That was my line of reasoning for the most part. But if we're going to make the argument that viewing footage of non-simulated violence actively encourages it, I have no idea where the line for that would be drawn. Do we draw the line at footage produced by the attackers, a la the New Zealand shooting, or ISIS mass execution videos? What about footage captured by someone/thing else, a la security camera footage? What about material concerning prior attacks, like a 9/11 documentary? I have no cogent approach to that.
Estimates are all over the place due to the way Tor operates, but researchers suspect its most popular hidden services are all related to child porn.