colregs 0.3.0 → 0.3.1

package/docs/part-b-invariants.md

@@ -34,9 +34,9 @@ suffix naming what distinguishes them.
 - Insertion appends a suffix and never renumbers; a suffix names something and
   is not an ordinal (`REQ-INV-2`).
 - The paragraph is the unit (ADR 0001): an invariant reading several takes the
-  id of the one stating its operative content, as entry `14b` does.
+  id of the one stating its operative content, as entry `rule:14b` does.
 - `INV-` is a type prefix on a citation-derived name — the one departure from
-  `docs/identifiers.md`, because `13a` is already an entry id. Jurisdiction is a
+  `docs/identifiers.md`, because `rule:13a` is already an entry id. Jurisdiction is a
   dimension (`REQ-SCOPE-2`, `Q-8`); only `intl` is populated.
 
 ---
@@ -82,10 +82,10 @@ not overtaking); its edges are the forward edges of Rule 21(c)'s sternlight
 arc, which 13(b)'s second limb states in light terms.
 
 - **Undetermined term.** "Coming up with" is a speed comparison the model cannot
-  express (`Q-46`); `13b-overtaking` substitutes `pair:geo:tcpa_s > 0`, which
+  express (`Q-46`); `rule:13b` substitutes `pair:geo:tcpa_s > 0`, which
   admits a pair closing because the vessel ahead stopped. This states the rule.
 - **Not stated by 13(b).** Risk of collision: `Q-50`.
-- Entries: `13b-overtaking`, `13b-overtaken`;
+- Entries: `rule:13b`;
   `situation.constants.overtaking_sector_from_deg`/`_to_deg` (ink). Settled by
   P2.2's Alloy sector model: a partition, or a bearing that falls in both.
 
@@ -100,9 +100,11 @@ overtaking vessel bears forward of her beam.
 
 - **Why separate.** A formalisation carrying encounter type per vessel
   classifies the overtaken side as a crossing, two types on one pair.
-- Entries: `13b-overtaking`, `13b-overtaken`, each other's `when` with subjects
-  swapped. Settled by P4.2 carrying the type on the pair and TLC finding no
-  state where the subjects disagree.
+- **2026-09-16.** The two-entry form this section recorded is reversed by
+  ADR 0015 as rewritten: 13(b) is one symmetric entry, the same sector object
+  read on either subject under `any_of`. The invariant is unchanged.
+- Entries: `rule:13b`. Settled by P4.2 carrying the type on the pair and TLC
+  finding no state where the subjects disagree.
 
 ### INV-13a — the overtaking vessel keeps out of the way
 
@@ -116,7 +118,7 @@ roles displace any role Rules 4–18 would otherwise assign to either vessel.
 well as Rules 14 and 15.
 
 - **Not stated by 13(a).** Risk of collision: `Q-50`.
-- Entries: `13a`, overriding all eleven Rule 18 and all three Rule 12 entries.
+- Entries: `rule:13a`, overriding all eleven Rule 18 and all three Rule 12 entries.
   Settled by P4.2's role assignment: the override is or is not needed to keep
   "never both give-way" true.
 
@@ -142,8 +144,8 @@ until own is finally past and clear, the encounter type of the pair is
   model's is its segmentation, and no bearing or range is invented here.
 - **Readings in doubt.** Two: what arms the latch, `Q-51`; what it forbids,
   `Q-52`.
-- Entries: `13d`, reading `own`/`other:hist:was_overtaking` and no geometry; the
-  `was_overtaking: false` gates on `14b`, `15a-crossing`, `15a-give-way`
+- Entries: `rule:13d`, reading `own`/`other:hist:was_overtaking` and no geometry; the
+  `was_overtaking: false` gates on `rule:14b`, `rule:15a:crossing`, `rule:15a:keep_out_of_the_way`
   implement `Q-52`'s *broad* reading. Settled by `Q-51`, `Q-52`, then TLC on a
   three-state trace of an overtaking drawing out onto the bow.
 
@@ -159,7 +161,7 @@ the duty by its second clause, and a formalisation deriving role from encounter
 type alone cannot distinguish `Q-52`'s readings.
 
 - **State remembered.** As `INV-13d`.
-- Entries: `13a`'s `any_of` second limb, `own:hist:was_overtaking: true` — role
+- Entries: `rule:13a`'s `any_of` second limb, `own:hist:was_overtaking: true` — role
   asserted from the latch directly, keeping the limbs separable. Settled by
   `Q-52`.
 
@@ -186,7 +188,7 @@ in [0°, 11.25°] ∪ [348.75°, 360°).
   that way.
 - **Not formalised.** The night and day observation limbs are stated as the
   geometry they encode; a vessel that cannot see the lights has 14(c).
-- Entries: `14b`. Settled by `INV-13b`'s partition sweep plus a decision on the
+- Entries: `rule:14b`. Settled by `INV-13b`'s partition sweep plus a decision on the
   constant.
 
