Photo by Jessica Kantak Bailey on Unsplash

There is something almost unfair about seeing Versailles for the first time. We arrive knowing how the story ends. We know about bread and debt, revolution and executions. We know the monarchy will fall, just as we know the palace will eventually become something closer to a monument than a seat of government, its rooms preserved for strangers who purchase tickets to walk where kings once expected petitioners to wait.

That knowledge follows us through the gates.

Still, Versailles is magnificent. The gardens stretch outward with geometric confidence. Hallways terminate in rooms that seem designed less to accommodate human beings than to communicate something about the people permitted to occupy them. Gold catches the light almost everywhere, and the ceilings turn government into mythology. The Hall of Mirrors reflects windows, chandeliers, paintings, tourists, and itself until the distinction between architecture and spectacle becomes difficult to maintain.

It is tempting, from this side of history, to see only excess. That interpretation is convenient, and incomplete.

Versailles did not begin as an attempt to build a monument to political blindness. Louis XIV inherited a hunting lodge from his father and transformed it over decades into something… more. In 1682, Versailles became the headquarters of government, gathering court and administration around a monarchy that had become increasingly centralized around the sovereign himself.

It was not only a palace. It was an organizational design.

France was large, and its interests were distributed. Nobility possessed independent sources of wealth, influence, and regional power, ministers had their own responsibilities, and provincial administrators their own realities. Armies fought at distances the king could not personally observe. Taxes, commerce, agriculture, religion, infrastructure, diplomacy, and law generated more information than any single individual could directly handle.

The kingdom needed roads to the center, and Versailles became one of them. Then, gradually, many of the roads began ending there.

That distinction matters, because organizations rarely become isolated when people stop communicating. Sometimes they become isolated because the people at the center become extraordinarily good at communicating with one another.

Before the Palace

Every large institution begins with distance. Sometimes that distance is physical, but more often, today, it is functional: finance understands the books, sales understands the customers, operations understands production, technology understands systems, legal understands obligations, human resources understands policy, engineering understands machinery, and procurement understands suppliers.

Each group learns its own language because specialized language is useful. Each develops its own processes because repeated work becomes more efficient when people agree how it should be performed. Each builds systems to store the information it needs and relationships with the people who help it succeed. None of this is inherently dysfunctional. A silo is often simply specialization viewed from the outside.

The problem emerges when the walls become more durable than the purpose they were built to serve.

The people inside one group begin solving problems whose consequences belong to another. Information accumulates locally, while decisions cross boundaries more easily than context. Two departments discover that they maintain different versions of the same truth. Three others develop separate solutions to the same problem. A project requiring five teams becomes less a project than a negotiation among five independent governments.

Soon, the organization has islands. Some are friendly, and build bridges. Some trade regularly, while others stopped speaking after a dispute nobody currently employed can fully remember. Senior leadership sees the consequences eventually, because nearly every difficult problem begins acquiring the same strange shape: the individual teams may be competent, and the whole is not coordinated.

So leadership does what leadership is supposed to do: it centralizes.

Not necessarily authority at first; communication holds that position. Bring the kingdom together. Create common forums, open channels, ask questions, and listen to the answers. Map what is happening, identify duplicated work, and discover where people are blocked. Find risks nobody at the center knew existed because every individual function assumed someone else had already reported them.

For a time, this can feel transformative. The sovereign leaves the palace, or perhaps more accurately, opens its doors. People who previously communicated almost exclusively through their hierarchy are suddenly invited to speak beyond it. Questions once filtered through layers of management can be asked directly. Leaders explain what they have learned. Employees begin seeing evidence that somebody with the authority to act has encountered the same problems they have.

The effect can be profound. Not because every problem is solved, but because the distance between experience and authority becomes visible. Communication creates something institutions often lack long before they realize it: a shared map. Here is the broken road, and there, the village nobody has visited. Here are the soldiers waiting for supplies, and here, the bridge everyone assumed someone else maintained. Here are the competing maps we have been using to describe the same territory.

The center learns as the edges begin to recognize one another. And for perhaps the first time in years, the institution can see itself.

This is why centralization is so seductive: sometimes it works.

