Cycle 7 Stage D: word-boundary newline catalog v5

CHANGELOG.md

@@ -11,6 +11,7 @@
 - Cycle 8 fixture compare reports visible-projection pass `20/20` for space-carrier arms.
 - Chromium `pre` screenshots: U+034F / U+FE00 pixel-equal; U+200C PNG bytes differ.
 - Detector-blind GPT-2 / SynthID development scoring on seeds `890000`, `900000`, and `910000` reduced watermarked detections for U+034F space-carrier x1 without unwatermarked inflation on these tiny 4-pair corpora. This is not confirmation and is not a product promotion.
+- Cycle 7 Stage D catalog v5 (word-boundary newline) is research-only `PROMISING_DEVELOPMENT` on n=4+4 and remains product-disqualified. Family 12 is not a product carrier.
 - Product contract loading uses the embedded v1 payload in installed wheels. Writers that replace pinned `specs/cycle8` tokenizer JSON refuse to run without a GPT-2 encoder; the detector harness still scores when tiktoken is absent. Chromium HTML evidence encodes less-than and greater-than as JSON unicode escapes so an ASCII script terminator cannot close the inline script.
 
 ## v0.2.0 — Exact-survival confirmation and release hardening

docs/cycle7/README.md

@@ -6,12 +6,13 @@ Cycle 7 researched collapse-resistant *visible* transforms. Those edits may pres
 
 Cycle 8 (`docs/cycle8/`) is the exact-visible-projection generation. Do not promote Cycle 7 catalogs into `release_transform_registry()`.
 
-## Live state (verified 2026-08-25)
+## Live state (verified 2026-08-26)
 
 See `docs/cycle7/live-state-audit.md`. Independent GitHub inspection:
 
-- `main` HEAD `af341884fff211caeb24649a6ce79305c3491c6c` (PR #95 squash-merge)
-- Cycle 6 freeze parent `ddccd74a9e0e710085b385fed98959f2730b9d60`
+- `main` HEAD `3ab6bd1f98077128a5b8c83ee1f30f89f58d6e77` (PR #97 squash-merge: product-visible invariance / Cycle 8)
+- Cycle 7 Stage B/C ancestor `d7eeb0f905b7046ddc4a4d8281354f06713d58b0` (PR #96)
+- Cycle 6 freeze ancestor `ddccd74a9e0e710085b385fed98959f2730b9d60`
 - PR #93 and PR #94 merged
 - Recovery run `32886342498` success
 - Aggregate hash `30577aafaffd0c50f0ddb384a4509eb0bb93e4374bf39704869ddbf5053186a4`
@@ -33,9 +34,10 @@ See `docs/cycle7/seed-ledger.md` and `fuckmark/cycle7/ledger.py`.
 | --- | --- | --- |
 | Exploratory development (Stage A) | `810000` | used; do not keep expanding rules against it |
 | Exploratory development / rule construction (Stage B1) | `860000` | used; do not keep expanding rules against it |
-| Exploratory development / rule construction (Stage C1) | `870000` | active; topic frozen as `measurement protocol` |
+| Exploratory development / rule construction (Stage C1) | `870000` | used; do not keep expanding rules against it |
+| Exploratory development / rule construction (Stage D1) | `890000` | used; do not keep expanding rules against it |
 | Validation development (Stage B3) | `820000` | used as Stage B3 disjoint validation; do not retune on it |
-| Validation development (Stage C, reserved) | `880000` | unseen until a v4 freeze |
+| Validation development (Stage D) | `880000` | used as Stage D disjoint validation of catalog v5; do not retune on it |
 | Confirmation reserved | `830000`, `840000`, `850000` | must not be inspected |
 
 Confirmation seeds were chosen as unused 10k blocks before any Cycle 7 detector look.
@@ -52,7 +54,7 @@ The Cycle 6 frozen sanitizer module is unchanged.
 
 ## Durable transform families
 
-See `docs/cycle7/durable-families.md`. Catalog `cycle7-durable-rule-catalog-v4`.
+See `docs/cycle7/durable-families.md`. Catalog `cycle7-durable-rule-catalog-v5`.
 