 ### INV-14a — both vessels alter to starboard
@@ -205,8 +207,8 @@ way; both are directed to act, and the duty is symmetric.
 - **Why it matters formally.** The one Section II encounter with no `stand-on`
   vessel, and why `INV-17a1-scope` has content.
 - Entries: none; `conduct`, in `known_omissions`, no conduct shape (`Q-45`);
-  `14a` is retired in `data/deprecated-identifiers.json`. Settled by the first
-  conduct monitor.
+  `14a` was never a name and is gone with the rest of the citation-derived ids
+  (ADR 0015). Settled by the first conduct monitor.
 
 ---
 
@@ -228,7 +230,7 @@ encounter type is `crossing` exactly when it is neither `overtaking` under
 - **Known incompleteness.** Two sailing vessels get no encounter type (Rules 14
   and 15 are gated on power, Rule 12 has no deeming paragraph): the Rules'. A
   pair with the history fact absent gets none: the model's (`Q-43`).
-- Entries: `15a-crossing`.
+- Entries: `rule:15a:crossing`.
 
 ### INV-15a-give-way — the vessel with the other to starboard gives way
 
@@ -241,7 +243,7 @@ other holds `stand-on`.
 - **Undetermined term.** "On her own starboard side" has no sector in the
   paragraph; the data reads `own:geo:rel_bearing_deg` in (0°, 112.5°], its upper
   edge 13(b)'s constant, so it is checkable against the partition.
-- Entries: `15a-give-way`, plus six `rel:overrides` from Rule 18 entries, since
+- Entries: `rule:15a:keep_out_of_the_way`, plus six `rel:overrides` from Rule 18 entries, since
   Rule 18's chapeau excepts Rules 9, 10 and 13 and no others.
 
 ### INV-15a-single — at most one give-way vessel in a crossing
@@ -277,7 +279,7 @@ advancing position.
   segment ("ahead of" a moving vessel), and the state the role attached.
 - **Undetermined term.** "If the circumstances of the case admit" has nowhere to
   live — `modality` is one closed value, an action has no `effect` — `Q-31`.
-- Entries: none; a `gap` on `15a-give-way`.
+- Entries: none; a `gap` on `rule:15a:keep_out_of_the_way`.
 
 ---
 
@@ -443,7 +445,7 @@ discharge her obligation.
 - **Why separate.** The natural automaton takes the give-way duty as an input to
   its transitions; 17(d) says the arrow does not run back, so model two
   obligations holding concurrently, not one machine with one obligation.
-- Entries: none; nearest is 8(f)(ii)'s parallel for `shall-not-impede`, `8f3`.
+- Entries: none; nearest is 8(f)(ii)'s parallel for `shall-not-impede`, `rule:8f_iii`.
 
 ### INV-17-phases — the phase structure
 
@@ -485,9 +487,9 @@ governs.
 - **Why both directions.** "To nothing else" is what runs the `rel:overrides`
   edges from Rule 18 *to* Rules 12 and 15; got wrong once, found by a sweep
   (`Q-40`).
-- Entries: overrides on `13a` against all Rule 18 and Rule 12 entries; on
-  `18a1`–`18a3`, `18c1`–`18c2`, `18f1` against `15a-give-way`; on the Rule 18
-  entries that can meet two sailing vessels against `12a1`–`12a3`; a derived
+- Entries: overrides on `rule:13a` against all Rule 18 and Rule 12 entries; on
+  the 18(a)(i)–(iii) entries, the two 18(c) entries, `rule:18f_i` against `rule:15a:keep_out_of_the_way`; on the Rule 18
+  entries that can meet two sailing vessels against the three Rule 12 entries; a derived
   check asserts the hand-list of six against Rule 15 is complete.
 
 ### INV-18-order — the rank order
@@ -510,7 +512,7 @@ The subject holds `give-way`; the object `stand-on`.
   (`Q-32`) is the model's decode onto these ranks. A vessel constrained by her
   draught is a power-driven vessel for 18(a) and is ranked by it.
 - **Not stated by Rule 18.** Risk of collision and encounter type: `Q-50`.
-- Entries: `18a1`–`18a4`, `18b1`–`18b3`, `18c1`–`18c2`.
+- Entries: the four 18(a) entries, the three 18(b) entries, the two 18(c) entries.
 
 ### INV-18-partial — the order is partial, and the gaps are the Rules'
 
@@ -542,7 +544,7 @@ discharging `INV-16` requires of her.
 
 - **Why stated.** Read as a condition on the antecedent, it gives a roleless
   pair exactly when compliance is hard and the roles matter.
-- Entries: `18c1`, `18c2`, modality `shall-if-practicable` with effect
+- Entries: `rule:18c_i`, `rule:18c_ii`, modality `shall-if-practicable` with effect
   `give-way`/`stand-on` — `Q-31`'s shape, resolved only because the duty is a
   role.
 
@@ -563,8 +565,8 @@ is not relieved (8(f)(ii)).
   protected vessel a Rule 17 duty no paragraph confers — 8(f)(iii)'s content,
   and why `none` is explicit.
 - **Undetermined term.** "Exhibiting the signals in Rule 28" is a
-  display-compliance fact `18d1` does not read; the entry is wider (`Q-34`).
-- Entries: `18d1`, `8f3`.
+  display-compliance fact `rule:18d_i` does not read; the entry is wider (`Q-34`).
+- Entries: `rule:18d_i`, `rule:8f_iii`.
 
