Cycle 8 mix freeze one-shot confirmation 0/192

CHANGELOG.md

@@ -21,6 +21,17 @@
 - `source_verified_release_transform_registry` keeps verified visible-preserving rules instead of dropping them after promotion checks.
 - `BayesianDetectorEvidence.confirmatory_score()` refuses UNVERIFIED status. `raw_score` stays the frozen v1 field.
 - `load_bayesian_checkpoint` rejects JSON larger than 64 MiB.
+- Seed `880000` is `PUBLICLY_EXPOSED` by closed unmerged PR #98 and is not eligible as unseen validation. Added `global-seed-ledger-v1` and reserved Cycle 8 scale seeds `930000` / `940000` / `950000` before scale generation.
+- Cycle 8 scale exploratory seed `930000` U+034F x1: 0/16 then 0/32 raw transformed WM, then **1/64** on the 64-pair expansion (matched UW 0, visible projection pass, sanitizer matrix matches raw). Do not rewrite 1/64 as zero. Independent replication seed `940000` n=64 is **0/64** raw transformed WM (max score 0.557052 versus threshold 0.557099). Combined 64-pair corpora: **1/128**. Product registry fail-closes carrier sites that would violate frozen hard invariants.
+- U+034F is still not product-authorized. Do not generate `950000`. Do not inspect `830000` / `840000` / `850000`.
+- Reserved Cycle 8 density exploratory seed `960000` (`carrier density follow-up`) before generation, then scored a detector-blind U+034F space x1 versus space plus word-final letter x1 comparison. Both arms are **1/16** raw transformed WM on the same residual row. Density did not beat space x1. Not product-authorized.
+- Cycle 8 letter-x1 (U+034F after ASCII letters, visible-word invariants, quote-interior carrier policy, detector-blind selected-site cap 192) diagnostically rescored seen `930000` n=64, `940000` n=64, and `960000` n=16 to **0** raw transformed WM with clean unwatermarked controls and exact visible projection. Independent reserved seed `970000` (`intra-word carrier follow-up`) is **0/16** then **0/64**. Experimental letter-x1 **0/192** = seen `930000` n=64 plus seen `940000` n=64 plus independent `970000` n=64 (**128/192 seen**, **64/192 independent**). Space-x1 1/64 and 1/16 remain frozen.
+- Cycle 8 letter-x1 system benchmark reserved seeds `980000` and `990000` before generation. Fresh independent corpora: letter-x1 **0/128** raw transformed WM, space-x1 **1/128** on the same texts, letter max score 0.554066 versus threshold 0.557099. Local fixtures: visible projection 21/21, protected spans 21/21, Chromium `pre`/textarea/contenteditable pixel-equal on measured samples. Formal confirmation readiness is `NOT_READY`. Letter-x1 is not product-authorized and is not confirmation. Do not generate `950000`.
+- Cycle 8 letter-space margin seeds `1000000` and `1010000` were reserved before generation. Combined fresh letter-space is **1/128** on the same residual geometry; do not rewrite that as zero.
+- Cycle 8 letter-mix margin reserved seeds `1020000` and `1030000` before generation. Detector-blind U+034F/U+FE00 alternating letter mix is fresh **0/128** raw transformed WM, mix UW **0/128**, worst mix score 0.513691 versus threshold 0.557099 (gap 0.043407). Letter-x1 on those same corpora is 0/128 with max 0.527389. Chromium mix fixtures were pixel-equal on the measured `pre`/textarea/contenteditable surfaces. Mix is not product-authorized and is not confirmation. Formal confirmation readiness remains `NOT_READY`.
+- Cycle 8 letter-mix scale reserved seeds `1040000` and `1050000` before generation. Combined with `1020000`+`1030000`, mix is **0/256** raw transformed WM, mix UW **0/256**, worst mix score 0.519522 (gap 0.037577). The two-corpus 0/128 scorecard is not rewritten. This 0/256 is development scale, not confirmation.
+- Froze development mechanism `u034f-ufe00-letter-alt-v1` as `cycle8-mix-freeze-v1`. Public CLI remains empty.
+- Cycle 8 mix freeze one-shot confirmation on preregistered seeds `830000` / `840000` / `850000`: mix **0/192** raw transformed WM, mix UW **0/192**, visible **192/192**, frozen sanitizers matching raw zeros, worst mix score 0.524300 versus threshold 0.557099 (gap 0.032798). Those seeds are spent. Do not rerun looking for zero. Mix is not product-authorized. Do not generate `950000`.
 
