The Structures That Govern Us Also Shape What We Can See

What if the organization you work in is not merely a machine for producing results, but an instrument that quietly changes the people who operate it?

That question makes organizational design more consequential than a choice between hierarchy and decentralization. A management system does not simply distribute authority. It trains attention, rewards certain kinds of perception, and leaves a residue on everyone who passes through it. Like light bouncing off a surface, work reflects the structure that shaped it. The reflection may be subtle, but it is never neutral.

This is why decentralized management is both more promising and more difficult than its advocates often suggest. Giving people autonomy can release intelligence that hierarchy suppresses. But autonomy alone does not create a healthy organization. It can also produce fragmentation, invisible power, endless coordination, or a culture in which the loudest internal process overwhelms the people the organization exists to serve.

The deeper challenge is not how to remove hierarchy. It is how to design a structure that enlarges human perception without losing contact with reality.

Every Structure Is a Camera

Holacracy begins with an important conceptual separation: a person is not identical to a job description. One person may hold several roles, and each role carries explicit accountabilities, authority, and constraints. Teams, or circles, govern themselves through defined processes. Authority is distributed through a system of nested units that are autonomous in some respects and dependent in others.

This resembles a camera with multiple lenses. Instead of asking, “Who is the boss?” the system asks, “What role is responsible for this kind of work, and what information does that role need?” The advantage is clarity. A person can act with authority because the authority belongs to the role, not because the person has accumulated status.

But every camera also excludes something. A lens brings certain objects into focus while leaving others at the edges. A role can see its accountabilities clearly while missing the emotional atmosphere of a team. A circle can become excellent at resolving internal tensions while losing sight of the customer. A governance process can integrate objections from relevant roles while quietly defining relevance so narrowly that inconvenient voices never enter the room.

This is the first principle of organizational design: authority determines not only who can act, but what the organization is capable of seeing.

A traditional hierarchy tends to make the chain of command highly visible and the informal network obscure. A decentralized system may make roles and policies visible while leaving other forms of power hidden. People can still control access to information, shape agendas, determine who counts as relevant, or become indispensable because they alone understand the process. The formal structure becomes flatter, but the actual structure may remain steep.

The difference is similar to the difference between a map and a landscape. A map can be precise about roads and still omit the river that floods every spring. A governance chart can identify accountabilities and still fail to register fear, status, fatigue, customer disappointment, or the quiet withdrawal of talented people.

A structure is not neutral scaffolding. It is a filter that decides which signals become actionable and which signals disappear.

The question, then, is not whether an organization has a structure. It must. The question is whether its structure helps people perceive the world it is responsible for.

The Residue of Autonomy

There is a useful metaphor for this process: everything we encounter leaves an imprint on us, even the light that bounces off us. The metaphor is physical, but its organizational meaning is profound. Work systems imprint themselves on judgment, language, posture, ambition, and identity.

A command hierarchy may imprint caution. People learn to wait for permission, anticipate the preferences of superiors, and treat initiative as a risk calculation. A poorly designed decentralized system may imprint constant vigilance. People learn that every problem requires another meeting, another negotiation, another act of social navigation.

The stated philosophy may be freedom, but the lived experience may be exposure. When people hold multiple roles across several circles, they may gain flexibility, yet lose a stable sense of where they belong. Constant connectedness can become a form of enclosure. There is no single boss to escape because the whole system is always present.

This helps explain why a management system can look liberating from the outside and feel exhausting from within. Autonomy is not the same as relief. It means that decisions move closer to the work, but it can also mean that individuals absorb more ambiguity, coordination costs, and emotional responsibility. If the system does not provide enough protection, prioritization, and space, autonomy becomes another name for distributed strain.

Consider two product designers working in different organizations.

The first reports to a department head who approves every meaningful decision. Over time, she becomes skilled at reading authority. Her design work may be technically strong, but she learns to optimize for approval rather than discovery. She becomes a careful interpreter of power.