 ### INV-18f1 — the WIG craft keeps well clear
 
@@ -575,7 +577,7 @@ in flight near the surface, she holds `keep-clear` with respect to every other
 vessel — keeps well clear and avoids impeding their navigation — and the other
 holds `none`. `keep-clear` is outside the give-way/stand-on pairing and Rule 16.
 
-- Entries: `18f1`.
+- Entries: `rule:18f_i`.
 
 ---
 
@@ -601,9 +603,9 @@ not jointly exhaustive (`INV-19a-third-state`); Rule 19 supplements Section I
   cross it and back inside one encounter, and no paragraph says what becomes of
   a 13(d) latch or a Rule 17 phase: `Q-55`.
 - **Undetermined term.** "In or near an area of restricted visibility" has no
-  fact (3(l)'s atmospheric condition is nowhere in `data/facts.json`), so `19a`
+  fact (3(l)'s atmospheric condition is nowhere in `data/facts.json`), so `rule:19a`
   selects Section III for any pair not in sight — wider — and records a gap.
-- Entries: `11`, `19a`, complementary on `pair:geo:in_sight`.
+- Entries: `rule:11`, `rule:19a`, complementary on `pair:geo:in_sight`.
 
 ### INV-19a-noroles — Section III has no give-way and no stand-on vessel
 
@@ -633,7 +635,7 @@ excludes Section II, 19(a) excludes Rule 19, and only Section I governs.
 - **Why written down.** A hole forced by both scope paragraphs, which "not
   Section II, so Section III" will not have. The model closes it by dropping
   19(a)'s second conjunct — a decision, not a reading: `Q-56`.
-- Entries: `19a`'s `gap`.
+- Entries: `rule:19a`'s `gap`.
 
 ### INV-19b — safe speed, and engines ready
 

package/docs/requirements.md

@@ -79,7 +79,7 @@ Neither consumer lives in this repo.
   exists. That bar is about suppression, not licensing: it is independent of
   REQ-PROV-2 and ADR 0010, and it binds whether a jurisdiction's text ships or
   is withheld. A delta that only *adds* entries is exempt: it suppresses nothing,
-  so the Q-11 hazard cannot arise from it (ADR 0008, `30a-buoy`/`30b-buoy`).
+  so the Q-11 hazard cannot arise from it (ADR 0008, `rule:30a:mooring_buoy`/`rule:30b:mooring_buoy`).
   A delta that suppresses or replaces an `intl` entry still waits.
 - **REQ-SCOPE-4** — Adding a jurisdiction MUST be additive. It MUST NOT require
   a schema change or edits to existing `intl` entries.
@@ -156,7 +156,7 @@ Four layers, each independently addressable.
   identifiers do not.
 - **REQ-MODEL-4** — **Applicability entries**: `when` (predicate over facts) →
   lights or refs → modality → citation → jurisdiction. Every entry MUST have a
-  stable id derived from its paragraph path (`25b`, `25d1`).
+  stable id, a paragraph key in the `rule:` namespace (`rule:25d_i`; ADR 0015).
 - **REQ-MODEL-5** — Gates MUST be expressed as predicates over facts
   (`fact:length_m < 7`), never as pre-enumerated tuples or configuration counts. Any
   count of "configurations" is an output of evaluation, never an input to the
@@ -187,8 +187,8 @@ Four layers, each independently addressable.
   ADR 0005 §4.
 - **REQ-MODEL-12** — `rel:conditional_includes` currently carries three
   distinct shapes under one relation name: a bare `one_of` alternative set
-  (`25d2`), a gated alternative set (`when` + `one_of`, `27f`), and a gated
-  import with its own citation (`when` + `rel:includes` + `cite`, `29a`). Which
+  (`rule:25d_ii`), a gated alternative set (`when` + `one_of`, `rule:27f`), and a gated
+  import with its own citation (`when` + `rel:includes` + `cite`, `rule:29a`). Which
   behaviour applies is inferred from which keys are present. This is a
   **soft** requirement — the data is correct today and the tests cover it,
   so nothing is broken. It is recorded because a third jurisdiction adding
@@ -211,7 +211,7 @@ Four layers, each independently addressable.
     denoted, MUST be marked deprecated in data with the version that
     deprecated it, and MUST NOT be removed in the same major version.
   - **Mutating** one is forbidden. This covers the obvious case (renaming
-    `25b`) and the dangerous quiet one: an identifier keeping its spelling
+    `rule:25b`) and the dangerous quiet one: an identifier keeping its spelling
     while changing what it denotes — a paragraph path repointed at
     different text, a fact value narrowed, a light id reassigned. A
     consumer cannot detect this, and every stored citation becomes silently
@@ -225,29 +225,28 @@ Four layers, each independently addressable.
   removing an identifier is a major version (REQ-PKG-4); repointing one is
   not a version event at all, because it is not permitted.
 
