Twitter users aren't known for staying quiet when they see something that's flat out wrong, or with which they disagree. So why not harness that energy to solve one of the most vexing problems on social media: misinformation?
With a new pilot program called Birdwatch, Twitter is hoping to crowdsource the fact-checking process, eventually expanding it to all 192 million daily users.
"I think ultimately over time, [misleading information] is a problem best solved by the people using Twitter itself," CEO Jack Dorsey said on a quarterly investor call on Tuesday.
With Birdwatch, users provide context and information, linked directly below a questionable tweet.
"The idea is that people would be able to come away from Twitter better informed," Keith Coleman, Twitter's vice president of product, told NPR.
Twitter has taken many steps over the last year to curb misinformation: It's slapped warning labels on false claims about election fraud and the coronavirus pandemic. It's added links to credible news outlets that debunk those claims. And it's booted users who repeatedly break its rules.
But scaling up those efforts has been a challenge. False information can spread rapidly, but Twitter's labeling process takes time. And the company thinks some tweets that don't break Twitter's rules could still benefit from more context and information.
With Birdwatch, the company is testing out whether crowdsourcing can augment the fight against misinformation.
"We think the community can respond really quickly [and] that they can cover cases that go beyond Twitter's policies," Coleman said. "Today Twitter intervenes or annotates in cases that are in conflict with our policies. But the community could do that on cases that may be in grey areas."
How Birdwatch works
Birdwatch combines the crowdsourcing of information, familiar to users of Wikipedia, with ratings similar to Reddit's system of voting up or down on posts.
Users can write notes on tweets, flagging them as misleading or false. They can even add links to their sources of information.
They also rate each other's notes — a key part of the program. Twitter uses those ratings to put the most helpful notes at the top of the list, and to build a reputation profile for Birdwatch users.
In this pilot phase, Birdwatch notes only appear on a separate section of Twitter. Eventually, the idea is to make Birdwatch part of Twitter's main platform.
With just a thousand participants in the pilot phase right now, it's hard to tell whether the project is working as Twitter intends.
Some people are using it to correct misleading information, like linking to research showing masks help prevent the spread of COVID-19 on a tweet claiming otherwise.
But there's also a lot of partisan bickering and opinion, especially on tweets from politicians.
"It's sort of just replicating what we see on Twitter," said Madelyn Webb, a researcher at the nonprofit First Draft, which combats misinformation. "Things go viral, everybody wants to talk about them and then the rest of it sort of falls to the wayside."
Twitter's Coleman says the early phases of the program may not reflect the company's vision for Birdwatch.
"We knew we would get a mix of quality," he said. "The key is, how can we identify the best of it? How can we provide a rating system that allows the community to provide feedback so we can pull out the best of it? And ultimately, how do we encourage more of that over time?"
Who checks the fact checkers?
Perhaps the biggest challenge for Twitter is ensuring that Birdwatch does not recreate some of the problems it's trying to address — like becoming a venue for misinformation.
"What if there's an accurate tweet and someone fact checks that to say something inaccurate, then who fact checks that, and then who fact checks that?" asked Molly White, a long-time Wikipedia editor.
White is also wary of the potential for harassment and "brigading" — when people gang up on a particular post or user.
"The first thing I thought of when I saw the announcement about Birdwatch is, like, how on earth are they going to try to protect these people?" she said.
Coleman says Twitter shares those concerns about manipulation and harassment.
"We do expect to see brigading and things like that happening on Birdwatch," he said. "But we believe that we can design it to work differently than Twitter in a way that can handle that."
That means changing the way social media incentivizes users. On Twitter, people are rewarded for creating posts that go viral. They get a lot of reactions, such as replies, retweets and likes. And they gain more followers, and visibility.
In contrast, Birdwatch users are encouraged to develop reputations as being helpful and credible. Birdwatch plans to display prominently the notes and ratings from the most reputable users.
Ultimately, it's an experiment designed to find out whether Twitter users can trust each other more than they trust Twitter itself to verify information.
"We know that not everyone trusts a single tech company or any singular institution to make the decisions about what context to add and when," Coleman said.
Can you build consensus when people disagree about the truth?
Some critics of tech companies say it's encouraging that Twitter is trying something new to address the fraught fight over misinformation and who gets to set limits on online speech.
"It is good to see companies getting creative beyond, 'Should this content be up or down?' or 'Let's remove more content faster,'" said Evan Greer, deputy director of the nonprofit Fight for the Future.
"But like with anything else, the devil will be in the details," she said. "Will we be able to see how decisions were made, who made them? And will there be due process and mechanisms in place to prevent the system from being abused?"
What's more, for Birdwatch to succeed, it will have to overcome an uncomfortable truth not just about Twitter, but all social media in a bitterly divided nation.
"People on Twitter do not agree on what truth is. They do not agree on what real news is. And that's a problem," said Tiffany Li, a Boston University law professor who studies technology.
"If they don't agree on what the truth is, they're going to have different opinions on whether or not a tweet or a post is truthful. And I don't think that Birdwatch can really solve this divide."