diff --git a/.claude/board/LATEST_STATE.md b/.claude/board/LATEST_STATE.md
index 6c04f3329..856029e00 100644
--- a/.claude/board/LATEST_STATE.md
+++ b/.claude/board/LATEST_STATE.md
@@ -1,4 +1,8 @@
-## 2026-10-16 — D-CGW-1: CE64 bits 43..45 are the orientation (in PR)
+## 2026-10-17 — CE64 one layout; Moore tenant as synapse, membrane and local NARS (board, #1595 in PR)
+
+- Entry `2026-10-17-ce64-one-layout-moore-synapse-membrane.md`: eleven decisions with their basis (CE64 = 64-bit ISA contract in three regions, one layout aligned by migration, Moore reuses it minus what locality answers, the witness is two neighbours agreeing on a membrane, Hebbian wiring = SPOFC evidence with local NARS), the CE64 reader catalog, open items. No code.
+
+## 2026-10-16 — D-CGW-1: CE64 bits 43..45 are the orientation (#1594, merged `8f50cd9e`)
 
 `causal_edge::Orientation` (Core + the seven Haspelmath dative functions) is the one meaning of 43..45; `direction()` → `orientation()`, `pack`/`pack_v2` take it. The S/P/O pathological triad and `detect_simpsons_paradox` are removed (no production writer). Contract `MooreDirection` replaces `DirectionReading`; `lift_ce64` refuses an oriented edge. #1593 merged (`124eb6cf`). Entry `2026-10-16-d-cgw-1-orientation-contract.md`.
 
diff --git a/.claude/board/PR_ARC_INVENTORY.md b/.claude/board/PR_ARC_INVENTORY.md
index b2feb2ee2..d40a60197 100644
--- a/.claude/board/PR_ARC_INVENTORY.md
+++ b/.claude/board/PR_ARC_INVENTORY.md
@@ -1,3 +1,10 @@
+## 2026-10-17 — lance-graph PR #1594 (merged `8f50cd9e`, branch `ccr-9f3b8075-74u47l-cgw1`, head `0ee072cb`, 10 commits) — D-CGW-1: CE64 bits 43..45 are the orientation
+
+- **Added:** `causal_edge::Orientation` (Core + seven Haspelmath dative functions) as the one meaning of 43..45; `orientation()` / `set_orientation()`; `pack` / `pack_v2` take an `Orientation`; contract `MooreDirection { slot, polarity }` and `MooreRefusal::Oriented` (`lift_ce64` refuses an oriented edge and writes nothing); `causal-edge/tests/orientation_contract.rs`; ~85 call sites migrated across 8 crates; entry `2026-10-16-d-cgw-1-orientation-contract.md`; arc entry for #1593.
+- **Locked:** removed the S/P/O pathological triad, `concern_level` and `CausalNetwork::detect_simpsons_paradox` (no production writer of the triad).
+- **Deferred / open:** D-CGW-9 (activation off 46..49), the 50..52 ruling, two edge-facet writers on the 12+4 reading, WordNet G1–G3.
+- **Confidence:** orientation_contract 5 tests (red first); moore_tenant h/h2 with a disable run; causal-edge, contract, deepnsm-v2, planner, shader-driver, osint, p64-bridge, thinking-engine tests pass; KJV `bible_wave` exits 0. `intervene_counterfactual` not run locally (disk), left to CI.
+
 ## 2026-10-16 — lance-graph PR #1593 (merged `124eb6cf`, branch `ccr-9f3b8075-74u47l-synergy-1591`, head `526a0bc5`) — synergy map since #1579; config note reword; D-ARG-1..4 queued
 
 - **Added:** entry `2026-10-16-synergy-map-since-1579.md` (§1–§9: revision of every base point, CE64 vs identity/edge tenants and the 24×i4 loci, triangle join as presence-plane intersection); `.cargo/config.toml` note on the avx and v2 arms; STATUS rows D-CGW-13/14 and D-ARG-1..4.
