It actually sounds pretty simple at first: according to a Twitter blog post Tuesday, the company will simply organize conversations on Twitter differently based on "behavioral signals" created to root out trolls in "communal areas" of the social network.
In order to identify the behavior, the company has picked out certain patterns of behavior which it associates with troll-like accounts.
In an announcement earlier today, Twitter said they were taking steps to limit "behaviors that distort and detract from the public conversation" by downranking content that exhibits such behaviors from search results and "public conversations".
Among select users in markets around the world, the changes resulted in a 4% drop in abuse reports from search and 8% fewer abuse reports from conversations, according to Twitter.
The changes are a part of Twitter's efforts to increase the "health" of its platform.
"There are many new signals we're taking in, most of which are not visible externally", the post reads. On one hand it's been upsetting to users who have been personally targeted, on the other hand it's just taken away the utility of poring through the conversations that Twitter enables in the first place. Twitter will look for signals such as unconfirmed email addresses, multiple accounts opened by the same user, and repeated interactions with accounts that don't follow them. "Because this content doesn't violate our policies, it will remain on Twitter, and will be available if you click on "Show more replies" or choose to see everything in your search setting".
"These signals will now be considered in how we organise and present content in communal areas like conversation and search".
But it adds: "Our work is far from done". Our goal is to learn fast and make our processes and tools smarter.
The company says it will deploy a screen saying "show more replies" in front of responses that its systems adjudicate as vexatious, cynical or calculated to offend. "We are making progress as we go", Dorsey said.