-  **Immutability baseline: `0.1.1`.** The prohibitions above bind every
-  identifier present in the first version released *after* `colregs@0.1.1`,
-  and every identifier introduced from then on. Identifiers as they stood
-  in `0.1.1` and earlier are outside the baseline. The reason, recorded so
-  it is not mistaken for convenience later: `0.1.1` was published
-  2026-08-29, the day the package was seeded, before the identifier review
-  this requirement itself calls for had been done and before any consumer
-  existed. Read without a baseline, the requirement froze the vocabulary at
-  the moment of its first accidental publication and forbade the one review
-  it was written to make possible — including the vocabulary type-prefixing
-  that resolved a live namespace collision (`docs/identifiers.md`). That is
-  a defect in the requirement, not a licence to skip the review.
-
-  The baseline is **set exactly once**. It MUST NOT be moved, raised,
-  re-stated in a later version, or joined by a second baseline clause. This
-  clause is the whole of the exception; there is no mechanism for granting
-  another. Without that, "move the baseline" is a silent escape hatch from
+  **Immutability baseline: `1.0.0`.** The prohibitions above bind every
+  identifier present in `colregs@1.0.0` and every identifier introduced
+  from then on. Identifiers as they stood before `1.0.0` are outside the
+  baseline: while the package is pre-1.0 the schema is being designed, and
+  an identifier may be renamed or discarded without a deprecation record.
+  The baseline was first set at `0.1.1` on 2026-08-30, the day after the
+  package was seeded, so that the identifier review this requirement calls
+  for could happen at all. It was moved to `1.0.0` by Solace's ruling on
+  #121 (ADR 0015, 2026-09-16), which reshaped every entry id into a
+  paragraph key in the `rule:` namespace.
+
+  From `1.0.0` the baseline is **fixed**. It MUST NOT be moved, raised,
+  re-stated in a later version, or joined by a second baseline clause.
+  Without that, "move the baseline" is a silent escape hatch from
   REQ-MODEL-10 and the exception becomes the pattern — a specification that
   can suspend its own prohibition by editing one number is advisory, not
-  normative. One recorded exception is a correction; a second is a policy.
+  normative.
 
   `test/data.test.mjs` pins the baseline literal and asserts it is stated
-  exactly once. A build has no access to git history, so it cannot see the
+  exactly once. There is no identifier diff against the last release: ADR 0006
+  proposed one, ADR 0015 removed it pre-1.0, and it returns with the baseline
+  tag if at all. A build has no access to git history, so it cannot see the
   number being *edited* in place; a test that reconstructed history to
   check would cost more than it is worth and would still pass on a rewritten
   history. What it can refuse is a **second** baseline clause, which is the
@@ -274,14 +273,18 @@ Four layers, each independently addressable.
     names, not a modelling convenience of this package.
   - **`nuc`, `cbd`, `ram`, `ram_underwater`** — kept unspelled as terms of
     art; see `docs/identifiers.md` for the reasoning and the trap in `ram`.
+
+  ADR 0015 (Solace's ruling on #121, 2026-09-16) reopened the first two
+  bullets: an entry id is a paragraph key in the `rule:` namespace now, and
+  the suffix taxonomy went with it. They stand as the audit's record.
 - **REQ-MODEL-11** — Deprecated identifiers MUST be recorded as data — a
   registry naming each retired identifier, what it denoted, the version that
   deprecated it, and its replacement where one exists. Prose in a changelog
   MUST NOT stand in for it: a consumer pinned to an old version needs to
-  resolve a stale identifier mechanically. The registry is
-  `data/deprecated-identifiers.json`, schema-validated
-  (`schema/deprecated-identifiers.schema.json`, ADR 0006); it ships empty
-  until the first identifier is retired.
+  resolve a stale identifier mechanically. **The registry is absent until
+  `1.0.0`.** ADR 0015 deleted it, and the diff that read it: nothing before
+  the baseline is immutable, so nothing before it can be retired, and an
+  empty file with a dormant test is a mechanism that cannot fire.
 