 ## v0.2.0 — Exact-survival confirmation and release hardening
 

README.md

@@ -15,6 +15,16 @@ FuckMark is a deterministic, detector-blind research harness and CLI for disrupt
 
 The public CLI currently fail-closes: no product-authorized invisible carrier has been promoted yet, so `FuckMark` returns the original text unchanged rather than applying contractions or other visible edits. That is intentional. Historical visible-edit research remains replayable through explicitly named historical registries.
 
+Cycle 8 U+034F space-carrier x1 is a `PROMISING_DEVELOPMENT` / `HYPOTHESIS`. Tiny 4-pair corpora are 0/12 transformed WM. Scale exploratory seed `930000` is 0/16, 0/32, then **1/64** raw transformed WM. Independent replication seed `940000` is **0/64**. Combined large-N on those two 64-pair corpora is **1/128**. Density seed `960000` n=16 compared space x1 with space plus word-final letter x1: both **1/16** on the same residual row. Do not rewrite those space-x1 residuals as zero.
+
+Cycle 8 U+034F letter-x1 (after ASCII letters, visible-word invariants, quote-interior carriers, detector-blind cap 192) is a stronger `PROMISING_DEVELOPMENT` / `HYPOTHESIS` on the same visible contract. Diagnostic rescore of seen corpora: `930000` n=64 **0/64**, `940000` n=64 **0/64**, `960000` n=16 **0/16**. Independent reserved seed `970000` is **0/16** then **0/64**. Experimental letter-x1 **0/192** = those two seen 64-pair diagnostics plus independent `970000` n=64 (**128/192 seen**, **64/192 independent**).
+
+A later system benchmark reserved seeds `980000` and `990000` before generation and scored letter-x1 versus space-x1 on two new independent 64-pair corpora: letter-x1 **0/128** raw transformed WM (max 0.554066, gap 0.003032 below threshold 0.557099) and space-x1 **1/128** on the same corpora. Visible projection passed on every scored letter row and on 21/21 local fixtures.
+
+Letter-space margin seeds `1000000` and `1010000` were reserved before generation. Combined fresh letter-space is **1/128**; do not rewrite that as zero.
+
+Letter-mix margin seeds `1020000` and `1030000` were reserved before generation. Detector-blind U+034F/U+FE00 letter-alt is fresh **0/128** raw transformed WM (max 0.513691, gap 0.043407) with mix UW **0/128**. Independent scale seeds `1040000` and `1050000` were reserved before generation; combined mix is **0/256** (worst max 0.519522, gap 0.037577). The two-corpus 0/128 scorecard is not rewritten. Letter-x1 on those same corpora is also 0/256 (worst max 0.543903). Chromium mix fixtures were pixel-equal on the measured surfaces. The mix mechanism is frozen as `cycle8-mix-freeze-v1`. One-shot confirmation on seeds `830000` / `840000` / `850000` is **0/192** raw transformed WM, mix UW **0/192**, visible **192/192**, worst max 0.524300 (gap 0.032798). Those seeds are spent. It is still not product-authorized. Seed `880000` is publicly exposed by closed unmerged PR #98 and is not eligible as unseen validation. See [`docs/cycle8/mix-freeze.md`](docs/cycle8/mix-freeze.md), [`docs/cycle8/letter-mix-margin.md`](docs/cycle8/letter-mix-margin.md), [`docs/cycle8/letter-system-benchmark.md`](docs/cycle8/letter-system-benchmark.md), and [`docs/seeds.md`](docs/seeds.md).
+
 See [`docs/product-contract.md`](docs/product-contract.md) and `specs/fuckmark-user-visible-invariance-v1.contract.json`.
 
