Retraction and correction checks before you cite
Lesson 3 of 6 · Citations that survive
1 · Learn the move · Recipe pattern
A reference can pass every existence check and still sink you: the paper is real, the DOI resolves, the authors match - and it was retracted two years ago. Retractions, corrections and expressions of concern are status changes on a real paper, so they need their own pass, and the pass has a date. The recipe: check the Crossref metadata for retraction or correction records, check the Retraction Watch database, and open the publisher page, where notices are posted. Record the date you checked. VERIFIED always means verified-as-of; a list checked in March is stale by a September submission. Rerun the pass at submission, when it is cheap, not after acceptance, when it is not.
Build me a retraction-check recipe for this reference list: [list]. For each entry: (1) Crossref metadata check for retraction/correction/expression of concern, (2) Retraction Watch database lookup, (3) publisher page inspection. I run the checks and report; you record status and check date per entry. A notice of any kind moves the entry out of VERIFIED until I have read the notice itself.
2 · Your turn. You write the prompt
Your revision leans on a 2016 paper for the central mechanism claim, and a colleague mentions hearing that the group had a paper corrected last year. You do not know if it was this paper, this claim, or this group at all. The resubmission deadline is in nine days.
Remember: the AI sees only your prompt, not this page. If the situation isn't in your prompt, it doesn't exist.
Optional. These shape the output when you run your prompt below, not your score.