>>9800866Tack broder!
When detecting shills many metrics can be used.
This graph shows REPLIES PER HOUR PER TOPIC, meaning the reply count on threads tagged with each topic.
I noticed that the Bernie threads had more much more replies and discussion than bloomberg, indicating that the bloomberg posts during the time were likely 95% shill posts.
Bloomberg even surpassed trump and bernie in terms of thread creation, but the reply density was really, really, really low. As well as a low "word count" indicating copy pasta/bot spam.