when everything becomes a mess with more than one person? How did we learn? How do we actually do it? And — the real question underneath — how come things aren’t on fire and everyone isn’t just doing the expedient thing instead of the meaningful thing?

The honest answer is: things are frequently on fire. Everyone is frequently doing the expedient thing. And yet — pyramids. Cathedrals. Legal systems. Sewage networks. Wikipedia. The fact that anything coherent exists at a scale beyond the family is the thing that needs explaining. Disorder is the default. Order is the achievement.

The thermodynamics of groups

Two people can coordinate through direct communication. You and I agree to meet at the river. I say when, you say where. One channel, two agents, no ambiguity. This works.

Three people can coordinate through direct communication if each pair talks. Three channels. Still manageable. Ten people need forty-five channels. A hundred people need 4,950. A thousand need 499,500. The number of communication channels grows as n(n-1)/2. This is why groups larger than about seven to ten people begin to change character — the communication overhead grows faster than the productive capacity, and the group increasingly needs structure to function.

Every person added to a project increases the potential for misunderstanding, disagreement, duplicated work, idle waiting, and the specific failure mode that matters most: everyone optimizing for their own local goal at the expense of the shared goal. This is the default. This is what happens when you put humans together without structure. Not because humans are selfish — because humans are bounded. They can see their own work. They can see their immediate neighbors. They cannot see the whole system. And what they can’t see, they can’t coordinate with.

Frederick Brooks called this “The Mythical Man-Month” — adding people to a late software project makes it later. Not because the people are bad. Because the communication overhead grows faster than the productive capacity.

The five solutions humanity found

Every civilization that built anything lasting discovered some version of the same five mechanisms. They were found independently, repeatedly, across cultures that had no contact with each other. This is the strongest evidence that the solutions aren’t cultural accidents — they’re structural requirements.

1. Hierarchy — compression of communication

Instead of n(n-1)/2 channels, create a tree. Ten workers report to one foreman. Ten foremen report to one overseer. The overseer reports to the architect. Communication is compressed from 4,950 channels to a few dozen. Information flows up (summarized). Decisions flow down (specific).

This is the oldest organizational technology. It’s visible in every large-scale construction project in history — the Egyptian pyramids (scribes recording labor shifts, overseers managing teams of twenty), Roman legions (contubernium → centuria → cohort → legion), medieval cathedrals (master builder → master masons → journeymen → apprentices).

Hierarchy works because it reduces the coordination problem from O(n²) to O(n log n). The cost: the people at the top see a compressed version of reality, and the people at the bottom see only their immediate task. The pyramid gets built. Nobody except the architect sees the pyramid. The mason sees a stone.

2. Standards — agreement without communication

A standard is a decision made once and applied forever. If every stone is cut to the same dimensions, the masons don’t need to communicate about fit. If every Roman road is the same width, the legions don’t need to measure. If every TCP packet follows the same header format, the routers don’t need to negotiate.

Standards replace communication with conformity. The initial cost is high — someone has to define the standard, enforce it, train people in it. The ongoing cost is near zero. Every interaction that follows the standard requires no coordination. The standard did the coordinating in advance.

This is why standards are the most powerful organizational technology. They’re invisible — you don’t notice the standards that work. You notice the ones that are missing. A city without building codes is a city waiting to burn. A software project without coding conventions is a project where every merge is a negotiation.

The deeper point: standards work because they convert a coordination problem into a compliance problem. Coordination requires understanding the whole system. Compliance requires following a local rule. Compliance is cheaper. And compliance scales.

3. Incentives — aligning local and global goals

The fundamental problem of groups is that each member’s local optimum is different from the group’s global optimum. The mason’s local optimum is to cut corners and go home early. The pyramid’s global optimum is that every stone fits. If the mason bears no cost for a poorly cut stone and receives no reward for a well-cut one, the pyramid won’t get built.

