lv-mccarthy D2: entity map + stoplist, both audits green — and audit_stoplist was scanning its own rationale

Entity map at ~/mccarthy-corpus/entities.json. 123 surfaces after a 107-surface stoplist.

  entities.py        27/27 controls -- 19 positive (Glanton, Toadvine, Rawlins, Blevins,
                     Alejandra, Chigurh, Moss, Bell, Boyd, Holden, Tobin, Magdalena,
                     Eduardo, Parham, Socorro, Webster, Redbo, Niño, Franklin) and 8 negative
  audit_stoplist     PASS -- no stoplisted surface is ever addressed as a person
  audit_entity_map   PASS -- positive `boy` 0.89, negative band tops out at Riddle 0.17,
                     all 5 remaining flags on the read-and-cleared list

⚠⚠ A DEFECT IN audit_stoplist.py ITSELF, latent for every corpus before this one. It built
its surface set from every list value in the stoplist JSON -- including `_why`, which by
convention is a LIST OF PROSE LINES. Every sentence of the rationale went into the matcher,
and the empty separator line matched the honorific pattern 139 times, printing a flag with no
surface name at the top of the report, above the one real catch. It now skips `_`-prefixed
metadata keys and empty strings.

THE ONE REAL CATCH WAS A CONTRADICTION INSIDE MY OWN FILE. `Franklin` sat in the geography
list because it is the old name for El Paso, while the same file's context note recorded
'I'm here to see Mr Franklin' -- a lawyer in All the Pretty Horses. The honorific audit found
the contradiction between the two halves of the file. Franklin is now renameable.

A SECOND SELF-INFLICTED ONE: the fragments list was a speculative A-Z, which stoplisted `I`
and `A` -- ordinary English words -- and `Sir I dont think I can do that` duly tripped the
honorific audit. It is now the four letters actually MEASURED as entities (E, H, T, K).
Stoplist what the entity map produced, not the alphabet.

Everything ambiguous was read in context before placement, and the reasoning is in the file:
  Socorro is the ranch COOK in Cities of the Plain, not the New Mexico town -- renameable
  Webster, Jackson, Harlan, Lamar are Glanton's men and lawmen, not places -- renameable
  Niño, Keno, Redbo are HORSES, the author's inventions -- renameable, the `Inglés` precedent
  Mangas, Travis, Venada, Moderno are genuinely dual-use -- renamed, the safe direction
  Santa, Varas, Griffin, Eagle, Avenue, Calle, Terrell are real geography -- stoplisted
  Yaqui and Gilenos are real peoples; Ford and Hashknives are a brand and a real outfit
  Ed (Ed Tom Bell) and JC are short but are names, read and kept renameable

Sensitivity floor, stated because it is part of the result: the top 170 of 199 surfaces were
classified. The bottom 29 were not individually read, so a rare real-world referent may be
renamed -- the safe direction, an accepted cost, not an oversight.

