Audit Panel Terms

What does Editorial impact mean?

Likely question

What does the Editorial impact label on the page really mean?

Short answer

It is an editorial priority estimate, not a direct truth claim.

Longer answer

Editorial impact tells how important the pipeline currently considers a signal in terms of markets, policy, infrastructure, or operational change.

Safe formulation

Editorial impact indicates prioritization, not degree of truth.

What does Source confirmation mean?

Likely question

What does Source confirmation measure?

Short answer

It measures how well multiple sources support the same signal.

Longer answer

Source confirmation combines source count, independence, source health, and possible contradictions. It is a visible summary of how strongly the pipeline considers the signal to be supported by multiple sources.

What does Broad confirmation mean?

Likely question

What does Broad confirmation mean in practice?

Short answer

It means that more than one independent source supports the same signal.

Longer answer

Broad confirmation is shown when the system finds several sources on the same topic that are not just the same domain or copied text.

What does confidence mean?

Likely question

Where does confidence 84 come from?

Short answer

It is a heuristic confidence score, not a statistical truth metric.

Longer answer

Confidence is built from factors such as source count, source weight, source health, possible contradictions, and whether the signal has broader support.

What does Selection audit mean?

Likely question

What does Selection audit show?

Short answer

It shows how the final publication was formed from the candidate pool into a finished signal.

Longer answer

Selection audit makes it visible that the system does not publish raw AI output directly. It shows how many candidates came in, how they were filtered, what AI shortlisted, and what was finally published.

What do candidate pool, diversified pool, AI shortlist, and final picks mean?

Likely question

What do these selection-pipeline terms mean?

Short answer

They describe different stages of the selection pipeline.

Longer answer

• candidate pool is the set of all found candidates
• diversified pool is a more balanced subset built from it
• AI shortlist is the smaller set recommended by the model
• final picks are the final published selection

What does Live run audit mean?

Likely question

What does Live run audit show?

Short answer

It shows the quality and integrity of a single production run.

Longer answer

Live run audit shows, for example:

• how many rows were in the active run
• whether there were ranking gaps
• whether duplicates appeared
• whether AI output matched the final publication

What do Published, OpenAI output, and Reconciled final mean?

Likely question

What do Published, OpenAI output, and Reconciled final tell us?

Short answer

They compare the AI proposal against the final published result.

Longer answer

OpenAI output shows what the model produced. Published shows what was actually released. Reconciled final shows the final adjusted list where model output has been merged with system rules, audit logic, and possible corrections.

What do Public snapshot, Public object, and Manifest mean?

Likely question

What do Public snapshot, Public object, and Manifest mean?

Short answer

They tell which published artifact, which JDBIN object, and which manifest correspond to the visible state.

Longer answer

Public snapshot refers to the public JSON snapshot the frontend reads quickly. Public object refers to the active JDBIN object from which the snapshot is derived or which live reads use. Manifest is the control layer that describes the base/delta chain and the active state.

What do Open research snapshot and Open live public JSON do?

Likely question

What do those links do?

Short answer

They open the public artifacts that matter for auditing.

Longer answer

Open research snapshot opens the research or audit view of what the pipeline has just produced. Open live public JSON opens the public JSON snapshot used by the frontend.

Safe summary of the whole audit panel

Likely question

How should all of this be summarized in a meeting?

Short answer

The audit panel shows how sources, AI selection, and the published result connect to each other.

Longer answer

The best short answer is that the audit panel does not claim perfect truth. It makes the key publication stages visible: what material came in, what AI suggested, what was published, on what source basis, and whether there were gaps or anomalies in the run.

Safe formulation

The goal of the audit panel is not to claim perfect certainty, but to make the publication path visible, inspectable, and reasoned.