 ### 4.1 Rule categories and the situation record
 
@@ -801,7 +804,7 @@ Tracked here until resolved; each becomes an ADR.
 - **Q-4** — Two upstream SignalK spec asks are outstanding and independent of
   this package: a making-way indicator, and `design.maxSpeed`.
 - **Q-5** — REQ-VERIFY-5 asks for boundary fixtures on every numeric gate.
-  Three gates (`23a2`, `26b-mast`, `30c`'s `fact:length_m` thresholds) live only in
+  Three gates (`rule:23a_ii`, `rule:26b_ii`, `rule:30c`'s `fact:length_m` thresholds) live only in
   `modality_by`, not in the entry's `when` — they flip `shall` to `may`, not
   which entries apply. The fixture format only asserts applying entry ids, not
   expected modality, so there is no way to fixture these three without
@@ -814,8 +817,8 @@ Tracked here until resolved; each becomes an ADR.
   id asserts nothing, so `applicability-fixtures.json` stays byte-identical
   and needs no migration; REQ-CAT-7 states the rule and
   `fixtures/situation-fixtures.json` carries the worked example. What is left
-  is not a decision but the work: writing the boundary fixtures for `23a2`,
-  `26b-mast` and `30c` in the new form. Pencil, so a session that finds a
+  is not a decision but the work: writing the boundary fixtures for `rule:23a_ii`,
+  `rule:26b_ii` and `rule:30c` in the new form. Pencil, so a session that finds a
   better shape may change it, logging the change; settled for good by those
   three fixtures actually landing.
 - **Q-6** — The treaty-language facts behind §5 (en/fr authentic, es/ru
@@ -873,7 +876,7 @@ Tracked here until resolved; each becomes an ADR.
   `24(d)`, `24(f)`, `24(g)`, `24(i)`, `25(d)(i)`, `25(d)(ii)`, `25(e)`,
   `26(d)`, `30(e)`), plus one structural path-mismatch (`23(d)(i)` — the
   Inland equivalent sits at bare `23(d)`, not `23(d)(i)`), one whole-rule
-  clean absence (`28`, "[Reserved]" in Inland), and two clean intl-only
+  clean absence (`rule:28`, "[Reserved]" in Inland), and two clean intl-only
   absences. Secondary claim (amendment history): also refuted — the WIG
   amendment (A.910(22), 2001) explicitly renumbers, displacing the
   pre-existing `23(c)` (small-vessel alternative lights) to `23(d)`; a 1981
@@ -962,6 +965,11 @@ listed here, one line each, because the ADR is what makes them live. Most are
   invariant that does not fit. *(engine)*
 - **Q-17** — What separation distance *d* defines the "safe" level? Fixed to
   one value to start; settled by the sensitivity matrix of Q-22. *(engine)*
+  One data point: `docs/maritime-sources.md`'s MAIB *Polesie*/*Verity* report
+  cites UK case law putting close-quarters at a ~12-minute TCPA and adapts a
+  suggested-TCPA table (its Figure 16) from *A Guide to the Collision
+  Avoidance Rules* — not a d in metres, but a real-world anchor for the
+  sensitivity matrix.
 - **Q-18** — What is the dynamics model, and what is the list of dynamics
   classes (tanker, ferry, yacht, …)? Settled by the first two-vessel
   computation; the class list is a data question once it stabilises.
@@ -983,8 +991,10 @@ listed here, one line each, because the ADR is what makes them live. Most are
   it have to be computed at runtime? Settled by the first two-vessel
   computation's cost. *(engine)*
 - **Q-23** — Is Rule 2(b) a duty where departure is necessary, and does the
-  departing vessel bear the burden? Settled by reading the cases (*The Bywell
-  Castle*, *Boy Andrew v St Rognvald*, both unverified), not by advocacy.
+  departing vessel bear the burden? Partly settled — yes and yes — by
+  `docs/maritime-sources.md`'s *Ever Smart* and *Crowley* entries; the
+  taxonomy of qualifying circumstances, and *The Bywell Castle* and *Boy
+  Andrew v St Rognvald*, are still unread. *colregs-engine* issue #83.
 - **Q-24** — Are the four worked illustrations (R0, R1, R2,
   `inconclusive-in-model`) actually in the regions they are said to be in?
   They fix meanings, not numbers; settled by computing them. *(engine)*
@@ -1120,8 +1130,8 @@ named that gap before the class existed.
   protected-vessel property is a per-vessel fact, consumer-supplied like
   `pair:env:narrow_channel` and `pair:env:traffic_lane` are.** `fact:confined_to_channel`
   and `fact:following_traffic_lane` name it, in the same shape as the existing
-  `fact:near_channel` and `fact:obstruction_exists`; entries `9b`, `10i` and
-  `10j` now read `other:fact:confined_to_channel`/`other:fact:following_traffic_lane`
+  `fact:near_channel` and `fact:obstruction_exists`; entries `rule:9b`, `rule:10i` and
+  `rule:10j` now read `other:fact:confined_to_channel`/`other:fact:following_traffic_lane`
   and their `gap` fields are closed. `9(d)` has no entry yet, so it inherits
   the same fact when one is written. **18(d)(i) is deliberately left open**:
   "exhibiting the signals in Rule 28" is a display-compliance fact rather than
@@ -1129,7 +1139,7 @@ named that gap before the class existed.
   CLAUDE.md "Coverage") — settling it is a narrower, separate decision.
 - **Q-35** — 8(f)(iii)'s antecedent is "a vessel, the passage of which is not to
   be impeded" — the *output* of another norm, not a fact. A precedence
-  predicate reads only facts, so entry `8f3` reads the risk-of-collision half
+  predicate reads only facts, so entry `rule:8f_iii` reads the risk-of-collision half
   alone and applies to every pair with risk of collision. Settled by deciding
   whether a norm may read another norm's effect, which is the same question a
   `conduct` monitor will ask about role and phase.
@@ -1141,22 +1151,22 @@ named that gap before the class existed.
   Settled by reading 8(f)(ii) against the cases, not by picking a role.
 - **Q-37** — 13(a) is `classification` in ADR 0005 §1 and in the proposal's
   first-cut table, but it is the one paragraph of Rule 13 that assigns a role
-  and a `classification` entry has nowhere to put one. Entry `13a` is
+  and a `classification` entry has nowhere to put one. Entry `rule:13a` is
   `precedence` here, with 13(b)'s sector test left for the `classification`
   entry that would set the `hist:was_overtaking` latch. Settled with the rest
   of `Q-14`, paragraph by paragraph.
 
