board: population-law cross-check — #1311 witness vs real-data fold experiment

[AdaWorldAPI]

Analysis plus one test-only probe. There is no production code change.

What this does

Compares two independent witnesses and finds the execution law both of them support:

• merged frontend parity: Gremlin and SQL meet at the same Quack Query (test-only witness + parity matrix) #1311: the Gremlin/SQL frontend parity witness;
• an independent real-data population-fold experiment over academic_20k.csv (20,845 rows). Its headline counts were re-derived from the CSV.

Files

• .claude/plans/population-law-crosscheck-v1.md: the report, sections A–N. Each claim is graded MEASURED, STRUCTURAL, VERIFIED-IN-CODE, ANALYZED or UNPROVEN.
• crates/lance-graph-quack/tests/result_operand_probe.rs: the test-only probe, 4 tests (new).
• .claude/plans/frontend-parity-witness-v1.md: correction notes on three frontend parity: Gremlin and SQL meet at the same Quack Query (test-only witness + parity matrix) #1311 conclusions.
• Board files: the entry, the D-PLX rows, INTEGRATION_PLANS, and the regenerated indexes.

Main results

• Law. Execute over the population that already carries the multiplicity. Reach other data by functional reads. Switch anchors instead of building target populations. A phase boundary exists only when a later pass needs, per row, a completed fold over several rows of an earlier pass.
• frontend parity: Gremlin and SQL meet at the same Quack Query (test-only witness + parity matrix) #1311 revised. A second fan-out is not a barrier by itself: re-anchoring costs read depth (G1). The real barrier is a carried fan-in aggregate.
• Probe (MEASURED): OUTCOME A.
  • Phase 1 counts posted lines per partner into the normal Count sink.
  • A checked, test-only copy moves that sink into a u32 lane.
  • Phase 2 reads it per line through partner_id, using both the existing EqU32Via and GroupKey::Via.
  • Both match a host oracle. Phase 2 allocated 0 bytes, and the extra state is O(K), not O(N).
  • Three disable runs each turned the test red.
  • No result handle and no key metadata are needed for this consumer.
• Remaining gap: scalar width (B). The Count fold writes i64 while the *_via readers take u32 tables. Removing the copy is queued as D-PLX-2, a u32 Count sink. An i64 lane is not implied.
• Value seam vs key seam. Reusing values needs only that slot j of the result means the fk's target row j. Refolding by a component of the result's own key (for example Pair.hi) needs the producing key's geometry. lance-graph-report's CellSpace keeps that geometry, host-side.
• Corrected after review. The I→S presence fold over a bounded domain is a missing per-group OR fold state, not a phase dependency.
• The five frontend parity: Gremlin and SQL meet at the same Quack Query (test-only witness + parity matrix) #1311 gaps. G1 is a confirmed general gap. G2 splits. G3 and G5 are simple missing operators. G4 was a mis-specified witness.

Verification

• cargo test -p lance-graph-quack --test result_operand_probe: 4 passed.
• clippy -D warnings and fmt are clean.
• Disable runs: removing the via predicate, reading the key locally, and altering one oracle count each turned the test red.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Eo5eHGADmeKNE3xYhgdK76

---

Generated by Claude Code

[coderabbitai]

Navigate logical layers of code changes, visualize relationships, and explore their blast radius.

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: 4957a554-d523-4589-aa4a-53dea758c9db

📥 Commits

Reviewing files that changed from the base of the PR and between 2f2b67c and a325583.

📒 Files selected for processing (6)

• .claude/board/INTEGRATION_PLANS.md
• .claude/board/STATUS_BOARD.md
• .claude/board/entries/2026-10-03-population-law-crosscheck.md
• .claude/board/entries/README.md
• .claude/plans/frontend-parity-witness-v1.md
• .claude/plans/population-law-crosscheck-v1.md

 _____________________________________________________________________________________________________________________
< Use a project glossary. Create and maintain a single source of all the specific terms and vocabulary for a project. >
 ---------------------------------------------------------------------------------------------------------------------
  \
   \   (\__/)
       (•ㅅ•)
       / 　 づ

• 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.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a325583450

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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[coderabbitai]

Autopilot could not be updated. Open Coding to check access and billing.

[AdaWorldAPI]

This PR merged at its first commit, a325583. The commits named in my two review replies (2ab6221, ac49bd0) were pushed after the merge, so they never reached main. They are rebased into #1313, together with a wording correction ("a K-sized group domain" instead of "a population over its own coordinate space"; the exposure mechanism is stated as open).

---

Generated by Claude Code