It's a good point, but by then we'll have software to model the relationships between all of the known sources - New York Times Journalists, Reddit Trolls, etc., and filter out some of the craziness.