   **Closed in pencil 2026-09-04 (PR #26).** 13(b) is now written, as *two*
-  `classification` entries rather than one — `13b-overtaking` and
-  `13b-overtaken` — because the encounter type belongs to the pair and reading
-  aspect alone classified the overtaken vessel's side of the same encounter as
-  a crossing. 13(a) stays `precedence`, and its predicate now carries both
+  `classification` entries rather than one — one per subject — because the
+  encounter type belongs to the pair and reading aspect alone classified the
+  overtaken vessel's side of the same encounter as a crossing. 13(a) stays `precedence`, and its predicate now carries both
   halves of "any vessel overtaking any other": the 13(b) sector and the 13(d)
   latch, as an `any_of`, which closes the gap the entry recorded. 13(d) is a
   third `classification` entry, reading history and no geometry at all. The
   split confirms the thing ADR 0005 §1 was really defending — one category per
-  paragraph — and costs one paragraph two entries, which the entry-id suffix
-  scheme was already built for.
+  paragraph. **Reversed in part, 2026-09-16:** ADR 0015 as rewritten makes
+  13(b) one symmetric entry, `rule:13b`, reading the same sector object on
+  either subject under `any_of`. 13(a) staying `precedence` is untouched.
 - **Q-38** — 9(d), 18(d)(ii), 18(e), 18(f)(ii), 1(a)–(e) and 20(b)–(c) are
   recorded in `known_omissions` rather than modelled: 9(d) needs the channel's
   axis, 18(e) needs a fact for being a seaplane, 20(b)–(c) need time of day,
@@ -1183,7 +1193,7 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
   or an overtaking; Rule 12 says which of them gives way, not what kind of
   meeting it is. 12(b) is a `definition` and is the cite on `kin:wind_side`
   rather than an entry. The second half of the question is narrower and worse:
-  12(a) says "two sailing vessels", which is 3(c), but entries `12a1`–`12a3`
+  12(a) says "two sailing vessels", which is 3(c), but entries the three Rule 12 entries
   read `rule18_class:sail` instead, so a vessel engaged in fishing under sail
   falls out of Rule 12 — because letting her in would have given her a give-way
   duty under 12(a) and a stand-on role under 18(b)(iii) with no override
@@ -1203,7 +1213,7 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
   it: Rule 18's chapeau excepts Rules 9, 10 and 13 and no others, so every
   Rule 18 norm that can be in force between two sailing vessels — 18(b)(i)–(iii)
   from the ordinary sailing vessel's side, 18(c)(i)–(ii) from the fishing
-  vessel's — carries `rel:overrides` against `12a1`–`12a3`. 13(a), being
+  vessel's — carries `rel:overrides` against the three Rule 12 entries. 13(a), being
   "notwithstanding" the whole of Sections I and II, overrides them too, which
   it had to: two sailing vessels in an overtaking held 13(a)'s and 12(a)(ii)'s
   roles at once and nothing had noticed. Two fixtures and a sweep over sailing
@@ -1217,21 +1227,21 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
 
