This paper establishes that a targeted policy backfires because it reveals information about non-targeted units. In the world's largest fishery, the regulator attempts to reduce the harvesting of juvenile fish by temporarily closing areas where the share of juvenile catch is high. By combining administrative microdata with biologically richer data from fishing firms, I isolate variation in closures that is due to the regulator's lower resolution data. I estimate substantial temporal and spatial spillovers from closures. Closures increase total juvenile catch by 48% because closure announcements implicitly signal that fishing before, just outside, and after closures is high productivity.