They tried to quantify how much hate was necessary to shift opinion. Ten individual pieces? Twenty?
Creating a few dozen dummy accounts to use as a cannon fodder army of vitriol-spewing robots was easy enough, then populating their output feeds with a solid string of updates so as to humanize them was simply a matter of time and labor.
Before long they had a large enough collective that they could manipulate simple things, like the lesser-known content of minor players in their respective spheres. A short comment on a video, then a burst of upvoting, and they owned the top public response slot, the first thing that many viewers would read either during or after consuming the content. Thus they could inform, and direct, the opinion of certain segments of the populace.
In time the army of dozens grew to hundreds, then thousands. Scripts were written to handle the manual labor of clicking on little thumbs-up and -down icons, but the initial hatred was always left to a human being. People weren’t that far gone that they couldn’t tell the difference between something written by a machine and the angry remark of a real person. Perhaps in time they would be, but for the first few years of the project it was left to actual people to push and prod the public mindshare.
And slowly, ever so slowly, they gained control.