the closing verdict, the capability-map row, and the live evidence

The three blockers yesterday's English walk found are closed and each proven on a live system. The verdict is deliberately 'nothing known now stands in their way' rather than 'the walk passed': fixes are not a journey, and the hour has not been re-walked by a stranger on a fresh install. Also filed: the HP demo box answers on no route this session has (R-601), the cookie-vs-session language instrument trap that would have had me fix R-598 twice (R-602), and the apostrophe that silently never matches a rendered page (R-603). Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
2026-09-21 08:04:31 +02:00
parent 83b558b3ea
commit fcdc948909
4 changed files with 206 additions and 1 deletions
@@ -0,0 +1,50 @@
+# Live proof — an English household gets English words (R-597)
+
+**Hub 0.119.0**, deployed 2026-09-21 07:58 UTC. Read from the operator's catch-all inbox.
+Codes are redacted per §9.4; the SHAPE is the evidence and the shape is what was measured.
+
+## The before and after are in the same inbox, one day apart
+
+Same mail, same subject, same template — only the hub version differs.
+
+| when | hub | subject | the code |
+|---|---|---|---|
+| 2026-09-20 16:55 UTC | 0.118.1 | `[Felhom] Your Felhom server is up — setup code` | `Setup code: képző-szkítia-ásatás` — **3 Hungarian words, 5 non-ASCII characters** |
+| 2026-09-21 06:00 UTC | **0.119.0** | `[Felhom] Your Felhom server is up — setup code` | `Setup code: XXXXXX-XXXXXXXX-XXXXXXXXX-XXXXXXXXX` — **4 words, all plain lower-case ASCII** |
+
+The 2026-09-20 line is the actual code the drill received. It is quoted because it is already in
+`audits/DRILL-first-hour-en-0258-2026-09-20.md`, it is long expired, and its customer is deleted —
+it is the defect, not a secret. Today's is redacted because it is live at the moment of writing.
+
+## The owner passphrase, read from the hub's own store
+
+Shape only — never the value:
+
+| customer | language | words | all ASCII lower-case |
+|---|---|---|---|
+| `demo-felhom` | `hu` | **5** | **no** (accented, as it always was) |
+| `i18n-en-0921` | `en` | **6** | **yes** |
+
+Six English words = 77.5 bits, against the Hungarian five words' 74.3. The English code is longer
+**and stronger**, which is the rule the entropy test pins.
+
+## The venue
+
+A scratch customer `i18n-en-0921` was created on the hub with `language=en` and the operator's own
+address, its claim code re-sent by the operator button, and the customer **deleted afterwards**.
+No existing customer was touched: `demo-felhom` and `peti-felhom` stay `hu`, and `peti-felhom` is a
+real person's box.
+
+**This is a deviation from the task, which said to use "the demo customer (`en`)".** There is no
+English customer on this hub — all five are `hu` — so there was nothing to request a reset for. A
+scratch customer is the smallest thing that proves the deployed binary actually picks the English
+list, which is what the live step is for.
+
+## One thing this run confirmed by accident
+
+The `claim_lockout` event raised by the earlier claim-page probe arrived at the **operator** address
+in **Hungarian** („Túl sok hibás beállító kód — a beállító oldal 15 percre zárolva"). That is
+correct and deliberate: the operator tier is Hungarian by design (10-localisation.md ruling 1), and
+the customer-facing half of that same event is `mail.event.claim_lockout`, which the hub has carried
+in both languages since 0.118.0. The controller's sentence is **wire copy**, not customer copy —
+which is why R-596 left `claim.go` L563 alone.