Friday, May 17, 2019

Keep Up with the Falsehoods


US court orders Bayer to pay $2bn in Roundup cancer case FT

Uber Shares Are Quickly Approaching ‘White Knuckle’ Territory Bloomberg (DL). Pardon my schadenfreude.

AI can now defend itself against malicious messages hidden in speech Science. We keep hearing nobody knows how AI “deep learning” works; this article says it several times. What could go wrong?

Keeping up with the falsehoods

We are now 18 months from the United States presidential election, but the challenges facing reporters and fact-checkers assigned to cover Donald Trump’s re-election bid are already clear.
Some of them are challenges journalists have faced before, but taken together they add up to an escalation that illustrates how hard it will be to ensure readers are getting the truth in the run-up to next year’s voting.
Among them:
  • The growth in the number of the president’s falsehoods is accelerating. As Daniel noted in a piece on how The Washington Post’s Fact Checker tallies Trump’s false or misleading claims, there were 3,251 falsehoods from Trump between January 2017 and June 2018. President Trump crossed the 5,000 milestone 601 days into his term, in September.  It took him only 226 more days to cross the 10,000 mark, according to The Fact Checker’s database.
  • The president and his allies are continuing to retweet from the fringe. Last weekend, his Saturday morning tweets included some propaganda from far-right groups, including one video from an account called Deep State Exposed that said, “The ‘elite’ proclaim America must submit to Islam or else!!!” A senior campaign aide this week also tweeted a 2015 video showing a Ukrainian rocket launch, suggesting the footage was of rockets being fired into Israel from the Gaza Strip and using it to criticize Rep. Ilhan Omar (D-Minn.).
  • Even when something Trump has said or tweeted is proven false, the president is doubling down rather than backing off. A good example came this week when Trump repeated a false claim that the U.S. has “given” Puerto Rico $91 billion in aid, which has been checked and debunked by The Post, the Associated Press and the Orlando Sentinel, among others.
  • Marginalized groups like Latino, Muslim and Jewish communities are disproportionately targeted by misinformation online, according to a new study covered by BuzzFeed News this week. When messages are co-opted from these groups by far-right extremists, vulnerable communities are less likely to engage in politics online.
  • Fact-checking might not always help. That’s according to another study covered in Nieman Lab this week, which found that “dueling fact perceptions are rampant, and they are more entrenched than most people realize.” And fact checks rarely reach the people who need to see them.
The repetition of false claims — often talking points politicians love to repeat because they fire up a specific slice of their base — and their amplification by online extremists is something we’ve noted before. And it’s likely to continue as the campaign heats up.
Combined with the other two phenomena — a growing number of falsehoods and the potential for them to arise from conspiracy theorists — the repetition means fact-checkers and reporters will need to be especially diligent about not only debunking misinformation, but also putting it in context. Why for example, does Trump want you to believe the United States has sent $91 billion to Puerto Rico? Where does that falsehood come from?
The need for context was highlighted in a study this week from the nonprofit group Media Matters for America, which monitors misinformation on the right. The report said major media outlets often just amplify Trump’s claims by “passively” repeating them in a headline or tweet, even if the claims are debunked in the body of a story. The group used The Post’s Trump database to conduct the research.
It’s lost on no one that providing this kind of context takes resources and time, both of which are hard to come by in today’s newsrooms. Eventually, automated systems such as those being developed by Bill Adair and his team at the Duke Reporters’ Lab, which was profiled in The Atlantic this week, could help us plow through the mountain of misinformation that continues to come out of Washington.
In the meantime, as Kessler told Daniel, the number of falsehoods has “become bit of a burden because it consumes so much time.” Kessler, too, is looking at an onslaught as the 2020 race heats up.
“I don’t know what we’ll do when it comes to campaign season and he’s holding three rallies a day,” Kessler said.

… technology

  • In an expansion of Facebook’s partnership with fact-checking outlets, Instagram has started limiting the spread of false posts, Daniel reported Monday. Posts that are debunked by fact-checkers will be removed from Instagram’s Explore tab and hashtag result pages. But Gizmodo wants to see more data about how the fact-checking project is already working on Facebook — and so do we.
  • Speaking of Facebook, on Monday, the company announced that it had removed 97 Facebook accounts, pages and groups for coordinated inauthentic behavior originating from Russia and focusing on Ukraine. The removed accounts, pages and groups had a combined following of more than 100,000 users.
  • Following the fire at the Notre Dame cathedral in Paris last month, misinformation spread far and wide on social media — much of it aimed at villainizing Muslims. This week, the Agence France-Presse profiled two students who became the targets of online hate after a photo of them spread with a caption wrongfully blaming them for the fire.

