Produced by MapLight, the Decoder is a newsletter to help you track the most important news, research, and analysis of deceptive digital politics. Each week, we'll send you coverage of the webs of entities that seek to manipulate public opinion, their political spending ties, and the actors working to safeguard our democracy. Know someone who might be interested? Ask them to sign up!
- Facebook bans vaccine misinfo: The United Kingdom begins vaccinations for COVID-19 this week, but concerns about people refusing a vaccine due to misinformation persist. To address the issue, last Thursday Facebook announced that it would begin removing false claims about vaccines on Facebook and Instagram. In The Verge, Kait Sanchez details why the change may come too late, while worries about Facebook’s ability to actually enforce the problem abound (especially in light of recent failures addressing Georgia electoral disinformation). The Oxford Internet Institute’s Project on Computational Propaganda also has a timely new paper out on how big tech has failed to curb (and in some cases enabled) the proliferation of occasionally lucrative health misinformation. CNN has a handy list of how other platforms are handling misinformation about vaccines.
- The electoral disinformation threat: Tuesday, a Republican election official blasted Trump as well as GOP senators for “inspiring people to commit potential acts of violence.” The official, Gabriel Sterling, even said that a noose had been left at the door of a Dominion Voting Systems contractor after a video based on the President’s conspiracy theories circulated among QAnon adherents. A Washington Post survey conducted Saturday found that just 27 congressional Republicans out of 249 acknowledge Joe Biden’s presidential election win. In MIT Technology Review, Abby Ohlheiser points out that Trump’s disinformation may be leading us to a future where trust in democracy is diminished, and legislation isn’t always based on fact.
- Ethical AI?: Also on Tuesday, Dr. Timnit Gebru, the co-lead of Google’s ethical AI team, announced her firing via Twitter. Dr. Gebru is known for her groundbreaking research showing facial recognition software is less accurate at identifying women and people of color. Dr. Gebru’s ouster was apparently the result of a dispute over a research paper.
- The Fallout: The MIT Technology Review, which has seen a copy of Dr. Gebru’s paper, reports that it outlined substantial environmental costs and the potential for disinformation inherent in large language models (one of which powers Google search and also recently GPT-3). Gebru’s team at Google is rallying around her, and have contradicted claims by Google’s head of research, Jeff Dean, about her ouster. More than 1,500 employees have also signed a Walkout petition in support of Dr. Gebru.
- Race matters: Facebook is overhauling its algorithm in order to try to police hate-speech more effectively, reports the Washington Post. Previously, the company took a ‘race-blind’ approach so the statement ‘men are pigs’ would be treated the same as racist or anti-semitic slurs. Twitter also announced that it would be tightening its policies on hate speech to prohibit dehumanizing language. Not to be left out, Youtube announced that it would be testing a filter which will hold potentially harmful content for review, as well as launching a feature that will warn users when their content may be offensive to others.
- Further reading: For those interested in how disinformation targets and affects minorities in particular, this interview with Professor Keesha Middlemass of Howard University on her research into disinformation and its impacts on Black Washingtonians is well worth a read.
- Oversight board takes a case: Facebook’s Oversight Board announced last Tuesday that it was taking its first cases. An analysis by Emily Bell in the Columbia Journalism Review observes these are fairly low-risk choices for the board (CJR also has a good overview of who is on the board, and its purview, for those who would like a refresher). Interestingly, most of the first cases are non-English language, a known weak point for the company’s content moderation policies.
- Twitter & India: Speaking of non-English content moderation, on Wednesday Twitter applied a warning label to a prominent Indian politician -- possibly the first time it has done so. This is notable because critics have been asking social media companies to apply the same content moderation standards around the world as they do in the US for a while now (see for example Ali Ahmadi’s recent article in First Draft, or Rasmus Nielsen’s article in Scroll).
- EU policies: On Thursday, EU Commission vice-president Vera Jourova announced the bloc is planning to introduce legislative measures to fight disinformation. The planned measures, which were unveiled with the launch of the Democracy Action Plan, include proposing legislation on political advertising transparency (including paid ads, branded content, and microtargeting), promoting media literacy by focusing on journalist safety and decreasing the abusive use of SLAPP lawsuits, as well as strengthening the existing EU toolbox for countering foreign interference and disinformation. Politico Europe has a helpful roundup of the most important parts of the announcement.
- NDAA and 230: On Tuesday, President Trump threatened to veto the National Defense Authorization Act (NDAA) unless the Section 230 liability shield for tech companies was repealed. However, even if Trump follows through on the threat, Congress seems likely to override the veto. This doesn’t mean that Section 230 is safe though, as the newly-elected Republican head of the House Energy and Commerce Committee, Rep. Cathy McMorris Rodgers (R-Wash) told Politico that addressing 230 should be “at the top of the list” for Republicans.
- Simington and FCC: The Senate Commerce Committee voted Wednesday to advance the nomination of Nathan Simington to the Federal Communications Committee. The step is important, as Simington’s appointment to the committee would lead to a 2-2 deadlock when Chairman Ajit Pai steps down in January, potentially stymying any policy measures the new administration wishes to push through. However, Axios points out that Simington’s confirmation is still unlikely.
- Problem with Fleets, II: We talked in last week’s Decoder about some of the troubles with Twitter’s newly-introduced ‘fleets’. Alex Howard of the Digital Democracy Project pointed out another issue this week: what about fleets that should be public record?
