Coordination Decision Rights and Autonomy Matrix¶

A decision-rights artifact — instantiates Distributed Coordination Architecture

Maps, for each class of coordinated decision, who may commit, decide, execute, veto, stop, and review — drawing the line between legitimate local autonomy and choices that require joint control.

Coordination Decision Rights and Autonomy Matrix is the artifact that answers "who gets to decide this?" for every class of decision that has collective consequence. For each decision class it records the distinct authorities — who may commit the group, who decides, who executes, who can veto, who can stop, who reviews, who remedies — and it maps actual control rather than titles. Its defining job is to set the width of the autonomy envelope: local discretion should be wide where actions are reversible and their side effects bounded, and it should narrow toward joint or escalated control as coupling, irreversibility, scarcity, and risk rise. Done right, it kills both shadow hierarchy (someone quietly deciding things they have no legitimate claim to) and fictitious decentralization (autonomy on paper that evaporates the moment a choice matters).

Example¶

Two automakers run a jointly owned battery-cell plant: two parent organizations, one facility, endless ambiguity over who may decide what. The trigger is a near-miss — an engineer from one parent approved a cell-chemistry substitution that the other parent's warranty team had never cleared, and it nearly shipped. They build a Coordination Decision Rights and Autonomy Matrix. The rows are decision classes: cell-chemistry change, capital spend over a threshold, a safety stop, hiring above a grade, disclosing joint IP. The columns are authority types: commit, decide, execute, veto, must-be-consulted, must-be-notified.

The filled matrix reads, for example: a cell-chemistry change is a joint decision, and either parent's safety lead holds a unilateral stop; tooling spend under ≈$50k is local to the plant general manager, notify-only. The effect is two-sided. A supplier substitution can no longer happen without the warranty veto being consulted — the coupling that caused the near-miss is now gated — while the plant GM stops queuing for permission on the small, reversible things that were never anyone else's business.

How it works¶

Rows are decision classes, not tasks. It sorts decisions by kind and consequence, so the same person can be autonomous on one class and bound on another.
Columns are authority types richer than a RACI. Commit, decide, execute, veto, stop, review, remedy — the semantics of coordination authority, not just "responsible/accountable."
It records actual control, not nominal titles. A matrix of who really can stop a line beats an org chart of who is nominally in charge.
Envelope width scales with risk. Wide local discretion for reversible, bounded actions; joint or escalated control as coupling, irreversibility, and scarcity rise — with explicit emergency and good-faith safety exceptions.
It is tested against edge cases. The hard, ambiguous decisions are walked through before they happen, not litigated in the moment.

Tuning parameters¶

Envelope width — how much reversible action stays purely local. Wider buys speed and morale; narrower buys control over couplings, and the two trade directly.
Decision-class granularity — a few broad classes versus many fine ones. Finer is more precise but heavier to maintain and easier to game.
Authority-type richness — a light responsible/accountable scheme versus the full commit/veto/stop/remedy set. More columns capture real coordination authority but cost clarity.
Escalation default — what happens to a decision no class covers: default to local, default to joint, or default to escalate.
Emergency-exception scope — how broad the safety-stop override is, and who holds it, since an over-broad override quietly re-centralizes everything.

When it helps, and when it misleads¶

Its strength is that it lets people act locally without asking permission while protecting the specific couplings where a local choice would damage the whole — the exact balance the archetype is after. It is the antidote to both the shadow decider and the org whose "decentralization" is a fiction.

Its failure modes are a static matrix that ossifies as the coupling structure changes underneath it, and — the weak version the archetype warns against — a generic responsibility chart with no real decision semantics, which is too thin to resolve anything. It maps titles instead of actual control, or it is drafted to legitimize an existing power grab ("the matrix says I decide"). The discipline that guards against this is to test the matrix against real edge cases and against who actually holds control, and to revise it as coupling, scale, and risk change rather than freezing the first draft.^[1]

How it implements the components¶

actor_autonomy_and_authority_map — it records actual control: the authority to commit, decide, execute, veto, stop, review, and remedy, and where autonomy is legitimate versus where a local choice creates collective consequence.
decision_rights_and_local_autonomy_envelope — it separates local, consultative, joint, approval, notification, and escalation decisions by class, sizing the envelope to reversibility and risk and including the emergency safety exceptions.

It does not chart the interdependencies that tell you where joint control is warranted — that is the Dependency and Interaction Map; it does not run the exception path when rights are exceeded or conflict — that is the Exception and Escalation Protocol; and it does not bind the shared outcome and roles into a signed agreement — that is the Joint Operating Agreement.

Instantiates: Distributed Coordination Architecture — the authority layer that says who may decide each class of coordinated choice.
Consumes: Dependency and Interaction Map — the coupling it reveals is what justifies tightening control from local toward joint.
Sibling mechanisms: Dependency and Interaction Map · Exception and Escalation Protocol · After-Action Coordination Review · Commitment and Dependency Register · Coordination Health Review · Distributed Planning and Reconciliation Session · Event-Driven Coordination Channel · Interface Control Document or Service Contract · Joint Operating Agreement · Liaison and Integrator Role · Shared Coordination Board · Synchronization Checkpoint

Notes¶

The matrix must record actual authority, not nominal. A grid of job titles is worse than nothing, because it launders the fiction that whoever holds the title holds the control — the near-miss usually comes from the gap between the two.

References¶

[1] A RACI chart (Responsible, Accountable, Consulted, Informed) is the familiar responsibility-assignment matrix. This mechanism deliberately goes beyond it: coordination needs the harder authority distinctions RACI omits — who may commit the group, who holds a veto or a stop — and it fails in exactly the way the archetype notes when it is reduced to a bare RACI with no decision semantics. ↩