Provenance Lookup Before Publication¶

Publication-gate protocol — instantiates Use-Time Source Attribution Calibration

A last-gate check that, claim by claim, traces a draft back to where each piece actually came from and credits anything borrowed before it goes public.

A finished draft is a commingled thing: some sentences record what the author saw, some repeat what a colleague said, some come from a half-remembered source, some were lifted almost verbatim and then forgotten. Provenance Lookup Before Publication is the checkpoint that runs at the last reversible moment — just before the material crosses into public view — and, item by item, resolves where each claim actually came from and whether anything borrowed still needs crediting. Its defining move is timing and direction: it fires at the point of external release, and it works backward from the artifact to its sources rather than forward from a memory. It does not grade how sure you feel (that is the rubric's job) or decide observed-versus-imagined from cognitive cues; it pulls the receipt for everything about to be printed.

Example¶

A newsroom editor takes a reporter's finished feature through the pre-publication provenance gate. Each factual assertion in the piece is matched to a retrievable origin: a named source note, an on-record interview, a document in the file. Three things surface. A vivid figure is attributed only to "a study" with no link — it is traced until a real, citable source can be attached, and when none can be found the line is cut rather than published. A striking sentence turns out to echo a competitor's article almost word-for-word — it is either quoted-and-credited or rewritten in the reporter's own framing. And a background claim everyone "just knew" traces back to nothing but newsroom lore — it is downgraded to attributed opinion.

The output is not a verdict on whether the story is good; it is a story in which every public claim has a recoverable origin and nothing borrowed travels uncredited. What was cheap to fix in the draft would have been a correction — or a plagiarism complaint — once printed.

How it works¶

Working from the artifact backward, the gate:

Inventories the claims. Sweep the draft for every load-bearing assertion, quotation, figure, and distinctive phrasing — the items whose origin matters if challenged.
Looks each one up. Follow the backlink to a retrievable origin: the note, record, interview, or source it came from. A claim that resolves to nothing but "I think I read it somewhere" fails the lookup.
Routes borrowed material to credit. Where wording or a specific idea traces to an external source, send it down the citation/credit path before release, not after.
Blocks on the untraceable. An origin that cannot be recovered is held out of publication — cut, hedged, or re-reported — rather than shipped on trust.

What distinguishes it from its siblings is the release boundary: it is indifferent to how the item feels and cares only whether its origin can be retrieved at the moment publication makes the attribution permanent.

Tuning parameters¶

Claim granularity — every sentence versus only the load-bearing claims. Finer catches more laundered facts but slows the gate; coarser is fast but lets minor assertions through unchecked.
Trace depth — accept the first source that appears, or follow the chain to the primary origin. Deeper catches nth-hand claims that each link treated as already-verified, at real cost in time.
Blocking strictness — hard-stop any untraceable claim versus flag-and-allow with a hedge. Stricter protects credibility but can stall a release; looser keeps things moving but leaks unverified material.
Borrowing threshold — how much verbatim or structural overlap trips the credit path. Lower catches paraphrase-plagiarism; higher tolerates common phrasing but risks uncredited lifts.
Publication scope — what counts as "public": internal circulation, client delivery, or open release. Sets how high the bar sits, since irreversibility rises with reach.

When it helps, and when it misleads¶

Its strength is catching the uncredited borrow and the untraceable "fact" at the last instant they are still cheap to fix — before a claim becomes a correction and a lifted phrase becomes a plagiarism complaint. It is the mechanism that converts a vague "we're pretty sure of our sourcing" into a claim-by-claim record.

Its honest limit is that it verifies a source exists, not that the source is right: a claim can be perfectly traceable to an origin that is itself wrong, and the gate will pass it. Its classic misuse is running backward — attaching an authoritative-looking citation to launder a claim the author already decided to publish, rather than to test whether the claim holds. And the specific borrowing failure it guards against is cryptomnesia,^[1] where the content survived but its source did not, so a genuine memory of someone else's words feels like your own. The discipline that keeps it honest is to hold "has a source" and "the source supports the claim" as two separate checks, and to treat an untraceable load-bearing claim as a stop, not a speed bump.

How it implements the components¶

Provenance Lookup Before Publication fills the trace-and-credit side of the archetype's machinery — the components a release gate can actually operate:

commingled_item_inventory — the claim sweep enumerates the assertions, quotations, and figures in the draft whose origin must be recovered.
provenance_backlink_or_trace — the lookup itself: each inventoried item resolved to a retrievable origin, or flagged when it resolves to none.
borrowed_material_credit_path — closely-borrowed wording or ideas are routed to citation and credit before the piece is released.

It does NOT attach a source_confidence_annotation grading how sure the attribution is (the Source Attribution Confidence Rubric does), reconstruct source from the source_cue_feature_set when no external record exists (the Reality Monitoring Checklist does), or attach a generated_output_source_label to machine-written passages (the Generated Content Disclosure Gate does).

Instantiates: Use-Time Source Attribution Calibration — this protocol supplies the release-gate that stops mis-sourced material at the public boundary.
Consumes: Chain-of-Custody or Lineage Check for claims whose origin needs a deep, multi-hop trace rather than a single lookup.
Sibling mechanisms: Borrowed Idea Attribution Scan · Chain-of-Custody or Lineage Check · Generated Content Disclosure Gate · Hallucination Intrusion Triage · Memory Source Probe · Observation Recheck or Replication · Reality Monitoring Checklist · Source Attribution Confidence Rubric · Source Attribution Training Set · Source Confusion Matrix Review · Source-Label Preserving Summary Template

Notes¶

The gate can only look up what was recorded earlier: if provenance was never captured at the moment of writing or gathering, there may be nothing to retrieve at publication time. That makes it a downstream partner of capture-time source labeling, not a substitute for it — a pre-publication lookup is cheapest when the source notes it queries already exist.

References¶

[1] Cryptomnesia — reproducing another person's words or ideas while sincerely believing them original, because the memory of the content survived but the memory of its source did not. It is the specific borrowing failure a pre-publication provenance lookup exists to catch. ↩