A well-known finding in organizational research holds that teams whose members report high psychological safety, the shared belief that the group will not punish someone for raising a concern or admitting an error, tend to outperform teams that lack it, particularly on tasks requiring rapid learning from mistakes. This finding is frequently cited by consultants as grounds for a specific intervention: managers should visibly reward the reporting of errors, on the theory that public reinforcement builds the shared belief fastest. Recent replication work complicates this recommendation. Several of the original studies measured psychological safety and team performance at the same point in time, using a single survey wave; from that design alone, a team's psychological safety score cannot be shown to precede, rather than follow from, its performance. It is entirely possible that teams begin performing well for unrelated reasons, a favorable market or a stronger technical lead, and that success itself makes members feel safer voicing errors, since a winning team has less to lose from an admission. A smaller set of longitudinal studies, which measured psychological safety before any performance outcome was known, still found a positive relationship, but a substantially weaker one than the cross-sectional studies reported, and the relationship held only for teams working on genuinely novel tasks, not for teams executing well-understood routines. For routine work, the longitudinal studies found no reliable relationship at all. The practical upshot is narrower than the original finding suggested: publicly rewarding error reporting is a defensible intervention for teams facing unfamiliar problems, where mistakes carry real diagnostic information, but there is little evidence it improves outcomes for teams running established processes, where the same intervention may simply consume time without changing what the team learns.