Collective Agency

Collective agency is the increased capacity to choose and act that becomes possible when distinct participants coordinate well. Its ethical form is Coagency: coordinated agency without surrendered agency.

The group does not become one mind. Each participant keeps perspective, judgment, and responsibility while learning to read the same living field. Difference is not friction to erase. It is how the group detects blind spots, shares capability, and corrects itself.

Coagency is a reciprocal learning loop

Coagency begins with shared awareness: what each participant perceives, which perspective they hold, what matters from that position, and what reality now permits. That awareness makes useful prompting possible.

The Zone of Proximal Development (ZPD) is the next valuable action someone cannot yet take alone but can take with support. A More Knowledgeable Other (MKO) is the person, agent, example, tool, or resource with the missing knowledge for that move.

MKO is a relationship for one context, not a permanent rank. The role moves. One participant may know the purpose, beneficiary, or lived reality. Another may know the technique, system, or precedent. Both become more knowledgeable when consequences return.

awareness: perception + perspective + context
  -> value: what matters, to whom, and what must be protected
  -> intent: the consequence we mean to create
  -> ZPD: the next reachable capability edge
  -> MKO: the participant with the missing knowledge for this move
  -> scaffold: the least support needed for effective action
  -> bounded action
  -> observed consequence
  -> reciprocal feedback
  -> updated intent, belief, capability, and next action

Prompting before awareness is guesswork. The same words can help, overwhelm, or misdirect depending on the learner's present model, condition, values, and available attention. The MKO first listens and reflects their picture back. Then the learner can correct it before action.

Two questions close every cycle

Feedback must test both direction and execution.

Was the intent right?

• Did we understand what mattered from the beneficiary's perspective?
• Did the intended consequence serve the right person or system?
• Did we protect what should not be traded away?
• Does the intention remain meaningful in the context reality revealed?

An efficient action serving the wrong intent is efficient misdirection.

Was the action as effective as it could be?

• Was the shared model of reality accurate enough?
• Did the prompt fit the actor's ZPD?
• Did the MKO supply enough support without taking over?
• Did the chosen lever create the expected consequence within its bounds?
• Can equal or greater value now be created with less support, cost, or risk?

A good intention without effective action leaves value unrealised.

Five conditions

Condition	What it requires	Failure signal
Shared intent	Participants can name the beneficiary and valued consequence in their own words.	Activity continues after purpose is no longer shared.
Skilled autonomy	Each participant has the capability and permission their position requires.	Avoidable permission replaces judgment.
Mutual awareness	Participants read one another, the field, uncertainty, and changing context.	Prompts are issued from an untested picture of reality.
Live feedback	Consequences return quickly enough to correct intent and action.	Output is celebrated while effects remain unknown.
Returned power	Authority and value return to the people creating and receiving the outcome.	The helper or institution becomes harder to leave or query.

Rugby makes the relationship visible

In flowing rugby, one player sees space, another runs the line, and a third passes before the opening is obvious. Separate choices share one direction. The ball and the consequence provide fast feedback. The player best placed to read the next move becomes the contextual MKO; the role shifts with the phase.

The try belongs to everyone, but nobody disappears. Each decision still matters. That is agency strengthened through relationship.

Games, sport, music, simulations, and other practices make useful training grounds because action and consequence sit close together. Participants can learn to perceive, prompt, act, and correct before the stakes become existential.

The coagency test

Before designing or defending a team, human–AI relationship, institution, or workflow, ask:

1. Purpose: Can participants name the shared outcome in their own words?
2. Perspective: Can each participant correct how the others perceive their reality and values?
3. ZPD: Is the next challenge reachable with support rather than trivial or overwhelming?
4. MKO: Does the knowledgeable role follow the missing knowledge instead of fixed status?
5. Scaffold: Is support contingent, proportionate, and designed to fade?
6. Feedback: Can consequences travel back and change both intent and action?
7. Return: Do authority, capability, and value return to the participants?

If agency falls, feedback cannot correct power, or the helper needs dependence to preserve its role, stop calling the relationship coagency.

Proof

The mechanism is present when awareness, ZPD calibration, contextual MKO selection, bounded action, and returned feedback can be observed. That does not yet prove the relationship improved anyone.

Outcome proof requires a participant to act more effectively with less support in a later comparable situation while the beneficiary appreciates the consequence and protected constraints remain intact. The strongest transfer proof is success in a second context without the original MKO.

Context

• depends-on Agency — supplies the capacity to choose and act that coordination must protect.
• depends-on ZPD and MKO — calibrates the capability edge and knowledgeable relationship.
• applies-to Shared Wisdom Loop — transfers knowledge through contingent support, fading, and returned evidence.
• pairs-with Flow — describes aligned attention and action but does not by itself prove increased agency.
• proved-by Performance — compares intended and observed consequences.
• pairs-with Rugby — makes contextual roles, mutual awareness, and live feedback visible.

Questions

Which participant should become more capable, what can they do independently now, and what is the smallest support that would help them cross the next valuable edge without surrendering judgment?