D-PUZZLE-0 step 2: crosswords on the same population algebra as Sudoku

[AdaWorldAPI]

Step 2 of D-PUZZLE-0: does a second domain run on the population algebra from step 1 (#1382) without changing it?

How "same algebra" is checked

• 4ed5e4e: the domain-free half moves out of the Sudoku probe into cognitive-shader-driver/examples/shared/population_fold.rs. That half is the propagation reading (given / forced / entailed-not-forced / candidate), the five questions as facts_population masks, the filter / histogram / decode folds, the group fold and the declaration gate. A fence test fails if domain vocabulary (sudoku, crossword, chess, digit, slot, cell, letter, grid, board, move) appears in it. The Sudoku probe's counts are identical before and after the move (1,000,066 edges).
• 0ac5d6b: crossword_population_fold_probe.rs holds only the domain. It uses a 5×5 grid with ten crossing slots, generated fills over a 3-letter alphabet, 10 distractor words per length, and two given slots. The law is the forced single, and every instance has exactly one solution (backtracking). Questions, folds and oracle are unchanged from the shared module.
• cc887e3, d182301 (step 2b): crossword_real_words_probe.rs fills the same template with real words: every 4- and 5-letter word of the vocabulary.
  • English is exactly what DeepNSM reads: v1's 4096-word table (word_rank_lookup.csv, ranks 1..=4096) united with v2's academic_20k.csv, both committed. cc887e3 used the top 1000 word_forms.csv surfaces; d182301 replaces that, because DeepNSM does not use that list as a vocabulary.
  • German is the 20,000 most frequent DeReKo-2014 forms, read only when DEREKO_PATH names the file at run time. DeReKo is CC BY-NC 3.0, so no German word or derived artifact is committed, and the German run is skipped without the variable.
  • Given slots are added in random order until the fill is unique.

Measured, synthetic alphabet (1,000,052 edges, 22,455 instances)

question	edges
every asserted claim	224,550
given	44,910
forced, chain known	13,745
entailed, not yet forced	165,895
live candidates	775,502

Every count is equal three ways (filter, histogram, per-edge decode) and to the instances' own counters. Every instance holds exactly 10 asserted claims.

path	crossword ns/edge	Sudoku ns/edge
one population, filter	0.30 – 0.43	0.35 – 0.39
32-bin histogram	1.67 – 1.75	0.93 – 1.07
per-edge decode (oracle)	3.9 – 4.1	3.8 – 4.0

Release build, avx2, the same 4-core shared Xeon, median of 7, three runs; one machine's numbers. The filter and the oracle cost the same in both domains. The histogram is slower on the crossword lane, where 77 % of edges share one code; repeated increments of one bin are the likely cause, not measured.

What differs is the law, not the algebra. Forcing is rare in crosswords (13,745 of 177,640 non-given slots) against Sudoku's naked singles, so most true words stay IndirectUnknown × Causes.

Measured, real words (one run each)

	English (v1 4096 + academic)	German (DeReKo top 20k)
dictionary, 4 / 5 letters	1,250 / 1,697	802 / 1,466
edges / instances	1,005,146 / 1,209	1,000,074 / 1,390
build	7.4 s	6.7 s
given / forced / entailed	5,370 / 1,278 / 5,442	6,118 / 1,400 / 6,382
filter / histogram / oracle, ns/edge	0.35 / 2.23 / 4.01	0.36 / 2.10 / 3.79

Counts are equal three ways in both languages.

Falsifiers (disable-verified red)

• A domain word planted in the shared half → fence test red.
• A uniqueness check that always says "one" → its test red (both crossword probes).
• A law that forces any slot with a candidate → "forced word is the true word" fails, 3 tests red.
• A bitset index that ignores fixed letters → 4 tests red, including the index-against-scan test.
• NE rows kept in the DeReKo parser → the parser test red.
• The v1 4096 rank cut removed, or the academic list dropped → the English vocabulary test red.
• The German 20k cut removed → its test red.

All three probes are registered as [[example]] test = true, and clippy -D warnings is clean.

Open

• CI covers English only; German needs the DeReKo file.
• Chess goes to stockfish-rs (GPL; shakmaty supplies legality). How it reaches the shared half is undecided: copying it there makes a second copy, and promoting it into lance-graph-contract is a contract change.

Entries: .claude/board/entries/2026-10-07-puzzle-0-crossword-same-algebra.md, 2026-10-07-puzzle-0-crossword-real-words.md.

🤖 Generated with Claude Code

https://claude.ai/code/session_0177nSSyQaXqbNuGR1vMXXN6

[coderabbitai]

Note

Currently processing new changes in this PR. This may take a few minutes, please wait...

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Essentials
• Run ID: ebea73d6-b1a7-4402-8126-873e119b34be

📥 Commits

Reviewing files that changed from the base of the PR and between bccc46e and 94499c3.

📒 Files selected for processing (10)

• .claude/board/STATUS_BOARD.md
• .claude/board/entries/2026-10-07-puzzle-0-crossword-real-words.md
• .claude/board/entries/2026-10-07-puzzle-0-crossword-same-algebra.md
• .claude/board/entries/README.md
• crates/cognitive-shader-driver/Cargo.toml
• crates/cognitive-shader-driver/examples/README.md
• crates/cognitive-shader-driver/examples/crossword_population_fold_probe.rs
• crates/cognitive-shader-driver/examples/crossword_real_words_probe.rs
• crates/cognitive-shader-driver/examples/shared/population_fold.rs
• crates/cognitive-shader-driver/examples/sudoku_population_fold_probe.rs

 _______________________________________________________________________________________________________________________________
< Always code as if the person who ends up maintaining your code is a violent psychopath who knows where you live. - John Woods >
 -------------------------------------------------------------------------------------------------------------------------------
  \
   \   (\__/)
       (•ㅅ•)
       / 　 づ

✨ Finishing Touches

📝 Generate docstrings

• Commit to this branch
• Create a new PR

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Warning

Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption.

---

_{Comment @coderabbitai help to get the list of available commands.}

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: 66ddcd5b-f513-42ed-9c4f-39361ec033a1)

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: 807876d8-e7d7-4cdb-9281-03cc238046ab)

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: 5f403cb7-4459-4494-ac24-8ec1927c386b)

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: ef007150-4652-4b0d-8ab3-034ab278c06c)

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: b44551db-e07a-42f3-b353-04e37a990d22)

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-07T16:49:06.579015Z	94499c3	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: f63e9f65-9565-46ab-8ce6-7d952c21d878)