scripts/mccarthy-corpus/stoplist_mccarthy.json

@@ -0,0 +1,173 @@
+{
+  "_why": [
+    "Surfaces excluded from renaming because they are REAL-WORLD referents, not the author's",
+    "invented proper nouns. The R49 rule is that ambiguous cases are deliberately RENAMED --",
+    "renaming is the safe direction and leaving is the leaking one -- so this list is kept to",
+    "categories that are unambiguously real: geography, demonyms, brands, foreign common nouns,",
+    "and extraction fragments that are not words at all.",
+    "",
+    "⚠ SENSITIVITY FLOOR, stated because it is part of the result: the entity map holds 199",
+    "surfaces and the top 170 by count were classified, with every ambiguous one READ IN",
+    "CONTEXT before being placed. The bottom 29 were NOT individually read, so a rare",
+    "real-world referent may be renamed. That is the safe direction and an accepted cost.",
+    "",
+    "⚠ McCarthy is a hard case for the same reason Hemingway was and Bronte was not: the Border",
+    "Trilogy is half set in real Mexico, so real towns, real states and real Spanish sit beside",
+    "invented characters on every page. The 1,411 non-ASCII letters measured on this corpus are",
+    "the same fact from the other side."
+  ],
+  "_caught_by_reading_in_context": {
+    "Socorro": "NOT the New Mexico town. She is the COOK at Mac's ranch in Cities of the Plain -- 'Socorro came and took the plate of biscuits'. A character; left renameable.",
+    "Webster": "NOT the dictionary or the city. 'the outrider Webster', 'A Tennessean named Webster' -- one of Glanton's men. Character.",
+    "Jackson": "'two men named Jackson, one black, one white, both forenamed John'. Characters, not the city or the president.",
+    "Harlan": "'a man named Harlan from Texas'. Character, not the Kentucky county.",
+    "Franklin": "KEPT RENAMEABLE. 'I'm here to see Mr Franklin' -- a lawyer in All the Pretty Horses. ⚠ He was in the geography list on the first pass, because Franklin is also the old name for El Paso, and audit_stoplist.py caught the contradiction against this very note.",
+    "Lamar": "'called Lamar on the mobile ... he can have his unit back' -- a lawman in No Country. Character, not the street or the Texas county.",
+    "Griffin": "STOPLISTED. 'Fort Griffin', 'we pulled out from Griffin for a last hunt' -- the real Texas buffalo-hunting town.",
+    "Santa": "STOPLISTED. Never standalone: the fragment of 'Santa Rita del Cobre' and 'Santa Cruz'.",
+    "Varas": "STOPLISTED. The fragment of 'Las Varas', a real Chihuahua town on the Madera road.",
+    "Eagle": "STOPLISTED. 'Eagle Pass' (real Texas border town) and 'the Eagle Cafe'.",
+    "Avenue": "STOPLISTED. A street-name component -- 'Juárez Avenue'. A common noun, not a name.",
+    "Niño": "KEPT RENAMEABLE, deliberately. A Spanish common noun, but here it is the NAME OF A HORSE ('That Niño horse'). The author's invention even though the word is not -- exactly parallel to Hemingway's `Inglés`.",
+    "Keno": "KEPT RENAMEABLE for the same reason -- 'their horse Keno'. A gambling game's name used as a horse's.",
+    "Redbo": "KEPT RENAMEABLE -- John Grady's horse in All the Pretty Horses.",
+    "Mangas": "KEPT RENAMEABLE. Based on the real Apache chief Mangas Coloradas, but presented as a character who speaks ('The man's name was Mangas and he was cordial'). Ambiguous, so it takes the safe direction.",
+    "Travis": "KEPT RENAMEABLE. Dual use -- real 'Travis Park' in San Antonio AND a character who speaks ('said Travis'). Ambiguous, so it is renamed.",
+    "Venada": "KEPT RENAMEABLE. 'La Venada', a Juárez cantina. McCarthy's Juárez venues are largely invented and this cannot be settled from the text, so it takes the safe direction.",
+    "Jc": "KEPT RENAMEABLE. Not a fragment -- 'JC pulled up in front of Maud's' is a character in Cities of the Plain who is named by his initials.",
+    "Ed": "KEPT RENAMEABLE. Ed Tom Bell, the sheriff and first-person narrator of No Country for Old Men.",
+    "E / H / T / K": "STOPLISTED as fragments. E, H and T were the small-caps drop-cap defect repaired in D1; K is the fragment left by the weapon `H&K`. A bare initial is never a name -- the `G` class from the Hemingway build, where `G` was about to be renamed to a surname 248 times.",
