Twitter is stepping up its efforts to clean up conversations on its platform.
The company is using a combination of human review processes, machine learning, and policies to find and hide tweets that “distort and detract from the public conversation.”
Specifically, it is seeking tweets that come from accounts that don’t have any followers, or from people who sign up for multiple accounts at once, or that haven’t verified their email address.
Twitter says tweets from such accounts don’t always violate its policies, so they won’t be removed from Twitter. Instead, these tweets will not be visible in conversation threads and other places on Twitter. Twitter warns that they may become visible when people click “show more replies” to see an entire thread or conversation.
In its early tests, Twitter says it saw a 4% drop in abuse reports from search and 8% fewer abuse reports from conversations. Twitter says it will continue to refine its review processes in order to make Twitter more enjoyable for people to use.
