Descriptive-Layer Follow-On Council Roadmap

Context and Problem Statement

ODR-0022 (Session 023) decided to import the descriptive layer by property-category rather than as a 1:1 leaf walk. In doing so it confirmed that several categories carry pattern-level ontological commitments (new classes with identity criteria) that ODR-0022 — a kind: architecture strategy record — deliberately routes rather than discharges. Those commitments need Council treatment under ODR-0001 A9 (a kind: pattern record MUST state a UFO meta-category + an identity criterion over hard cases), and several were already pre-named as spawn triggers in ODR-0008 §Q2a/§Q4a before this session.

These follow-on decisions are currently scattered across three places — ODR-0008 §Q2a(a) (UFO-axis sub-modules), §Q2a(b) (authority-retrieved artefacts), §Q4a (Building/Room conditional promotion), and ODR-0022 §4 (fixtures inclusion-as-Mode). Without a single citable record they will be re-discovered and re-litigated each time the descriptive layer is revisited (the failure mode ODR-0021 was created to stop for the form layer). This ODR is the descriptive-layer analogue: one trigger-indexed roadmap of the follow-on Council sessions, each linked to the source trigger and the records that realise it, so the WG can sequence them and reviewers can see what is scheduled versus trigger-gated. It schedules; it does not itself make the ontological commitments (each row’s session does).

Considered Options

• Option A (chosen) — Maintain a descriptive-layer follow-on Council roadmap. One trigger-indexed roadmap of four trigger-gated pattern-modelling sessions, each routed to its realizing records, format tier, and steward.
• Option B — Leave the four items in ODR-0008 §Q2a/§Q4a + ODR-0022 §4 (no consolidating record). Rejected: they stay scattered across three records and get re-discovered each revisit; the failure mode ODR-0021 was created to stop, here for the descriptive layer.
• Option C — Discharge the commitments inside ODR-0022. Rejected: forbidden by A9 — ODR-0022 is kind: architecture (strategy/granularity); minting opda:RiskAssessment + its IC there would be a pattern commitment in an architecture record (the exact mismatch the pre-flight scope check rejects).
• Option D — Spawn all four ODRs now. Rejected: premature for R2/R3 (their triggers have not fired; running them is guessing) and wasteful (the G walk reshapes R2’s input).
• Option E — Track in ADR-0005 §G. Rejected: wrong corpus (these are ontology-modelling pattern decisions, ODR-side) and §G is the engineering-work register, not a Council roadmap.

Decision Outcome

Chosen option: “Option A — Maintain a descriptive-layer follow-on Council roadmap”, because the commitments are real pattern/IC decisions that A9 forbids burying inside an architecture strategy record, and a single trigger-indexed roadmap prevents both premature work (on gated items) and lost work (on the ready one).

Maintain a descriptive-layer follow-on Council roadmap of four trigger-gated pattern-modelling sessions — R1 Authority-Retrieved Artefacts (ready now), R2 UFO-axis sub-modules (gated on the Category-G walk), R3 Building/Room class promotion (gated on a sub-Property consumer query), R4 fixtures inclusion-as-Mode (a soft confirmation) — each routed to its realizing records, format tier, and steward, and ratified-as-scheduled behind a named trigger — chosen because the commitments are real pattern/IC decisions that A9 forbids burying inside an architecture strategy record, and a single trigger-indexed roadmap prevents both premature work (on gated items) and lost work (on the ready one).

Consequences

• R1 (ODR-0008d) is the live next Council action for the descriptive layer. Convene it (Reduced or Full per a pre-flight scope check) to model the authority-retrieved-artefact family internals + opda:RiskAssessment + the peril scheme, resolving Kendall’s held-as-live four-way. It unblocks the Category-E emission in ODR-0022 and the internal structure of the five already-emitted artefact classes.
• R4 confirms before the Category-D fixtures emission ships, so opda:inclusionStatus is attached to a transaction-scoped fixtures list (ODR-0007), not to opda:Property.
• R2 and R3 remain trigger-gated; this roadmap is the place a reviewer checks before proposing them, and the place that records the trigger when it fires.
• When a session runs, strike its row here and cite the new record (ODR-0008d / ODR-0008a-c / the Building-Room amendment). This ODR shrinks to empty as the roadmap is executed; an empty roadmap is the signal the descriptive-layer pattern work is complete.
• No emission or generator change results from this ODR — it is a scheduling record. Its effect is to make the four follow-on commitments citable, sequenced, and trigger-disciplined.