+    "fragments": "Only the four letters MEASURED as entities on this corpus. A blanket A-Z was the first pass and it stoplisted `I` and `A`, which are ordinary English words -- `Sir I dont think I can do that` then tripped the honorific audit. Stoplist what the entity map actually produced, not the alphabet.",
+    "Ford / Hashknives": "STOPLISTED. 'a Model A Ford truck'; 'went to work for the Hashknives' -- the real Aztec Land & Cattle outfit. Brands and real companies.",
+    "Calle": "STOPLISTED. Spanish for 'street' -- 'the Calle de Noche Triste'. A common noun, like Avenue.",
+    "Yaqui / Gilenos": "STOPLISTED. Real indigenous peoples -- 'lay clear of the Yaqui country', 'all trace of the Gilenos faded'. Demonyms.",
+    "Terrell": "STOPLISTED. 'nineteen felony charges were filed in the Terrell County Court' -- a real Texas county."
+  },
+  "geography": [
+    "America",
+    "Animas",
+    "Antonio",
+    "Ardmore",
+    "Arizona",
+    "Arkansas",
+    "Avenue",
+    "Babícora",
+    "Baviacora",
+    "Bavispe",
+    "Bexar",
+    "Boquilla",
+    "Caborca",
+    "California",
+    "Calle",
+    "Casas",
+    "Chihuahua",
+    "Cloverdale",
+    "Coahuila",
+    "Diego",
+    "Durango",
+    "Eagle",
+    "El",
+    "Encantada",
+    "Francisco",
+    "Grandes",
+    "Griffin",
+    "Huisiachepic",
+    "Janos",
+    "Langtry",
+    "Madera",
+    "Mexico",
+    "Missouri",
+    "Nacozari",
+    "Namiquipa",
+    "Odessa",
+    "Oklahoma",
+    "Paso",
+    "Pilares",
+    "Purísima",
+    "Saltillo",
+    "San",
+    "Sanderson",
+    "Santa",
+    "Sierra",
+    "Sonora",
+    "States",
+    "Tennessee",
+    "Terrell",
+    "Texas",
+    "Torreón",
+    "Ures",
+    "Uvalde",
+    "Varas",
+    "Vietnam",
+    "Yuma"
+  ],
+  "demonyms_and_peoples": [
+    "American",
+    "Americans",
+    "Apache",
+    "Apaches",
+    "Arabian",
+    "Chiricahuas",
+    "Comanche",
+    "Comanches",
+    "Delaware",
+    "Delawares",
+    "Gileno",
+    "Gilenos",
+    "Mennonite",
+    "Mexican",
+    "Mexicans",
+    "Mormon",
+    "Tarahumara",
+    "Vandiemenlander",
+    "Yaqui",
+    "Yaquis",
+    "Yumas"
+  ],
+  "brands_and_products": [
+    "Blanca",
+    "Bronco",
+    "Carta",
+    "Colts",
+    "Ford",
+    "Hashknife",
+    "Hashknives",
+    "Ramcharger",
+    "Tecate",
+    "Winchester"
+  ],
+  "foreign_and_common_nouns": [
+    "Adios",
+    "Bueno",
+    "Claro",
+    "Como",
+    "Cuates",
+    "Dios",
+    "Gracias",
+    "Hacendado",
+    "Mam",
+    "Nada",
+    "Nosir",
+    "Pues",
+    "Senor",
+    "Senora",
+    "Vamos",
+    "Yessir"
+  ],
+  "fragments": [
+    "E",
+    "H",
+    "T",
+    "K"
+  ],
+  "_audit_allow": {
+    "_note": "Surfaces audit_entity_map.py flags as article-preceded that are nonetheless genuine names, READ and cleared. Pass as --allow when re-running the audit. They are article-preceded because English puts a surname attributively behind 'the': 'the Bailey horse', 'the Muñoz woman', 'the Borginnis woman'. That is the detector's known blind spot, not an error.",
+    "surfaces": [
+      "Bailey",
+      "Muñoz",
+      "Borginnis",
+      "Venada",
+      "Moderno"
+    ]
+  }
+}

scripts/r49-corpus/audit_stoplist.py

@@ -53,7 +53,14 @@ def main() -> int:
 
     text = load_text(Path(a.corpus))
     blob = json.loads(Path(a.stoplist).read_text())
-    surfaces = sorted({s for v in blob.values() if isinstance(v, list) for s in v})
+    # ⚠ SKIP `_`-PREFIXED KEYS. By convention a stoplist carries its rationale in `_why` and
+    # `_caught_by_reading_in_context`, and `_why` is a LIST OF PROSE LINES. Scanning it as
+    # surfaces put every sentence of the rationale into the matcher, and its empty separator
+    # line matched the honorific pattern 139 times on McCarthy -- a flag with no surface name
+    # attached, at the top of the report, above the one real catch. A metadata key is not a
+    # surface set; found 2026-09-17 and latent for every corpus before it.
+    surfaces = sorted({s for k, v in blob.items() if not k.startswith("_")
+                       and isinstance(v, list) for s in v if s})
     cleared = {s.strip() for s in a.allow.split(",") if s.strip()}
 
     flagged = []