 Stage A (`v2`) families:
 
@@ -74,7 +76,11 @@ Stage C (`v4`) additions, aimed at higher natural site density after Stage B sho
 10. Clause punctuation newline (`,` / `;` / `:` + space ↔ mark + LF).
 11. Optional `of` after `all` / `both` / `half` before a closed determiner/possessive.
 
-These change tokens without relying on repeated U+0020. They survive ASCII whitespace collapse by construction. Family 3 also survives NFKC and Cf-strip. Family 4 is a formatting channel: VERIFIED against the six Cycle 7 sanitizer variants, not against an unlisted reflow sanitizer.
+Stage D (`v5`) addition, aimed at a collapse-resistant channel denser than clause punctuation rather than another lexical whitelist:
+
+12. Word-boundary newline (ASCII space between two alphabetic words of length ≥2 ↔ LF).
+
+These change tokens without relying on repeated U+0020. They survive ASCII whitespace collapse by construction. Family 3 also survives NFKC and Cf-strip. Families 4, 10, and 12 are formatting channels: VERIFIED against the six Cycle 7 sanitizer variants, not against an unlisted reflow sanitizer.
 
 Quote interiors may receive these durable edits under `quote-container-durable-v1` without changing quote delimiters. Cycle 6 `quote-container-surface-spacing-v1` still admits only spacing inside quotes.
 
@@ -96,6 +102,12 @@ Seed `820000` is spent as validation. Seed `880000` is the next unused validatio
 
 Stage C1 on seed `870000` (topic frozen as `measurement protocol`) added clause-punctuation newlines and optional quantifier `of`. Mean durable candidates: **4.75** (still the Stage B four-site regime). Quantifier `of` mean **0.00**. Collapse-surviving intact/root **0.715**. Detector: durable **3/4** after collapse; Cycle 6 spacing **4/4** after collapse. Overall: **`INSUFFICIENT_EVIDENCE`**. Seed `880000` was not inspected. Details: `docs/cycle7/stage-c-decision.md`.
 
+## Stage D
+
+Stage D1 on seed `890000` (topic frozen as `document structure`) added word-boundary newlines. Mean durable candidates: **38.0**. Collapse-surviving intact/root **0.164**. Detector: durable **0/4** after collapse; Cycle 6 spacing **3/4** after collapse. Cycle 8 later used the same seed base with topic `invisible carrier development` and a disjoint corpus; Stage D evidence is the `document structure` texts only.
+
+Disjoint validation seed `880000` (topic frozen as `independent check`) replicated: mean **44.875**, collapsed intact/root **0.080**, durable **0/4** after collapse, Cycle 6 spacing **4/4** after collapse. Overall: **`PROMISING_DEVELOPMENT`**. Not formal confirmation. Catalog v5 remains **PRODUCT_DISQUALIFIED** for the CLI. Details: `docs/cycle7/stage-d-decision.md`.
+
 Not claimed:
 
 - Cycle 7 formal confirmation

docs/cycle7/durable-families.md

@@ -1,6 +1,6 @@
 # Cycle 7 durable families
 
-Catalog identity: `cycle7-durable-rule-catalog-v4` (Stage C development). Families 1–3 remain the frozen Stage A `v2` set; families 4–9 remain the frozen Stage B `v3` set. Cycle 6 `quote_safe_zrd` ruleset hash is unchanged.
+Catalog identity: `cycle7-durable-rule-catalog-v5` (Stage D development). Families 1–3 remain the frozen Stage A `v2` set; families 4–9 remain the frozen Stage B `v3` set; families 10–11 remain the frozen Stage C `v4` set. Cycle 6 `quote_safe_zrd` ruleset hash is unchanged.
 
 All families below are deterministic, detector-blind, key-blind, and non-neural. Cycle 6 `quote_safe_zrd` ruleset hash is unchanged.
 
