build-a-source-log

# build-a-source-log The log is the control plane for research: it shows what was searched, what was found, and which claims each source can support. ## When to use - Use for papers, briefs, investigations, literature reviews, fact checks, and long-lived research. - Start before collecting many sources; reconstruction later loses search and decision history. ## Procedure 1. Choose a stable system with export support and assign each source a unique ID. 2. Record title, creator, publisher, date, version, source type, durable locator, and access date. 3. Capture discovery path and exact search query so important searches can be repeated. 4. Store notes as attributed quotation, faithful paraphrase, data extract, interpretation, or follow-up question. 5. Add precise locators such as page, section, table, timestamp, row, or archive snapshot. 6. Link each source to the research questions and draft claims it supports, challenges, or contextualizes. 7. Record credibility, limitations, conflicts, retraction or correction checks, and access restrictions. 8. Deduplicate versions without deleting meaningful publication history. 9. Run a regular audit for orphan claims, missing locators, unresolved sources, and inaccessible links. ## Failure plan - Never mix copied text with original notes without visible attribution. - Do not store sensitive data or copyrighted full text beyond what authorization permits. - If metadata is uncertain, label it unresolved instead of normalizing a guess. - Preserve excluded sources and exclusion reasons when completeness or reproducibility matters. ## Done - Every used claim is verified against a source ID and precise supporting location - Searches, versions, credibility notes, and exclusions are recorded for reproduction - The log exports cleanly and no unattributed text can drift into the draft