The second works in a decentralized system with broad authority to act. She can test a prototype without waiting for permission, but she belongs to several overlapping circles. Each has its own meetings, policies, and tensions. She becomes skilled at navigating interfaces. Her design work may be inventive, but her attention is fragmented by governance.

Neither structure simply liberates or oppresses her. Each trains her to become a particular kind of person.

This is why the evaluation of a management system should include a question rarely asked in performance reviews: What habits of mind does this system produce after three years?

Does it create people who notice customers sooner? People who can make decisions under uncertainty? People who understand tradeoffs across the whole organization? Or does it create specialists in permission seeking, process defense, or internal diplomacy?

The long term effects matter because organizational culture is not mainly what an organization says about itself. Culture is the accumulation of behaviors that its structure repeatedly makes sensible.

Why Self Management Needs an Outside

Decentralized systems are often praised for being biased toward action. That bias is valuable. In complex environments, waiting for perfect consensus can be more dangerous than making a reversible decision with incomplete information. A process that focuses discussion on the concrete next step can prevent meetings from becoming contests of personality or abstract debate.

Yet action has an orientation. Action toward what?

An organization can become highly efficient at responding to internal tensions while becoming less responsive to external reality. It can solve the problem brought by the person in the room while neglecting the problem experienced by the person not in the room. A governance process may be excellent at integrating objections from roles, but customers do not always have a role. Nor do future users, excluded communities, suppliers, or the public consequences of a decision.

This is the missing outside.

A self organizing unit is not self sufficient. The word “whole” can mislead us. A circle may be a whole for one purpose and a part of a larger system for another. Its autonomy is real, but so is its dependence. A team that optimizes its own metrics may damage the customer journey. A function that protects its own capacity may slow the organization. A company that improves internal coordination may still produce something nobody wants.

The artistic metaphor clarifies this. An artist is supposed to open a window onto the world, but a window cannot reveal what the artist has never tried to see. Likewise, an organization can only respond to realities it has built practices for noticing. If every meeting begins with internal metrics, every agenda is generated by existing roles, and every objection must be translated into a current accountability, then the system will reproduce its own field of vision.

The customer is not merely another stakeholder to add to a list. The customer is a test of whether the organization’s internal reality corresponds to the external world.

A useful governance system therefore needs two forms of sensing:

1. Internal sensing, which detects tensions in roles, workflows, authority, and resources.
2. External sensing, which detects changes in customer needs, competitive conditions, social impact, technology, and the lived experience of people outside the organization.

Internal sensing asks, “What is obstructing this role from doing its work?” External sensing asks, “What are we failing to notice because our roles make it inconvenient to notice?”

The first keeps the organization functional. The second keeps it relevant.

Self management without external perception is not freedom. It is a well organized form of self reference.

The Hidden Design Problem: Who Gets to Define Relevance?

One of the most important questions in any participatory system is also one of the easiest to overlook: who decides who is relevant?

If every person must be consulted on every decision, the organization freezes. If only formally assigned roles can object, the organization becomes blind to signals that have not yet been encoded. The challenge is to distinguish between noise and early warning.

Imagine a customer support specialist noticing that users are confused by a new feature. The product circle may say the specialist has no authority over product design. Formally, that may be true. Epistemically, it may be absurd. The specialist is close to a pattern that the product team cannot see from its dashboards.

Relevance is therefore not identical to jurisdiction. The person responsible for a decision is not necessarily the only person capable of seeing its consequences.

A mature organization needs a relevance protocol, not merely a list of roles. Such a protocol might ask:

• Who experiences the consequences of this decision directly?
• Who has unusual proximity to the relevant reality?
• Who has a track record of identifying similar problems early?
• What evidence would change our mind?
• Which absent perspective would make this decision look obviously incomplete?

These questions prevent two opposite errors. The first is indiscriminate consultation, in which everyone is invited and nobody can think. The second is procedural exclusion, in which only recognized authorities can introduce information.

