"We suck at dealing with abuse and trolls on the platform and we've sucked at it for years," Dick Costolo, the CEO of the microblogging service, conceded in a recently leaked staff memo.
"We lose core user after core user by not addressing simple trolling issues that they face every day," he added in one of two internal emails obtained by the technology and culture website The Verge.
Costolo blamed himself for letting down victims of online harassment, but his pledge to fix the problem rang somewhat hollow for feminist writer Lindy West, whose experiences reportedly prompted the CEO's vow to be "more aggressive" about stopping harassment.
"I'll believe it when I see it. And I want to believe it," West said from Seattle.
"I'll screen-grab [the abusive messages] and block the user and report it, but there's so much out there that I'm sure, cumulatively, it eats up a lot of time."
West, who has endured online threats for years for writing about what she calls "big, fat, bitchy" topics, reached out to her "cruellest troll" — a man who created a bogus Twitter account posing as her recently deceased father — and interviewed him in a recent episode of the U.S. radio program This American Life.
She welcomed the discourse that followed, but the online vitriol polluting her social media feeds continued.
Twitter won't budge on anonymity
As in the cases of an untold number of women and girls who, according to a Pew study, receive a disproportionate amount of internet harassment, West felt there was little she could do about it.
"Most of my reports [to Twitter moderators] of abuse are rejected," she said. "I could spend all day collecting hundreds of links. Now, I'll just report it if it's a graphic rape threat or something. If it's ‘Kill yourself, bitch,' I don't even bother."
Filtering online abuse on Twitter is a challenge, in large part because of the platform's commitment to anonymity — an option not available to Facebook or GooglePlus users, who are required to sign up using real names.
Another component is an apparent shortage of Twitter staff to process harassment complaints.
The volume of complaints Twitter handles from its 288 million users a month would be "overwhelming," said Andrea Weckerle, founder of CiviliNation, a non-profit organization taking a stand against online hostility.
"This goes back to needing a bigger team to give quicker responses, because people's lives are impacted," she said.
Compiling the evidence is time-consuming, too.
Twitter's abusive-user form requires a complainant to collect and send URLs of specific threatening tweets for evaluation.
"Imagine hours, days, weeks. It's daunting to compile it all," digital strategist Shireen Mitchell said.
The turnaround might take a day or two, West added, "and you could get a reply saying we looked at your report and we don't care."
The nastiest tweets directed to Mitchell, who is black, might concern her race, her status as a female gamer, or her work advocating for young girls of colour to embrace futures working in tech, she said.
"But when I'm engaging them in this back and forth, they're actually deleting those tweets, so the URLs are gone," she said.
Banned users also resurface under new account names.
To fulfil the need for better blocking tools, third-party developers have created filters such as Block Together, The Blockbot and Twitter Quick Blocker.
Racist comments also posted to Facebook
Some victims of harassment have called for reduced anonymity, so abusers have to use their real identities.
Mitchell points out, however, that even Facebook failed to deter people from making openly racist comments attacking web developer Adria Richards, whose tweets outing a male developer for sexist comments inadvertently led to him getting fired in 2013.
"The amount of people using their full names on Facebook and threatening her, oh my goodness, I actually did screenshots," Mitchell said.
Richards has shared her own recommendations for Twitter to clamp down on trolling, which include making trending harassment patterns available via the application programming interface (API) and allowing users to subscribe to blacklists to exclude words from their timelines.
Since December, Twitter has introduced new tools to streamline abuse reports.
Richards has noticed the changes over time, such as the ability to report on behalf of another tweeter, and more prominently displayed buttons to the ticketing form.
"It's still a very manual process, but at least you can now report multiple tweets," she said.
West believes the technical tweaks can only go so far.
"The solution really is to change the culture so that when men disagree with a woman, he doesn't have to immediately reach for a gendered slur," she said.
As for arguments about protecting free speech — even for those making violent threats or hateful comments — West doesn't buy it.
"Staying neutral on hate speech affects my free speech," she said. "It's important to realize that taking steps to safeguard the online experiences of marginalized groups is actually an action that protects speech."