Skip to main content

Three takeaways from a day at Twitter

Three takeaways from a day at Twitter


Can democracy withstand the speed of our tweets?

Share this story

Photo by Amelia Holowaty Krales / The Verge

On Wednesday, Twitter invited reporters to its headquarters in San Francisco for an update on improvements to the product. It’s somewhat unusual for the company to invite reporters in — I’ve visited headquarters around three times a year for the past five years or so — but it’s not unusual for Twitter to provide updates. Providing updates is, in some ways, Twitter’s favorite thing to do.

In January we talked about the many times Twitter has substituted talking for doing, an approach championed by CEO Jack Dorsey this year on his lengthy podcast tour. I actually find it charming how readily Twitter employees will stipulate to any criticism of the product — usually while throwing in a few of their own for good measure. There is a rare candor among Twitter employees that makes me think fondly of the company even when things go wrong there, which they do with a regularity unusual for a company of its high profile.

The occasion for Wednesday’s event was to discuss its work on creating better conversations, fighting abuse and harassment on the platform, and letting users more easily follow their interests. All of that has only recently become possible, said Kayvon Beykpour, Twitter’s charming and unusually durable head of product. Twitter became profitable for the first time last year; its daily users are increasing even as monthly users decline; and it has shipped enough updates to fight harassment that it can now talk meaningfully about improvements.

Here are three things I took away from the event.

One, a new feature designed to let users follow their interests could open up Twitter to a big new audience. It could also amplify the worst stuff on the platform. I wrote about the feature, which Twitter calls “interests,” for The Verge. It’s now in testing on Android:

Twitter will begin allowing users to follow interests, the company said today, letting users see tweets about topics of their choosing inside the timeline. When the feature goes live, you’ll be able to follow topics including sports teams, celebrities, and television shows, with a selection of tweets about them inserted alongside tweets in your home feed.

Topics will be curated by Twitter, with individual tweets being identified through machine learning rather than editorial curation, the company said. For now, only sports-related interests can be followed, said Rob Bishop, a Twitter product manager. 

Twitter has long believed that if the average person could just find the tweets that interest them, they would become a daily user for life. And yet finding those tweets involves knowing which accounts to follow, and regularly adding to and pruning the list, and it turns out that most people don’t want to take on that burden. And so it has tried doing that job for users in a variety of ways — a suggested user list, Twitter Moments, a “temporary follow” feature for seeing tweets from sporting events — but none has been wildly successful.

Interests represent the company’s latest effort to crack that nut. Certainly it seems useful for following something like a sports game — Twitter can round up popular tweets on the subject from lots of accounts, using machine learning to identify relevant tweets even if they come from an unknown account or don’t use a particular hashtag.

But use the same feature for news and it starts to feel worrisome. For starters, it’s a new attack vector for bad actors. Figure out how to get your tweets included in the set shown to people interested in “politics” or “elections,” and you’ve got a fun new 2020 project for Russia’s Internet Research Agency. And even if state actors don’t get involved, you’ve still got a product that seems destined to amplify the hottest, most polarizing takes, no matter what the subject. If the current test shows that the feature takes casual fans of the Golden State Warriors and turns them into diehard season ticket holders, that could be reason to worry.

(Twitter’s response to all this is that yes, it knows, and that’s why it’s starting with something dumb like sports.)

Two, Twitter thinks it can save its much-maligned box of trending topics. I wrote here earlier this week that it’s time to end “trending” modules, which are easily gamed and provide little benefit to users. This seemed particularly evident over the weekend, when an explosion of conspiracy theories related to the disgraced financier Jeffrey Epstein overtook Twitter trends.

I asked Yoel Roth, who leads platform integrity efforts at Twitter, whether Twitter wouldn’t have more platform integrity were a thing of the past. He told me that he thought trends ought to be fixed, rather than canceled. “It’s a major area of investment for us to make trending topics more reliable,” he said.

One reason all those conspiracy theories appeared in the trending box is that ... many people really were tweeting them, Roth said. The company did not find large-scale efforts to manipulate trends, such as with bots. There was a genuine and sincere spike of insanity on Twitter and it was accurately reflected in the trending box.

Still, Roth allowed, “the experience wasn’t what we wanted.” When multiple hashtags about a single topic trend, Twitter usually stitches them together into one. (The whole thing would have seemed less terrible if the trend was “Jeffrey Epstein” rather than dueling hashtags over which president was behind his death.) But that didn’t happen this time, due to bugs.

The company also argues that Twitter trends provide a useful view of what people are talking about on the platform, even when what they’re talking about is extremely dumb or wrong, and that it’s one of few places on the platform where people can easily escape the echo chamber of their timeline.

I don’t find any of this particularly convincing, but if you wonder why Twitter trends still endure, there you go.

Finally, Twitter should think about whether it’s too fast for democracy to withstand. I asked Beykpour whether Twitter’s product teams were thinking at all about speed, and what that does to the body politic. I was influenced by two good pieces I read recently on the subject — Robin Sloan’s argument that there should be an upper limit on how many people can read a tweet; and Margaret Sullivan’s call for a new “slow news” movement.

