D-SAG-CF-1: inverted temporal ablation × ONA/Hyperon reasoning oracles

[AdaWorldAPI]

Question

Can inverted temporal ablation test whether a conclusion survives the removal of evidence, a rule, an interpretive method or a shared upstream assumption, using the existing lance-graph/OGAR/Quack substrate rather than a second graph engine?

Scope: D-SAG-CF-1 research RFC and synthetic proof-DAG oracle, building from the merged #1509. No production adapter, causal-certification claim or measured performance improvement. Draft intentionally.

What is in this PR

• .claude/plans/inverted-temporal-ablation-reasoner-oracles-v1.md: full two-world contract W0/W_X, source/rule/method/assumption/scheduler cuts, provenances, result receipts, output categories, model-relative vs do(X), temporal horizon pinning, explicit OGAR ClassView/MethodKey/reading/CE64 boundaries, 16 adversarial falsifiers F0–F15, phased W0–W7 execution plan, performance/STOP thresholds and a Claude Code implementer prompt.
• .arxiv/2026-10_inverted_temporal_ablation_reasoner_oracles.md: short paraphrased primary-source excerpts, testable implications, links to Jøsang fission (0712.1182), Wang/NARS self, ONA vs OpenNARS, Goertzel/Hyperon (2310.18318), Chockler actual causality (1608.07879), provenance semirings, Differential Dataflow and Soufflé.
• .claude/tools/inverted_temporal_ablation_oracle.py: dependency-free synthetic Boolean/DAG proof-term oracle, 12 stdlib unittest cases for singleton/joint cuts, independent vs shared ancestry, aliased evidence roots, rule ablation, and explicit cycle/identity refusals. Not a NARS F/C or SCM implementation.
• .claude/board/entries/2026-10-12-inverted-temporal-ablation.md: discoverable Claude Code handoff with blockers and acceptance gates.

Key rulings from reading the shipped code

1. planner/src/temporal.rs is temporal and epistemic deinterlacing, not a complete dependency closure; a concrete QueryReference::at is required for strict horizon semantics.
2. chain_counterfactual.rs already cuts one step through the same replay kernel, with a detached counterfactual (-6) tag. It is not complete source/method retraction.
3. SAG D-SAG-1: software-awareness graph — lance-graph's own methods, programs and evidence as Quack-queried rows #1509 provides method/provenance rows queried in one hop. Do not introduce an illegal Quack population intermediate or claim multi-hop execution exists.
4. Shared ancestry must not be mistaken for independent evidence; bounded NARS stamps and compressed CE64 F/C are insufficient for unconditional exact fission.
5. Boolean persistence, support multiplicity, numerical belief revision, decision changes, Gadamer revision and structural causality are different outputs, not one truth flag.

Oracle roles

Independent system	First diagnostic
ONA (C)	Temporal event + anticipation/procedure under clean-room input ablation
OpenNARS (Java)	Evidential stamps; revision vs choice; shared roots
Hyperon / MeTTa	Rule/method-as-data ablation; reinterpretation
Differential Dataflow	Correctness and cost of signed-update incremental recomputation
Soufflé	Proof/explanation provenance oracle
lance-graph Quack/R2IL	Proposed typed sparse execution, compared with scratch replay

Validation and limitations

• Suggested runnable local W0 smoke test: python3 .claude/tools/inverted_temporal_ablation_oracle.py.
• I have not executed the test suite or compiled foreign engines in this PR creation environment; 12 tests are supplied, not reported as passing. This is a documentation + isolated stdlib script change; no Rust production files modified.
• Adapters, exact retraction, recursive Quack traversal, CE64 numerical inverse, causal identification and benchmark speedups are not implemented.
• Next tiny PR: pin ONA upstream commit, deterministic baseline fixture, one source-suppression arm with explicit partial provenance, negative controls (F0/F1/F6/F8/F15). Expose unsupported internal ancestry rather than inventing it.

Review questions

• What counts as the complete evidential root ID?
• Can ONA expose full derivation ancestry, or is the first adapter necessarily black-box/Partial?
• Is TruthOnly distinct from both necessary and dispensable at the Gadamer admission gate?
• What legal relation carrier handles the second Quack hop?
• Which sparsity/fanout regime makes the indexed path genuinely faster after setup costs?

References: #1509, #1496, #1497, #1498, existing GADAMER-ABLATION-RESEARCH.md and dismech-causal-replay-v1.md.

[coderabbitai]

Warning

Review limit reached

Your organization has reached its usage spending cap. Adjust your spending cap in the billing tab.

Next included review available in 38 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available. Your 62 included PR review attempts over the past 7 days set your current allowance at 1 review per hour.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Essentials
• Run ID: ccdf8ad1-1864-48a6-a85f-4b3f193d3123

---

📥 Commits

Reviewing files that changed from the base of the PR and between c1ce262 and 91791c4.

---

📒 Files selected for processing (4)

• .arxiv/2026-10_inverted_temporal_ablation_reasoner_oracles.md
• .claude/board/entries/2026-10-12-inverted-temporal-ablation.md
• .claude/plans/inverted-temporal-ablation-reasoner-oracles-v1.md
• .claude/tools/inverted_temporal_ablation_oracle.py

---

• 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.

---

_{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 2026-10-12T19:49:24.837091Z	91791c4	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: 91791c47eb

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]

Enforce versioned rule identities

When a caller constructs a world with an ID such as Rule("R", ...), this validation accepts it even though the error message and RFC contract require rule_id@version. The resulting proof labels and cut selectors cannot distinguish revisions of the same rule, so comparisons can report an exact rule ablation without pinning which implementation was removed; validate the version component here or represent it as a separate required field.

Useful? React with 👍 / 👎.

[AdaWorldAPI]

Superseded by #1523. Every commit from this branch is merged there unchanged, including the oracle. The capstone only adds a pointer.

The capstone lists three tests in the oracle as "claimed, unverified" (§7):

• test_f0_pristine_replay_is_exact compares a pure function with itself.
• test_f7_contingency… repeats F3.
• F8 has no test.

Closing in favour of the capstone.

---

Generated by Claude Code