More Information

• Strategy that surfaced these: ODR-0022 — Descriptive-Layer Import Strategy & Property Categorisation (§1 category routing — B/E/D/G; §4 fixtures inclusion-Mode = R4) and its Council session-023.
• Source triggers: ODR-0008 — Property Descriptive Attributes §Q2a(a) (UFO-axis sub-modules = R2), §Q2a(b) (Authority-Retrieved Artefacts = R1), §Q4a (Building/Room conditional promotion = R3; the five definite class promotions already ratified).
• Realizing records the sessions cite: ODR-0007 (transaction Relator/Mode — R1, R4); ODR-0009 (PROV-O provenance — R1); ODR-0011 (peril scheme + FixtureItemScheme — R1, R4); ODR-0017 (succession/lifecycle rules — R1); ODR-0018 (PII co-annotation — R1, R2); ODR-0005 (Property identity — R3).
• Pattern precedent: ODR-0021 — Deferred Form/Profile-Layer Enhancements (the form-layer trigger-gated register this adapts for the descriptive layer); ODR-0003 — Programme & Work Breakdown (the programme anchor this implements).
• Methodology: ODR-0001 §A9 (per-kind discipline — why each row needs a session), §Format tiers (Reduced/Full/Author-only selection), §Pre-flight scope check (run per row before convening).
• Data: the 12-peril × 2 risk-result structure + the five emitted artefact classes verified against source/00-deliverables/semantic-models/data-dictionary-canonical.json and source/03-standards/ontology/opda-descriptive*.ttl during Session 023.

Rules

The roadmap (four follow-on sessions)

Run status (2026-05-30 — R1–R4 run in parallel)

All four convened in parallel (agent-fan-out, no hive-mind) per the configuration below. R2/R3 ran under a directing-authority override of their gates (provisional, framework/criteria-level; the override discharges convening, not the substantive trigger).

• R1 → RUN. session-024 → ODR-0008d (Q1–Q6 7–1 / 8–0 FOR opda:RiskAssessment as a UFO Information Object + 12-member opda:PerilScheme; the five classes retro-corrected to Information Object; Cagle DA held-as-live for reuse-Search with a named re-open trigger). Row struck.
• R2 → framework ratified, spawn deferred. session-025 — the 3-axis split framework + stewards + per-axis load-bearing trigger ratified; ODR-0008a/b/c not spawned (Davis held: §Q2a(a) trigger unfired; mint nothing). Surfaced a binning-rule defect — enum-bearing Quale-in-Region G-concepts (currentEnergyRating, councilTaxBand, builtForm, tenureKind) mis-routed to Category C; fix logged against ODR-0022 §1 / ADR-0030 §G1. Update 2026-05-31 — trigger FIRED: the gating event (the curated G walk) has landed — 185/188 emitted-or-collapsed, council-reviewed (session-028 → ODR-0024, honest 179/239) — and the S025 binning-rule defect is now fixed (ODR-0024 R5, the structural C-vs-G rule, candidate-G 188→239). R2 is therefore convene-able: it can now judge §Q2a(a) “operationally load-bearing” over a UFO-typed leaf-set; only the a/b/c spawn awaits that judgement. Update 2026-05-31 — ADJUDICATED (session-029, Full panel): the judgement is NOT load-bearing (0–6–0) — the decisive finding is that the walk typed leaves by bearer Kind (rdfs:domain), not by UFO meta-category (no gufo:Quality/Mode typing exists in the corpus), so the partition the trigger presupposed was never built; and on the flagship leaves the UFO key fails OntoClean disjointness (priceQualifier straddles Mode/Quality, ownershipType is quality-by-type but legal-estate-by-bearer — Guarino). ODR-0008 stays monolithic; nothing minted. ODR-0008c is struck from the UFO axis (it reduces to rdfs:domain opda:LegalEstate = Kendall’s S008 attachment-class four-way, re-routed to its original encumbrance-cardinality trigger). a/b deferred on the sharpened conjunctive trigger (committed gUFO typing pass with straddlers adjudicated AND a biting query/validator). The UFO insight is preserved as a deferred gated gUFO rdf:type annotation pass (uncontested leaves only, annotation graph) — which is itself conjunct (i) of the re-open trigger. Framework (S025) affirmed; row updated, not struck.
• R3 → RESOLVED (2026-05-31). session-026 HELD (eventual IC affirmed genuine +O); reconvened session-030 (Full Council, 8 voices) under a directing-authority override of the trigger → neither opda:Room (8–0) nor opda:Building (7–1) is promoted to a class; the roomDimensions.rooms[] repeating group is modelled as an anonymous by-value opda:RoomDimension structure on opda:Property (ODR-0024 R10; reuses the R3 opda:MonetaryAmount pattern). Building’s +O IC is genuine but latent/unexercised. Row struck; emission executed via ADR-0033 (239/239 — curated Category-G walk complete); promotion re-opens only on an identity fact (a Building instance / a stable room token), not a calendar gate.
• R4 → CONFIRMED. session-027 — opda:inclusionStatus is a sale-transaction Mode/Relator (ODR-0007), not a Quality of opda:Property; Category-D emission unblocked. Row struck.

