docs(adr-0006): revalidate the blocked publisher sources, 24 days on

[seonghobae]

Fleet revalidation

• protected base: main@a080297d2546bb61e89520d637cabc202db331ec
• exact head: a7dbbeed09f615d3af0ed2bfdb9df6169160c627
• lifecycle: Draft / documentation evidence needs terminology repair / exact-head hosted evidence non-terminal
• scope remains one ADR, documentation only.

Valid revalidation evidence

The 2026-09-05 probe still supports the narrow operational conclusion that the two publisher-original retrieval routes remain blocked from this environment:

Source	2026-09-05 observation	Earlier recorded observation
Jones–Loe (SAGE Open)	HTTP 403; 5,549-byte anti-bot response rather than a PDF	HTTP 403
Iannario (De Gruyter Brill)	HTTP 202; zero-byte body	HTTP 202; zero-byte body

This is sufficient to keep the publisher-original attachment Goal open. It is not sufficient to call every response an "identical signature": the earlier Jones–Loe record shown in this ADR preserved the status but did not preserve the same byte-count/body signature.

Repair finding: HTTP 429 attribution

The new ADR text says the archived-snapshot 429 Too Many Requests is "client-side rate limiting". That attribution is not supported by HTTP semantics. RFC 6585 §4 defines 429 as a response indicating that the user/client has sent too many requests in a period; it deliberately does not define how the responding server identifies the requester or counts requests. A 429 observed by this client therefore establishes a rate-limit response affecting this request path, not that rate limiting was implemented on the client side.

Primary standard: RFC 6585 §4, 429 Too Many Requests — https://www.rfc-editor.org/rfc/rfc6585.html#section-4

The Wayback availability API reporting the archived snapshot as available can remain separate evidence. It does not establish publisher-original provenance and does not satisfy the Zotero attachment-parent acceptance criterion.

RED → GREEN acceptance

Before Ready, repair the ADR itself so that:

1. 429 is described as a rate-limit response encountered on the archive retrieval path, without inventing a client-side/server-side implementation attribution that the response cannot prove;
2. identical signatures is narrowed to the evidence actually retained — same status/block class for Jones–Loe, and same 202 + zero-byte result for Iannario — unless byte-identical prior evidence is produced;
3. publisher WAF evidence, archive availability, OA/license metadata, byte provenance, and Zotero parent-attachment state remain separate fields/claims;
4. the current exact head is re-reviewed and all required hosted checks are terminal-valid.

Fresh workflows on a7dbbeed09f615d3af0ed2bfdb9df6169160c627 are non-terminal: Security Scan 33956669757, SAST Semgrep 33956669815, CodeQL PR 33956669753, Security and Quality 33956669776 are queued. Current inline review-thread inventory is empty; CodeRabbit also reported review-capacity exhaustion rather than substantive approval. No predecessor evidence transfers.

Do not merge this documentation while the factual attribution remains in the canonical ADR. No source-neutral retrigger, self-approval, bypass, force-push, destructive rebase, or gate weakening.

[coderabbitai]

Warning

Review limit reached

Next included review available in 7 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available.

You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Team

Run ID: 4fb6c03a-c4a3-4efb-a848-38f091b79b57

📥 Commits

Reviewing files that changed from the base of the PR and between a080297 and a7dbbee.

📒 Files selected for processing (1)

• docs/planning/adrs/0006-polytomous-llm-judge-bias-calibration.md

---

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.}

[seonghobae]

@jules Please repair the canonical ADR text on the current branch without changing scope or adding new claims. Two factual attributions are unsupported by the retained evidence:

1. Replace reproduced the recorded blocks with identical signatures with evidence-scoped wording. Jones–Loe has the same HTTP 403/block class, but the earlier record did not retain the same byte-count/body signature. Iannario did reproduce HTTP 202 with a zero-byte body.
2. Replace 429 ... which is client-side rate limiting with neutral wording such as a 429 rate-limit response was encountered on the archive retrieval path. RFC 6585 §4 does not identify where rate limiting is implemented; the observed response cannot establish a client-side implementation.

Keep publisher-original WAF evidence, archive availability, OA/license metadata, byte provenance, and Zotero attachment-parent state as separate claims. Do not weaken the publisher-original Goal. Update the PR exact-head evidence after the normal descendant; no force-push or source-neutral retrigger.