 **Current release: v0.3.0**  
@@ -134,7 +144,7 @@ Cycle 7 Stage A/B/C researched collapse-resistant *visible* edits (contractions,
 
 ## Cycle 8 status
 
-Cycle 8 is the exact-visible-projection generation. See [`docs/cycle8/README.md`](docs/cycle8/README.md). U+200C remains a diagnostic baseline (stripped by Cf). Combining grapheme joiner U+034F is an early durable-track *hypothesis* under frozen sanitizers; it is not a release mechanism.
+Cycle 8 is the exact-visible-projection generation. See [`docs/cycle8/README.md`](docs/cycle8/README.md). U+200C remains a diagnostic baseline (stripped by Cf). Combining grapheme joiner U+034F letter-x1 remains a strong visible-invariant development arm (`HYPOTHESIS`): fresh reserved-before-generation **0/128** on seeds `980000`+`990000` with a thin 0.003 gap, plus a separate experimental **0/192** (128 seen + 64 independent). The later U+034F/U+FE00 letter-mix arm is currently stronger on fresh independent corpora: **0/128** on seeds `1020000`+`1030000` with worst score 0.513691 (gap 0.043407), plus independent scale **0/256** including `1040000`+`1050000` with worst score 0.519522 (gap 0.037577). Mix is frozen as `cycle8-mix-freeze-v1` and is not a release mechanism. One-shot confirmation is **0/192**. Public CLI remains empty.
 
 ## Reproduce the research environment
 

docs/cycle7/README.md

@@ -35,7 +35,7 @@ See `docs/cycle7/seed-ledger.md` and `fuckmark/cycle7/ledger.py`.
 | 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` |
 | 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 C, reserved) | `880000` | **PUBLICLY_EXPOSED** by closed unmerged PR #98; not eligible as unseen validation |
 | Confirmation reserved | `830000`, `840000`, `850000` | must not be inspected |
 
 Confirmation seeds were chosen as unused 10k blocks before any Cycle 7 detector look.
@@ -90,11 +90,11 @@ Details: `docs/cycle7/stage-a-decision.md`.
 
 Stage B1 on seed `860000` looked **`PROMISING_DEVELOPMENT`** on a tiny detector snapshot (durable 2/4 after collapse). Disjoint validation seed `820000` reproduced density/geometry but **not** the detector reduction (durable 4/4 after collapse). Overall Stage B decision: **`INSUFFICIENT_EVIDENCE`**. Details: `docs/cycle7/stage-b-decision.md`.
 
-Seed `820000` is spent as validation. Seed `880000` is the next unused validation split and has not been inspected. Seeds `830000` / `840000` / `850000` remain unseen confirmation reserves.
+Seed `820000` is spent as validation. Seed `880000` is **PUBLICLY_EXPOSED** by closed unmerged PR #98 and is not eligible as unseen validation. Seeds `830000` / `840000` / `850000` were unseen during Cycle 7 and were later spent once as Cycle 8 `cycle8-mix-freeze-v1` confirmation.
 
 ## Stage C
 
-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 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 by Stage C on `main`. Closed unmerged PR #98 later exposed it. Details: `docs/cycle7/stage-c-decision.md`.
 
 Not claimed:
 

docs/cycle7/seed-ledger.md

@@ -1,6 +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`).
+
+The v3 reservation of seed `880000` as unseen Stage C validation remains historically true. Closed unmerged [PR #98](https://github.com/byte271/FuckMark/pull/98) later publicly generated and scored `880000`. Current product/research status: **PUBLICLY_EXPOSED**, not eligible as unseen validation. Do not merge PR #98. Do not copy its newline mechanism onto the product path.
 
