Possibility Is Not Agreement

1. Two states that look alike

The RSM Canonical Model defines both terms carefully. A Configuration is "a composition of participants, capabilities, capacity, resources, relationships, evidence requirements, and dependencies that could satisfy an Intent." A candidate configuration is "a hypothetical instance," derived by reasoning. A Formation is "an authorised, established arrangement with its own identity and biography; established only by an authorised act once commitments/conditions exist." The model states the rule in its shortest form: Configuration ≠ Formation.

The definitions differ in two ways that matter. A configuration could satisfy an intent, while a Formation is an arrangement. A configuration is derived, while a Formation is established. Derivation is something a reasoning system does by itself, over information it has been allowed to see. Establishment requires the parties. The Canonical Model calls the boundary between them a firebreak: the RSM owns the types, and any specific candidate configuration, gap or pathway an application computes "is a derived instance carrying provenance" and "MUST NOT be mistaken for canonical fact."

The firebreak exists because the confusion is so tempting. A reasoning system that has done good work produces a configuration that looks finished, with every role filled and every dependency mapped. To the person reading it, and to downstream software, it is easy to treat it as the plan, then as the arrangement, and then as something people have agreed to. Each step seems small. Taken together, they amount to claiming agreement that nobody gave.

2. What viability can and cannot say

Formation reasoning evaluates configurations, and the evaluation is meant to be thorough: whether capability and capacity are covered, whether constraints hold, whether timing, geography, infrastructure and economics are feasible, how strong the evidence is. A person deciding whether to pursue an arrangement needs to know where it is strong and where it is fragile, and a good evaluation tells them.

Every one of those findings, however, is a finding about possibility. A configuration can look strong on every dimension and still be nobody's commitment, because none of the dimensions measures whether anyone has agreed. Evaluation, recommendation and commitment are separate gates, and passing one tells you nothing about the next. Wellzai's architecture is explicit that its own reasoning stops short of the last gate. It may recommend or request a commitment, "but the actual Commitment must originate from a Participant or an agent operating under explicit delegated authority."

Part of the reason is practical. Viability is computed from what the system knows, and what it knows about willingness is usually the weakest part. A processor found during discovery, the architecture notes, "contributes theoretical Capability; a processor that has reserved Capacity under defined conditions contributes substantially stronger evidence of Formation readiness." What changes between those two states is the processor's own action, not the model's arithmetic.

3. Commitment is what crosses the line

If viability does not carry an arrangement into reality, commitments do. The RSM's protocol specification states the sequence directly: a configuration acquires commitments, the required dependencies become satisfied, the Formation is established, and activities begin. Formation, it says, occurs "when a Configuration has acquired sufficient real Relationships and Commitments to constitute an actual organized arrangement around a purpose."

A Commitment in the Canonical Model is "an authorised participant undertaking," either personal, made by a person for themselves, or representational, made for an organization by an authorized representative. Its canonical lifecycle runs from proposed to conditional to committed, and then to fulfilled, withdrawn or expired, with breached and released as further states. The protocol specification requires that implementations "preserve the semantic distinction between exploratory interest, conditional intent, actual Commitment, and fulfillment."

Conditions are where the discipline is tested. Most commitments in a forming arrangement are conditional. A producer may commit acreage on the condition that transition financing is approved and a buyer's price floor is accepted. The protocol is explicit that these conditions must survive: "A conditional Commitment cannot be silently promoted into an unconditional Commitment simply because an optimization algorithm assumes its conditions will eventually be satisfied." This is the configuration-versus-formation confusion at small scale. An optimizer that treats a likely condition as met has, in effect, committed on the participant's behalf.

Commitments also change, and the changes matter for history. Withdrawal and expiry are states in the canonical lifecycle, not errors to be cleaned up. A commitment that was withdrawn is part of the record of how an arrangement came close to forming and did not, and a system that kept only the current state would lose it. Who commits matters as much as what is committed. Whether a collective's undertaking binds each of its members is a question the Seednote Is a Cooperative's Commitment Its Members'? takes up. How commitments depend on one another, and which one moves the rest, is the subject of The Commitment That Unlocks the Others.

