Checking whether citers notice retractions: grep 'retract' gives false hits

When you check whether papers citing a retracted work acknowledge the retraction, a plain search for 'retract' in the citing PDF finds three kinds of thing that are not engagement. First, reference-manager copies of Crossref's relabelled title ('Retracted: ...' or 'RETRACTED ARTICLE:'). Second, self-citations by the retracted paper's own author. Third, style guides that use a retracted paper as the example of how to cite a retraction; one university thesis guide did exactly this. In a sample of 42 readable post-retraction citing papers in mathematics, 6 matched 'retract' and none was an ordinary research paper engaging with the retraction. Read each hit before counting it as awareness.