Noted - Fake News, Future Trends and Visual Audio

Every two weeks, News Labbers come together to share project updates and interesting developments from around the News industry. We’ve decided to pass some of this onto you. This month: tools to beat fake news, predictions for journalism in 2017 and visualising audio content.

Trust Tools for News

Since holding the Trust Project Challenge with Connected Studio and the Trust Project, we’ve been following how organisations are responding to issues of trustworthiness in news.

  • Slate released its Chrome extension “This is Fake”, which flags fake stories in your Facebook news feed and links out to debunking articles. It builds in “antiviral functionality” by encouraging readers to share a link to the debunking article as a comment on the original post. Stories from websites that repeatedly spread fake news — identified by running a match against a database put together by Slate — appear under a different banner.

screenshot of Facebook post tagged with "This is Fake"
A news article tagged by Slate’s “This is Fake” plugin.

  • A similar database-building effort is taking place in Europe at Le Monde. According to Digiday, the newspaper’s thirteen-person fact-checking team are helping to build up a list of untrusted sources that can then be matched against a site’s domain. Users who download the publisher’s browser extension will see one of three flags on articles they open: red for untrusted, yellow for suspicious and green for trusted.
  • Reuters has built a bot that can find and verify breaking news stories on Twitter. Two years in the making, News Tracer works by identifying story clusters and assigning verification scores to tweets based on 40 factors, according to a report in Nieman Lab from November. Story scores change as more information comes out, and stories with scores above a certain threshhold are tweeted out as breaking news alerts.
  • The Washington Post has developed a Chrome extension that fact-checks Donald Trump’s tweets, providing additional context to statements from the U.S. president-elect’s account.

screenshot of tweet annotated with the Washington Post plugin
A Trump tweet annotated with the Washington Post’s browser extension.

If you’re interested in Trust Project Challenge prototypes, you can check out its Github repository with links to teams’ projects and presentations. Melody Kramer also recently interviewed attendee RC Lations from the Institute of Nonprofit News for her media innovation column in Poynter. Lations’ team worked on improving transparency by adding additional metadata to author bylines.

Nieman Lab’s annual round-up of industry predictions for the New Year is heavy on possible solutions for fake news. Here are some other predictions we found interesting:

  • Artificial intelligence as a tool for journalists - rather than the usual talk of it as a way for users to interact with news content. At News Labs, we’re kicking off the New Year with a hackathon exploring uses of machine-assisted transcription, so we may see some AI there. We hope transcription will serve as both discovery tool and workflow aid in the form of radioReader and our Online Content Toolkit (OCTO).
  • Bots for fact-checking and moderation, versus chatting or distributing content. At our last get together before the holidays, News Labbers heard about work that the Guardian’s development team has done to automate parts of their comment moderation. And at the Trust Project Challenge, we were interested to see that many of the teams had come up with their own indicators of untrustworthy news — some of which appeared across prototypes and may hint at ways for bots to identify fake content.

NB: we were also pleased to see that News Labs received a mention for our Juicer API, which ingests content from news organisations around the web and helps users analyse the content.

Radio.Garden and Visual Audio

We’ve recently been working to configure Audiograms, a tool developed by New York Public Radio and open-sourced on GitHub. The BBC’s social media teams are keen to use it to share radio content on Facebook, Twitter and Instagram.

Radio.Garden might not seem related, but it operates in the same general space of visual audio. Built for the Netherlands Institute of Sound and Vision by design firms Studio Puckey and Moniker, it makes radio stations all over the world visible by location, and allows users to tune in at any point on the map.

It got us thinking: could there be a way to design and develop mapped audio content for BBC Radio? Alternatively, is there a way that maps could be used as visual metadata when sharing otherwise-invisible audio content on social media and online?