Repository navigation
docs(research): versioned semantic candidate resolution across 64 thoughts - #1604
Conversation
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configuration
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. Comment |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
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[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6cadb6a868
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".
| | `EvidenceBase` | independently attested source stamps/roots supporting or contesting a claim | not thought count, translation count or audience count | | ||
| | `Perspective` | compiled thought program, front end, rules and interpretive lens | a named philosophical lens is not a historical individual's actual judgment | | ||
| | `ReadSet(t, v)` | addresses/fields the thought inspected, plus negative dependencies | must support rebase invalidation when prior reads change | | ||
| | `CandidateOutcome` | derived result: Absent / Admissible / Contested / Preferred / Established / Abstained / Superseded | a view of versioned evidence, NOT a CE64 bit or write-side fact rewrite | |
There was a problem hiding this comment.
Add a distinct rejected outcome (or define which existing value represents it). The RFC requires rejected hypotheses to remain distinguishable from superseded ones and says rejection must preserve its witnesses, but the proposed CandidateOutcome vocabulary cannot encode that state: Superseded is defined only for a previously preferred candidate displaced by new evidence or context. An implementation following this vocabulary would therefore have to misclassify an adjudicated-false candidate or lose the rejection during replay.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Fixed in #1618 (f95c72fe): CandidateOutcome now has a separate Rejected outcome that keeps its witnesses. It is distinct from Superseded.
Generated by Claude Code
| }, | ||
| "expected": { | ||
| "v2_derived": [], | ||
| "must_rederive_from_scratch": true, |
There was a problem hiding this comment.
Do not require the implementation to rederive the entire result from scratch here. The RFC explicitly permits either DRed or an independent from-scratch reference, and the inherited thought-git contract retracts only derived cells affected by R ∧ M before rederiving them. Consequently, a conforming DRed implementation can correctly produce an empty v2_derived set while failing this must_rederive_from_scratch expectation; the fixture should assert the observable retraction and use a from-scratch rebuild only as the oracle.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Fixed in #1618 (f95c72fe): C-09 now asserts the retraction itself ("retracted": ["B"]). It accepts DRed, and a from-scratch rebuild serves only as the oracle.
Generated by Claude Code
1 participant
Why this PR
The Language LEGO research in #1603 establishes an intended multilingual grammar-to-query ABI. The 64-thought/git lifecycle in #1599 already defines branch/rebase/squash/land/blame. The epistemic and self-model groundwork in #1597/#1602 maps the common NodeGuid frame and perspectives.
Missing contract: what a semantic candidate is, how to keep incompatible proposals distinct, and how a sparse, versioned thought merge can prefer, preserve, abstain, contest or revise candidates without mistaking masks, thought agreement or lexical priors for independent evidence and truth.
Changes
docs/research/versioned-semantic-candidates-64-thoughts-rfc.mdClaimIdentity,QuestionContext,CandidateSet,EvidenceBase,Perspective,ReadSet,CandidateOutcome,ResolutionReceipt(not new Rust types/storage).AlphaOverlay(not yet a production value ledger), immutable reference bakes, separate grammatical vs NAL confidence, source independence, open-world negation, CE64 layout migration, thought slot reuse, and no foreign gather.docs/research/versioned-semantic-candidates-fixtures.jsonScope / dependencies
This is design + fixtures, no behavioral changes, CE64 relabeling, new crate/opcode, benchmark result, or claim of AGI. It does not rebuild the thought git API.
Implementation steps are gated on D-TGA-1/3/4 (#1599), the grammar DTO from #1603, and independently verified semantic/reasoning oracles. Most important initial falsifiers: C-01/02 (real identity versus hallucinated closure), C-05 (candidate multiplicity), C-06/07 (stamp independence), C-09 (rebase deletion).
Source-of-truth files read
.claude/plans/thought-git-api-v1.md,.claude/knowledge/thinking-tables-wiring.md,.claude/knowledge/epistemic-frontier-capstone.md,.claude/knowledge/reference-frame-vs-motion.md,crates/lance-graph-contract/src/alpha.rs,crates/lance-graph-nal/src/lib.rs,crates/lance-graph-relalg/src/ir.rs, D-HOLO H14c/H14d plans.