Beykpour told me that the team had talked about speed, but noted that speed has historically been Twitter’s signature advantage over its competitors. Other places slow the pace of posts via algorithms that make it harder for individuals to reach their audiences; Twitter delivers a real-time feed of information that makes it addictive for many of us.

In the wake of (*gestures broadly at everything*), and while Twitter is still rethinking its core incentives, speed seems ripe for reconsideration. 


FTC Chief Says Willing to Break Up Companies Amid Big Tech Probe

The chairman of the Federal Trade Commission says he won’t rule out a breakup, David McLaughlin reports. But he doesn’t seem very excited about:

FTC Chairman Joe Simons, who is leading a broad review of the technology sector, said in an interview Tuesday that breaking up a company is challenging, but could be the right remedy to rein in dominant companies and restore competition.

“If you have to, you do it,” Simons said about breaking up tech companies. “It’s not ideal because it’s very messy. But if you have to you have to.”

Tom Steyer Close to Qualifying for Next Democratic Debates

Edward-Isaac Doveere examines how rich Democratic candidates for president can buy their way into the debates using Facebook ads:

The entire first month of Steyer’s campaign was geared toward getting into the debates. Using the data from his two groups, his campaign has produced 16,000 variations of digital ads, aides told me, including those that are adjusted automatically by software to more effectively target viewers by their interests. The aides acknowledged that they built the early phase of the campaign specifically to fit the DNC’s requirements, postponing other voter-engagement efforts until the fall. “We are creating and pushing out new [ad] campaigns every day, and adjusting for efficiency as we learn about the best ways to reach people who respond to Tom’s message,” said Martha Patzer, one of Steyer’s deputy campaign managers, who previously worked at Need to Impeach (she’s one of the many staffers at Steyer’s groups who transitioned to his presidential team).

An aide on a rival 2020 campaign took a different view: Digging at how much money Steyer has spent on targeted advertising on Facebook, the aide said, “This amounts to a wealth transfer between Tom Steyer and Mark Zuckerberg.”

AWS Adds “Fear” To Facial Recognition Repertoire, Draws Immediate Fire

Ed Targett has the most dystopian news of the day. Is this really the kind of product you get when you work backward from delighting the customer

“We have added a new emotion, ‘fear’”, the company said in a service update on several accuracy and functionality improvements to Rekognition, adding that it now identifies eight emotions, including “angry”, “disgusted”, and “surprised”.

AWS is a major US government contractor, and protesters have targeted the company for providing infrastructure that underpins immigration operations.

China Said It Closed Muslim Detention Camps. There’s Reason to Doubt That.

Grim story from Chris Buckley and Steven Lee Myers. It includes an interview with a man put into an indoctrination camp for eight months for watching videos about Islam on his phone:

In late July, the government said most detainees had been released from the indoctrination camps built to eliminate what it described as the threat of Islamic radicalism and antigovernment sentiment among the overwhelmingly Muslim population of Uighurs in the Xinjiang region in China’s northwest.

But reporters from The New York Times found, over seven days of traveling through the region, that the vast network of detention camps erected by the government of China’s authoritarian leader, Xi Jinping, continues to operate, and even expand.


Exclusive: Automattic CEO Matt Mullenweg on what’s next for Tumblr

Nilay Patel and Julia Alexander interview Automattic CEO Matt Mullenweg about the future of Tumblr, which he just bought in a fire sale:

MULLENWEG: I guess we’re still a corporate parent, but we’re a very friendly one, and we’re all about blogging, innovation, publishing communities. So I would love for Tumblr to become a social alternative. That’s in line with Automattic’s values around privacy and freedom of speech and publishing, but it has the fun and friendliness of some of the other networks we use, but without that democracy destroying… oh, I don’t know what you want to call it.

10,000 Rounds Of Ammo And 25 Guns Were Seized From A Teenager Posting Far-Right Memes On iFunny And Discord

Ryan Broderick reports on a potentially awful situation that was fortunately averted:

An 18-year-old Ohio man charged Monday with threatening a federal officer allegedly posted extensively online about mass shootings, specifically targeting Planned Parenthood. According to an affidavit, the bulk of Justin Olsen’s radicalized content was posted under the name ArmyOfChrist on meme-hosting app iFunny.

Federal agents seized 15 rifles, 10 semiautomatic pistols, and 10,000 rounds of ammunition during the arrest earlier this month.

The Meme App iFunny Is A Huge Hub For White Nationalists

Broderick followed up the above piece with a look at the rise of extremism on the iFunny app, where the arrested man allegedly posted his threats. iFunny is apparently popular with teenage boys:

“I’ve been using iFunny since it came out in 2011,” the anonymous user said. “These kind of things really picked up at the height of offensive conservative counterculture in 2016. I guess the ideas remained for a lot of the underground users, so as time went on they evolved to have a community that was similar to 8chan.”

