D-HOLO H6c results; H14 pre-registered (full-row writes vs connected lane deltas)

[AdaWorldAPI]

Two things in .claude/plans/d-holo-address-indexed-algebra-v1.md: the H6c results, and a new pre-registration, H14.

H6c (measured 2026-10-14, v4)

Probe: holo_tile_crossover_probe (lance-graph-mask-risc example); 5 tests against the row oracle.

Question	Result, three runs	Verdict
Q1: P/D per query crosses 1 between g = 4 and 16 in both layouts	crosses at g = 8–12	HOLDS
Q2: a group-count switch at most 1.25× D everywhere	G = 8 worst query 1.21–2.28× D; byte switch 9.8–15.1×	not shown: the max-over-fixtures rule is too noise-sensitive
Q3: built from a declaration, P or H at most 0.5× D at N = 100	0.32–0.42×	PASS

The terminal selector stays held.

H14 (pre-registered, not run)

For one classid × thought × 64k-row tile, a dirty row is written in one of three ways:

• A: the full 512-byte SoA row.
• B: one record per dirty (thought, row) holding the row key, the lane mask and the changed lanes.
• C: a per-lane table keyed by row. C is a control, kept only to measure what it costs to re-attach lanes to their rows.

Two levels are kept separate: a session holds thoughts (one thought mask per session), and a thought holds lanes (one lane mask per thought and row). A lane is never stored apart from its row.

Rule: B becomes the default only if, at k ≤ 4 changed lanes, its read and session read stay within 1.25× of A and its bytes are at most half of A's. Otherwise A is the default. C is reported, never chosen.

The plan also fixes the grid, the metrics (v4 only, nanoseconds per row) and the disable runs before any probe exists. STATUS_BOARD has a new D-HOLO-H14 row.

Rebased onto main. LATEST_STATE and PR_ARC_INVENTORY were merged by keeping both sides. The entries index was regenerated (--check passes) and SUPERSESSION-INDEX is unchanged.

🤖 Generated with Claude Code

https://claude.ai/code/session_011BMJnDBAfzcuQh5HeT39Rh

---

Generated by Claude Code

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 86f2dc32-671c-4890-b3c9-95ce1ea16fe1

• Autofix · 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.

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[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 2026-10-15T21:20:43.394435Z	96d0837	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.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 96d0837283

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]

Add Carrot credits or activate Agent usage billing to use Autofix