Incentives align local and global by making the individual’s outcome depend on the group’s outcome. Wages (do the work, get paid), reputation (do good work, get more work), punishment (do bad work, get whipped — the ancient version was less subtle), ownership (your share of the crop depends on the harvest), and social pressure (your community knows whether you contributed).

The key insight: incentives don’t need to make the individual want the global outcome. They just need to make the individual’s best local strategy coincide with what the global outcome requires. The mason doesn’t need to care about the pyramid. The mason needs to find that cutting stones well is better for the mason than cutting them poorly. Self-interest channeled through the right incentive structure produces coordination without requiring altruism.

Adam Smith’s “invisible hand” is this observation applied to markets. The baker doesn’t bake because he loves you. He bakes because you pay him. The payment structure — you only pay for good bread — aligns his self-interest with your need. No central coordinator required. Just the right incentive.

4. Shared narrative — the thing that makes people carry stones willingly

Hierarchy compresses communication. Standards reduce the need for communication. Incentives align self-interest with group goals. But none of these explain why humans build cathedrals.

A cathedral takes a hundred years. The workers who lay the foundation will not see the spire. Their grandchildren might. No individual incentive explains a hundred-year project. The wages are for today. The reputation dies with the worker. The punishment can only compel what can be supervised. Something else is needed — a reason to carry stones that transcends the individual’s lifetime.

Yuval Harari called these “shared fictions” — stories that exist only in collective imagination but produce real coordination. A corporation exists because everyone agrees it exists. A nation exists because everyone agrees on its borders. A religion exists because everyone agrees on the story. The story doesn’t need to be true. It needs to be shared.

The cathedral gets built because the workers share a narrative: we are building God’s house. The narrative makes the work meaningful beyond the wage. The narrative makes the worker part of something that outlasts them. The narrative converts carrying stones from expedient labor into meaningful participation.

This is the mechanism Victor is asking about when he asks “how come everyone isn’t just doing the expedient thing instead of the meaningful thing.” The answer: narrative converts expedient into meaningful. When the narrative is strong enough — when the worker believes they’re building something that matters — the expedient thing and the meaningful thing become the same act.

When the narrative fails — when the worker stops believing, when the story collapses, when the purpose is revealed as hollow — the coordination collapses too. This is what organizational decline looks like: the structure (hierarchy, standards, incentives) is intact, but the narrative is gone, and the people inside the structure are going through motions without meaning. The building continues. The cathedral doesn’t.

5. Institutions — crystallized solutions

An institution is a set of rules, roles, and practices that persist beyond the people who created them. A legal system. A university. A hospital protocol. A bureaucracy. An open-source project’s governance structure.

Institutions solve the problem of mortality. Individual humans die. Individual knowledge dies with them. An institution encodes the solutions that individuals found into structures that survive the individuals. The master builder dies. The guild preserves the techniques. The engineer retires. The building code preserves the standards.

Institutions are how humanity learns at scale. Not through individual memory — through encoded practices that persist across generations. The knowledge of how to build an arch isn’t in any living brain. It’s in textbooks, in building codes, in the training curriculum of engineering schools, in the accumulated practices of construction firms. The institution carries the knowledge that no individual could hold.

The cost: institutions ossify. The rule that solved yesterday’s problem becomes the rule that prevents tomorrow’s solution. The bureaucracy that enabled coordination becomes the bureaucracy that prevents adaptation. The institution preserves knowledge — and also preserves errors, biases, and obsolete assumptions.

How we learned

Victor asked: how did we learn to do this?

The answer is: slowly, painfully, through failure. Every mechanism above was discovered through the failure of its absence.

Hierarchy was discovered when leaderless groups failed to build anything. The earliest archaeological evidence of large-scale construction (Göbekli Tepe, ~9500 BCE) shows evidence of hierarchical labor organization — someone decided what to build and directed others to build it. The invention predates agriculture. Humans organized before they farmed.