   **Decided in pencil 2026-09-05 (PR #46): Rule 15 reads 3(b), and Rule 18
   overrides it too.** The same question answered the other way for Rule 15.
-  `15a-give-way` kept the four Rule 18 ranks — NUC, RAM, fishing and WIG —
+  `rule:15a:keep_out_of_the_way` kept the four Rule 18 ranks — NUC, RAM, fishing and WIG —
   out of Rule 15 by negating them on both subjects in its own predicate; the
   negation is gone, both subjects gate on `fact:propulsion: propulsion:power`
-  which is 3(b), and `18a1`–`18a3`, `18c1`–`18c2` and `18f1` carry
+  which is 3(b), and the 18(a)(i)–(iii) entries, the two 18(c) entries and `rule:18f_i` carry
   `rel:overrides` against it. The gate was not merely the relation said in
   the wrong place. It produced the right roles wherever Rule 18 speaks and
   *no* role at all where Rule 18 is silent: two vessels engaged in fishing
   under power, and a vessel not under command crossing one restricted in her
   ability to manoeuvre under power, are pairs Rule 18 does not order, and
   neither vessel took a helm role from any entry. Both are fixtures now, and
-  both failed on the data before the change — `15a-crossing` classified the
+  both failed on the data before the change — `rule:15a:crossing` classified the
   encounter and nothing assigned a role. What it leaves is Rule 12's residue
   again: on those unordered pairs Rule 15 lays a give-way duty on a vessel
   that may be unable to discharge it, which is Rule 2's region and is
-  recorded, not gated. `13a` needs no override against `15a-give-way` — the
+  recorded, not gated. `rule:13a` needs no override against `rule:15a:keep_out_of_the_way` — the
   entry excludes every overtaking by the `hist:was_overtaking` latch and by
   13(b)'s sector — and a test pins that absence so it stays a reason rather
   than an oversight. The hand-list of six is asserted complete by a derived
@@ -1272,7 +1282,7 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
   judgement. Settled by deciding whether the vocabulary gains one — `Q-31`'s
   neighbour, not its duplicate.
 - **Q-43** — **The partition needs history to be present, and says nothing when
-  it is absent.** `14b` and `15a-crossing` are gated on `hist:was_overtaking`
+  it is absent.** `rule:14b` and `rule:15a:crossing` are gated on `hist:was_overtaking`
   being `false` on both subjects, which is 13(d)'s requirement and the only way
   the three encounter types stay disjoint once the latch is set. Because absent
   is absent, a situation that omits the fact is classified as *no encounter at
@@ -1310,15 +1320,14 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
   effect of its own. It produces the only effect the entry has: "such a
   situation shall be deemed to exist when…" is the deeming test, and
   `{"encounter": "head-on"}` is what a deeming test yields — exactly as
-  `13b-overtaking` and `13b-overtaken` cite 13(b), the deeming paragraph,
+  `rule:13b` cites 13(b), the deeming paragraph,
   and not 13(a). 14(a) states the situation in words ("reciprocal or nearly
   reciprocal courses") and imposes the duty ("each shall alter her course to
   starboard"); the duty is `conduct`, and the words are what 14(b) makes
-  checkable. So the entry is `14b`, cites 14(b), and reads nothing 14(b) does
+  checkable. So the entry is `rule:14b`, cites 14(b), and reads nothing 14(b) does
   not fix; 14(a) joins `known_omissions` as conduct beside 18(d)(ii), and
-  `14a` is retired in `retired_entry_ids` and, because v0.1.3 shipped it,
-  recorded in `data/deprecated-identifiers.json` with `14b` as its
-  replacement — the registry's first record (REQ-MODEL-11). The
+  the `14a` id it displaced is gone with the rest of the citation-derived ids
+  and leaves no record behind it (ADR 0015). The
   general question — whether a paragraph with no effect of its own gets an
   entry — is `Q-39`'s and is untouched; this paragraph turned out not to be an
   instance of it.
@@ -1326,7 +1335,7 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
   vessel to be overtaking when *coming up with* another from more than 22.5°
   abaft her beam. That is `own:kin:sog_kn` against `other:kin:sog_kn`, and the
   predicate language compares a fact to a constant and never one fact to
-  another — the same wall `12a2` hits on "both have the wind on the same side",
+  another — the same wall `rule:12a_ii` hits on "both have the wind on the same side",
   which it gets over only because that comparison has two values to enumerate
   and this one has infinitely many. `pair:geo:tcpa_s > 0` stands in: the pair
   is closing, and closing from abaft the beam is coming up. It excludes the
@@ -1338,7 +1347,7 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
 - **Q-47** — **13(d)'s latch never clears.** The paragraph runs "until she is
   finally past and clear", nothing in the fact vocabulary carries that, and so
   in this model `hist:was_overtaking` is set by a consumer and cleared by a
-  consumer while entry `13d` classifies an overtaking for as long as it stands.
+  consumer while entry `rule:13d` classifies an overtaking for as long as it stands.
   Not approximated with a range or a bearing: past and clear is a seamanship
   judgement of the same kind as risk of collision, and a threshold for it would
   be a number invented rather than declared. Settled by whatever settles the
@@ -1349,7 +1358,7 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
   test is suppressed on the other, newly-gaining vessel — once overtaking,
   always overtaking, per 13(d)'s own text, so a geometry test that would
   newly name a second give-way vessel must not fire while the first still
-  holds the role by history alone. `13a`'s sector branch now also reads
+  holds the role by history alone. `rule:13a`'s sector branch now also reads
   `other:hist:was_overtaking: false`; the closing fixture in
   `fixtures/situation-fixtures.json` is the case this closes — a vessel
   drops back onto the latch-holder's own stern, and only the latch-holder
@@ -1361,7 +1370,7 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
   fixture — can state a pair of bearings that no two headings produce, and the
   entries will classify it without complaint. It bites in one visible place:
   where both vessels have the other on the starboard side, which cannot happen
-  on a collision course, `15a-give-way` applies to both of them and the "never
+  on a collision course, `rule:15a:keep_out_of_the_way` applies to both of them and the "never
   both give-way" property fails. The fixtures added here are built from a
   heading and a bearing so that they are consistent by construction, and the
   file says so; the model is not. Settled by deciding whether the record gains
@@ -1384,7 +1393,7 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
   and speeds. Two cases could not keep their story: the R1 head-on had both
   vessels with the other to starboard, which no speeds make a steady bearing,
   and is recast as one; the 13(d) latch case is 400 m abeam with the CPA
-  already past, and lost `7d1`. The partition sweep now constructs positions
+  already past, and lost `rule:7d_i`. The partition sweep now constructs positions
   and headings for every bearing pair and is consistent at every point. The
   "never both give-way" property is asserted over a sweep of steady-bearing
   geometries — every relative bearing, several speed pairs, both intercept
@@ -1401,12 +1410,12 @@ written up in `docs/identifiers.md` §"Effects"; what it could not is here.
   the relative quantities as derived facts stays open in the block's
   `settled_by`.
 
