Inside Twitter’s effort to change conversations with topics

Twitter’s plan to nudge users into interacting with accounts they don’t follow is starting to take shape.

Three months after introducing topics, which allows Twitter users to follow interests the same way they’d follow individual accounts, the feature is rapidly expanding. While initially limited to sports-related topics, there are now nearly 1,000 topics available to follow. 

Astrology Twitter. Knitting Twitter. Bird Watching Twitter. Bobsledding Twitter. Cat Twitter. With these topics, it’s easier than ever for Twitter users to browse new (and sometimes strange) corners of the platform they might not normally view. 

Topics is the latest evolution of a years-long effort to help make Twitter more accessible. It’s also one of Twitter’s first major attempts to make its platform feel less like one big echo chamber. According to Pew Research, 80 percent of tweets come from just 10 percent of Twitter’s users. By helping Twitter users more easily find and follow their interests, the thinking goes, they’ll perhaps be more likely to become engaged users. 

“By injecting other people into your home timeline you wouldn’t normally talk to, there’s opportunity for conversation, replying, and follows that you might not have been able to see,” Twitter product trust partner Savannah Badalich told Mashable. “The big thing is we want you to participate.” 

But that potential for enhanced engagement breaks down if topics become spammy, abusive, or just plain irrelevant. 

Badalich says that each topic goes through a quality assurance process that reviews topics both before they are pushed live in the app and after users start to follow them. 

“We look at a bunch of different signals to tell us whether we’re surfacing healthy content,” she said.

Those signals can include reporting and blocking, but also whether or not specific tweets are causing a topic to be unfollowed. Or even whether someone opts to see a topic “less often.”

Still, it can be an imperfect process and Badalich’s team has already had to adjust topics that aren’t working as intended.

At one point, the biology topic began surfacing blatantly transphobic tweets to some users. The tweets weren’t necessarily in violation of the company’s rules, but they were hardly representative of the “healthy conversation” Twitter says it wants to encourage. 

“For biology, for example, there are a lot of tweets like that we’re sort of on the cusp of, not necessarily violative, but obviously folks didn’t want to see it and so they were reporting it,” Badalich says. “And so as we saw things like those types of reports, we can tweak it.”

In the case of the biology topic, it was eventually paused so Twitter could make those “tweaks.”

The biology topic isn’t the first time one of Twitter’s algorithms has amplified the wrong kind of tweet. But for Twitter, which has pledged to make dialogue on its platform healthier, it’s an example of how the company might be learning from past mistakes. By taking an active role in moderating what tweets get amplified, the company’s better able to make those kinds of adjustments more quickly.