Trace tier2 kind-8 recipe sequencing

artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-named-availability-note.md

@@ -57,6 +57,13 @@ The current comparison shows `18` differing names:
 - `Tool And Die`: `Louisiana = 0`, `Dutchlantis = 1`
 - `Uranium Mine`: `Louisiana = 0`, `Dutchlantis = 1`
 
+The broader `compare-candidate-table` surface agrees with the same conclusion:
+
+- common semantic family remains `scenario-named-candidate-availability-table`
+- `Warehouse05` stays `1 / 1`
+- `difference_count = 43`, but those differences are still driven by the wider industrial mix,
+  header words, and the zero-trailer-name set rather than by any unique `Warehouse05` bit
+
 ## Current Implication
 
 The Tier 2 candidate/world-state rebuild strip remains plausible in general because it owns named

artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-recipe-book-note.md

@@ -0,0 +1,89 @@
+# Runtime Effect Kind-8 Tier2 Recipe-Book Note
+
+This note records the current recipe-book comparison between the strongest title-overlap pair:
+
+- `Louisiana.gmp`
+- `Dutchlantis.gmp`
+
+The comparison uses the checked `compare-recipe-book-lines` surface.
+
+## Immediate Result
+
+Unlike the named-availability table, the recipe-book surface differs heavily.
+
+Observed high-level result:
+
+- `matches = false`
+- `content_matches = false`
+- both files stay on `source_kind = recipe-book-summary`
+- both files keep `book_count = 12`, `line_count = 5`, `book_stride = 0x4e1`, `line_stride = 0x30`
+
+## Louisiana profile
+
+`Louisiana.gmp` is sparse in this surface:
+
+- only `book00` currently shows mixed line-area content
+- the rest of the line areas are currently `zero`
+- inside `book00`, only `line02` currently shows a nonzero mode word:
+  - `mode_word_hex = 0x00080000`
+  - `supplied_cargo_token_word_hex = 0x00004080`
+  - `demanded_cargo_token_word_hex = 0x00000000`
+
+## Dutchlantis profile
+
+`Dutchlantis.gmp` is much broader:
+
+- many later books and lines remain `mixed` where `Louisiana.gmp` stays `zero`
+- current explicit mode-word differences include:
+  - `book01.line02 = 0x001a0000`
+  - `book02.line02 = 0x00040000`
+  - `book03.line02 = 0x001a0000`
+  - `book06.line02 = 0x00010000`
+  - `book07.line02 = 0x00070000`
+- multiple supplied-token lanes are also nonzero where `Louisiana.gmp` stays zero
+
+## Current implication
+
+This is the first strong Tier 2 discriminator between the strongest title-overlap pair:
+
+- `Warehouse05` availability is **not** different (`1 / 1`)
+- but the scenario-side recipe-book runtime inputs differ substantially
+
+That strengthens the current bias toward the
+`0x00435630 -> 0x00412d70`
+side of the Tier 2 strip:
+
+- `0x00435630` rebuilds the port/warehouse cargo recipe runtime tables from scenario-side recipe books
+- `0x00412d70` then rebuilds candidate runtime records from scenario state
+
+So the next Tier 2 recovery question should treat recipe-book/runtime-record sequencing as a more
+likely differentiator for `Louisiana.gmp` than either title overlap or the direct named
+`Warehouse05` availability bit.
+
+## Same-condition-family contrast
+
+The current recipe-side bias does not collapse to the shared condition family alone.
+
+`Louisiana.gmp` and `Texas Tea.gmp` both currently sit on the same add-building condition family:
+
+- `Louisiana.gmp`
+  - cluster:
+    `nondirect-ge1e-h0001-0007-0000-5200-0200-p0000-0000-0000-ffff :: [7:0]`
+  - label: `Add Building Warehouse05`
+  - row shape: `[0:8:0,0:8:0,0:8:0,1:2:-10,2:2:500000]`
+  - nonzero recipe modes: only `book00.line02 = 0x00080000`
+- `Texas Tea.gmp`
+  - cluster:
+    `nondirect-ge1e-h0001-0007-0000-6d00-0200-p0000-0000-0000-ffff :: [7:0]`
+  - label: `Add Building Port01`
+  - row shape: `[0:2:25,0:8:0]`
+  - nonzero recipe modes:
+    `book00.line02 = 0x000a0000`,
+    `book01.line02 = 0x000a0000`,
+    `book02.line02 = 0x000a0000`,
+    `book04.line02 = 0x00050000`
+
+So the current evidence says:
+
+- shared `[7:0]` condition family is not enough to explain the shipped add-building row
+- the recipe/runtime side is still map-specific even inside that same condition family

artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-sequencing-note.md

@@ -0,0 +1,73 @@
+# Runtime Effect Kind-8 Tier2 Sequencing Note
+
+This note extracts the current late bringup sequencing around the Tier 2
+candidate/world-state rebuild strip beneath
+`world_entry_transition_and_runtime_bringup` `0x00443a50`.
+
+## Earlier recipe-runtime rebuild checkpoint
+
+The earlier post-load recipe-runtime rebuild call at `0x00443ebc` runs:
+
+- immediately after the named candidate-availability collection at `[world+0x66b2]`
+  has been restored from fixed `0x22`-byte rows
+- before the neighboring candidate filter/count rebuilds
+  `0x00412c10 / 0x00412bd0`
+- before the year-derived follow-ons
+  `0x00434130 / 0x00436af0`
+
+That means the earlier checkpoint is already candidate/world rebuild work, not the final
+kind-`8` materializer by itself.
+
+## Later `0x197` checkpoint
+
+The later explicit `0x197` checkpoint at `0x00444ac1` sits:
+
+- after `world_publish_shell_controller_progress_scalar_from_year_thresholds_or_selector_overrides`
+  `0x004354a0`
+- after the territory-side sweep
+  `world_region_border_overlay_emit_segment_geometry_from_region_grid`
+  `0x00487de0`
+- and then falls through into
+  `world_preseed_named_candidate_availability_records_from_live_pool`
+  `0x00437737`
+- followed by the later candidate-side availability refresh
+  `0x00412c10`
+
+So the late Tier 2 strip begins with named-availability preseed and latch refresh, not with the
+shell progress or territory overlay helpers that precede it.
+
+## Internal Tier 2 rebuild relationships
+
+Current grounded internal relationships:
+
+- `0x00437737`
+  - pre-seeds scenario-side named candidate availability records from the live pool
+  - upserts records through `0x00434f20`
+- `0x00434f20`
+  - writes a boolean named-availability override bit
+  - immediately re-enters `0x00412c10` when the live candidate pool exists
+- `0x00412c10`
+  - refreshes `[candidate+0x7ac]` from named availability and then tails into
+    cargo-economy filter refresh `0x0041eac0`
+- `0x00435630`
+  - rebuilds port/warehouse recipe runtime tables from scenario-side recipe books
+  - then re-enters `0x00412d70`
+- `0x00412d70`
+  - rebuilds candidate runtime records from scenario state
+  - does not consult the scenario-side recipe-book name at `[state+0x0fe8]`
+- `0x00412fb0`
+  - broader collection load owner that re-enters `0x004120b0`, then `0x00412d70`,
+    then `0x00412ab0`, and only after its later catalog rebuild re-enters `0x00412c10`
+
+## Current recovery implication
+
+The strongest remaining Tier 2 question is sequencing, not naming:
+
+- whether the shipped `Add Building Warehouse05` row in `Louisiana.gmp`
+  depends on the order or interaction of
+  `0x00437737 -> 0x00434f20 -> 0x00412c10`
+  versus
+  `0x00435630 -> 0x00412d70`
+  and the broader load-side rebuild owner
+  `0x00412fb0`,
+  rather than on a direct `Warehouse05` availability bit or recipe-book display-name leak.