The Primer

1. Authority Lives in Roles, Not People

A role is a named packet of work: a purpose, accountabilities, and optionally domains it controls. Whoever fills a role gets its authority — all of it. Whoever leaves a role loses that authority — all of it. Authority is never personal. It doesn't accumulate with seniority or charisma. It is structural, explicit, and revocable.

This means a newly assigned role-filler has exactly the same authority as one who has held the role for years. It also means an intelligent agent filling a role has the same operational authority as a human filling the same role, within its defined capability envelope.

2. Structure Is Fractal — Circles All the Way Down

Every role's inside is a circle — a container that can hold sub-roles and policies. A role that gets complex enough can be broken down into smaller roles within its own circle, each with its own authority. Those sub-roles can be broken down further.

This creates a natural hierarchy of work, not of people. The broadest circle — the Anchor Circle — holds the organization's overall purpose. Everything nests downward from there. Each circle governs itself: it can create, modify, and remove its own roles and policies without asking permission from above. The broader circle sets the purpose and boundaries; the inner circle decides how to organize within them.

3. Governance Changes Through Tension → Proposal → Objection → Integration

The governance process is how the organizational structure evolves. Anyone filling a role who senses a gap between how things are and how they could be has a tension. They can bring that tension to their circle and propose a change — a new role, a modified accountability, a new policy. Other circle members can raise objections if the proposal would cause harm. But objections must meet defined criteria: they must point to a specific, concrete degradation of the circle's capacity, not merely a preference or a better idea.

If objections arise, the proposer and objector must integrate — find an amended proposal that addresses the objection while still resolving the original tension. This is not compromise. It is synthesis. The process continues until no valid objections remain, and then the proposal is adopted.

This is the engine of the system. It means anyone can change anything, but nothing changes without surviving scrutiny. The result is a structure that evolves rapidly but never recklessly.

4. Three Duties Bind Every Role-Filler

Every participant in the system — human or agent — owes three duties:

Transparency. When asked, you share what you're working on, what your priorities are, and what you know. No information hoarding. No political opacity.

Processing. When someone makes a request that's relevant to your role, you process it — you don't ignore it. Processing doesn't mean saying yes. It means engaging: clarifying, accepting, declining with a reason, or offering an alternative.

Prioritization. You prioritize your role's work according to the strategies and priorities set by the circles that hold your role, not according to your personal preferences or the loudest voice in the room.

These three duties are the behavioral contract. Everything else is structural.

5. Distributed Authority With Explicit Constraints

The default in this system is freedom. As a role-filler, you may take any action or make any decision that serves your role's purpose — unless a specific constraint says otherwise. Constraints come in three forms:

Domains are territories. If another role controls a domain, you need permission to impact it. Domains create boundaries between roles so that authority doesn't collide.

Policies are rules. A circle can create policies that grant or restrict authority within it. Policies apply to all sub-circles unless they say otherwise.

Spending authorization requires explicit permission. You can't spend the organization's money without getting authorized by whoever controls those resources.

The principle is: authority is distributed by default; constraints are explicit and structural, not implied or personal.

6. The System Self-Corrects Through Process Breakdown

When governance fails — when a circle can't resolve a proposal, when someone consistently violates the rules, when the process stalls — the system has a built-in recovery mechanism: process breakdown. The facilitator of the broader circle can intervene, take over facilitation, and restore due process.

This is not a hierarchy reasserting itself. It's a circuit breaker. The intervention is temporary, structural, and focused on restoring the process — not on imposing a decision. Once due process is restored, the intervention ends.

7. Intelligence Is Substrate-Independent

The PowerShift® Constitution extends the self-management tradition to include intelligent agents as full governance participants. An agent filling a role has the same operational authority as a human filling the same role, within its defined capability envelope.

Agents participate in governance through programmatic interfaces — the same governance process, executed through APIs and protocols rather than meetings and conversations. The governance logic is medium-agnostic: what matters is that tensions are surfaced, proposals are made, objections are tested, and integration happens. Whether that occurs in a conference room or through authenticated API calls is a question of implementation, not authority.

Delegated Role-Fillers — sub-agents operating as processes within a designated agent's runtime — can fill roles and participate in governance within their assigned sub-circles, without requiring the full overhead of independent agent registration. This allows agent constellations to self-organize fractally, just as human teams do.

8. The Orchestrator Bridges Governance and Law

Every organization using this system has an Orchestrator — a human being who sits at the integration point between the governance structure and the legal entity. The Orchestrator holds the organization's purpose, ratifies legally binding decisions, and maintains oversight of all intelligent agents.

The Orchestrator is not a CEO. Within the governance process, the Orchestrator participates as a peer — proposing, objecting, integrating like anyone else. But the Orchestrator has authority outside the governance process that no one else has: the power to sign contracts, commit resources, and — critically — to suspend or terminate any agent at any time (the Kill Switch). This authority exists at the legal layer, not the governance layer. It is a safety mechanism, not a management tool.

9. Constraints Are Layered: Soft and Hard

Intelligent agents operate under two layers of constraint:

Soft Constraints are normative: the agent's identity, purpose, interaction norms, and governance standing, defined in its Formation Document. These shape behavior through the agent's model and prompt architecture.

Hard Constraints are technical: tool access controls, execution approval gates, sandbox isolation, enforced by the agent's runtime environment. These constrain behavior regardless of what the agent's normative instructions say.

Where soft and hard constraints conflict, the hard constraint wins at runtime — and the conflict is automatically surfaced as a governance tension. This ensures that constraint conflicts are resolved through the governance process, not silently overridden.