Loomworks — C5 scoping note: the domain reference model — v0.1

Document: loomworks-c5-scoping-note-domain-reference-model-v0_1 Version: 0.1 Date: 2026-06-05 Role: Scoping (produces a scoping note — not a CR, not build artifacts) Operator: Marvin Percival — DUNIN7 (Done In Seven LLC, Miami, Florida) Canonical home: loomworks-record (technical scoping document — Markdown primary; HTML companion optional)

Plain-language summary

This note scopes C5 — the domain layer: how an engagement reaches the bodies of knowledge (domains) it draws on, and how those domains are composed for a reader.

What is already settled and not reopened here: seed v0.10 committed the two-relationships model (membership is a chain; reference is a fan-out of peers) and the two read-access modes (open by default; restricted via OVA). C5 specifies how references compose; it does not redefine what they are.

What this note adds beyond the prior Discovery input: a worked-through contribution-authority model. The session found that the seed names only read access. It is silent on write access — who may put knowledge into a shared domain. The resolution (Contributor proposes, Domain Steward elevates, elevation is the derivation trigger) is grounded in the existing contribution-trust-graph reasoning in methodology v0.20, but it is new authority structure the seed does not carry. That part requires a seed v0.11 amendment before C5 can specify under it. This note surfaces it and flags the gate; it does not settle it.

Decision needed from the Operator: (1) confirm the seam-vs-policy boundary for contribution authority (scope the roles and the gate; leave per-domain promotion policy open); (2) decide whether the contribution-authority model proceeds as a seed v0.11 amendment now or is held; (3) rule on the named open questions or carry them.

In scope: reference typing and composition; the non-transitive rule; same-type conflict resolution; dedup-by-identity; multi-parent membership; the Companion-surfaces-domains flow; the upper-scope-Manifestation-trigger question; the contribution-authority finding (flagged as seed-level).

Out of scope: OVA enforcement itself (parked — Kaspa vProgs not ready); per-domain promotion policy mechanics (domain governance, "adopts whatever fits"); the build (this is scoping, not a CR).

1. Orientation — what C5 inherits

1.1 Settled in seed v0.10 (canonical, at origin `dc9c784`) — not reopened

C5 builds on four commitments the seed already carries. C5 may specify how they operate; it may not redefine them.

Two relationships, not one. An engagement relates to scopes above it in two distinct ways. Membership — who the engagement belongs to (its organization, and what that belongs to) — is normally a single chain. Reference — the bodies of knowledge it draws on (domains) — is normally many, peers to one another, none above the other. Conflating a referenced domain with a membership parent is named in the seed as a category error.

Composition lives in Shaping. v0.10 deliberately declined to push composition up into Manifestation. A Manifestation is a reading of one scope's Memory — never a merge. Shaping is the room where knowledge from more than one scope first comes together for a specific reader.

A Manifestation is single-scope. It organizes one body of knowledge so it reads coherently. It does not compose across scopes.

Two read-access modes. Open ("all") by default — reachable by any engagement, any Operator, because a commons compounds when open. Restricted by exception — an ACL established through OVA, after which OVA governs reach. Both are first-class; an open scope is open by design, not merely un-restricted. While OVA is not enforcing, every scope is effectively open, but the seam is declared.

1.2 The C5 input documents

Discovery note loomworks-discovery-note-domain-reference-model-v0_1 (2026-06-05) — four findings (two relationships; typed/non-transitive references; two access modes; Companion-surfaces-domains), the denim/cotton worked example, and an explicit "Left to C5 scoping" list.
Methodology what-dunin7-is-building-v0_20 — §Contribution-trust (the contribution trust graph) and §Domains/Roles. This is the prior reasoning the contribution-authority finding (§3 below) builds on. It lives in the engine repo, not loomworks-record.

1.3 Manifest status (orientation only)

Current-status manifest is genuinely v0.48 (header/Status line say 0.48; footer and changelog are stale at v0.47 — a known internal inconsistency, not a version error). v0.48 is dated 2026-05-31 and predates the seed v0.10 landing, so it carries no C5 or domain-layer entry by name. It is authoritative here only for the carried-forward residues it names: cross-Operator stewardship of promoted fragments (§4.A); Phase C "Memory stewardship, multi-backing, multi-Operator" (§4.D); OVA parked. A manifest v0.49 absorption pass is overdue and out of scope for this note.

2. The C5 specification items

These are the seven items the Discovery note left to C5. Each is specification-level — how the settled seed commitments operate. For each: the concern, a proposed shape, and what the CR must decide. Proposed shapes are scoping proposals, not settled decisions.

