‘Junk news’ still rising, study finds, as US midterms near

The spread of “junk news” on Twitter and Facebook has increased since the 2016 election, despite the companies’ investments in people and technology to try to control disinformation that could influence next week’s US midterm elections.

A new report from the Oxford Internet Institute found that the proportion of “junk news” about politics — what the researchers call deceptive, extremist and conspiratorial content — increased on Twitter by five percentage points since Donald Trump was elected president.

Twitter users posting about politics shared more “junk news” than professional news outlets, the study showed, and about 25 per cent of all links shared on Twitter now lead to sites the researchers identified as junk, up from 20 per cent two years ago.

Only 19 per cent of the links were to the sites of known, legitimate news organisations, the institute found. Researchers did not measure how many people these links reached.

The study said a large number of the Twitter accounts linked to the spread of misinformation during the 2016 US elections were still actively spreading junk news.

On Facebook, the audience for junk news had spread beyond supporters of President Trump and the far-right to include mainstream conservative readers, the study found.

There was also now a small but increasing population on the far-left consuming junk news on Facebook, for example, spreading misinformation about Trump administration policies towards the LGBTQ community.

Philip Howard, director of the Oxford Internet Institute, said more of the problematic content appeared to originate from the US, rather than the disinformation campaigns conducted by Russia and other foreign powers.

“So much of this junk news is actually homegrown, it is not from the crazy Venezuelan government, it is not coming from Iran, it is from the US: massive networks of junk news, multiple sites, sharing the same kind of content: radical, extremist, sensationalist,” he said.

The institute’s researchers used a sample of 2.5m tweets based on a curated list of hashtags and usernames, collected over 30 days, a period which included the hearings for the nomination of Supreme Court Justice Brett Kavanaugh, when partisan sentiment peaked in the US. The study used Twitter’s public streaming application programming interface (API), which represents no more than 1 per cent of the global Twitter stream.

As Facebook does not make as much information public, the study relied on information from almost 7,000 Facebook pages from the same period.

Other studies have come to different conclusions. One from researchers at NYU and Stanford published this month said user interactions with false content had fallen sharply on Facebook since the end of 2016, while continuing to rise on Twitter.

Another from the University of Michigan said there had been a gradual decline in the proportion of misinformation on Facebook since mid-2017.

Twitter contested the Oxford Internet Institute findings, arguing that many of the links deemed “junk” by the researchers are “established media outlets reflecting views that exist within American society”.

A spokesman said: “Banning them from our service would be a knee-jerk reaction and would severely hinder public debate, the potential for counter narratives to take hold, and meaningful discussion of news consumption.”

The spokesman added that users sharing such content were not foreign, not bots, and, for the most part, not co-ordinated.

A Facebook spokesperson called the conclusions of the study misleading. “We’ve seen Facebook-specific takeaways from other academic bodies that help paint a clearer picture of the state of false news on Facebook,” he said.

Both companies have been trying to clean up their platforms, shutting down fake accounts and bots, working with outside bodies such as researchers and fact-checking organisations, and launching news literacy projects. This week, Twitter partnered with Unesco, the United Nation’s cultural organisation, to promote information literacy.

But Mr Howard said he did not think their “tweaks” had had a “systematically positive impact”.

“If two years’ worth of initiatives has amounted to a slight increase in junk news, then I think there may be some deeper solutions we need to consider,” he said.