Convening configuration (per council)

Per ODR-0001 §Format tiers + §Consensus-mode framework. All four default to agent-fan-out — the runtime is Agent Teams (TeamCreate + parallel Agent spawns + SendMessage cross-talk), the pattern S021/S022/S023 used. Hive-mind is avoided by default: it is materially more expensive (persistent hive state + programmatic consensus rounds), and none of these councils needs it — per-question votes are independent tallies, and the two-artefact structured vote table is composed by the Queen from the teammates’ working-file votes, with no consensus engine. The swarm init flags are advisory bookkeeping (the CLI’s maxAgents init is also currently bug-prone); the deliberation actually runs on Agent Teams.

R1 panel (the ready one): Queen Kendall (owns the four-way) or Guizzardi (IC); DA the genuinely-opposed voice — Allemang (“keep them flat datatype bags; don’t over-promote”) or Cagle (“reference data, not classes”); panel Guizzardi+Guarino (IC), Moreau (PROV-O / ODR-0009 provenance), Knublauch (SHACL node-shape over a lean class), Cagle (taxonomy / peril scheme), Baker+Isaac (the 12-member peril/dataset SKOS scheme), Pandit (search-result PII regime). ~8 voices / ~6 teammates — the S023 shape.

Hive-mind escalation (avoid unless a trigger fires). Escalate a council from agent-fan-out to a hive-mind consensus-mode ONLY if, at the pre-flight scope check, its verdict is genuinely conditional (one question’s verdict contingent on another → hive-mind/byzantine) or must emit a structured object a downstream tool consumes (hive-mind/typed-output). For R1’s four-way this is not expected — a four-way alternative is resolved by deliberation + per-question vote, exactly as S023’s held dissents were under agent-fan-out. If a scope check does escalate, record the fired trigger in that session’s convening block. CLI only if ever needed: hive-mind init -c byzantine (or -c quorum --quorum-preset supermajority); the default agent-fan-out path needs no hive-mind init at all.

Findings carried from Session 023 (the evidence each session inherits)

• R1 — the structure is regular and already half-built. The six-field result block (riskIndicator, actionAlertRating, result, summary, recommendations, datasetAttribution) recurs across exactly 24 parents = 12 environmental perils × 2 (the peril + its riskSubcategories[]): flooding, coalMining, nonCoalMining, radon, groundStability, contaminatedLand, coastalErosion, climate, energy, infrastructure, planning, transportation — plus localSearches (185 annotated leaves, CON29-style) and valuationComparisonData (23). The five §Q4a classes are already emitted as bare classes with identity-key SHACL shapes (opda-descriptive.ttl / opda-descriptive-shapes.ttl); what R1 adds is their internal property structure, the new opda:RiskAssessment class (the per-peril result — does not exist today), and the 12-member peril SKOS scheme. So R1 is not a new class from nothing; it is the internals + one new class + one scheme. The session must resolve Kendall’s held-as-live four-way alternative and the IC questions: RiskAssessment as a distinct class vs a structured datatype on Search; peril as a SKOS scheme vs subclasses; environmental-search vs local-authority-search as one class or two; the riskSubcategories[] recursion.
• R2 — gated, not deferred-forever. The G walk (ODR-0022, ~181 curated descriptive properties) is the event that makes the Quality/Mode/Substance-Kind-label axes countable; only then can §Q2a(a) judge whether the sub-module split is “operationally load-bearing.” Running R2 before the G walk would be guessing.
• R3 — narrow and consumer-driven. A single named BASPI5 round-trip query that needs sub-Property (room/building) reasoning is the whole trigger; absent it, Building/Room stay as descriptive datatype properties on opda:Property (Davis dissent, §Q4a).
• R4 — small but real. ODR-0022 §4 asserts inclusion is a transaction Mode/Relator (the same boiler is “included” in one sale, absent the next) — an ontological claim that touches ODR-0007. It needs one confirmation pass (or an ODR-0007 amendment) before the FixtureItemScheme + opda:inclusionStatus are emitted, so the inclusion property is not mis-attached to the brick-and-mortar.

Sequencing and discipline

• Run order: R1 first (trigger fired; unblocks the Category-E emission and the search/survey/EPC/valuation families). R4 next (small; unblocks the Category-D emission). R2 and R3 stay gated — do not convene until their named trigger fires (R2: the G walk; R3: a sub-Property consumer query).
• Each session discharges its own A9 obligations (UFO meta-category + IC over named hard cases) per ODR-0001; this roadmap records that they are owed and when, not the commitments themselves.
• Anti-duplication. This record is the ODR-side roadmap for descriptive-layer pattern/Council work only. It does not replace ADR-0005 §G (the cross-corpus deferred-engineering-work register) or the adoption record’s Pending: list (the running session log); it consolidates the modelling-decision spawn-triggers that ODR-0008 §Q2a/§Q4a and ODR-0022 §4 created, so they are citable in one place. When a row’s session runs, its row is struck here and the new ODR (ODR-0008d, etc.) is the live record.
• Anti-pattern. Do not convene a gated session (R2, R3) without recording which trigger fired (mirrors ODR-0021’s discipline). Do not fold R1’s pattern commitments back into ODR-0022 or ODR-0008 (§Q2a keeps ODR-0008 monolithic and spawns on this trigger).