diff --git a/docs/re/data/jp-difficulty-not-reached.txt b/docs/re/data/jp-difficulty-not-reached.txt
new file mode 100644
index 00000000..3d8b1abc
--- /dev/null
+++ b/docs/re/data/jp-difficulty-not-reached.txt
@@ -0,0 +1,42 @@
+# Are GP_DIALOG 2/3 specifically ENGLISH and JAPANESE? ❔ STILL NOT SETTLED.
+# 2026-08-31. The run did not reach DIFFICULTY, and what it did produce is below.
+#
+# THE QUESTION: 2/3 differ in 2.77 % of bytes while sharing every element name --
+# what a language pair looks like -- but "English and Japanese" rested on the
+# disc's convention rather than on a capture of THIS screen
+# (dialog-0-1-is-a-duplicate.txt). The untested step was a `ja` capture.
+#
+# 🔴 WHAT FAILED: the reach probe's round trip. It reached the JP main menu, moved
+# the cursor, pressed Ⓑ to the title, pressed Ⓐ -- and never saw the menu again,
+# sitting at glyph 11654 until it timed out. The sweep then started with the game
+# off-menu and timed out too ("never identified the main menu"), so NEW GAME was
+# never pressed and DIFFICULTY was never opened.
+#
+# ⚠️ NOT A DETECTOR PROBLEM, which is what I assumed while watching it. The glyph
+# 11654 in the log is a LATER phase. The JP menu itself detects perfectly:
+#   1-F1  ring y 225.5  NEW GAME   glyph 320  is_main_menu True
+#   2-F2  ring y 385.5  TUTORIAL   glyph 320  is_main_menu True
+# against the English menu's 327. The 250..420 band covers both.
+#
+################################################################################
+# ✅ WHAT THE RUN DID ESTABLISH -- captures/menu-nav/live-jp-main-menu.png
+#
+# THE LOCALE TOOK, and this is the first JP MAIN MENU capture in the corpus:
+#   新規 / ロード / チュートリアル / オプション / エクストラ, footer 選択 / 決定.
+#
+# ✅ JP INITIAL FOCUS IS 新規 -- NEW GAME, the top item, ring y 225.5. The SAME
+# item and the SAME row as English (focus-does-not-survive-a-reboot.txt measured
+# NEW GAME at 225.5 across six English boots). So initial focus is not
+# locale-dependent, on one JP boot.
+#
+# ✅ AND THE MENU LAYOUT IS IDENTICAL ACROSS LOCALES: ring rows 225.5 and 385.5
+# match the English rows exactly, so the JP build places its buttons where the
+# English one does and differs only in the glyphs. That is the language-pair
+# structure confirmed at the MENU (GP_TITLE 5/8) -- ⚠️ which is NOT the dialog
+# pair the question is about, and does not transfer to GP_DIALOG 2/3 by itself.
+#
+# ⚠️ REACH: one JP boot. The DIFFICULTY screen was not reached, so the question
+# this run was launched for is exactly where it was.
+#
+# 📌 The locale was restored on exit and verified back at language = 1, by the
+# trap that runs on ANY exit including this failure.