/identify-shared-priors

name: identify-shared-priors
description: "Trace supposedly independent evidence/reasoning paths to shared priors, datasets, models, prompts, framings, assumptions, or upstream sources that induce correlated errors."

identify-shared-priors

Purpose

Trace supposedly independent paths to shared priors, datasets, models, prompts, framings, assumptions, or upstream sources.

Input contract

required: [evidence_paths, provenance_records]
optional: [prompt_records, model_records, upstream_sources]
constraints: [shared dependency identity and path membership must be explicit]

Procedure

1. Inventory each path's inputs and assumptions. 2. Match shared priors, data, models, prompts, framing, and upstream evidence. 3. Produce dependency clusters and unresolved provenance.

If shared priors and dependence groups are explicit, consider `estimate-effective-evidence-count` as the next tactic.

Output contract

produces: [shared_prior_map, dependency_clusters, unresolved_links]
delta_fields: [findings, evidence_updates, uncertainties, open_questions]

Quality gates

• Similar conclusions alone are not evidence of shared priors; trace the dependency.

Failure and counterexamples

Mark provenance unknown rather than inferring common origin from naming similarity.

Provenance map

• resolved: independent-convergence-audit