Bring the Kingdom to the King

Louis XIV understood something similar. After the death of Cardinal Mazarin in 1661, he chose not to appoint another first minister. Instead, he placed himself more directly at the head of government, surrounded by ministers, secretaries of state, officers, and councils through which policy could be discussed and royal decisions expressed.

Versailles eventually brought much of the political world together around him.

This had practical value. The nobility could be observed. Ministers could be consulted. Petitions could arrive and orders could depart. Ambassadors could be received. Political rivalries that might otherwise develop far from royal supervision unfolded within sight of the crown. As such, the palace itself became an interface between ruler and ruled – a communication technology made from stone.

And like many successful technologies, its success changed the behavior of the people using it.

Royal favor mattered, and as a result, so did access. Presence mattered because access depended upon it. The Palace of Versailles describes a court that could include thousands of people, ordered by strict hierarchy and etiquette. Regular attendance could bring lodging, financial rewards, invitations, appointments, and influence. Rules governed who could approach important figures, where they might do so, and under what circumstances.

Proximity to the king became useful, so rational people pursued proximity to the king.

This is not particularly mysterious. Give any system a scarce resource and people will learn how the resource is allocated. If decisions are made in a room, people will try to gain admission. If initiatives gain funding through a council, people will learn how the council thinks. If advancement depends upon executive visibility, people will become visible to executives.

If strategic importance is determined through presentations, people will become excellent at PowerPoint.

Eventually an organization develops people who are extraordinarily skilled at operating within the mechanisms created to coordinate it. That can be good, but it can also be the beginning of the court.

Too Many Letters

There is a practical problem hiding beneath every dream of direct communication: scale. A leader can speak directly with ten people, perhaps a hundred – with considerable effort. Ten thousand cannot have meaningful individual access to one person, no matter how sincere the invitation. The arithmetic eventually wins.

Questions multiply while reports lengthen and meetings consume the calendar. Every problem arrives with history, context, stakeholders, constraints, personalities, risks, dependencies, and unresolved arguments. The very success of open communication creates more information than the center can process.

So the institution adapts. It creates advisors, councils, committees, steering groups, working groups, transformation offices, centers of excellence, executive sponsors, portfolio managers, and representatives. None of these are evidence of failure; in fact, most are evidence that somebody recognized a real limitation and attempted to solve it.

The Ming dynasty discovered a version of the same problem centuries ago through a very different system. In 1380, the Hongwu Emperor eliminated the chief ministerships, concentrating greater responsibility around the emperor. Yet removing the intermediary did not remove the need for mediation. A gap emerged between throne and bureaucracy, and structures including the Grand Secretariat developed to bridge it. Cambridge’s history of the period explicitly connects the removal of the chief ministership with the later growth of both secretarial and inner-court structures.

The lesson is not that emperors should keep chief ministers. It is that complexity does not disappear when we remove the layer responsible for interpreting it. The work remains, and eventually someone performs it.

This is how the direct road becomes a checkpoint.

At first, the checkpoint helps. Not every message deserves the sovereign’s attention, and not every problem requires an executive decision. Someone, somewhere, should combine duplicates and identify common themes. Someone should separate inconvenience from existential risk, and coordinate three departments before escalating a dispute that they can resolve themselves.

The advisor protects the center from noise, the council creates order, and the representative makes scale possible.

Twenty reports become five themes. Five themes become three priorities. Three priorities become one page. The page reaches the center, and efficiency improves.

And something is lost. Not necessarily anything important, and not at first. But compression is never neutral.

Someone decides what the summary contains. Someone decides which examples are representative. Someone decides whether a concern is isolated or systemic. Someone decides which risk is urgent. Someone decides what context can safely be removed. Someone decides whether disagreement represents a legitimate challenge or merely resistance to change. Every layer between observation and authority makes an interpretation and reshapes the message accordingly.

The letter still reaches Versailles. It simply arrives shorter than when it was written.

Eventually, the center may know everything that has been reported to it while knowing remarkably little about everything that has actually happened.

The Court Forms

Courts do not require kings. They require only three things: access must be scarce, access must matter, and some people must possess more of it than others. Once those conditions exist, a social structure begins forming around the scarce resource.

