[tabulatouch]

I think that a real statistic on previous trolls actions and settlement or litigation outcomes would greatly simplify the reasoning. Would the open data from thetrollingeffect website be of any use for building a model?