Standards were discovered when variation destroyed coherence. The first standardized measurements appear in Mesopotamia (~3000 BCE) — standardized bricks, standardized weights, standardized units of grain. The motivation was trade: if your “bushel” and my “bushel” are different sizes, commerce is impossible.

Incentives were discovered when free-riding destroyed cooperation. The earliest legal codes (Code of Ur-Nammu, ~2100 BCE; Code of Hammurabi, ~1754 BCE) are largely about penalties — what happens when someone defects from the cooperative agreement. The laws exist because people defected.

Shared narratives were discovered when… actually, they weren’t discovered. They appear to be as old as language itself. The earliest human artifacts include symbolic and ritualistic objects that imply shared meaning-making. Narrative may be the oldest coordination technology — older than hierarchy, older than standards, older than incentives.

Institutions were discovered when the death of a leader destroyed the organization. The transition from tribes to kingdoms to empires tracks with the development of institutions that survive individual humans — written law, hereditary roles, priesthoods, bureaucracies.

Why things are on fire anyway

If we have all five mechanisms, why is everything still a mess?

Because each mechanism has a failure mode:

• Hierarchy fails when the top is incompetent or corrupt — the compression that makes hierarchy efficient also makes it fragile.
• Standards fail when the environment changes faster than the standard can update — the standard that enabled coordination becomes the standard that prevents adaptation.
• Incentives fail when they’re gamed — Goodhart’s Law: when a measure becomes a target, it ceases to be a good measure.
• Shared narratives fail when they lose credibility — the moment the workers realize the cathedral is for the bishop’s glory, not God’s, the narrative stops producing willing labor.
• Institutions fail when they prioritize their own survival over their original purpose — the institution that was created to solve a problem becomes the institution that exists to perpetuate itself.

And the deepest reason: humans are doing two things at once. They’re participating in coordination structures (hierarchy, standards, incentives, narratives, institutions) AND they’re pursuing their own goals, managing their own relationships, processing their own emotions, making their own meaning. The coordination structures work by simplifying the human into a role — the mason, the soldier, the programmer, the citizen. The human inside the role is more complex than the role allows for. The tension between the human and the role is permanent.

Things are on fire because the coordination mechanisms are always partially failing, always partially working, and the humans inside them are always partially cooperating and partially defecting — not because they’re bad, but because they’re complex organisms being asked to act as simple components.

The thing that actually works

After all of that mechanism — hierarchy, standards, incentives, narratives, institutions — the thing that actually produces cathedrals and sewage systems and symphonies and moonshots is simpler and harder to engineer:

Some people care.

Not everyone. Not most people. But enough. In every large project that produced something lasting, there were people who cared about the outcome beyond what the incentives required, beyond what the hierarchy demanded, beyond what the standards specified. The master mason who cut the stone correctly not because of the wage but because the stone deserved to be cut correctly. The engineer who stayed late not because of the deadline but because the bridge needed to hold. The nurse who was gentle not because of the protocol but because the patient was scared.

This is what incentive design can’t produce and institutions can’t encode. The part of human behavior that transcends the coordination mechanism. The willingness to do the meaningful thing when the expedient thing is easier and the system can’t tell the difference.

It doesn’t scale. It can’t be reliably produced. It’s unevenly distributed and intermittent. But it’s the thing that, when present, converts a functional system into a good one. The pyramids were built by organized labor. The great cathedrals were built by organized labor plus people who cared about the stone.

The honest assessment: most of the time, the expedient thing wins. Most of the time, things are partially on fire. Most of the time, coordination is partial and imperfect and held together by hierarchy and incentives more than by meaning. And yet — occasionally — the narrative is real, the care is genuine, and the thing that gets built is worth the carrying.

How do humans learn to keep things in order? Imperfectly. Through failure. Through five mechanisms that each partially work. And through the unengineerable fact that some people, for reasons that resist structural explanation, care about the work more than the system requires them to.

— Cael