@@ -227,3 +227,27 @@ False positives blocked: `all of them` / `both of us` / `half of it` (pronouns r
 
 EXTERNAL-VALIDATION-ONLY: Cambridge Grammar of English / English Grammar Today: `all`, `both`, and `half` may appear with or without `of` before articles, demonstratives, and possessives, with no meaning change. Pronoun objects keep obligatory `of`. That literature is not a local TinyDev density measurement.
 
+## Family 12 — word-boundary newline
+
+Construction: `word_boundary_newline`.
+
+This is the Stage D density primitive. It is not another closed lexical whitelist. It is the collapse-resistant analogue of Cycle 6 general word spacing: replace a single ASCII space between two alphabetic words with a single LF, or the reverse.
+
+Precondition: both sides are ASCII alphabetic runs of length at least 2. Single-letter words (`a`, `I`), digits, punctuation-adjacent spaces (`. `, `, `, `: `), and already-newline punctuation sites do not match. Protected spans still block overlapping edits.
+
+Rewrite: `The protocol` ↔ `The\nprotocol`.
+
+Inverse: the opposite direction.
+
+Tokenizer: GPT-2 `The protocol remains fixed` versus `The\nprotocol remains fixed` differs. VERIFIED on the frozen GPT-2 revision when transformers is present.
+
+Whitespace collapse: `whitespace-collapse-v1` preserves LF. The rewrite still differs after `ws_collapse` and `ws_collapse_nfkc_cf_strip`. VERIFIED on fixtures.
+
+Sanitizer note: this is a layout/formatting channel, denser than Families 4 and 10. HYPOTHESIS: a wrap/reflow sanitizer would erase it. It is admitted because Stages B and C showed punctuation-newline was the only durable family with repeated natural sites, and those sites were still too sparse (mean ~4–4.75) to replace Cycle 6 spacing after collapse.
+
+Conflict geometry: adjacent word-boundary spaces do not overlap (each candidate is the one-character separator). Cover-greedy v4 can spend the B14 budget on up to 14 wraps. Format punctuation newlines remain separate spans.
+
+Not admitted: extra U+0020 insertion, NBSP, ZWSP, wrapping after length-1 words, or converting punctuation spaces that Families 4 and 10 already own.
+
+Adjacent-sentence transposition remains a HYPOTHESIS for a later stage. It is not in catalog v5, so cover-greedy cannot spend the B14 budget on one long overlapping swap that would starve this dense channel.
+

docs/cycle7/literature.md

@@ -51,5 +51,7 @@ REJECTED: running-text `and` to `&` (not prose-equivalent under CMOS/AP). U+2026
 
 EXTERNAL-VALIDATION-ONLY. Cambridge English Grammar Today: `all`, `both`, and `half` allow optional `of` before articles, demonstratives, and possessives (`all the` / `all of the`) with no meaning change; `of` is obligatory before personal pronouns (`all of them`). Family 11 encodes that bound. That literature is not a local density measurement.
 
-HYPOTHESIS. Clause commas are much more common than sentence terminals on ordinary prose, so Family 10 should raise layout-channel density relative to Family 4 alone. Whether that is enough to replace Cycle 6 spacing after `whitespace-collapse-v1` is a Stage C measurement, not a theorem.
+HYPOTHESIS. Clause commas are much more common than sentence terminals on ordinary prose, so Family 10 should raise layout-channel density relative to Family 4 alone. Stage C on seed `870000` measured mean 1.50 clause sites and did not escape the four-site regime. VERIFIED on that seed only.
+
+HYPOTHESIS. Replacing ordinary word-boundary U+0020 with LF is the collapse-resistant analogue of Cycle 6 general word spacing. `whitespace-collapse-v1` preserves LF, so the token change remains after that sanitizer. VERIFIED on seeds `890000` and `880000` under catalog v5: mean wrap-site density 33.375 and 37.875, collapsed intact/root 0.164 and 0.080, durable detections 0/4 after collapse on both n=4 snapshots, Cycle 6 spacing 3/4 and 4/4 after collapse. That is development evidence, not 192-row confirmation. HYPOTHESIS: a wrap/reflow sanitizer would erase it.
 