 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.
 
@@ -21,8 +23,8 @@ Seeds were assigned **before** Cycle 7 detector scoring. Seed `860000` and topic
 | 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` |
 | 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 |
-| Confirmation reserved | 830000, 840000, 850000 | **no** until a later frozen confirmation protocol |
+| Validation development (Stage C, reserved) | 880000 | **PUBLICLY_EXPOSED** by closed unmerged PR #98; not eligible as unseen validation |
+| Confirmation reserved | 830000, 840000, 850000 | spent by Cycle 8 mix freeze confirmation; do not retune |
 
 Do not promote 810000, 820000, 860000, 870000, or 880000 into confirmation after seeing scores.
 

docs/cycle7/stage-c-decision.md

@@ -4,6 +4,8 @@
 
 This is development-only. It is not a Cycle 7 formal confirmation. Seeds `830000` / `840000` / `850000` were not inspected. Seed `880000` remains unused validation and was **not** generated.
 
+Current status after this writeup: closed unmerged PR #98 later publicly generated seed `880000`. That exposure does not change the Stage C1 numbers below. Seed `880000` is no longer eligible as unseen validation.
+
 Catalog: `cycle7-durable-rule-catalog-v4`. Scheduler: unchanged cover-greedy v4. Detector threshold: frozen Cycle 6 value `0.5570987654320988`. Model: `openai-community/gpt2` revision `607a30d783dfa663caf39e06633721c8d4cfcd7e`.
 
 v4 adds Family 10 (clause punctuation newline) and Family 11 (optional quantifier `of` before a closed determiner/possessive). Families 1–9 are unchanged from catalog v3.

docs/cycle8/README.md

@@ -17,11 +17,11 @@ Cycle 8 is a new scientific generation. The mechanism class is hidden Unicode re
 
 ## Seed ledger
 
-See `docs/cycle8/seed-ledger.md`. Seeds `890000`, `900000`, `910000`, and `920000` were assigned before Cycle 8 generation. Do not inspect `830000` / `840000` / `850000`. Do not inspect Cycle 7 reserved `880000`.
+See `docs/cycle8/seed-ledger.md` and `docs/seeds.md`. Seeds `890000`, `900000`, and `910000` are spent tiny Cycle 8 evidence. Scale seeds `930000` / `940000` / `950000` were reserved before scale generation. Density seed `960000` was reserved before density generation and scored n=16: both space x1 and space-wordfinal are 1/16. Letter seed `970000` was reserved before letter generation; independent letter-x1 is 0/16 then 0/64. Diagnostic letter rescored of seen `930000` n=64 and `940000` n=64 are 0/64. Experimental letter-x1 0/192 is 128 seen plus 64 independent. System-benchmark seeds `980000` and `990000` were reserved before generation; fresh letter-x1 is 0/128 and fresh space-x1 is 1/128. See `docs/cycle8/letter-system-benchmark.md`. Mix seeds `1020000` and `1030000` were reserved before generation; fresh letter-alt is 0/128 with worst score 0.513691. Mix scale seeds `1040000` and `1050000` were reserved before generation; combined mix is 0/256 with worst score 0.519522. Mix is frozen as `cycle8-mix-freeze-v1`. One-shot confirmation on seeds `830000` / `840000` / `850000` is **0/192** and those seeds are spent. See `docs/cycle8/letter-mix-margin.md` and `docs/cycle8/mix-freeze.md`. Letter-space seeds `1000000` and `1010000` remain 1/128. Seed `880000` is **PUBLICLY_EXPOSED** by closed unmerged PR #98 and is not eligible as unseen validation.
 
 ## Carrier research
 
-See `docs/cycle8/carrier-research.md`.
+See `docs/cycle8/carrier-research.md`, `docs/cycle8/letter-mix-margin.md`, and `docs/cycle8/mix-freeze.md`.
 
 U+200C is the diagnostic baseline. It is not durable under Cf-strip.