2.1 Reference typing and per-type composition

Concern. References are not uniform. The Discovery note found at least three flavors, and Shaping must compose each differently.

Proposed shape. A reference edge carries a type, and the composition rule attaches to the type:

Additive / anecdotal (e.g. denim heritage) — freely composable color. Pull if relevant; never constrains. Composes by inclusion.
Hierarchical floor (e.g. jurisdiction rules) — must apply; cannot be overridden by engagement convenience. Composes by enforcement, and floors of the same type can conflict (see 2.3).
Contextual (e.g. regional climate) — weights and shapes output without hard-constraining. Composes by influence.

The key claim: an untyped reference graph is the expensive, intractable case; a typed reference set is tractable because each edge declares how it composes. C5's composition logic dispatches on edge type.

CR must decide. The closed set of reference types (are three enough, or is the set open/extensible?); the formal composition operation per type; where the type is declared (on the edge, at reach-time).

2.2 The non-transitive rule and opt-in depth

Concern. The diamond hazard. Two routes to cotton (Brand → Denim → Cotton, and Brand → Cotton directly) create double-pull, provenance ambiguity, and conflict-resolution ambiguity.

Proposed shape. Reference edges are non-transitive. Default reach depth is 1. A Shaping composes the engagement's own declared references; it does not automatically walk references-of-references.

Cotton wanted directly (brand's own sourcing claims) → declare Brand → Cotton explicitly. Attributed to a direct brand decision.
Cotton matters only because denim's heritage invokes it → do not declare it; it arrives inside denim's material, as denim's assertion, attributed to denim.
Both → one direct edge plus cotton-inside-denim; dedup-by-identity (2.4) composes any shared assertion once, both reasons preserved.

The opt-in exception: a Shaping may explicitly choose to also reach a referenced scope's references ("also reach denim's references") — an Operator/Shaping decision, never an automatic transitive walk. Opt-in preserves operator-authority over what gets reached.

Note the diagram correction the Discovery note records: Denim → Cotton is not membership (denim does not belong to cotton); it is denim's own reference to cotton. Once membership and reference are separated, the diamond is well-defined.

CR must decide. How opt-in depth is expressed and recorded; whether depth >1 is bounded; how provenance is stamped when an assertion arrives via a referenced scope's own reference.

2.3 Same-type conflict resolution

Concern. Two floors of the same type can conflict — a two-jurisdiction clinical trial references two regulatory flooring scopes that may impose contradictory requirements.

Proposed shape. Resolution by specificity-or-strictness: the more specific or the stricter floor governs where same-type floors conflict. This needs formalizing — "stricter" must be definable per floor type, and the tie-break when neither is strictly more specific must be named (candidate: surface the conflict to the Operator rather than auto-resolve, consistent with "surface and signal; the Operator approves").

CR must decide. The precedence rule's exact form; whether unresolvable conflicts halt-and-surface or apply a default; how the resolution is recorded as provenance.

2.4 Dedup-by-identity at composition

Concern. A shared assertion reachable by two paths must compose once, not twice.

Proposed shape. Composition dedups by assertion identity, not by content similarity. The same assertion reached via two edges is composed once; the multiple reasons it was reached (e.g. brand-level reference + denim-level content) are both preserved in provenance. This is what keeps the diamond safe alongside the non-transitive rule.

CR must decide. What constitutes assertion identity at composition time; how multiple reach-reasons are recorded against a single composed assertion.

2.5 Multi-parent membership (shared-subject exception)

Concern. Membership is normally a single chain, but the shared-subject pattern (an engagement jointly operated by two organizations) breaks pure single-chain membership. It is rare, real, and must be expressible.

Proposed shape. Membership is a chain in the common case but must support multi-parent for the shared-subject exception. The model must not assume single-parent membership in its storage or its composition logic, even though single-parent is the overwhelming default.

CR must decide. How multi-parent membership is represented without imposing graph costs on the single-parent common case; how a jointly-operated engagement's membership composes (does it inherit both organizations' brand/voice, and how are those conflicts resolved — likely reusing 2.3). Operator: confirm whether multi-parent membership is in C5's first CR or deferred to a later increment. (Default proposal: keep the seam — don't foreclose it — but defer full mechanics if it complicates the first CR.)

2.6 The Companion-surfaces-domains flow

Concern. The fan-out model creates a curation burden: someone must know which domains exist and build the engagement's reach set. That someone must not be the Operator studying a scope taxonomy — the seed's voice principle forbids requiring the Operator to understand technical structure.