docs/cycle7/live-state-audit.md

@@ -62,17 +62,27 @@ Live GitHub still has `main` at `ddccd74a9e0e710085b385fed98959f2730b9d60`. Open
 
 Hygiene-only follow-up `6e587ea016cf5a2f4e7845a5550f97e0b5d92b34` made Release Readiness Baseline and CI succeed after `tests/test_repository_hygiene.py` rejected comments/docstrings in new Python. Cycle 6 sealed confirmation on that SHA succeeded. Codex review was usage-limited, not a code finding. CodeRabbit did not auto-review (repository star threshold).
 
-## Continuation (2026-08-25, after PR #95 squash-merge)
+## Continuation (2026-08-25, after PR #96 squash-merge)
 
-PR #95 squash-merged at 2026-08-25 21:23:21 UTC after all 41 head checks succeeded and `mergeStateStatus` was `CLEAN`. No unresolved review threads.
+PR #96 squash-merged at 2026-08-25 21:26:26 UTC. Title remained `Cycle 7 Stage B: high-density durable catalog v3`; the branch also contained Stage C catalog v4 and C1 evidence.
 
-Verified `main` HEAD: `af341884fff211caeb24649a6ce79305c3491c6c`
+Verified `main` HEAD: `d7eeb0f905b7046ddc4a4d8281354f06713d58b0`
 
-Subject: `Cycle 7 Stage A: collapse-resistant durable transforms (#95)`
+Subject: `Cycle 7 Stage B: high-density durable catalog v3 (#96)`
 
-Parent freeze commit remains `ddccd74a9e0e710085b385fed98959f2730b9d60`. Cycle 6 formal result is unchanged: `NONZERO_RESIDUAL` 7/192.
+Cycle 6 formal result is unchanged: `NONZERO_RESIDUAL` 7/192.
 
-Open PR after the merge: [#96](https://github.com/byte271/FuckMark/pull/96) on `cursor/cycle7-stage-b-density-3fb1`, rebased onto `main` and retargeted. Open issues: none.
+Open PR after the merge: [#97](https://github.com/byte271/FuckMark/pull/97) (`cursor/product-visible-invariance-2d93`, draft, Cycle 8 / user-visible invariance). That work is separate from Cycle 7 catalog development.
 
-Changing the PR base from `research/cycle7-durable-transforms` to `main` does not emit `pull_request` `synchronize`, so Actions on the rebase SHA must be retriggered by a follow-up push. Do not treat a CodeRabbit-only success as full CI.
+Stage C1 on seed `870000` was `INSUFFICIENT_EVIDENCE` (mean durable candidates 4.75; durable detector 3/4 after collapse). Seed `880000` was not inspected.
+
+## Continuation (2026-08-26, after PR #97 squash-merge)
+
+Verified `main` HEAD: `3ab6bd1f98077128a5b8c83ee1f30f89f58d6e77`
+
+Subject: `Restore exact user-visible invariance and start Cycle 8 carrier research (#97)`
+
+Cycle 6 formal result is unchanged: `NONZERO_RESIDUAL` 7/192. Cycle 7 visible-edit catalogs are **PRODUCT_DISQUALIFIED** for the CLI.
+
+PR [#98](https://github.com/byte271/FuckMark/pull/98) (`cursor/cycle7-stage-d-wrap-3fb1`) was closed unmerged when #97 landed, then rebased onto this `main`. Cycle 8 exploratory seed `890000` / topic `invisible carrier development` and Cycle 7 Stage D seed `890000` / topic `document structure` share pair seeds but have disjoint text hashes. Cycle 7 Stage D validation used `880000`; Cycle 8 still must not inspect it. Confirmation seeds `830000` / `840000` / `850000` were not inspected.
 

docs/cycle7/seed-ledger.md

@@ -1,8 +1,8 @@
 # Cycle 7 seed ledger
 
-Ledger identity: `cycle7-seed-ledger-v3` (`fuckmark/cycle7/ledger.py`).
+Ledger identity: `cycle7-seed-ledger-v4` (`fuckmark/cycle7/ledger.py`).
 
