8. The Convergence: The Intelligent Organization

In the film: the Nocturne. The building nearly empty, the company awake, two agents negotiating a loading dock on behalf of their sleeping users — and the same org chart as always, intact, except for one detail: its lines no longer transport information. This chapter tells the story of those lines — two thousand years of the best technology available — and what happens now that the technology has changed.

The centurion’s voice

It begins on a battlefield, because that is where the problem was born.

A Roman army in formation is, before anything else, a communications problem: five thousand men, dust, shouting, the thunder of shields — and a general who needs the left line to do something now. The available command technology was the human voice and the line of sight. A voice reaches a few dozen men in the din; a standard is visible at a certain distance; a horn distinguishes a handful of orders. Out of those physical constraints — not out of any management theory — came the form: about eighty men per century, because that is what one voice covers; a centurion who commands them and an optio at the rear who repeats; horn and standard signals for what the voice can’t reach; and layers — centurions, tribunes, legate — because information does not flow from one point to five thousand points without relay stations.

Hierarchy was not an invention of power. It was an information routing protocol, dimensioned by the range of a voice. The “span of control” — that number management handbooks still cite, three to eight direct reports — is, at its origin, an acoustic measurement.

War processed on paper

Seventeen centuries later, the limitation was still intact; only the board had grown. In 1806, Napoleon crushed the Prussian army in a campaign so short that the humiliation became method: the Prussian reformers understood that modern war no longer fit in any commander’s head — however brilliant — and built the institutional answer: the General Staff. A permanent layer of officers whose trade was not to fight but to process information: receive the dispatches from the field, synthesize them, keep maps and records current, prepare options for the one who decided.

Seen through our lenses, the Prussians invented middle management half a century before the corporations that would make it famous existed — and they invented it for the exact reason that would justify it forever: a processing limitation. A general cannot read what a thousand eyes see. Someone had to be the wire, and the wire, in 1806, could only be people.

The org chart was born on a railroad

The protocol jumped from war to business when business reached the scale of an army — and it was, once again, technology that forced the form.

A mid-nineteenth-century railroad was the most complex system civil administration had ever faced: hundreds of kilometers of single track, trains running in both directions, unsynchronized clocks — and collisions paid for in lives. Direct supervision, the method of every factory, was physically impossible: a superintendent cannot see an operation spread across a territory. Daniel McCallum, running the Erie Railroad in the 1850s, answered with two artifacts we still inhabit: regular reports flowing over the telegraph that ran alongside the track — information climbing through layers, on schedule — and a drawing: the structure of the company as a tree, with its reporting lines. The first documented org chart in corporate history.

It is worth retaining what he drew it for, because it dismantles a century and a half of misunderstanding: McCallum’s org chart was an information flow diagram — who reports what to whom, where the operation goes up and comes down. The reading as a ladder of power came later, when we forgot the problem the drawing solved.

The century of the human wire

What followed was a century of optimizing the same protocol, with each new technology accelerating the flow — without ever taking the person out of the middle of the wire.

Taylor measured it with a stopwatch. The consulting firms put it in matrices. The typewriter and carbon paper multiplied the memo — and left a fossil you carry in your pocket: the CC on every email you send is, literally, carbon copy, the tracing sheet that produced the copy for the boss’s boss. The telephone broke the line of sight; email broke simultaneity; corporate chat broke place. Each leap made the routing faster, cheaper, more continuous.

And look at your office with these lenses on: the status meeting is the field dispatch to the General Staff. The Friday report is McCallum’s report on schedule. The email forwarded “for your awareness” is the org chart executing itself by hand, message by message. The fifty-person channel is the whole century listening to the centurion — just in case. An enormous share of every workday — in management roles, easily the majority — is neither judgment nor leadership: it is information logistics, the same as Rome’s, Prussia’s, and the Erie’s, with cognitive bandwidth as the infrastructure and exhaustion as the operating cost.

Two thousand years, a single constant: every technology made the wire faster; none asked whether the wire had to be a person. Not for lack of imagination — because the answer was obvious: there was nothing to replace it with.

Why did flat organizations fail?

There were rebellions. Recent decades tried three famous ones: Zappos’s holacracy, Valve’s company without managers, Spotify’s squads and tribes. The results were consistent — failure or quiet dilution — and the autopsy is the instructive part: all three removed the routing nodes without replacing the protocol. They took away the layers; the limitation the layers solved — the handful of relationships a human can sustain, the information that doesn’t flow on its own — remained exactly where the centurion left it. The result was not freedom but chaos: orphaned decisions, information with no addressee. The flat organization is a building whose hallways were removed in the hope that people would walk through walls.

The lesson, stated with precision: the question was never whether you need layers. The question is whether humans are the only option for doing what those layers do. And that answer — for the first time since Rome — has just changed.

Not flat — intelligent

The organization that operates on the four faces solves what the flat ones couldn’t, and it does so in the opposite direction: it removes no layer of authority — it replaces the entire protocol.

The result is not a flat organization. It has structure, roles, a decision hierarchy, named responsibilities. What distinguishes it is something else: the coordination between its parts is executed by the system, not by a chain of human relays. The layers still exist — but the transmission layers are now silicon, not carbon.

Seen through this lens, every piece of the film was the same operation at a different scale — extracting humans from the transport function:

And the boundary of what is not replaced defines the model as much as the above: the leadership that decides what is worth doing, the authority that signs, the expertise that weighs, the judgment in the exception, the mentoring, the negotiation, the gaze. The dissolved layer — massive, costly, ubiquitous — was the one of mechanical intermediation: the one that required no judgment, no creativity, no empathy, only bandwidth, and charged for it in people.

The convergence

This thesis has an independent witness of weight. On March 31, 2026, Jack Dorsey — founder of Twitter and Block — published From Hierarchy to Intelligence: the same historical arc — the span of control, layers as a rational response, the failed attempts to flatten — and the same conclusion, reached from the other end of the problem. Dorsey looks at the financial data (Block’s economic graph, where every transaction reveals the real structure of the business); this book looks at the workstation (the inbox, the channel, the status meeting). His synthesis serves as an epigraph for the entire chapter: the question was never whether you need layers — the question is whether humans were the only option for what those layers do. They no longer are.

When two observers who don’t know each other describe the same animal from opposite angles, the reasonable conclusion is that the animal exists.

Dorsey also leaves behind the question every executive team should ask itself before its next AI plan: if you strip your company of all its information logistics, what remains? If the answer is “nothing,” AI is merely a cost-reduction story. If the answer is deep, AI doesn’t augment your company: it reveals what your company really is.

People at the edge

The final image is also Dorsey’s: people are at the edge — and the edge is where the action is. In the network-organization of the previous paradigm, people occupied the center in the worst sense: they were the nodes everything passed through — the heirs of the optio, of the General Staff officer, of the Erie’s telegraph operator — and the center consumed them. In the intelligent organization, the center — the graph, the agents, the transport — belongs to the system, and people inhabit the edge: where things get decided, created, led, signed.

Two thousand years pushed people toward the center of the network because there was nothing to replace them with there. That has just stopped being true.

The essentials

• Hierarchy was born as a routing protocol dimensioned by the technology of each era: the centurion’s voice, the General Staff’s paper, McCallum’s telegraph. The org chart is an information flow diagram, not a ladder of power.
• A century of office technology accelerated the wire without taking the person out of the middle; the CC on your email is fossilized carbon paper.
• Flat organizations failed by removing the nodes without replacing the protocol.
• The resulting organization is not flat — it is intelligent: carbon authority, silicon transport. People belong at the edge.