Why Regenerative Systems Need a Shared Model

1. The coordination question needs a common grammar

The RSM Concept Paper begins from an observation that recurs throughout Wellzai's work: regenerative food, agriculture, health and community systems rarely fail because a necessary participant is absent. More often the ingredients for a desired outcome already exist but remain fragmented across people, organizations, communities, places, living systems, infrastructure, knowledge, capital, evidence and operational systems. Why that fragmentation makes regeneration a formation problem rather than a marketplace problem is the subject of Why Regeneration Is a Formation Problem. The point here is narrower and prior to it. Before anyone can ask what configuration of people, organizations, places, capabilities, resources, relationships, evidence and commitments must come together to make this outcome possible, the systems holding those fragments must agree on what the words in that question mean.

That agreement has to cover more than most software attempts. The fragments the Concept Paper lists are of very different kinds: people and the organizations and communities they form, places and the living systems within them, infrastructure, knowledge, capital, evidence, and the operational systems that record what happens. A model adequate to coordination has to represent all of them, together with the relationships, commitments and history that connect them, without flattening them into one generic notion of a participant or a record.

The canonical model compresses this into a single governing statement: a regenerative system is a persistent network of sovereign Selves, organizations, communities, places and living systems, with their capabilities, capacities, expressions, relationships, evidence and history, within which outcomes form around purpose. Every clause of that sentence is a commitment about representation. Sovereign Selves means a person can take part without an institutional wrapper. Living systems means a soil or a watershed is something the model can observe and reason about, not merely a location field. History means the model must remember trajectories, not just current state. And outcomes form around purpose means the model must be able to describe arrangements that do not yet exist, because the coordination question is about what could come together, not only about what already has.

2. When the primitives outgrow the application

There was a moment, recorded in the RSM documents themselves, when this became unavoidable. The concepts were first written inside Wellzai's own application. The current version of the canon describes the change plainly: the thesis is unchanged, and what changed is ownership. Self, Person, Organization, Community, Place, Living System, Capability, Capacity, Relationship, Evidence, Intent, Commitment, Formation, Outcome and Biography had become foundational, and in the Concept Paper's words they "stopped belonging to any one application. They are the primitives of regenerative systems."

The reasoning is a dependency argument. If those primitives were owned by one application, every other system would face two bad choices: depend on that application merely to understand basic semantics, or redefine the concepts and create drift. Both choices are familiar. The first produces a platform that everyone must route through to know what a commitment is. The second produces many systems that each use the word commitment and mean slightly different things, so that an arrangement which looks agreed in one system is still exploratory in another. RSM inverts the dependency instead. It defines meaning, and applications consume that meaning without redefining it. An application owns only the concepts specific to its own job.

That division of labour, stated in the RSM Canonical Model & Protocol, is what lets the foundation stay stable while the reasoning above it keeps changing. The protocol deliberately does not define how Wellzai ranks candidates, searches configurations, analyses gaps or sequences dependencies; those functions consume and produce RSM objects but sit outside the foundation. The same document draws the consequence in one sentence: this creates the foundation for Wellzai without making Wellzai itself the foundation. Wellzai is free to improve how it discovers, configures and coordinates, precisely because it is not allowed to change what a Capability or a Commitment is.

3. Two planes: what persists and what forms

The most important structural decision in RSM is that a regenerative system is described as two intersecting planes rather than one.

The first is the persistent network: what exists between outcomes. It holds people (as Persons representing Selves), organizations, communities, places, living systems, facilities and resources; their capabilities and capacity; their offers, seeks and demand expressions; their relationships, affiliations and memberships; their claims, evidence and assertions; and their accumulated biography. These persist independently of any single outcome. One capability may contribute to many outcomes, and one relationship may gain meaning over years. This is also what makes it possible for a participant not to start from zero each time: what they have described, demonstrated and done remains, under governance and consent, so that a future intent begins in a richer and truer network.

The second is Outcome Formation: what forms around purpose. An Intent opens an Outcome Context, a bounded space for reasoning about one possibility. Within it, discovery, configuration, viability, gaps, dependencies, pathways and recommendations assemble and evaluate what could work. Commitments accumulate. A Formation, an established arrangement with its own identity and biography, comes into being only by an authorized act; the line between a computed possibility and that act is the subject of Possibility Is Not Agreement. Execution itself happens in specialized systems that already exist for procurement, logistics, laboratory work, settlement and certification. RSM consumes the resulting activities, observations, evidence and outcomes and writes them back as biography.

The separation is not a diagramming convenience. The decisive phrase in the Concept Paper is that an Outcome Context references, but does not own, the persistent objects it reasons over. A farmer's capability does not become the property of the first opportunity that discovers it. A relationship that took several seasons to build is not dissolved when one formation ends. If outcomes owned their participants, every new purpose would have to rebuild the network from scratch, and every completed one would take part of the network with it. Keeping the planes apart is what allows the network to persist while outcomes come and go within it.

4. Four seams, and a firebreak

Two planes are only useful if it is clear how they meet. RSM names four seams. Reference: an Outcome Context points to persistent objects instead of copying them. Discovery query: reasoning asks the network what exists and what could contribute. History write-back: formations and their observed outcomes enrich the biographies of the participants, relationships and places involved. Learning read: future discovery reads those biographies as priors. The canonical model adds the rule that holds the whole arrangement together: reasoning operates between canonical objects; it does not redefine them, and independently governed network state remains independently governed.