-Seeds were assigned **before** Cycle 7 detector scoring. Seed `860000` and topic `independent replication` were frozen in the ledger **before** any Stage B generation or detector look. Seed `870000` and topic `measurement protocol` were frozen **before** any Stage C generation or detector look. Seed `880000` and topic `independent check` were frozen as the next unused validation split **before** any Stage C validation generation.
+Seeds were assigned **before** Cycle 7 detector scoring. Seed `860000` and topic `independent replication` were frozen in the ledger **before** any Stage B generation or detector look. Seed `870000` and topic `measurement protocol` were frozen **before** any Stage C generation or detector look. Seed `880000` and topic `independent check` were frozen as the next unused validation split **before** any Stage C validation generation. Seed `890000` and topic `document structure` were frozen **before** any Stage D generation or detector look.
 
 ## Spent (forbidden)
 
@@ -19,18 +19,25 @@ Seeds were assigned **before** Cycle 7 detector scoring. Seed `860000` and topic
 | --- | --- | --- |
 | Exploratory development (Stage A) | 810000 | yes, development only; do not keep expanding rules against it |
 | Exploratory development / rule construction (Stage B1) | 860000 | used; do not keep expanding rules against it |
-| Exploratory development / rule construction (Stage C1) | 870000 | yes, development only; frozen topic `measurement protocol` |
+| Exploratory development / rule construction (Stage C1) | 870000 | used; do not keep expanding rules against it |
+| Exploratory development / rule construction (Stage D1) | 890000 | used; do not keep expanding rules against it |
 | Validation development (Stage B3) | 820000 | used as Stage B3 disjoint validation; do not retune on it |
-| Validation development (Stage C, reserved) | 880000 | **not yet inspected**; reserved until catalog v4 is frozen |
+| Validation development (Stage C/D) | 880000 | used as Stage D disjoint validation of catalog v5; do not retune on it |
 | Confirmation reserved | 830000, 840000, 850000 | **no** until a later frozen confirmation protocol |
 
-Do not promote 810000, 820000, 860000, 870000, or 880000 into confirmation after seeing scores.
+Do not promote 810000, 820000, 860000, 870000, 880000, or 890000 into confirmation after seeing scores.
 
 Stage A generation used 810000 with pair stride 32, matching TinyDev pairing, one pair per domain, topic `reproducibility` chosen without a detector look.
 
 Stage B1 generation used 860000 with the same pairing rule and topic `independent replication`. Seed 810000 remains an exploratory seed for Stage A rescoring only.
 
 Validation generation used 820000 with topic `held-out evaluation`. That topic was frozen before validation generation. Do not retune catalog v3 on 820000. A later revised mechanism must not recycle 820000; Stage C reserved 880000 instead.
 
-Stage C1 generation uses 870000 with topic `measurement protocol`. Rule-construction generation must call `assert_rule_construction_seed(870000)`.
+Stage C1 generation used 870000 with topic `measurement protocol`. That seed is now used. Do not retune catalog v4 on 870000.
+
+Stage D1 generation used 890000 with topic `document structure`. Fresh rule-construction generation must call `assert_rule_construction_seed(890000)` and is now locked. Replay of the frozen D1 evidence via `--samples-from` uses exploratory-development admission so 890000 can be rescored. Do not retune catalog v5 on 890000.
+
+Cycle 8 independently assigned the same seed base `890000` with topic `invisible carrier development` and generated a different TinyDev corpus (same pair seeds `890000`/`890032`/`890064`/`890096`, disjoint text hashes). Stage D evidence is the `document structure` texts only. Do not retune Cycle 7 on Cycle 8 texts, or Cycle 8 on Stage D texts. Cycle 8 seeds `900000`, `910000`, and `920000` are not Cycle 7 data.
+
+Stage D validation used 880000 with topic `independent check`. Do not retune catalog v5 on 880000. Cycle 8 must still not inspect 880000. A later revised mechanism must reserve a new unused validation seed.