A new report indicates that the ecosystem of false information is still flourishing, two years after the 2016 elections

A few weeks before the congressional elections, the report prepared for the John S. and James L. Knight Foundation, which supports research on journalism, the arts and other subjects, revealed that over 80% of Twitter accounts mentioned frequently The links shared with false information during the 2016 election remain active. As a group, they publish more than a million tweets a day.

The researchers also found that 65% of the links to false information and conspiracy contained 10 prolific websites, which calls into question claims that the creation of such content is too widespread and diffuse for technology companies can fight effectively.

"False news that matter is not organic, small-scale, or spontaneous," says the 60-page report. "Most of the false information on Twitter points to some established conspiracy and propaganda sites, and coordinated campaigns play a crucial role in spreading false information."

Twitter, which received an advance copy of the report, said it was working to suspend accounts it considered fake or producing spam. Sometimes the company also "locks" suspicious accounts, which means other users can not access them before they can prove to Twitter that they are legitimate.

"As a unique open service, Twitter is a vital source of real-time antidote against daily for-profit shops," said Del Harvey, global vice president of trust and security for Twitter, in a statement. "We are proud of this use case and work diligently to ensure that we show the diverse context and perspectives of the people who engage in a civic debate about our service."

The Knight report researchers – Matthew Hindman, associate professor of public relations and media at George Washington University, and Vlad Barash, scientific director of network analytics company Graphika – have reviewed Twitter account links with more than 600 sites "regularly posting unverified stories or blatant lies. They declined to name key accounts, but said their global list came from an open source list of fake news and conspiracy sites kept on opensources.co.

Over 73,000 accounts linked to these sites 10 or more times in the 30 days leading up to the 2016 elections.

Although the authors have found much evidence of the importance of "robots" in the disinformation networks they have studied, accounts led by real humans may have been even more important. One-third of the most-watched Twitter accounts related to misinformation appeared to be bots, but more were human.

These users who spread false information on Twitter in some cases did so without their knowledge because they believed this information, said Barash. He added that this complicates efforts by Twitter and other technology platforms to fight the spread of misinformation.

"Many of their users consume this information, and cutting it wholesale is not necessarily a source of satisfaction for them," Barash said. "It's a very complicated situation."

The report of the Knight Foundation tells an apparent success in the fight against false information and conspiracies. Therealstrategy.com, one of the leading production sites for this content in 2016, has seen its account suspended by Twitter and other platforms. The website has apparently since been taken offline and the amount of content it has produced has dropped from over 700,000 links during the pre-election period to about 1,500 after.

The website, whose owners have not responded to efforts to contact them by e-mail and phone, still has an active site on Facebook, which said it was reviewing the allegations made against it after receiving Washington Post questions about it.

"The false news is not hundreds of accounts, and [fighting it] is not Whack-a-mole, "said Hindman. "These are a few dozen persistent sites that do it all day, every day."

The report, because it was prepared for a private client, was not subjected to a formal peer review process that would be a routine for published academic work, but the authors indicated that it had been examined informally by other academics.