Fleets, Twitter's latest feature, are tweets that allegedly exist for 24 hours—although even that is in question. Like Instagram and Facebook's stories (in fact, exactly like Stories) users on the app now see a row of circle-cropped profile pictures of people they follow, where their Fleets live. You can tap a user's circle and see a slideshow of text, photos, re-fleeted(?) tweets they've put in there, all live for 24 hours only. After that, fleets expire and disappear.
Fleets has been available in some regions since June, but as of yesterday, the ephemeral-tweet feature launched for users around the world—and as happens any time a social media platform makes a big change, people freaked out.
But Fleets aren't just another superficial tweak like changing the like button from a star to heart: they have the potential to worsen the platform's already insidious harassment problem.
“What was stopping people from transitioning from their passive observation to being more actively engaged?” Twitter head of research Nikkia Reveillac said in a press briefing for the launch. “What we learned when we talk to people is that tweeting and engaging in conversation can honestly be incredibly terrifying."
At the briefing, director of design Joshua Harris gave another reason for Fleets: users saving and then never tweeting draft posts. Like an online store that emails users when they've left something in their carts for later, Twitter's mission is to nudge you to send those tweets. Twitter knows your drafts folder is full of awful takes and would rather you let them fly. If they were temporary, maybe they'd be less risky.
But instead of making the app less terrifying in ways users have been begging for—like banning users who spread hate, including Donald Trump—Twitter gave us an impossible to moderate platform for more harassment and disinformation, and rolled it out half-baked.
On Wednesday evening, Twitter announced that it was slowing down the Fleets rollout, "to fix some performance and stability problems."
Because of the way Fleets launched—with little actual usage terms or moderation transparency—users have taken it upon themselves to put the feature through its paces in the wild.
As of last night, several users reported that you could view the Fleets of people who had blocked you, an obviously bad bug that could expose people to serious harm online, if they've blocked someone out of fear for their safety. Twitter told me that if you've blocked someone, they shouldn't be able to see your Fleets: Harris tweeted that this is something they are working on. We weren't able to reproduce the bug today.
Extremist researcher Marc-André Argentino has been testing Fleets' moderation practices, and found that there's essentially nothing you can't put in a Fleet—including hate speech, banned links, extremist propaganda, and disinformation about the U.S. elections or the pandemic. All of it stayed up, in his tests. Because Fleets are temporary, and because they're hidden behind another layer of access—you have to click through someone's profile picture to watch a slideshow to see what's in there—the jobs of researchers and moderators becomes more difficult, while letting bad actors target their own audience specifically from within a mainstream platform.
A Twitter spokesperson told me that Fleets must follow the Twitter rules, however, which prohibit the kinds of hate speech Argentino posted. (Argentino’s account was temporarily suspended on Wednesday.)
"Like Tweets, Fleets can have labels or warnings and labels from Tweets will carry over into Fleets. We also encourage people to report Fleets that may violate our rules directly to us by tapping on the ‘v’ icon on a Fleet. We’ll take enforcement action against any Twitter rule violations in Fleets accordingly."
Notifications are another issue. If you're tagged in a Fleet, you get a notification. Being alerted to when you're tagged is the whole point of tagging. Twitter says Fleets follow the notification preferences users have set for the rest of the platform, and that there are no specific Fleet notification options. However, when one of your tweets is added to someone else's Fleet, you don't get notified—allowing users to be targeted without their knowledge, and their tweets spread through backchannels they're not aware of.
"I would say it's very evocative of a Silicon Valley to have an idea, internally test it, internally ask around, but then not ask, for example, celebrities or activists who deal with harassment or researchers who study these things… who could easily provide use cases or examples of when this could go wrong," Caroline Sinders, an online harassment expert and researcher, told me.
There's a long history of platforms coming up with ideas that no one asked for, Sinders said. "And we should ask, is this because you can't figure out the other thing we want you to do, and so you feel like you have to iterate? Or are you afraid of building the other thing we want?" She referenced the user demand for an edit button—a feature that people actually want, and that would pose similarly difficult problems of networked harassment and disinformation that Twitter would need to address. But instead, we got Fleets.
"I understand the need to iterate and push a product forward, especially if you're trying to think of innovative ways to engage your audience. And people want innovative ways to tell stories. That's why we stay online," Sinders said. "We want creative outlets, but people also want basic safety features and enforcement of policy… It feels like a very tone deaf gift that no one asked for."
It's a tradition at this point to ask why Twitter continues to introduce new features that there wasn't a demand for (voice memos being another recent example), while failing to adequately address the serious issues of disinformation and hate speech on its platform. Fleets is another case that actually threatens to make some of these problems worse.