diff --git a/.claude/board/STATUS_BOARD.md b/.claude/board/STATUS_BOARD.md
index 1077ff18d..6df20770c 100644
--- a/.claude/board/STATUS_BOARD.md
+++ b/.claude/board/STATUS_BOARD.md
@@ -5,7 +5,7 @@ Plan: `.claude/plans/ce64-grammar-wiring-v1.md`. Entry: `entries/2026-10-16-gram
 | D-id | scope | status | gate / falsifier |
 |---|---|---|---|
 | **D-CGW-0** | `deepnsm-v2::wire`: CE64 0..63 from named sources; Gadamer gate before counting; W = Markov arc in `WitnessTable<64>`; activation = update-caused frequency change; 43..45 = Haspelmath function; Antecedent / TEKAMOLO / Quorum / Contradiction loci; qualia register; `wired_reasoning_probe` | In PR | 229 tests; probe 741 rows, 5 rules, 0 violations; disable runs red per entry §7–§11 |
-| **D-CGW-1** | one global meaning per CE64 field, held in the `causal-edge` contract (no per-tenant readings; superseded the same day); amend 43..45 to orientation with all triad readers migrated | In PR (branch `ccr-9f3b8075-74u47l-cgw1`): `causal_edge::Orientation` on 43..45, `pack`/`pack_v2` take it, the S/P/O triad removed (no production writer), Moore `lift_ce64` refuses an oriented edge; entry `2026-10-16-d-cgw-1-orientation-contract.md`. 46..49 (D-CGW-9) and 50..52 still open | one code table shared by accessor, writer and readers; field isolation 43..49. **Gate:** wire edges stay in the probe until this lands |
+| **D-CGW-1** | one global meaning per CE64 field, held in the `causal-edge` contract (no per-tenant readings; superseded the same day); amend 43..45 to orientation with all triad readers migrated | Shipped (#1594, `8f50cd9e`): `causal_edge::Orientation` on 43..45, `pack`/`pack_v2` take it, the S/P/O triad removed (no production writer), Moore `lift_ce64` refuses an oriented edge; entry `2026-10-16-d-cgw-1-orientation-contract.md`. 46..49 (D-CGW-9) and 50..52 still open | one code table shared by accessor, writer and readers; field isolation 43..49. **Gate:** wire edges stay in the probe until this lands |
 | **D-CGW-2** | frequency-ranked default reading (COCA, DeReKo) + antecedent-tested promotion of the alternative (relative pronoun) with agreement | Queued | `der Mann, der …` promoted; `der Mann gab der Frau …` silent; English `that` promoted without a COCA Rel tag |
 | **D-CGW-3** | case morphology decides argument order in the parser (`Den Hund biss der Mann`) | Queued | OVS flips, SVO unchanged, syncretic case unresolved; UD German GSD dev measured |
 | **D-CGW-4** | counterfactual lanes: `CausalCandidate` → `Causes` only with `IncreaseEligible` ∧ `Necessary` | Queued | necessary / dispensable / not-run triple |
diff --git a/.claude/board/entries/2026-10-17-ce64-one-layout-moore-synapse-membrane.md b/.claude/board/entries/2026-10-17-ce64-one-layout-moore-synapse-membrane.md
new file mode 100644
index 000000000..aea7b2ced
--- /dev/null
+++ b/.claude/board/entries/2026-10-17-ce64-one-layout-moore-synapse-membrane.md
@@ -0,0 +1,102 @@
+# 2026-10-17 — CE64 one layout; the Moore tenant as synapse, membrane and local NARS
+
+**Status:** DECISIONS recorded from the 2026-10-17 session (user statements quoted as
+BASIS). Code changes named here are NOT made; each waits for its own go.
+**Touches:** `causal-edge` (`layout.rs`, `isa.rs`, `edge.rs`),
+`lance-graph-contract/src/moore_tenant.rs`, `perturbation-sim` (Moore probes),
+ndarray `hpc/{causal_diff,p64_bridge,jina/causal,entropy_ladder}.rs`.
