In November, Twitter allowed Women Action & the Media to study how the social network responded to reports of harassment. The Cambridge group collected complaints from Twitter users who faced abusive language or threats and studied Twitter’s reporting tools.
The group released its results Wednesday, in a report that brings harassment on Twitter into sharper focus and reveals some of the shortfalls of the social network’s reporting tools. During its study period, the media organization collected 811 complaints and forwarded the most pressing ones to Twitter.
“I think the biggest thing for me was the problem of evidence,” said Jamia Wilson, executive director of the media group.
When private details such as a target’s home address or names of family members are shared by a harasser — a practice known as “doxxing” — victims can report the action only by supplying Twitter with the tweet. But researchers found that many harassers delete their threatening tweets, removing the evidence.
The group concluded that Twitter had a better record of suspending accounts that were flagged for violent or hateful language but was slower to act when private information was released.
Wilson also noted that a significant number of complaints went unaddressed because they did not fit the categories that Twitter had deemed objectionable.
For example, Twitter does not recognize “mobbing” — when an individual is targeted by hundreds of accounts — or instances when harassers reveal the former names of transsexual people, a behavior known as “deadnaming.”
Twitter was also missing the fact that harassers followed their targets across social networks, moving from Twitter to Facebook, for example, because existing tools did not record this trend.
In their report, the researchers recommended broadening the definitions of harassment so that these different kinds of threats do not fall through the cracks.
The group also recommended that Twitter change its policy to allow people to report violent users even if they cover their trail by deleting their tweets.
This year, Twitter made changes in its tools in an effort to make reporting easier.
The company also expanded its team dedicated to safety and added a tool that automatically flags abusive tweets and limits their visibility.
In the weeks leading up to the study, Twitter came under intense scrutiny for allowing hostile and threatening behavior to go unchecked. The group GamerGate had sent a Massachusetts developer, Brianna Wu, into hiding after she spoke up about misogyny in the gaming industry, and Robin Williams’s daughter, Zelda, made the first of a series of high-profile exits from Twitter because of threats and hostility following her father’s death.