The author confuses mobile web for mobile app traffic which isn't accounted for. One is the traffic from users using facebook.com via a browser on their phone versus using the installed application. Maybe the drop is real or Facebook has successfully converted more users to the Facebook app, Messenger app, etc.