+
+## Decisions (answered; do not re-ask)
+
+| # | DECISION | BASIS (user, 2026-10-17) | consequence |
+|---|---|---|---|
+| D1 | CE64 is an ISA contract, a lever read like BF16 mantissa bits; it is not storage and is not written back into | "Causaledge64 is not storage it's the 'TCP IP' ISA contract for AGI, a contract lever"; "Akin to BF16 mantissa bits" | no "result lands in 59..63" designs; no retiring of contract bits as storage |
+| D2 | CE64 stays 64 bits; no CE128 | user asked "Would it improve anything to switch to causaledge128 or would it dillute the mantissa reading as a wire format"; the session answered stay at 64 and the user did not reopen it | widening would dilute the mantissa reading of the wire |
+| D3 | There is ONE CE64 layout. Align every reader on it, migrate, catalog. No renames | "There is no other layouts / We just need to aligh on one / And migrate and catalog"; "Don't rename, use what exists and harden the bit layout to be canonical" | ndarray's three divergent layouts are migration targets, not alternatives |
+| D4 | The layout is three regions: bits 0..23 S/P/O payload (read as palette256 codes or as an ISA register), bits 24..39 F/C (palette256 or ISA register), bits 40..63 a 24-bit fixed contract with one meaning for every reader | "causaledge64 is 24 bit spo payload that can be read either as palette256 or ISA register / And FC that can be read either as palette256 or ISA register / And 24 bit fixed contract" | `causal_edge::layout` v2 is the canon; domains choose only what payload palette codes mean |
+| D5 | NARS16x8 / the Moore tenant reuses the same layout and omits the bits locality already answers | "NARS16X8 reuses it for Moore tenant / By omitting 8 Bits when locality already answers the 8 Bits" | `MooreNars16` = contract 24 − orientation 3 (43..45) − W 6 + polarity 1 = 16 bits (`moore_tenant.rs:107-219` on main at 124eb6cf). Since D-CGW-1 (#1594) 43..45 is `Orientation`; the Moore reading takes `Core` only and refuses an oriented edge (`MooreRefusal::Oriented`), its direction being `MooreDirection { slot, polarity }` |
+| D6 | The Moore witness is Cartesian addressing: a lane's partner is the neighbour's own Moore tenant at the opposite slot, found by coordinate, never stored | "In NARS16 the Moore tenant connects with the Moore tenant of the neighbors / So the witness is Cartesian addressing" | the module header's "one witness covers all eight lanes, held by the owner" and `lift_ce64`'s `MixedWitness` refusal encode the superseded reading |
+| D7 | The unit of Moore reasoning is the membrane: two neighbours agreeing on their shared boundary. Agreement IS the witness | "2 neighbors agreeing on a cognitive membrane" | a cell has 4 membrane orientations (8 slots / 2); agreement is a side-by-side compare of a plane with its shifted plane (the pair sits at one coordinate; no gather) |
+| D8 | Purpose: local reasoning akin to synapses, with excitation and inhibition | "The whole point of Moore tenant+ NARS16x8 is locality reasoning Akin to synapses and exhibition inhibition" | cell = neuron, lane = synapse, slot = which neighbour, polarity = which end |
+| D9 | Goal: a cheap local perturbation field with cross-node, Bayesian and Hebbian learning | "We want local cheap pertubationsfeld cross node learning Bayesian learning hebbian learning etc" | builds on `perturbation-sim` D-MOORE-PERT-AB-0 (#1427) and D-MOORE-OBSERVABLE-FIRST-0 |
+| D10 | Hebbian wiring IS SPOFC with local NARS reasoning: what fires together wires together as evidence | "The idea is that what fires together wires together SPOFC adding NARS reasoning locally" | co-firing = positive evidence, a lone firing = negative evidence (inhibition on the F side); strength is NAL truth `f = w⁺/w`, `c = w/(w+1)`, updated by `isa::truth::revision` |
+| D11 | Simpson's paradox is what SPOFC suffers under naive NARS accumulation; the architecture prevents it by SPOFC per SoA SPO, fine identity splitting and CE64, with 59..63 and revision as the backstop | "The Simpsons paradox is what spoFC suffers if you accumulate it as NARS usually does, we put it in every SoA SPO and differentiate a lot not to mention CE64 / so it can't dillute or spill and if it does we have 59..63 and revision" | on the Moore lattice, evidence crosses only an agreed membrane; the 8 synapses are never pooled blind. P-pool test on branch `ccr-9f3b8075-74u47l-simpson-coresearch` (`simpson_revision_pooling.rs`) pins the naive case |
+
+## WORKING-MODEL (follows from the decisions; not yet confirmed or built)
+
+- **SPOFC per synapse fills the three Moore tenants.** Per lane:
+  `MoorePalettePairs` (u8:u8) = P and O palette codes; `Nars16x8` (u8:u8) = F and C;
+  `MooreNars16` = the contract part; locality answers S (the centre cell), the direction (the
+  slot, 3 bits of orientation) and W (the membrane, 6). The 48 stored bits are 47 CE64
+  bits plus the Moore-local polarity bit, which has no CE64 home; locality answers 17
+  (S 8 + orientation 3 + W 6). 47 + 17 = 64, a full CE64 per synapse.