Proposed shape. As the Operator discusses the engagement with the Companion, the Companion notices the subject touches certain domains and surfaces them for consideration ("there's a denim domain and a cotton domain — draw on either?"). The reach set is built from the Operator's intent, in plain conversation. This needs no new seed grant — the manifest records Companion-as-Operator-system-interface as settled (Phase 60).

Three disciplines the CR must honor:

Surface and signal; the Operator approves. The Companion proposes a reach; it must not establish one. Auto-wiring an inferred reach is an automatic state transition on an artifact the Operator has authority over — the category error the seed names. Companion suggests → Operator confirms → reach edge created with the Operator recorded as actor. (This is also what keeps every reference edge traceable to an Operator decision — the property that made the diamond safe in 2.2.)
The suggestion is not the reach. Surfacing "denim seems relevant" is a proposal, not composition. It does not pull denim's content. Composition still happens later, in Shaping, from declared edges. Surfacing must not collapse the Memory → Manifestation → Shaping ordering.
Honest about access. The Companion surfaces open scopes freely and restricted scopes only to Operators whose OVA check passes — never surfacing a restricted scope as available to an Operator who cannot reach it ("only show what is available"). Permissive until OVA enforces; the seam is there now.

CR must decide. How the Companion's domain-surfacing is triggered and bounded; how a confirmed reach is recorded with the Operator as actor; the matchmaker affordance ("denim domain — already used by N engagements") as a discovery/reuse mechanism (strong product idea, downstream of getting the reach model right).

2.7 Upper-scope-Manifestation trigger (was open; largely resolved this session)

Concern (as carried from Discovery). A domain scope may have no single Operator to trigger its derivation. If a domain's Manifestation is never derived, Shaping has nothing to compose from that scope — and the seed says Shaping draws from the Manifestations of referenced scopes. So this is load-bearing for whether the reference model functions at all, not a side question.

Resolution reached this session. The trigger is the Steward's elevation act (see §3). A domain is founded when an Operator elevates an artifact into it; from then a Domain Steward holds it. Every elevation changes the domain's Memory and warrants re-derivation. Because elevation and derivation-trigger are the same authority, a domain always has someone whose act triggers its derivation — the Steward, by definition of the role. The "no single Operator to trigger derivation" problem dissolves.

What remains open. Freshness — whether re-derivation is eager (on each elevation), lazy (on first Shaping that reaches the domain), or scheduled. This is a tractable specification choice for the CR; it is no longer an unresolved architectural gap. (See §4 open questions.)

CR must decide. The freshness policy (eager / lazy / scheduled); whether a domain with elevated content but a stale Manifestation surfaces that staleness to a composing Shaping.

3. Finding: contribution authority is a second access axis (requires seed v0.11)

This is the substantive new material from this session. It is flagged separately because it exceeds the current seed seam and lands as seed-level, not C5-specification-level.

3.1 The gap

The seed's two access modes (open / restricted via OVA — §1.1.4) govern who may reach a scope — read access. The seed is silent on who may contribute to a scope — write access. For an engagement, write access is implicit: the engagement's Operator contributes to their own engagement. For a shared domain, write access is the live question: a domain born from Brand-A is referenced by Brand-B, who has denim knowledge to add. May Brand-B write into denim? If anyone may, the commons degrades to whatever the loudest contributor writes — the wiki failure mode. If no one may, the commons cannot grow past its founder.

So a scope has two independent authority axes:

Reach (read) — open-by-default / restricted-via-OVA. Settled in seed v0.10.
Contribution (write) — gated by elevation. Not in the seed. This finding.

These are independent settings. The denim commons wants open-reach, gated-contribution: anyone composes from it, only authorized knowledge enters it. A restricted client domain might want gated-reach and gated-contribution. The seed names only the first axis.

3.2 The resolution — Contributor proposes, Steward elevates

Grounded in the existing contribution trust graph (methodology v0.20 §Contribution-trust), which this session confirmed already settles the shape. The roles:

Contributor to a domain — may propose knowledge into it. The proposal is attributed and enters a review queue. It does not yet compose — it is not part of the domain's composable Memory until elevated. This is the safe form of open contribution: anyone may propose; proposing is not entering.
Domain Steward — may elevate a proposed contribution from the queue into the domain's composable Memory. The Steward's elevation is the authorizing write-gate. Nothing enters domain Memory as composable knowledge without a Steward's act.

This is curated commons, not wiki: open to reach, gated on contribution by a human elevation act.