At first, members of a council communicate with one another because they must. Then they communicate before the council meeting, because coordination makes the meeting more productive. Then afterward, because actions need to be aligned. Eventually, they develop shared vocabulary, and acquire common context. They learn which arguments succeed and what senior leadership worries about, and they become better at anticipating objections. They know which initiatives are likely to receive support and which will require additional framing.

Again, none of this requires corruption. These are the behaviors of competent people learning to operate effectively inside a system; and that fact is precisely why the pattern is dangerous. The court can be populated entirely by intelligent, diligent, well-intentioned people. In fact, it usually is.

But human beings allocate attention according to the environment around them. The more time representatives spend representing the edges to the center, the more of their working lives are necessarily spent with other representatives and with the center itself. Their information environment changes. Their peer group changes. Their vocabulary changes.

Eventually their incentives can change.

The question that began as what is happening in the organization? quietly becomes what does leadership need to know about what is happening in the organization? Those questions sound nearly identical. They are not.

The first is observational. The second is editorial.

Then another shift occurs: how should we communicate this to leadership? Then: how will leadership interpret this? Then: how does this affect the current strategic priority? The farmer has become a bullet point, the general a status indicator, and the damaged road? A dependency.

Reality is still present, but increasingly it appears only in translation.

Meanwhile, communication inside the court is flourishing. Meetings occur constantly, messages travel rapidly, and presentations circulate. Executives appear in newsletters, councils issue updates, and dashboards multiply. Employees receive more communication than ever before.

This produces one of the strangest failures available to an institution. It can become simultaneously overcommunicated and uninformed.

The Hall of Mirrors

Perhaps there is no room more appropriate to this problem than the Hall of Mirrors. Its western wall contains mirrors facing windows that overlook the gardens. Light enters, reflects, repeats, and fills the room. Here, everything seems multiplied. Stand in the right place, and the room appears larger than it is.

That is what an internally dense communication network can do to an institution.

Ask the center whether communication has improved and the evidence may be overwhelming. Look at the number of cross-functional meetings, the participation in strategic councils, the executive communications, the town halls, the dashboards, the governance structure, the surveys, the messages, the alignment.

Everyone appears to be talking to everyone. But network activity is not the same thing as information transfer.

A system can produce enormous quantities of communication while repeatedly circulating the same interpretation. The risk increases when people close to the center share similar information sources: An advisor hears from another advisor. A council receives a summary prepared by a working group. A senior leader receives the council’s conclusions. Another executive hears the same conclusion from the same council. Eventually, they compare notes, and agreement emerges.

From inside the room, the agreement feels independently corroborated. Four people saw the same thing.

Except perhaps they did not. Perhaps four mirrors reflected the same window.

This is where organizational consensus becomes dangerous. Not because consensus is undesirable, but because we frequently fail to ask whether apparently independent conclusions have independent origins. Ten leaders can agree about the health of an initiative while all ten ultimately depend upon the same two sources. A hundred presentations can repeat that a transformation is progressing successfully because every presentation inherited the designation from the previous reporting cycle.

A strategy can appear universally accepted because the people still invited into the room have learned the boundaries of acceptable disagreement. Communication continues, just as signal decays.

Outside the palace, something else is happening.

Beyond the Gates

Organizations rarely collapse because a council met too often. The consequences are less obvious than that.

A project requires a decision, and the decision is elevated. The relevant council has a larger strategic issue under review, so the project waits. Another priority arrives with executive sponsorship and resources move. The original project remains important, but no longer urgent.

Three months later, someone asks why it has not progressed, and a new plan is created.

Elsewhere, an experienced employee has spent two years explaining a structural problem. At first they are persistent, then frustrated, then pragmatic. Eventually they stop raising it at all, which may be celebrated as improved alignment. Six months later they accept another job, and their departure is categorized as regrettable attrition.

Elsewhere, a customer problem crosses three functions. Every function sees only the portion it owns. The council responsible for cross-functional coordination has already prioritized five larger themes. The customer leaves, and the organization records a commercial loss.