That rule is enforced through a small classification that every concept in RSM carries. Some things are canonical: facts that participants and their systems assert, observe, commit or do, with identity, provenance and authority. Some are derived: conclusions computed about canonical state, which must carry their inputs and the version of the reasoning that produced them and are never canonical fact. Some are projections: governed, authorization-scoped views such as a map, a profile or a biography, which are never an authority. And some are operational: the runtime bookkeeping of a particular reasoning session.

The canonical model calls the consequence a firebreak. RSM owns the types of formation: what a Configuration, a Coordination Gap, a Dependency and a Pathway are. When Wellzai computes a specific candidate configuration or a particular dependency graph, that instance is derived. It carries its provenance and must not be mistaken for a fact about the world. A configuration Wellzai proposes is a hypothesis about what could work; it cannot quietly become a record that certain parties are working together. Why that class of distinction has to be designed into the model rather than left to good intentions is the argument of Distinctions That Must Not Collapse.

The seams are also a way of being honest about what has been built. The first reference kernel of RSM, whose first external consumer was Wellzai, implements part of the persistent plane and deliberately defers most of the formation stack. Its plan records a principle worth repeating: defer capability, not semantics. Biography projections and learning-read priors could wait; the history required to construct biography could not, so it was mandatory from the first kernel. A seam that is designed in from the beginning can be completed later without redefining what flowed across it; the cost of getting that order wrong is described in the Fieldnote The First Plan Had Semantic Debt.

5. Reuse what exists; own only what nothing else can say

A shared model invites an obvious objection: the world already has vocabularies, and a new one risks becoming one more dialect. RSM's design rule is to reuse before inventing. Where an established vocabulary faithfully describes a person, an organization, a place, an observation, a unit of measure, a time interval or the provenance of a record, RSM uses it and does not shadow it. Its semantic architecture distinguishes five modes of reuse: adopting external semantics directly, profiling an external pattern with RSM constraints, mapping an RSM concept to a corresponding external one, referencing external identifiers as classification values, and owning a concept only where no faithful external term exists. Mappings are directional, so exporting to a standard does not imply that the standard is sufficient for internal reasoning, and an external mapping may never silently redefine one of RSM's distinctions.

What is striking is where the existing vocabularies stop. They describe what exists rather well. They have very little to say about what could form. In RSM's own mapping of its concepts to external standards, the concepts it must own outright are concentrated in exactly that territory: Intent, Configuration, Coordination Gap, Dependency, Pathway, the Outcome Context and Outcome itself. For several of them the mapping records that no external term is sufficient. That is not an accident of coverage. Most information systems were built to record states and transactions, so their vocabularies are rich in nouns for things that already are and poor in words for arrangements that are still becoming. A model for regenerative coordination has to supply that missing grammar of becoming, and it should supply nothing else that the world already provides.

The same discipline governs classification. RSM avoids a single universal taxonomy tree. Roles such as producer, processor, buyer, capital provider, evaluator or steward are contextual roles described in governed classification schemes, not fixed classes of actor, and the same entity can carry several independent classifications at once. Domain packs register the vocabularies and mappings appropriate to agriculture and food, nature and ecology, health, or capital and finance, and they extend the taxonomy without changing RSM's core semantics. The protocol makes the limit explicit: a domain pack cannot redefine a production practice as an outcome merely because the practice is considered desirable, and a market extension cannot redefine a procurement requirement as ecological evidence. Specialization adds words; it does not change what the shared words mean.

This is what would allow Wellzai to reason across domains. A capability in a soil laboratory, a capability in a processing facility and a capability in a lending institution are different in content, but they are the same kind of thing, distinct in the same way from the capacity behind them and the offer that expresses them. Because the kinds are shared, a formation that needs all three can be reasoned about as one problem.

6. What a shared model makes possible

The case for RSM is ultimately a case about what becomes possible when meaning is stable and ownership is not concentrated. Participants can keep their own systems of record and still become mutually intelligible. Their capabilities can contribute to many outcomes without being absorbed by any of them. An intent can open a space for reasoning that draws on the whole network and gives its history back when it closes. And the intelligence that reasons across all of this can keep improving without destabilizing the foundation every other system depends on.

The protocol summarizes the result as neither a marketplace nor a centralized data platform, but a federated foundation in which independently governed systems can describe capability and demand, ask questions of one another, evaluate evidence without indiscriminate disclosure, configure possibilities, establish commitments, observe consequences and accumulate shared learning. How that works without anyone surrendering their records is taken up in Answer the Question, Keep the Evidence; why the foundation begins with the person rather than the organization is the subject of Affiliation Enables Representation, Not Participation.

None of this makes coordination easy. It makes coordination expressible, which is the precondition for everything else. The Concept Paper closes with a sentence that serves as the shortest description of the whole model: Self is foundational. Association is voluntary. Authority is explicit. The network persists. Outcomes form within it. Intelligence connects the two.

Sources

• RSM Concept Paper v3.0, §1 The Problem RSM Addresses; §2 Why a Shared Model, Not a Single Application; §5 The Two Planes; §7 Formation Before Transaction; Persistence Before Repetition.
• RSM Canonical Model v3.0, §1 Governing Statement; §3.1 The four categories; §8 RSM Formation; §9 The Plane Seams.
• RSM Semantic Model and Taxonomy Architecture v3.0, §1 Approach; §3 Reuse Policy; §5 Mapping Matrix; §6 Taxonomy Framework and Domain Packs.
• RSM Canonical Model and Protocol Specification v1.0, §1.2 Normative Boundary; §18.2 Semantic Compatibility; §24.2 Architectural Consequence.