I went into the article expecting it to look like the old To Catch a Predator TV show, which really bordered on entrapment in many cases, trying to make good TV

I went into the article expecting it to look like the old To Catch a Predator TV show, which really bordered on entrapment in many cases, trying to make good TV happen, but that isn't the case here at all. They just put up an innocent profile and were immediately bombarded. Good stuff. Using these chats to train machine learning corpus on child grooming is a great idea too. Social networks could use that data trivially and at very low cost by having parents opt-in to having their childrens' communications automatically monitored and alerting the parents when a conversation trips a threshold. You've got to figure anyone harassing children on open channels isn't exactly a master criminal and will get caught sooner or later anyway. But harm done is substantial so you want it to be sooner, not later.