The iFunny user provided BuzzFeed News with a list of larger radicalized accounts. The content shared there is on par with anything that was being posted on the now-offline 8chan. Following news of Olsen’s arrest this week, many of these accounts labeled themselves “satire” and shared memes about the FBI.

LGBTQ YouTubers are suing YouTube over alleged discrimination

If nothing else, this case should make for some pretty popular YouTube videos:

A group of YouTube creators is suing YouTube for allegedly discriminating against their LGBTQ-focused videos by suppressing recommendations and making it difficult to earn ad revenue.

The lawsuit alleges that YouTube uses “unlawful content regulation, distribution, and monetization practices that stigmatize, restrict, block, demonetize, and financially harm the LGBT Plaintiffs and the greater LGBT Community.” The lawsuit also alleges that both YouTube’s machine learning moderation tools and human reviewers unfairly target channels that have words such as “gay,” “bisexual,” or “transgender” in the title.

Snapchat creators earn serious money from making AR filters

I don’t know how serious this money is, honestly. This is a story about a woman who earned 17 billion views for her Snapchat filters and made … $4,000. And she had to turn to Etsy to actually monetize her work. All of this is awful.

Since October 2018, Casciello has created about 30 Snapchat lenses, which overlay digital objects and special effects on photos and videos. One gives you a crown of smiling emojis, and a “Happy Vibe” filter gives you a sun-kissed look. Her creations have racked up over 17 billion views and helped her amass over 214,000 followers on the platform.

Casciello, who will attend Virginia Tech in the fall to study computer science and engineering, sells a packages of filters on Etsy for up to $49. They can be downloaded and edited for use on platforms like Instagram. She’s made more than $4,000 since launching her online shop in June.

Résumés Are Starting to Look Like Instagram—and Sometimes Even Tinder

Chip Cutter reports that people are adding Bitmoji to their resumes:

At a high school near Indianapolis, Ind., an applicant this spring sent a digital résumé for a teaching post with his bitmoji waving and the word “hi” at the top, says Erica Posthuma, a chemistry teacher at the school. Ms. Posthuma learned of the résumé when an assistant to a top administrator saw it, laughed and told Ms. Posthuma: “There’s a freaking bitmoji on the résumé.”

Ms. Posthuma describes her own CV as “dry as hell.” The school ultimately felt the bitmoji résumé was inappropriate.


Facebook now lets anyone make an Instagram face filter

Snap has a lead over Facebook in this regard, since it’s been letting people make their own filters for quite some time now. I’m curious how long it will take Instagram to catch up.

Facebook’s tool for building augmented reality effects, Spark AR, now allows anyone to make custom face filters and other effects for Instagram Stories. The platform was previously limited to approved creators, but today’s update will let anyone create and upload their own AR filters to Instagram. Instagram is also introducing an Effect Gallery tab on artists’ profiles, letting artists display their creations.

YouTube tests bigger thumbnails, and people hate it

I mean, I don’t hate it:

YouTube is experimenting with making video thumbnails much larger on its homepage, according to several screenshots that appeared on Twitter and Reddit this morning.

The new homepage design, seen in the images below, changes quite a bit about the current setup. Videos are no longer grouped by categories, fewer videos appear in a line for people to scroll through, and, yes, the thumbnails are noticeably larger. It’s unclear just how many people have been served the new layout. We weren’t able to get the design to appear ourselves, so it’s likely a limited test for now. A YouTube spokesperson told The Verge the company is “currently testing a new homepage layout to improve the watch experience for our users.”

YouTube tests letting ad bans be appealed by filming a YouTube video

This feels like the slowest possible way to handle these appeals?

YouTube is experimenting with a new program that will allow creators whose ad privileges have been revoked to appeal the company’s decision using “short videos.”

A review team will allow creators with demonetized channels to send in a video that discusses their channel’s content and explains their “creative process,” according to an email posted on a popular subreddit. The review team will then make an assessment within seven days about whether monetization should be reinstated, the email states.


Despite bias claims, Facebook is not censoring conservatives

Natalie Martinez saddles up on one of my favorite hobby horses:

A new 20-week Media Matters study on Facebook pages that regularly post about American political news again found that right-leaning pages and left-leaning pages have nearly identical engagement rates, while right-leaning pages on average earned more weekly interactions than left-leaning pages. 

Give the F.T.C. Some Teeth to Guard Our Privacy

I endorse this take from Jessica Rich, former director of the FTC’s Consumer Protection Bureau:

Congress has repeatedly declined to enact a broad-based federal privacy and data security law setting strong privacy standards, codifying penalties for wrongdoers and allocating the staff and funds necessary to enforce the law nationwide.

It’s not as if no one has thought about this issue. Since the late 1990s, consumer advocates, industry leaders and the F.T.C., among others, have at various times urged Congress to pass such a law, but to no avail. With a few exceptions, the F.T.C.’s legal authority over privacy is the same as it was before the internet was invented.

And finally ...

Talk to me

Send me tips, comments, questions, and trends: