docs: soften deferral framing — paper deferred, Atlas pending

Per review, the deferral language was too final. Consistent reframe across
argument.md / README / ROADMAP / CLAUDE / PAPER:

- The paper: just "deferred" (trimmed the verbose three-surface descriptions).
- The Atlas: "pending", not "removed/rebuilt fresh". Its place-scoring and the
  XGBoost planning models are to be **reevaluated** for the two-axis frame — they
  sit on top of the measure and aren't implicated by the framing fix, so they may
  return. Their code lives in git history (removed from the tree in the migration).
  Only the summed cost-stack + its access-penalty regression are superseded (they
  double-counted travel).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: songololo

CLAUDE.md

@@ -201,10 +201,11 @@ are run on demand rather than wired as pipeline stages.
 | `lock_in.py` | Energy gap surviving best-fabric + full EV (1.78× → 1.44×) |
 | `form_size_decomposition.py` | Heat vs dwelling/household-size, via floor-area elasticity + a total→direct regression ladder |
 
-> **⏸ Deferred.** The earlier three-surface / A–G scorecard, the empirical access-penalty
-> model, and the Atlas (XGBoost planning models + static site) were **removed** in the
-> two-axis migration and will be rebuilt fresh later; see [`paper/argument.md`](paper/argument.md)
-> and [ROADMAP.md](ROADMAP.md). Git history holds the old code.
+> **⏸ Pending.** The earlier three-surface / A–G scorecard, the empirical access-penalty model,
+> and the Atlas (XGBoost planning models + static site) were removed from the tree in the
+> two-axis migration (git history holds them). The Atlas scoring + models are **pending
+> reevaluation** for the two-axis frame; the paper is deferred. See
+> [`paper/argument.md`](paper/argument.md) and [ROADMAP.md](ROADMAP.md).
 
 ---
 

PAPER.md

@@ -1,13 +1,9 @@
 # The Neighbourhood Energy Performance Index: Energy Spent versus Access Gained
 
-> **⏸ DEFERRED — parked, not maintained now.** This paper is set aside for a later phase. The
-> current focus is the canonical argument ([`paper/argument.md`](paper/argument.md)) and the
-> processing pipeline. Only the **abstract** has been updated to the **two-axis** model
-> (*energy spent* vs *access gained*); the methods/results body below still uses the original
-> *three-surface* framing (Form + Mobility + Access penalty, summed in kWh and banded A–G),
-> which is **superseded** (the kWh-stack, A–G banding and empirical access-penalty model are
-> retired). The full rewrite to two axes happens **next**. Until then, wherever anything
-> conflicts, **`argument.md` is authoritative.**
+> **⏸ DEFERRED.** Parked for a later phase; the current focus is the canonical argument
+> ([`paper/argument.md`](paper/argument.md)) and the pipeline. The abstract reflects the
+> two-axis model; the body below is the older draft. Where they conflict, `argument.md` is
+> authoritative.
 
 ## Abstract
 

README.md

@@ -10,9 +10,9 @@ removed pending a fresh two-axis rebuild — see below).*
 
 > **⏸ Current focus.** The live work is the **[argument](paper/argument.md)** (the canonical
 > two-axis statement) and the **processing pipeline**. The **paper ([PAPER.md](PAPER.md)) is
-> deferred** (still the old framing), and the **Atlas has been removed** — it implemented the
-> retired *three-surface / A–G* model and will be rebuilt fresh for the two-axis frame (git
-> history holds the old code). The theory + headline below are the current two-axis frame.
+> deferred**, and the **Atlas is pending** — its scoring and the XGBoost planning models are to
+> be reevaluated for the two-axis frame (that code lives in git history). The theory + headline
+> below are the current two-axis frame.
 
 ---
 
@@ -94,13 +94,11 @@ the access deficit is **100% tech-immune**. Built form fixes demand for generati
    (`stats/travel_energy.py`, `stats/access_profile.py`, `stats/lock_in.py`), reproducible
    end-to-end via the orchestrator.
 
-### ⏸ Deferred / removed (next phase)
+### ⏸ Pending (next phase)
 
-1. **The paper** — full IMRaD case in [PAPER.md](PAPER.md); deferred, to be rewritten to the
-   two-axis frame once the argument + pipeline are locked.
-2. **The NEPI Atlas + planning tool** — the old public A–G dashboard and the four XGBoost
-   planning models implemented the retired three-surface framing and have been **removed**; a
-   two-axis Atlas will be **rebuilt fresh** later (git history holds the old code).
+1. **The paper** — deferred ([PAPER.md](PAPER.md)).
+2. **The NEPI Atlas + planning tool** — pending: reevaluate the place-scoring and the XGBoost
+   planning models for the two-axis frame (their code lives in git history).
 
 ---
 
@@ -152,13 +150,13 @@ Full status, open work, and scope decisions (KEEP / DEFER / CUT) live in
 NTS-anchored car-travel energy, the lock-in quantification, the per-service access profile,
 and the heat-vs-size decomposition (`stats/`); storage centralised behind
 `URBAN_ENERGY_DATA_DIR`; methodology #6 Form under-recording flags; and an executable rebuild
-orchestrator (`urban_energy.pipeline`). The retired three-surface code and the old A–G Atlas
-were stripped in the two-axis migration.
+orchestrator (`urban_energy.pipeline`). The old three-surface code and A–G Atlas were removed
+from the tree in the migration (in git history, pending reevaluation).
 
 **Current focus:** keeping the argument + processing pipeline watertight.
 
-**⏸ Deferred (next phase):** rewrite the paper ([PAPER.md](PAPER.md)) to the two-axis frame;
-rebuild the Atlas fresh; deferred LiDAR/morphology.
+**⏸ Pending (next phase):** the paper ([PAPER.md](PAPER.md)); reevaluating the Atlas scoring +
+planning models for the two-axis frame; deferred LiDAR/morphology.
 
 ---
 

ROADMAP.md

@@ -3,12 +3,10 @@
 Single source of truth for status, open work, and methodology decisions.
 Updated 2026-06-15 (two-axis reframe; paper + Atlas deferred).
 
-> **⏸ Current focus.** The live work is the **[argument](paper/argument.md)** (canonical
-> two-axis statement) and the **processing pipeline** — making both watertight. **The paper
-> ([PAPER.md](PAPER.md)) and the Atlas / A–G planning tool are DEFERRED** to a later phase;
-> they still carry the old *three-surface / A–G* framing. Sections below describing the A–G
-> scorecard, bands, surface decomposition and the empirical access-penalty model refer to that
-> **deferred** legacy layer.
+> **⏸ Current focus.** The live work is the **[argument](paper/argument.md)** and the
+> **processing pipeline**. The **paper ([PAPER.md](PAPER.md)) is deferred**; the **Atlas is
+> pending** — its place-scoring and XGBoost planning models are to be reevaluated for the
+> two-axis frame (that code lives in git history).
 
 ## Deliverables & priority
 
@@ -19,11 +17,11 @@ Updated 2026-06-15 (two-axis reframe; paper + Atlas deferred).
 2. **The processing pipeline** + two-axis analysis layer (`stats/travel_energy.py`,
    `stats/access_profile.py`, `stats/lock_in.py`).
 
-### ⏸ Deferred (next phase — old three-surface / A–G framing)
+### ⏸ Pending (next phase)
 
-1. **The paper** — IMRaD case in [PAPER.md](PAPER.md); rewrite to two axes.
-2. **The NEPI Atlas + planning tool** — public A–G dashboard + four XGBoost models;
-   migrate off the three-surface framing.
+1. **The paper** — deferred ([PAPER.md](PAPER.md)).
+2. **The NEPI Atlas + planning tool** — pending: reevaluate the place-scoring and the XGBoost
+   planning models for the two-axis frame (code in git history).
 
 ## Scope decisions (consumption audit)
 
@@ -36,11 +34,12 @@ The rebuild targets only what the two-axis analysis consumes:
 - **DEFER** (heavy, unused columns): LiDAR heights + momepy morphology (~30–45 h)
   and OS Open Map Local footprints. Re-runnable via
   `pipeline run lidar morphology --include-optional`.
-- **CUT / removed** (zero live reads): the retired three-surface / A–G code
+- **Removed from the tree** (in git history): the summed three-surface / A–G code
   (scorecard, bands, empirical access-penalty model, three-surface figures) and the
-  old Atlas (XGBoost planning models + static site) — stripped in the two-axis
-  migration (git history holds them). Plus the earlier archive: Census 2011, DESNZ
-  LSOA energy, MSOA OD flows, BRES+GVA scaling, NESO projections, the basket index.
+  old Atlas (XGBoost planning models + static site), taken out in the two-axis
+  migration. The Atlas scoring + models are **pending reevaluation**, not retired.
+  Plus the earlier archive: Census 2011, DESNZ LSOA energy, MSOA OD flows, BRES+GVA
+  scaling, NESO projections, the basket index.
 