Elsewhere, a critical system is aging. The people closest to it understand the risk intimately, but replacing it competes with an initiative whose strategic value is easier to explain in executive language. Maintenance continues, and the system survives. For now.

These are the corporate equivalents of neglected roads. No single damaged road destroys a kingdom. A farmer simply takes longer to reach market, or a courier arrives late. Perhaps a wagon breaks an axle, or a regiment receives supplies two days after it needed them.

Each incident is manageable. Collectively, however, roads determine whether the center actually governs the territory it colors on a map.

This is where the court becomes self-reinforcing. The worse the roads become, the harder reality becomes to transmit. The harder reality is to convey, the more the center depends upon representatives. The more the center depends upon representatives, the more valuable membership in the court becomes.

And the more valuable court membership becomes, the more organizational energy moves toward the center.

Eventually, everyone important is discussing the kingdom, but few of them are still standing in it.

This Is Not an Accident

It would be comforting to believe this happens only under poor leadership. Then the solution would be simple: hire better leaders, replace arrogant executives with humble ones, tell people to listen, or add another survey.

Unfortunately, the mechanism is more fundamental. Large systems naturally generate information faster than central decision-makers can absorb it, so they compress. That compression requires judgment, and judgment creates gatekeepers, which accumulate context. The same context makes gatekeepers more valuable, and valuable gatekeepers gain access, which eventually creates influence. Influence attracts attention, which in turn, creates communities around influence.

Those communities develop internal communication, and internal communication becomes faster than communication with the periphery because the members share vocabulary, relationships, context, and incentives.

Eventually, the shortest informational path through the organization runs through people near the center. It is at this point that the center becomes a place. Not necessarily a physical headquarters, but a network – a cluster of people who are disproportionately connected to one another.

The silos never disappeared. They migrated.

This is why destroying organizational islands does not guarantee integration. Destroy enough islands at the edge and, without deliberate counterpressure, you may simply build a continent at the center. The old silos at least knew they were silos; most courts still believe they represent the kingdom.

That belief is understandable – the court receives reports from everywhere. Its members speak with people across functions, and monitor enterprise dashboards. They participate in strategic initiatives and possess more organizational information than almost anyone outside it. The mistake, however, is subtle. Having information about everywhere is not the same as maintaining contact with everywhere.

Maps are essential. But eventually someone has to inspect the road.

Keep the Roads Open

The answer is not to abolish the palace. Large organizations need coordination, and leaders need advisors. Executives cannot personally investigate every operational concern, and councils can solve problems that individual departments cannot.

Hierarchy exists because unlimited direct communication does not scale. The goal is not to eliminate mediation, but to prevent mediation from becoming isolation. That, in turn, requires infrastructure.

First, the source must remain reachable. If information is summarized before reaching leadership, the summary should not sever the path back to the people who produced it. An executive does not need to interview every engineer, but when an engineering risk becomes material, someone in the decision chain should be able to speak directly with the engineer who understands it. The person who summarizes the warning should not exclusively control access to the person making it.

Second, leaders need uncurated contact with the edges. Not ceremonial contact, or a carefully selected panel of employees prepared by management, and certainly not a visit where everyone has been told what the executive hopes to discuss. Real contact: periodic exposure to customers, operators, engineers, analysts, technicians, salespeople, support staff, factory workers, project teams, and others whose incentives do not depend primarily upon maintaining the internal narrative. The goal is not for the sovereign to manage the farm, only to occasionally remember what soil looks like.

Third, councils need explicit limits. Every council should be able to answer why it exists, what decisions it owns, how its success can be measured, what information it requires, and under what conditions it should cease to exist. Without those limits, coordination structures have a tendency shared by nearly every durable bureaucracy: they discover additional reasons for their own continuation.

A council formed to address a crisis becomes a council responsible for preventing similar crises, then for governing the processes related to those crises, and eventually for reviewing requests affecting those processes. Eventually the organization can no longer remember how decisions in that area were made before the council existed.

Once this happens, the bridge has become a tollbooth.

Fourth, urgency should become less compressible as severity increases. Routine information can tolerate layers. Critical warnings should bypass them. The more severe the potential consequence, the fewer opportunities the organization should permit for the message to be reframed before reaching someone capable of acting. This does not mean every person who labels something critical receives an audience with the CEO. It means escalation architecture should be designed around consequence, not merely hierarchy.

