research: D-ORACLE-0 original-oracle fidelity and 64 auditable thoughts

[AdaWorldAPI]

Why

The research thread was drifting toward blending NAL, ONA, OpenPsi, Hyperon/MeTTa, Pearl projections, Shannon entropy and Gadamer admission into one un-auditable model. The operator's actual goal is original upstream implementations as independent executable oracles, with model-specific outputs preserved over 64 auditable thought alpha channels, and only later a mathematical/ISA normalization and end-to-end materialization comparison.

The DeepNSM close reading also conflated historically excluded scope (#1314) with the current main: belief/reason/tekamolo/hydrate modules are already present.

What this research-only PR adds

• .claude/plans/d-oracle-fidelity-matrix-v1.md: 144 proposed pre-registered fixtures covering OpenNARS Java, ONA C, OpenPsi/Rubikon, Hyperon Rust/MeTTa, independence/provenance, DeepNSM grammar, Pearl and symbolic methods/address algebra. The oracle-fidelity, audit, and H8/H9/H11 gates, explicit falsifiers and open architecture decisions.
• .claude/knowledge/oracle-fidelity-reference-ledger-v1.md: original repositories, code anchors and known unknowns; distinguishing current code from historic PR descriptions, measured from proposed.
• .claude/prompts/d-oracle-0-claude-handoff.md: read-first instructions for Claude Code, original-oracle version freezing, independent native fixtures, red-first reproduction and explicit stop conditions.

Critical unsolved falsifiers

• X17: derived beliefs may carry empty stamps while their premises contain existing evidence roots; source re-entry can incorrectly pool as independent. See board: coresearch map for evidence stance, dependence, reliability and pooling #1319 and current deepnsm-v2 belief.rs.
• CE64 near-unit confidence: known f32 weight-saturation overflow/NaN-to-zero in isa.rs.
• Syntactic SPO is not Pearl intervention/counterfactual: fsm::answered_questions uses CausalMask, while pearl.rs attaches causal ladder names to the same masks.
• Generic similarity is not transitively valid without a narrower contract.
• 64 channels must preserve stable hypothesis/source/event/premise identity across slot reuse.
• Class Methods remain valid; the old session pointer-chasing/scatter-gather executor was the performance problem, not the method abstraction. $var algebra precedes $var.think().
• No numeric truth conflation: literal counts, NAL F/C revision, ONA operations, interpretative admission, OpenPsi motivation, and Hyperon rewriting are distinct.

Boundaries

Docs only. 3 newly added Markdown files, zero production changes, zero oracle runs, zero performance or parity claims. The 144 entries are a fixture budget, not completed results. Original SHA/CLI/runtime/license needs to be frozen prior to testing. The PR is draft until the operator approves follow-on test/code ownership.

Research precedent: #1314, #1319, #1321, #1407, #1408, #1410, #1227, #1294, #981 and the current .claude/plans/dialectic-engine-v1.md.

[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: 594b288a-34c5-41d5-a7ce-ec4003665d15

• 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 2026-10-14T16:37:06.071750Z	5313190	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.

[coderabbitai]

Add Carrot credits or activate Agent usage billing to use Autofix

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 53131901a5

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".

[chatgpt-codex-connector]

Carry forward the existing oracle runs

The reviewed tree already contains .claude/knowledge/nal-recipe-audit/oracles.md, which pins ONA and OpenNARS to exact SHAs, records the successful ONA build and deterministic runs, and documents nine OpenNARS runs via a reproducible workaround. Marking both systems as not run or evaluated hides an existing baseline and sends the handoff toward duplicating work or choosing incompatible revisions; reference and qualify that audit here instead.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Pin the workspace evidence to an immutable commit

When main advances, this URL and the other blob/main/tree/main anchors in the ledger will display different code while remaining attached to statements described as evidence grounded on 2026-10-14. That makes the claimed provenance unreproducible and can cause a later executor to validate against a different implementation; use the inspected commit SHA in these links.

Useful? React with 👍 / 👎.