+  This would give the CANDIDATE `Nars16x8` (`moore_tenant.rs:8`) its reading.
+- **Local NARS is triangle closure inside the 3×3 neighbourhood.** Deduction A→B, B→C
+  gives A→C when `offset(s)+offset(t)` is a Moore offset; induction/abduction over a
+  shared neighbour. The valid `(s,t)→u` compositions are a fixed compile-time table;
+  each is a shifted-plane operation.
+- **PLAST (50..52) is consolidation, the weight is the F/C evidence.** Refines (does not
+  contradict) the canon row E-BASIN-NOT-EDGE-PLASTICITY: the Hebbian state stays on the
+  per-edge register, never basin-level.
+
+## CE64 reader catalog (migration targets under D3/D4)
+
+| site | reads CE64 as | migration |
+|---|---|---|
+| `causal_edge::layout` + `isa::Field` | v2, three regions | canonical; none |
+| contract `MooreNars16` | contract minus locality-answered bits | catalog only (D5) |
+| `causal-edge/src/edge.rs:139-151` header + v1 consts 183-199 | v1 (temporal at 52..63) | mark V1-legacy, point at `layout` |
+| ndarray `hpc/causal_diff.rs` `CausalEdge64` | own split (block/projection/verb/row/l1/f10/c10) | block/projection/verb → S/P/O palette codes; f/c → u8 F/C; row, L1 stay in `WeightEdge`; upper 24 bits contract default |
+| ndarray `p64_bridge::causal_edge_compat` | SRC 58 / TGT 52 / FREQ 44 / CONF 36 / LAYER 33; doc falsely claims compatibility | read v2 regions |
+| ndarray `jina/causal.rs` `pack_edge` | v1 (temporal 12 at 52) | v2 contract |
+| ndarray `entropy_ladder` | "spare bits [63:61]" (now Epi5 certification) | move off contract bits |
+| ndarray `ogit_bridge/.../CognitiveCell.ttl` | "per causal_diff.rs" | point at `causal_edge::layout` |
+| lance-graph (new) | — | test pinning ndarray's v2 constants against `causal_edge::layout` |
+
+Not CE64, out of scope: holograph `pack_edge`, contract `dawg::pack_edge`, ogar-obo
+`pack_edges`, `CausalEdgeV3`.
+
+## Also decided 2026-10-15/16, not yet on the board elsewhere
+
+| # | DECISION | BASIS (user) | consequence |
+|---|---|---|---|
+| D12 | SPOFC's 8-bit fields work like gas, gear and shift: granular enough to single out one thing in one cohort and one concept; once singled out, the other bits carry it as a fact, otherwise it goes to revision | 2026-10-17: "The 8 bit of spoFC can be used like gas and gear and shift in a car / It's granular enough to detect one thing in one cohort and one concept, which then needs to be carried by the other bits as a fact and otherwise revision" | the payload regions are a control surface, not a precision limit; identity splitting (D11) is what keeps a u8 F/C meaningful |
+| D13 | Hyperon atoms extract into classes and SoA: a class is head + arity + position shapes, its ClassView is its positions, one plane per position, row = instance. The IR (classes, positions, layout) stays separate from the masking algebra, which runs over planes at coordinates and never sees atoms, symbols or shapes. Datalog holds the abstraction; NAL, ONA and relalg are the reusable reasoning parts | 2026-10-16: "So an atom is basically the an array of structs?"; "Shouldn't it be easy to extract classes and SoA"; "And separate IR from masking algebra"; "We have datalog for the Abstraktion"; "And we have reusable nal ONA relalg etc" | D-ASM work lowers atoms to class planes first; no atom or symbol type enters mask-risc |
+| D14 | The edge facet exists to free edges from identity: a 6-tier HHTL path is too tight to spend on edges, which is why edges were rarely used; the 16-nibble DN identity plus a 16-slot edge facet (2×16, the existing 128-bit format) makes edge use independent of identity | 2026-10-16: "reason being is that 6 HHTL depth is extremely tight"; "and we hardly use edged because of that"; "2x 16 makes usage of edges independent of identity"; "we already introduced that format as an 128 optin / its already wired" | mechanism recorded in `2026-10-16-d-cgw-1-orientation-contract.md` (Decided 2026-10-16); this row keeps the reason |
+| D15 | The givens layer (thought layer 0, written only by hydration) is also the global reference lane, held behind a `LazyLock`, for later reference operations | 2026-10-15: "How about we use the first empty thought / For hydrate.rs and use that lane as a global lazylock variable for further reference operations" | layer 0 is recorded in `.claude/plans/ce64-hypothesis-layers-v1.md` (lines 688–701); the `LazyLock` reference lane is the part not yet written down |
+
+## OPEN
+
+- Remove `MixedWitness` from `lift_ce64` and reword the module header (D6), or keep
+  CE64 W as the corpus root that Moore never reads. Waits for a go.
