Noel-Storr and colleagues’ Cochrane Crowd evaluation (18 January 2021) involved platform-team members. One task identified reports of trials assigning people to treatment groups by chance, not whether conclusions were sound. Initial practice used 10–20 records without a passing score. Later platform expert status for this task required 1,000 classifications and at least 90% on both platform measures—finding eligible reports and rejecting others—not certification of subject expertise. Disagreements or unsure judgments went to resolvers: experienced contributors making final decisions, fewer in number. In this report-identification task, 16.6% of records needed resolution. Broad participation still depended on scarce final-decision capacity; unfamiliar AI-generated research was not tested.