-- **Q-49** — **Must `13d`'s effect cross the Section II/III boundary for Rule 19(d)(i) to be expressible?**
-  - *broad* — drop `13d`'s `pair:geo:in_sight` gate so `encounter: overtaking` is visible in Section III; breaks the scope invariant.
-  - *narrow* — `13d` stays gated; a 19(d)(i) entry reads `own:hist:was_overtaking`/`other:hist:was_overtaking` directly, as `14b` and `15a-crossing` already do.
+- **Q-49** — **Must `rule:13d`'s effect cross the Section II/III boundary for Rule 19(d)(i) to be expressible?**
+  - *broad* — drop `rule:13d`'s `pair:geo:in_sight` gate so `encounter: overtaking` is visible in Section III; breaks the scope invariant.
+  - *narrow* — `rule:13d` stays gated; a 19(d)(i) entry reads `own:hist:was_overtaking`/`other:hist:was_overtaking` directly, as `rule:14b` and `rule:15a:crossing` already do.
   - Default: narrow (nothing changes).
   - Recommendation: narrow — the test named below shows the fact resolves out of sight
-    (`Q-49: hist:was_overtaking resolves out of sight; 13d does not fire`, `test/data.test.mjs`).
+    (`Q-49: hist:was_overtaking resolves out of sight; rule:13d does not fire`, `test/data.test.mjs`).
 
 ### From the Part B invariants (P4.1)
 
@@ -1419,8 +1428,8 @@ nothing is blocked while open.
   *Wide:* they attach whenever in sight and the geometry holds — Rule 11 is the
   only gate Section II states — so a vessel overtaking a mile clear abeam is
   give-way and Rule 16 binds her. *Narrow:* Section II's duties attach only with
-  risk of collision; 14 and 15 are emphasis. Default **wide**: `13a`, `13b-*`,
-  `13d` and every Rule 18 entry read `in_sight` only. No recommendation.
+  risk of collision; 14 and 15 are emphasis. Default **wide**: `rule:13a`, `rule:13b`,
+  `rule:13d` and every Rule 18 entry read `in_sight` only. No recommendation.
 - **Q-51** — **What arms 13(d)'s latch?** *A:* 13(b)'s deeming, at the first
   state the geometry holds. *B:* 13(a)'s duty actually attaching. They differ
   where the geometry holds but a condition on 13(a) does not — `Q-50`'s
@@ -1428,7 +1437,7 @@ nothing is blocked while open.
   duty attaches. KIVELI [2025] EWHC 1185 (Admlty) holds the analogous Rule 14
   latch arms on risk of collision and then persists; no authority was found on
   overtaking geometry with no risk of collision. Free to reverse: four entries
-  (`13a`, `13b-*`, `13d`) and their fixtures, and nothing consumes the one
+  (`rule:13a`, `rule:13b`, `rule:13d`) and their fixtures, and nothing consumes the one
   state this changes the answer for. Settled by: a decision on Rule 13 itself,
   or Cockcroft & Lameijer on whether overtaking status needs risk of
   collision.
@@ -1438,8 +1447,8 @@ nothing is blocked while open.
   other Section II classification attaches. Both preserve the duty; they differ
   on encounter type, which Rule 17's phases and 14(a) hang off.
   Pencilled **broad** ✎ (Solace, 2026-09-14), which is also the data's
-  default: `13d` yields `encounter: overtaking` from history alone, `14b` and
-  `15a-crossing` gate on `was_overtaking: false`. 13(a)'s "notwithstanding"
+  default: `rule:13d` yields `encounter: overtaking` from history alone, `rule:14b` and
+  `rule:15a:crossing` gate on `was_overtaking: false`. 13(a)'s "notwithstanding"
   already displaces Rule 14; eCOLREGs states the broad reading as conventional.
   Settled by: a case or commentary on an overtaking becoming a head-on; none
   found.
@@ -1468,7 +1477,7 @@ nothing is blocked while open.
   Not in sight *and* not in or near restricted visibility (clear weather, beyond
   visual range, radar contact) is outside Rule 11 and 19(a): Section I only.
   *Hole:* represent it; a fact for "in or near restricted visibility" is owed.
-  *Closed:* treat not-in-sight as Section III. Default **closed**: `19a` drops
+  *Closed:* treat not-in-sight as Section III. Default **closed**: `rule:19a` drops
   the second conjunct, recorded as a `gap`. Recommend **hole**: the default is
   safe for a switching consumer and unsafe for a traceability claim.
 - **Q-57** — **The baseline for "keep her course and speed"?** 17(a)(i) fixes