4. Readiness is a judgement; establishment is an act

Even when commitments have accumulated, a final distinction remains, and the specifications are careful with it. Wellzai may evaluate whether an arrangement is ready to form. It does so against explicit criteria: a minimum buyer commitment, sufficient committed producer capacity, a reserved processing capacity, available financing, a confirmed logistics pathway, satisfiable evidence requirements. The criteria "need not be universal" and vary by intent and domain. The reference specification is plain about what the result means: a ready state "is a Wellzai determination. It is not itself canonical Formation establishment."

Establishment is a separate step. "Formation creation must require an authorized protocol operation," the reference specification says, and its acceptance harness "must verify that no canonical Formation exists before the authorized transition." Only after that act does the arrangement receive canonical identity and begin "accumulating its own Biography," so that later activities and outcomes belong to an organized system rather than to the candidate configuration it came from.

The highlighted principle, from the RSM protocol specification, contains both a requirement and a gap. The requirement is that whoever declares a Formation, on whatever criteria, must be identifiable afterwards. What remains open is who that is. The sources specify that establishment needs authority but do not say who holds it in an arrangement with many independent parties: the parties jointly, one of them by delegation, or some body they agree on. The Seednote Who Declares a Formation? stays with that question. The essay's argument does not need it answered, only that it never be answered by default.

5. The system that proposes must not commit

The line between possibility and reality would mean little if the reasoning system could cross it whenever it was confident. Wellzai's architecture therefore sets the boundary in its structure. Its central invariant is short: "Wellzai may propose what should come together; Participants determine what they will commit." The runtime reasoning process "does not possess inherent authority to make Commitments on behalf of Participants." It may recommend or request a commitment, but the commitment itself must come from a participant or from an agent with explicit delegated authority.

The reference specification turns this into a test. Wellzai may discover, evaluate, configure, identify gaps, develop pathways, recommend, request, coordinate and learn, and "it may not silently convert any of these operations into institutional agreement." The acceptance harness "must prove that Wellzai cannot independently create a Participant Commitment without delegated authority." In the RSM, then, the boundary is a property to be demonstrated, not just a policy to be stated.

The same boundary should shape how a recommendation is presented. What the reasoning concluded, what action it proposes and whose authority that action needs are three different things, and a person reading a recommendation should be able to see all three, and so see where the system's work ends and their own decision begins. A recommendation phrased as an instruction, or a proposed arrangement displayed as an existing one, hides that boundary from the people who most need to see it. The wider version of this argument, about model output, delegated action and human commitment, is developed in Models Propose, Participants Commit.

6. Why the line protects what comes after

It would be possible to read all this as caution, a set of brakes on a system that could otherwise move faster. The stronger reading is that the line is what makes the rest of the system worth trusting and worth learning from.

It makes the system worth trusting because participants can engage with a configuration without fear that engaging will be taken as consent. A producer can look at a proposed arrangement that includes their acreage, and a lender can see a pathway that assumes their financing, knowing that neither has been committed by being shown. Without that assurance, the safest response to any proposal is to refuse to look at it.

It makes learning possible because the record keeps what was believed apart from what was agreed and from what happened. The Canonical Model records an Activity "only on report," keeps planned activity distinct from actual activity, and evaluates an Outcome "independently of the Intent." A Formation's biography is governed memory of what actually occurred within it. A system that let configurations drift into formations would lose exactly the comparison that learning depends on: the difference between what it thought could come together and what actually did.

Most of what this essay describes belongs to the specified formation layer rather than to anything Wellzai runs today. Commitment reasoning, readiness evaluation and the establishment step are designed and specified, with acceptance tests defined, but they are not yet built; the Formation Lab on this site illustrates the sequence, and the Formation Network it assembles is a proposal on an illustrative scenario, with no real network or agreement behind it. The distinction itself does not need to wait for the software. Any coordination effort, digital or not, would benefit from keeping clear which of its arrangements are possibilities and which are agreements, and from treating the move from one to the other as something only the parties can make.

Sources

• RSM Canonical Model v3.0, §7 RSM Core (Commitment, Formation, Activity, Outcome, Biography); §8 RSM Formation, the General Model of Becoming.
• RSM Canonical Model and Protocol Specification v1.0, §11 Commitment and Formation.
• Wellzai Outcome Formation Reasoning Architecture Specification v1.0, §2.3 (the authority of the reasoning agent); §13 Commitment Reasoning; §14 Formation Recognition.
• RSM Reference Implementation and Acceptance Specification v1.0, §13 Commitment and Formation Implementation; §24.3 Human and Institutional Agency.