Fifth, dissent must have somewhere to go after consensus forms. This is harder than encouraging people to speak up – most organizations say disagreement is welcome. The useful question is what happens after the disagreement is expressed.

Does the dissenter have to persuade the same group whose decision they are challenging? Can materially new evidence reopen a settled decision? Is there an independent path for risk escalation? Does anyone periodically ask which assumptions would have to be wrong for the current strategy to fail?

A system that welcomes dissent but provides no mechanism through which dissent can alter a decision is not listening. It is collecting.

Finally, send the court back into the kingdom. Rotate membership. Return representatives to operational work. Allow people closest to execution to participate directly in decisions affecting it. Prevent strategic roles from becoming permanent substitutes for firsthand knowledge. In short, require the people designing roads to occasionally travel them.

This is not anti-leadership, but maintenance. Every communication architecture degrades: relationships concentrate, shortcuts become standard routes, and successful intermediaries accumulate additional responsibilities. Temporary structures become permanent, just as language becomes abstract, and the system drifts inward.

There is no perfect organizational design that solves this forever, but the roads must still be maintained; people will always find it easier to speak with those already standing nearby.

Letters to Versailles

In October 1789, Louis XVI and the royal family left Versailles for Paris, but the monarchy did not disappear that day. The Revolution had much further to travel, and reducing its causes to royal isolation would be absurd. France faced profound financial, political, social, and institutional crises that no communication architecture alone could have solved. But the court at Versailles was finished.

The palace, however, remained. It remains still.

That may be the most useful part of the story. Institutions do not usually experience their failures as revolutions. There is no crowd at the gates, and no dramatic final morning when everyone recognizes that the system has lost contact with the world it governs.

Corporate decline is usually more polite. A talented employee leaves, then another. A competitor takes a little market share, and a customer chooses someone easier to work with. A major project extends another quarter, or a transformation initiative requires a second transformation initiative to resolve problems created by the first. The organization restructures. A business unit is sold.

A new leader arrives, and another strategy begins.

Sometimes the company survives indefinitely, smaller than it once imagined itself becoming. Sometimes it is acquired, and the name remains on products for a while before becoming a brand inside another portfolio, governed from someone else’s palace.

There need not be villains anywhere in the story; in fact, that is what makes it worth studying. Louis XIV did not gather France around himself because he wanted his successors to misunderstand the country. Councils are not created because leaders wish to become insulated. Advisors are not appointed because executives hope to receive filtered information. Communication teams do not exist to prevent communication.

The structures usually begin as answers to real problems. So did Versailles. That is why the lesson cannot simply be do not build palaces. As humans have proven time and again, we will.

Complex organizations need centers – places, physical or otherwise, where information is collected, priorities are negotiated, and decisions are made. The question is whether the roads continue beyond them.

Whether a warning can still travel inward without becoming unrecognizable.

Whether a question from the edge can reach someone empowered to answer it.

Whether the people governing the organization still encounter people who do not spend their days governing the organization.

Whether advisors remain translators rather than gatekeepers.

Whether councils continue serving the problems they were created to solve rather than teaching the organization to serve the councils.

… And whether the people standing nearest the center remember that proximity gives them a better view of the palace, not necessarily a better view of the kingdom.

Today, millions of people visit Versailles because the court built something extraordinary. We walk through its rooms and look upward. We photograph the ceilings, admire the gardens, and inevitably, we enter the Hall of Mirrors.

It remains breathtaking. Light enters from the gardens and crosses the room, reflected again and again by mirrors opposite the windows. And for a moment, the room seems almost without boundary.

Perhaps that is the final warning: A room can be filled with light and still show you mostly what is already inside it. Just as an institution can be filled with communication and still know remarkably little about the world beyond its walls.

The letters may still arrive at Versailles. The roads may still technically reach the gates. The councils may read every page. The advisors may discuss them, even carefully.

The sovereign may receive an excellent summary.

And somewhere far beyond the mirrors, a farmer is still waiting for the road to be repaired.