Verifying social media content with the human algorithm

by Margaret Looney
Oct 30, 2018 in Social Media

With robots telling newsrooms when (and when not) to push out social media updates, how long before they take on the task of verifying the accuracy of content shared via social media?

Not so fast, said Claire Wardle, director of news services for Storyful. She tackled the issue of journalist vs. robot in a recent chat with IJNet.

Storyful uses an algorithm to discover relevant conversations on social media and uses journalists to verify it for other news organizations.

Wardle says some of Storyful’s manual verification techniques, such as conducting reverse image searches, could be automated, but a journalist’s overseeing eye is still needed for more contextual tasks.

“We often compare some of what we do with police work. It's about collecting clues. While some of it can be automated, I would be very worried about the trickier aspects of verification being automated,” she said. “I think discovery on social platforms could and should be automated, or improved using technology (and we're working on that!) but verification is a different matter.”

Wardle says the organization has more than 600 hand-curated Twitter lists that follow trustworthy sources worldwide, but they do use technology to monitor the noise levels.

"We can't be watching those streams all the time (our eyes would pop out of our heads!), so we use technology to monitor the speed at which the lists move, to be alerted when lists become more 'noisy,’” she says. “Sometimes lists start becoming more noisy because it's a news event, but sometimes it's because there's been a plot twist in a soap opera. We need to have journalistic instincts to be able to use the technology to get us so far and then do the rest.”

With more than 40 columns open on their TweetDeck accounts, Storyful journalists also rely on the help of developers to simplify their discovery process.

“When a breaking news story happens, our journalists will have multiple tabs open in their browsers searching YouTube, Twitter, Facebook, Instagram, Reddit," she says. "Our developers realized that they could build a pretty simple multi-search Google extension that would allow these searches to be done much more quickly.”

Wardle also points out a difference in verification standards that she ties to age. “I see a generational difference. With younger journalists, they tend to assume that what they're seeing is true, and grudgingly run checks, just in case it's false,” she says. “With older journalists, they're a lot more skeptical, and assume that what they're seeing is false, until they can prove otherwise.”

Image CC-licensed by Flickr via Mark Strozier.