+- `Nars16x8` = F:C per lane: confirm and ratify (contract PR).
+- What counts as "fired" per cell per step (the perturbation field crossing a
+  threshold?).
+- Evidence stamp per membrane hop so loops and triangle deduction do not double count;
+  where its bits live.
+- Hebbian saturation: inhibition on disagreement plus a decay rate (policy pin until
+  measured).
+- Palette index range in `causal_edge_compat` (6-bit `Palette64` vs palette256 codes).
+- PAL8 version gate when `causal_diff` moves to v2.
+- `moore_nars16_probe` / `moore_plasticity_probe` (PR #1406) not yet read; read before
+  the learning probe.
+- `ndarray::ce64::*` (2026-10-17: "Later on when accepting as a wire format / Would
+  that open a door? / ndarray::ce64::*"): unanswered. If the layout becomes a wire
+  format shared with ndarray readers, does the canonical register live in ndarray
+  (hardware tier, beside the SIMD it runs on) with `causal_edge::layout` re-exporting
+  it, or stay in `causal-edge`? Today ndarray carries three divergent layouts and
+  `causal-edge` the canon; the conformance test in the catalog above comes first
+  either way.
+- Ontology bridge, marked for later review (2026-10-16: "also mark the ontology
+  bridge for later review if the bindignspace dto quack thing can improve the old
+  hash to lance approach"): `lance-graph-ontology` and
+  `.claude/plans/sql-spo-ontology-bridge-v1.md` predate the BindSpace DTO / Quack
+  path; review whether their hash-to-Lance mapping can move onto it.
+- Simpson co-research (task #76): the map
+  `2026-10-17-coresearch-simpson-epistemic.md` and the P-pool test
+  (`lance-graph-planner/tests/simpson_revision_pooling.rs`) sit on branch
+  `ccr-9f3b8075-74u47l-simpson-coresearch`, not on main; the map still needs the
+  rows the architecture already covers marked as such (D11).
+- Promotion path: `EPIPHANIES.md` is generated from `epiphanies-canon.json`, whose row
+  sources must resolve to archive lines, so a post-baseline Eureka has no route into the
+  projection today. These decisions live here until that path exists.
diff --git a/.claude/board/entries/README.md b/.claude/board/entries/README.md
index c43e8da0d..52cd24fd3 100644
--- a/.claude/board/entries/README.md
+++ b/.claude/board/entries/README.md
@@ -25,10 +25,11 @@ index row, (3) no duplicate entry id. Checks 1 and 2 are deliberately
 opposite directions; the stranding this convention prevents shows up in
 exactly one of them, never both.
 
-363 entries, 2026-08-06 .. 2026-10-16.
+364 entries, 2026-08-06 .. 2026-10-17.
 
 | date | entry id | finding | file |
 |---|---|---|---|
+| 2026-10-17 | `ce64-one-layout-moore-synapse-membrane` |  | [2026-10-17-ce64-one-layout-moore-synapse-membrane.md](2026-10-17-ce64-one-layout-moore-synapse-membrane.md) |
 | 2026-10-16 | `wcoj-probes-1-3` |  | [2026-10-16-wcoj-probes-1-3.md](2026-10-16-wcoj-probes-1-3.md) |
 | 2026-10-16 | `wcoj-closing-atoms-council` |  | [2026-10-16-wcoj-closing-atoms-council.md](2026-10-16-wcoj-closing-atoms-council.md) |
 | 2026-10-16 | `synergy-map-since-1579` |  | [2026-10-16-synergy-map-since-1579.md](2026-10-16-synergy-map-since-1579.md) |