Can you spot the fake? Probably not! Earlier this week I covered a report that found that U.S.-native (i.e., not from Russia and the like) fake news is actually increasing on Facebook, a year in advance of the presidential election. The report, from Avaaz, calculated what it said were the 20 most-viewed fake news stories. As of Thursday afternoon, all 20 were still up on Facebook.

That’s not particularly surprising, since (a) Facebook says it demotes but doesn’t remove fake news, and (b) Facebook doesn’t identify most of the posts that Avaaz identified as fake, despite all of them having been debunked by various fact-checkers.

As a test, I tried sharing the top 20 fake stories that Avaaz found on Facebook, to see what got flagged. [Laura also asked her Nieman Lab colleagues to do the same, so apologies to any of our friends and family who thought we seemed suddenly conspiracy-friendly. —Ed.] I’d never tried to share fake news to Facebook before, and the process was illuminating.

In six out of the 20 cases, Facebook flagged the posts as containing “False Information” and noted that, if I went ahead and shared them, a fact-checking notice would be attached to my post. But when I went ahead and clicked “Share Anyway,” I found that the stories posted to my timeline without any warning attached to them. I would have expected some kind of overlay, like the one you now get when you try to share an image that Facebook’s fact-checkers have determined to be false. Instead, the posts look normal with the exception of a little “ⓘ” button that you can click on (but likely would never notice) for more info.

a little "i" that you might not notice, nothing else though. It's better for fake images. but this all seems to be post- those changes they made last month and i am surprised by un-blatant it is for news stories. pic.twitter.com/vQHaalU5rY

— Laura Hazard Owen (@laurahazardowen) November 7, 2019

Here are the stories where Facebook gave me a share warning:

1. Trump’s grandfather was a pimp and tax evader; his father a member of the KKK

8. Tim Allen quote: Trump’s wall costs less than the Obamacare website [Because this was an image, it was also overlaid with a “False information” warning]

10. Joe Biden Calls Trump Supporters “Dregs of Society”

12. Bernie Sanders Says Christianity Is An Insult To Muslims

14. MN Democrats vote for elementary school pornography

19. “President Trump is asking everyone to forward this email…” [image]

Facebook didn’t flag the remaining 14 stories as fake. Three of these “stories” were actually Facebook statuses where the status contained the false information.

2. Pelosi Diverts $2.4 Billion From Social Security To Cover Impeachment Costs

3. Ocasio-Cortez Proposes Nationwide Motorcycle Ban

4. Trump is now trying to get Pence impeached

5. Ilhan Omar holding secret fundraisers with Islamic groups tied to terror

6. BREAKING: Nancy Pelosi’s Son Was Exec At Gas Company That Did Business In Ukraine

7. Democrats Vote To Enhance Med Care for Illegals Now, Vote Down Vets Waiting 10 Years for Same Service

9. NYC coroner who declared the death of Jeffrey Epstein a suicide made half a million dollars a year working for the Clinton Foundation until 2015 [This was posted as a status, not a link, which made it extremely easy to re-share]

11. Mitch McConnell campaign essentially says boys will be boys after Alexandria Ocasio-Cortez blasts disturbing groping pic [This one’s sort of complicated; debunk here]

13. Democrats Wrote to Ukraine in May 2018, Demanding It Investigate Trump [this Breitbart story was shared by Trump on his own Facebook page]

15. Democrats don’t mind executing babies after birth [Trump status on his own Facebook page]

16. Olive Garden is funding Trump’s re-election in 2020 [posted as a status, with no link]

17. Alexandra Ocasio-Cortez said “A society that allows billionaires to exist…is wrong” [presented as an image; fact-check here]

18. A different time Trump said Democrats don’t mind executing babies after birth [Trump Facebook status]

20. Democratic party passes resolution against Christianity

Avaaz came up with an “estimated views” metric for each post, “based on the cumulative CrowdTangle data for each post across the pages, groups and profiles featuring the post on Facebook. (See page 15 here for more detail.) But it’s a metric that Princeton professor Andy Guess says is flawed. Based on Avaaz’s estimates of how many times each of these 20 items were viewed, the six posts Facebook flagged have been viewed about 45.6 million times. The 14 posts that Facebook didn’t flag have been viewed about 93.5 million times, per Avaaz. But skepticism of those figures seems warranted; they may be too high.

Issue: View stats not available for non-video posts.

Their solution: Compute a global ratio of views/engagements for videos only, then multiply by available metrics across the board to all posts.

Problems: Videos≠text, averages over HIGHLY skewed share distribution=inflated #s https://t.co/G8btPuNG6h

— Andy Guess (@andyguess) November 7, 2019

“Like the experts predicted back in 2016, we did end up heading down the dystopian path.” Researchers looked at the experiences of Muslim Democratic House Reps. Ilhan Omar and Rashida Tlaib running for Congress in 2018 and found that an organized Islamophobic social media campaign against them kicked off nearly as soon as they announced their candidacies. “These operations largely replaced Breitbart and other extreme-right media entities that were the primary source of anti-Muslim dialogue in the 2016 presidential campaign,” write Lawrence Pintak, Jonathan Albright, Brian J. Bowe, and Shaheen Pasha. The authors wrote in The New York Times:

The sheer number and proportion of negative tweets indicate that much of the targeting was done by people or organizations from far outside the districts in which the candidates ran. Our review of profiles of those accounts, which included 2,354 that attacked both women, bears this out.

But the most striking thing we uncovered happened in the months after the election. When we revisited these accounts in July, a significant portion of them were simply gone. Some had been suspended by Twitter for violating standards, such as posting inappropriate content or showing characteristics of bots. Others had been deleted by the account holders. Malicious actors will often remove the accounts that make up their bot networks — like drug dealers tossing burner phones — to cover their tracks. […]

This all suggests that this Islamophobic and xenophobic narrative largely originated with a handful of bigoted activists and was then amplified by vast bot networks whose alleged owners never existed. “Ordinary” account holders, many retweeting just one post, were then swept up in the rancorous energy of the crowd.

Jonathan Albright in Wired:

This is a new twist to electoral politics and democratic participation in 2020 and in the coming decade. Over time, and especially across disparate Twitter communities, groups, and hashtags, these tactics will continue to surface anger and emotional vitriol. They will connect political candidates’ identities to controversial issues, raising them in tandem, and then connecting them in the form of a narrative to real voters. This manufacturing of outrage legitimizes otherwise unsustainable rumors and ideas.