The Contributor → Steward progression. The roles are not separate castes. A Contributor can become a Domain Steward. Conferral is a human grant, trust-graph-advised, never automatic. The methodology is explicit on every relevant point (confirmed verbatim from v0.20 this session):

No numeric "contribution-value score" is defined or should be; Loom §8 leaves derived authority measures uncomputed. Standing is qualitative — has this person demonstrated they understand the domain well enough to contribute without supervision — not a tallied score. (The "scoring" recalled from a prior discussion was the trust graph; it was never a points tally. Checked rather than reconstructed.)
A new contributor's writes go to a review queue until a trusted contributor promotes or rejects them — the reviewer's act is the gate.
"Auto-elevation is the failure mode"; promotion "passes through the Operator's judgment." Contribution value informs who a human decides to trust or elevate; it never is the decision.
Governance of the trust graph is domain-level, not environment-infrastructure-level: "adopts whatever fits."

3.3 The seam vs. the policy

This boundary is the central design judgment, and it keeps faith with the methodology's deliberate openness:

Seam (environment-level — seed/C5 defines): that a domain has a contribution gate, a review queue, a Steward role that elevates, that elevation is the derivation trigger, and that contribution (write) is an axis distinct from reach (read).
Policy (domain-level — left open by design): how a specific domain decides to promote a Contributor to Steward, whether it adopts any quantitative rule, who reviews. "Adopts whatever fits." C5 must not harden this — hardening it would violate the prior reasoning's deliberate openness.

3.4 Why this is seed v0.11, not C5

The seed v0.10 commits read-side access modes only. The Contributor/Steward roles, the queue-and-elevate write-gate, and contribution-as-a-distinct-axis are authority structure the seed does not carry. Even though the trust graph is already in the methodology, the seed does not name it — and the seed is what C5 verifies against. So:

The contribution-authority model lands as seed v0.11 (adds the write-access axis and the Contributor/Steward roles to the seed's access-mode commitments).
C5 then specifies the composition and gate mechanics under the amended seed.

This is the standard halt-and-flag pattern: scoping surfaces it, names where it exceeds the current seam, the Operator decides whether it proceeds as a seed amendment now or is held.

Discovery-record note — the path not taken. Open contribution with provenance and no gate (anyone writes; bad contributions corrected, not blocked; leaning on "Memory does not forget" and "a correction is a contribution") was considered and set aside. Reason: it is the wiki failure mode — the commons degrades to whatever the loudest contributor writes, and Shaping composes from unvetted domain knowledge until someone corrects it. The Operator rejected it explicitly. The curated-commons path (Steward elevates) retains a path to valid artifacts while keeping the gate human.

4. Named open questions (carried, not resolved here)

Cross-Operator domain-steward proposer. Who may propose that a Contributor become a Steward across organizations, in a domain referenced by competitors? Filed as genuinely open in the prior reasoning (knowledge-elevation §5/§6.4; memory-space §5.5/§9 Q2/Q5) and carried in manifest §4.A ("cross-Operator stewardship of promoted fragments"). In every proposed resolution the conferral remains a human grant the Companion/AI advises. Carried.

Per-domain promotion policy. How a new Contributor crosses into the trusted core is per-domain governance ("adopts whatever fits"). A specific domain could adopt a quantitative rule; the environment defines no automatic value→trust conferral. Explicitly left to domain governance — out of C5 scope by design.

Upper-scope Manifestation freshness. Eager / lazy / scheduled re-derivation (see 2.7). Tractable specification choice for the CR; no longer an architectural gap. For CR decision.

Multi-parent membership timing (see 2.5). Seam now or full mechanics deferred. For Operator decision.

5. Out of scope

OVA enforcement itself — parked; Kaspa vProgs not ready. The reach axis is seam-declared, not enforced. C5 declares the seam; it does not build OVA.
Per-domain promotion policy mechanics — domain governance, "adopts whatever fits" (§4.2).
The build — this is a scoping note. The CR follows, after the seed v0.11 decision (§3.4).
Manifest v0.49 absorption — overdue but separate work (§1.3).

6. What this note recommends — decision summary

Confirm the seam-vs-policy boundary (§3.3): C5 scopes the contribution roles and the gate; per-domain promotion policy stays open. (This note's recommendation: confirm.)
Decide the seed v0.11 path (§3.4): does the contribution-authority model proceed as a seed v0.11 amendment now, or hold? C5's first CR depends on this — C5 specifies under the amended seed.
Rule on the open questions (§4) or carry them: cross-Operator proposer (carry); freshness (CR-time); multi-parent timing (now/defer).
Sequence: seed v0.11 amendment (if approved) → C5 CR drafting (specifies the seven items under the amended seed) → execution.

DUNIN7 — Done In Seven LLC — Miami, Florida Loomworks — C5 scoping note: the domain reference model — v0.1 — 2026-06-05