… politics

  • “The heroic efforts of India’s small band of independent fact-checking organizations should be bolstered with public and private support,” wrote Samir Patil of Scroll.in, an India news and fact-checking site, for The New York Times this week. Misinformation has plagued the election there for the past several weeks — much of it occurring on private messaging apps like WhatsApp, The Financial Times reiterated in a story this week.
  • In the Philippines, fact-checkers were attacked again — this time for an non-bailable charge. Daniel reported on Friday that both Rappler and Vera Files were wrongfully accused of coordinating a government destabilization plot by a newspaper that’s owned by a spokesman for President Rodrigo Duterte.
  • The Spanish election was last weekend — and fact-checkers were on the frontlines of misinformation about the campaign. IFCN associate director Cristina Tardáguila spoke with two outlets that debunked false claims about the election. And here’s how Maldita.es used a reader survey to bolster its work.

… the future of news

  • Full Fact, Africa Check and Chequeado are among the 20 international winners of the Google Organization’s AI Impact Challenge. The fact-checkers will work with the Open Data Institute for three years to “use artificial intelligence to dramatically improve and scale global fact-checking efforts.” The four organizations will share $2 million in funding.
  • Several American companies are hiring firms to analyze social media for misinformation about their business. Why? False posts can tank companies’ stock prices, NBC News reported.
  • Will TikTok be the next social media platform to get swarmed with conspiracy theories? It is billed as a fun and benign video app that’s full of memes, but The Atlantic’s Taylor Lorenz has found a few posts mentioning conspiracy theories, and Motherboard wrote in January about how the same algorithms that power other platforms struggling with misinformation also power TikTok.
Each week, we analyze five of the top-performing fact checks on Facebook to see how their reach compared to the hoaxes they debunked. Read more about this week’s numbers, and how false quotes are still among the most common types of misinformation on social media, here.
  1. Factcheck.org: “Social Posts Spin Harris’ Gun Control Proposal” (Fact: 2.5K engagements // Fake: 2.1K engagements)
  2. Faktisk: “This is not a Swedish woman who has been raped by immigrants” (Fact: 1.7K engagements // Fake: 432 engagements)
  3. Estadão Verifica: “When Disinformation Can Kill: Baking Soda Is Not Remedy Against Cancer” (Fact: 1.7K engagements // Fake: 71.6K engagements)
  4. Agence France-Presse: “Assimilation of immigrants: the Theodore Roosevelt quote falsely attributed to former Canadian PM Laurier” (Fact: 380 engagements // Fake: 256K engagements)
  5. Agência Lupa: “Gleisi did not say Senna was a ‘car racer and nothing else'” (Fact: 350 engagements // Fake: 4.1K engagements)
Misinformers frequently make things up about politicians and other public figures. Rarely do they fabricate an entirely new one.
But that’s exactly what happened in India, where a fictitious member of the legislative assembly (MLA) named Anil Upadhyay has become the subject of viral misinformation during this month’s general election.
Fact-checking site Boom Live debunked the bogus politician May 2, reporting that the character first appeared in a Twitter video posted April 8. The user falsely claimed that the politician beating a police officer in the video was Upadhyay when in fact it was a counselor from Uttar Pradesh.
Since then, Upadhyay has been blamed for everything from attacking a man in Gujarat to praising Prime Minister Narendra Modi as a member of the opposition party. Boom found that false posts mentioning the MLA are all over social media — but the politician doesn’t exist.
What we liked: In its analysis, Boom broke down the origin of the Upadhyay hoax, how it spread on social media and what kind of effect it has had on social media conversations about the election. The fact-checker took a pretty bizarre piece of misinformation and made it easy for anyone to understand.
  1. Last week, Poynter and the IFCN published a list of unreliable news sites based on five lists from fact-checkers and academics. Then, we retracted it. Here’s why.
  2. Facebook and Instagram banned several far-right personalities who have spread misinformation in the past, including InfoWars’ Alex Jones. But there are still loopholes.
  3. Writing for Wired, Paris Martineau spoke to someonline extremism and disinformation researchers who are burned out from the constant barrage of bad news.
  4. BuzzFeed News reported on how spammers have been hijacking Facebook pages with large followings to harvest advertising money.
  5. Speaking of ads, how does their transparency affect voter behaviors? Teyit analyzed that question in the Turkish context.
  6. Misinformation about New York schools is spreading in Chinese on WeChat, CJR reported.
  7. This Guardian longread about why people believe conspiracy theories is good for many reasons, but perhaps the best is the fact that the writer went on a cruise called “Conspira-Sea.”
  8. In Latin America, fact-checkers are collaborating to cover the ongoing crisis in Venezuela.
  9. More than half of Europeans might have seen some form of disinformation promoted by Russian actors on social media, Politico reported.
  10. Old and out-of-context photos and videos were sharedon social media after Cyclone Fani hit the Indian state of Odisha on Friday.
via Daniel and Susan