The goal is not consensus. Consensus treats disagreement as a problem to eliminate. Better governance treats disagreement as a diagnostic instrument. An objection is valuable when it reveals that a proposal creates a concrete risk, violates a responsibility, or ignores an important dependency. But a system must also allow observations that arrive before they can be neatly translated into a role’s formal needs.

That requires humility. The organization must admit that its current structure is a hypothesis about reality, not reality itself.

Designing Organizations That Leave Better Imprints

The most useful way to assess decentralized governance is not to ask whether it eliminates managers, meetings, or hierarchy. Ask instead whether it improves the quality of the imprint it leaves on people.

A good system should make people more capable of acting, more sensitive to consequences, and more connected to the purpose beyond the organization’s internal machinery. It should provide enough autonomy for initiative, enough structure for coordination, and enough contact with the outside world to prevent collective self absorption.

This suggests a practical model with three layers.

1. The action layer

Who has permission to act, and what decisions can be made without waiting? The default should be broad authority for reversible, local decisions. Restrictions should be explicit, intelligible, and tied to real risks rather than inherited status.

2. The learning layer

How does the organization discover that its current roles, policies, or assumptions are wrong? Governance should not only resolve tensions. It should regularly examine which tensions are being generated by the structure itself.

3. The contact layer

How does external reality enter the system? Customers, users, frontline workers, and affected communities need more than symbolic representation. Their experiences must have a path into priorities, resource allocation, and role design.

These layers correspond to three organizational questions:

• Can we move?
• Can we learn?
• Can we see?

An organization that can move but cannot learn becomes reckless. One that can learn but cannot move becomes bureaucratic. One that can move and learn but cannot see becomes brilliantly efficient at solving the wrong problem.

The transition to a new system is especially delicate. If old management practices remain as an unofficial shadow structure, employees must satisfy two constitutions at once. The formal system says authority belongs to roles. The informal system says real authority belongs to founders, executives, high performers, or those who control scarce information. This contradiction breeds cynicism because people are asked to trust a structure that their daily experience disproves.

Transitions also fail when they treat people as components to be rearranged. A new governance system changes identity. Someone who once derived security from managerial protection may experience role based autonomy as abandonment. Someone who once enjoyed influence through personal relationships may feel threatened by explicit policies. The implementation challenge is therefore not merely procedural. It is psychological and cultural.

Leaders should make the transition an experiment in collective perception. State what the new system is intended to help the organization see and do. Track not only speed and productivity, but also concentration, belonging, customer contact, decision quality, and the number of important signals that arrive from outside formal channels.

Key Takeaways

• Treat structure as a perception system. For every role, policy, or meeting, ask what it brings into focus and what it makes easy to ignore.
• Separate autonomy from abandonment. Give people authority to act, but also provide clear priorities, boundaries, recovery space, and support for coordination.
• Measure the imprint. Evaluate whether the system develops initiative, judgment, customer sensitivity, and whole system awareness over time.
• Create an external sensing practice. Regularly bring in direct evidence from customers, frontline workers, and people affected by organizational decisions, even when they do not occupy formal roles.
• Make relevance revisable. Allow people to challenge the assumption that only currently recognized roles can provide important information.

The most important organizational question may not be, “Who is in charge?” It may be, “What kind of person does this arrangement make it reasonable to become?”

Every governance system answers that question through repetition. It teaches people whether to wait or initiate, defend a territory or notice a dependency, optimize a metric or investigate a lived experience. It shapes the light that enters the organization, and it shapes what the organization reflects back into the world.

The aim of decentralization, then, should not be to dissolve structure. It should be to create a structure porous enough to receive reality and coherent enough to act on it. The best organization is not the one in which nobody has authority. It is the one in which authority, attention, and responsibility travel together.

That is the real test of self management: not whether people are free to do more, but whether the freedom helps them see more. Because whatever an organization learns to see, it eventually becomes capable of making. And whatever it repeatedly fails to see, it eventually teaches everyone inside it to treat as nonexistent.