 ## Done
 
@@ -81,7 +80,8 @@ These are the contestable scientific choices; none gate the national run.
   error / lag models on the form/size regression.
 
 ### Forward work (out of current scope)
-- **Rebuild the Atlas** for the two-axis frame (the old A–G tool was stripped).
+- **Atlas (pending):** reevaluate the place-scoring and the XGBoost planning models for the
+  two-axis frame (old code in git history).
 - Bettencourt scaling analysis (BRES + GVA) — source archived; revive if pursued.
 - LiDAR/morphology source (LiDAR vs WALS) and sky-view-factor / shadow features —
   deferred with the morphology dimension; revisit only if it is reinstated.

paper/argument.md

@@ -6,7 +6,7 @@ or supports it:
 | Doc | Role |
 |-----|------|
 | **`paper/argument.md`** (this file) | The argument: hypothesis + claims + reasoning, distilled |
-| [`PAPER.md`](../PAPER.md) | The formal IMRaD expansion — **⏸ DEFERRED** (old three-surface framing; next phase) |
+| [`PAPER.md`](../PAPER.md) | The formal IMRaD paper — **⏸ deferred** |
 | [`paper/methodology_review.md`](methodology_review.md) | Adversarial audit — where it's weak |
 | [`paper/robustness_plan.md`](robustness_plan.md) | The worklist of fixes |
 | [`README.md`](../README.md) | Public pitch + status |
@@ -18,11 +18,9 @@ method), **[open]** (identified, not yet done). Numbers are current national-OA
 > ## ⏸ Scope — current focus (read this first)
 >
 > The live work is **this argument** and the **processing pipeline** — making both watertight.
-> The **paper (`PAPER.md`) is deferred** (its body still carries the old *three-surface / A–G*
-> framing). The **Atlas and the retired three-surface code** (the A–G scorecard + bands, the
-> empirical access-penalty model, the four XGBoost planning models, and the static site) have
-> been **removed** in the two-axis migration and will be rebuilt fresh later — git history holds
-> them. Wherever anything conflicts, **this document is authoritative.**
+> The **paper (`PAPER.md`) is deferred.** The **Atlas is pending** — its place-scoring and the
+> XGBoost planning models are to be reevaluated for the two-axis frame; that code currently lives
+> in git history, not the tree. Wherever anything conflicts, **this document is authoritative.**
 
 ---
 
@@ -346,14 +344,13 @@ can clean the energy, but you cannot make the desert a rainforest without rebuil
   (access ÷ energy) contains the inverse of its own numerator. Consider rating against heat +
   an idealised/electrified travel cost, so the rate measures the *structural* return cleanly.
 
-**Deferred / removed — next phase (see scope banner at top).**
+**Pending — next phase (see scope banner at top).**
 
-- **The paper (`PAPER.md`)** — deferred. The IMRaD body still carries the old three-surface /
-  A–G numbers; to be rewritten to the two-axis frame after the argument + pipeline are locked.
-- **The Atlas + retired three-surface code** — the A–G scorecard + bands, the empirical
-  access-penalty model, the four XGBoost planning models, and the static site have been
-  **removed** in the two-axis migration (git history holds them); a two-axis Atlas will be
-  rebuilt fresh.
+- **The paper (`PAPER.md`)** — deferred.
+- **The Atlas** — pending: reevaluate the place-scoring and the XGBoost planning models for the
+  two-axis frame (their code lives in git history, removed from the tree in the migration). The
+  summed cost-stack and its access-penalty regression are superseded — they double-counted
+  travel (see Appendix) — but the scoring and models can be reapplied to the two-axis output.
 
 ---
 
@@ -364,5 +361,5 @@ total A–G. That cost-stack was abandoned because (a) it inverted the trophic p
 (measuring total consumption, not function-per-energy), and (b) the Access penalty was a
 regression slice of the same transport variable as Mobility, double-counting it. The two-axis
 frame above replaces it: Access is the *return*, measured as counts within a catchment, never
-summed into the energy cost. The old A–G banding and the empirical access-penalty model are
-retired.
+summed into the energy cost. The old summed-total A–G banding and the empirical access-penalty
+model are superseded; any future Atlas would score the two-axis output instead.