WM-13: Construct a preference family with accountable composition

First authored design, 2026-09-15. This elaborates exactly WM-13. It is a reviewable design, not implementation, independent acceptance or machine completion. The existing build loop continues under claude-2; these documents do not dispatch new jobs or override recorded decisions.

All WM cascades · Checklist item · Machine-readable cascade

The problem and existing evidence

WM-13 supplies preferences across grain and time, including their processual/terminal connection and actual FoldC composition. A terminal distribution, a family type or a constructed empty-fold proof does not instantiate that family for real work. This design preserves the wider obligation without inventing mission weights.

Governing checklist snapshot

WM-13 — C as a family across grain and time, connected to real work. OPEN: complete the grain-indexed family and processual/terminal connection at compatible observation domains with declared composition/revision authority. Grain and time indexing are different. PreferenceLadderDraft supplies family/bridge types, not production correspondence. Preserve communicable-witness existence as an outcome facet while excluding preferences over evidence content. FoldC must retain exact layer identities, order, composition operators, support and provenance at the actual consumer; FoldCWitness's empty runtimeFoldedLayers/constructed order example is not the present runtime fold. Mission layer remains undeclared; the general Holes.C declaration remains sorry. Terminal-only packet scope does not close this. Specific WANT-fact/softness proposals are not adopted by this checklist. Built evidence: Terminal seed is built; September 8 Items 14–17 and September 9 Item 19a explicitly require the broader family/derivation and connection.

The preceding evidence is the checklist’s dated account. It has not been re-audited as a current deployment census for this design.

Pattern applications

WM-13-C1 — Recover the concerns at each grain

Pattern: wm-purpose-and-account/recipient-benefit-shapes-the-model.

Identify participants, useful continuation and the costs/constraints relevant to institutional, mission and local decisions. Retain who can author or revise each preference. Evidence can reveal effects and conflicts; it does not by itself grant an agent authority to choose values.

WM-13-C2 — Declare members without conflating indices

Pattern: wm-pattern-design/preference-family-needs-composition.

Give each member a grain, time schedule, outcome domain, source and revision authority. Grain indexing and time indexing are distinct. Mark an undeclared mission member as owed construction; do not turn the terminal-only F13 packet into a global exemption.

WM-13-C3 — Justify processual-to-terminal connections

Pattern: wm-prediction/outcome-meaning-across-models.

Explain how local/processual outcomes relate to eventual work outcomes and how predictions share those domains. A bridge must have its own interpretation and evidence. Keep communicable-witness existence as a legitimate facet while excluding preferences over evidence content. Do not substitute a constant adapter as a discovered connection.

WM-13-C4 — Specify the actual ordered fold

Pattern: wm-pattern-design/preference-family-needs-composition.

Retain every active layer identity, order, support, composition operator, intermediate result and authority. Inspect actual runtime declarations rather than infer an empty fold from a Lean example. Different layers may play different composition roles; do not assume pointwise addition/multiplication or commutativity.

WM-13-C5 — Validate the constructed family

Pattern: wm-prediction/probability-laws-with-runtime-meaning.

Check each admitted member and composed result under its numeric/domain contract, including zeros and undefined combinations. Prove the properties actually used by downstream risk, not only empty-fold identity. Changes to a member require new dependent evidence; no silent repair of incompatible supports.

WM-13-C6 — Trace governance and consumer use

Pattern: wm-connected-learning/observation-to-next-choice.

Pass the actual folded member to WM-06 and retain which choice used it. A later preference revision retains its author, reason, affected grain and old evidence. Do not retrospectively reinterpret a past choice under the new C or claim a family exists because its type is populated by a dummy constant.

Composition and revision

These are design input/constraint relationships; they do not assert automatic runtime firing. Each contribution constrains the next actual artifact.

Build interfaces

Authority/member registry → explicit members and unresolved obligations. Bridge constructor → justified domain connection. Fold interpreter → ordered active layers and intermediate values. Validator → lawful composed member. WM-06 adapter → actual consumer identity. Family governance is retained alongside numerical outputs.

Each envelope must retain its schema, content identity, source/model revisions, task/occurrence and information boundary. Typed missing inputs remain missing; do not replace them with success-shaped defaults. Implementers may reuse existing compatible components, but must identify the real producer and consumer before crediting integration.

Worked situation

The receiver may need both a trustworthy final result and a process that preserves enough uncertainty to continue safely. A terminal “completed” preference cannot alone explain that processual concern. The family must say how the concern is represented and composed, without rewarding a favorable assertion merely because it is favorable.

This situation is illustrative, not a measured production result or an adopted probability table.

Build order and acceptance

1. Resolve the first two applications against the real work and exact authority. Preserve the unresolved decisions below; bell design questions before choosing dependent behavior.
2. Implement the middle producer/checker interfaces with explicit source and numeric contracts. Retain a meaningful failing example before repair and independent review after it.
3. Connect the final consumer and collect the particular evidence described above. Unit tests, source acceptance and ordinary use receive separate credit.

Distinguishing controls: Compare empty-fold identity with the actual nonempty declared runtime layers. Use order-sensitive examples, missing layer authority, incompatible support, revised member pins and forbidden evidence-content preference. Show processual/terminal correspondence and actual folded-member consumption; a standalone terminal test cannot close these controls.

Do not rerun the entire machine merely to validate a local edit. Select checks for the changed invariant; retain commits, commands and results. A future qualifying run still requires every required node and connection.

Dependencies and open decisions

WM-01 · WM-04 · WM-05 · WM-06 · WM-10 · WM-12

Open: mission-member construction, specific WANT-fact and softness proposals, layer composition authority and bridge adequacy. Retain the general discovery obligation and actual runtime/Lean fold discrepancy until reconciled. No new masses, weights or universal downward preference rule are adopted by this draft.

The dependency links express interface constraints, not a topological build order. A producer and consumer can jointly constrain an interface without requiring that all other implementations be finished first. Use a bounded packet once its relevant semantics are settled.

Sources and review boundary

The companion JSON pins the governing checklist, September rulings, current cascade-policy semantics and every applied pattern. Where the semantics source distinguishes a ruling, extrapolation or open proposal, that distinction controls. No generic finite witness is promoted to a machine instance. Existing patterns are reused within the application stated above; their other examples, thresholds and historical implementation claims are not automatically adopted.