![[PDF]](/pb-assets/Icons/adobe-pdf-icon-1498649896243.png){kind=icon}
In a magazine exclusive, April Anderson and Andy Lee Roth reveal original research showing how search engines are filtering out pro-LGBTQ stories and amplifying homophobic voices
FROM CENSORING CONTENT as “objectionable” to blacklisting keywords and restricting users’ advertising revenue, major media platforms are filtering online speech in ways that marginalise and stigmatise LGBTQ communities, as we reveal in research carried out exclusively for Index.
The same mechanisms that block LGBTQ-themed hashtags on Instagram or demonetise LGBTQ channels on YouTube, for example, also often permit – or even promote – anti-LGBTQ content.
Our research also showed that Google News was one of the worst offenders. We conducted a study of the news aggregator from 3 to 7 February 2020, using “LGBT” and related search terms. In that period we found that Google News consistently provided a prominent platform for evangelical Christian and far-right perspectives on LGBTQ issues.
On 6 February, for example, Google News featured an article from the Christian Post asserting that Disney and other Hollywood studios were “capitulating to character quotas” and “virtue signalling to the loud and influential LGBT lobby”. The following day, it highlighted an article by Tony Perkins, president of the arch-conservative Family Research Council, decrying an “LGBT agenda” that rejected “biblical teaching on sex and marriage”. Perkins’ organisation has been designated a hate group by the Southern Poverty Law Centre because of its record of defaming gays and lesbians “based on discredited research and junk science”.
Overall, 46% of the LGBTQ stories spotlighted by Google News in our research derived from conservative news outlets, including 32% from conservative Christian sources. By contrast, only 4% of LGBTQ stories featured by Google News – one from Mother Jones on former Democrat presidential contender Pete Buttigieg and another from Human Rights Watch about health policies in Tanzania – originated from progressive outlets. Despite an abundance of excellent news organisations dedicated to reporting LGBTQ issues, Google News featured none during the week studied. We asked Google for comments on the research, but it did not respond.
Our findings corroborate a study of Google News from September 2019 by Matt Tracy for Gay City News. Tracy conducted a systematic study of Google News’s anti-LGBTQ bias after noticing how often in the course of his work the engine returned “inflammatory, bigoted content” that seemed at odds with general attitudes on LGBTQ topics.
Our study compared the Google’s news aggregator to another search engine. Searching the same terms in the same week using the news service of DuckDuckGo produced contrasting results. Only 6% of DuckDuckGo’s LGBTQ articles derived from conservative, religious or more politically biased sources. Instead, it more frequently featured mainstream US news sources (56%) or international news sources (13%).
Google’s selection algorithms for its news aggregator are unavailable for public scrutiny beyond the general terms in which they are described in the company’s support portal and in interviews given by its executives. This makes it impossible to judge whether the prominence of anti-LGBTQ stories is due to in-built bias, skilful search engine optimisation by conservatives with homophobic agendas, or other factors.
A pending lawsuit filed in California in 2019 by LGBTQ content creators could force Google to make its powerful algorithms available for scrutiny. The class action complaint asserts that YouTube, the premier video sharing platform, and Google, its parent company, act as censors, regulating online content for their own gain.
The lawsuit contends that YouTube restricts LGBTQ content creators’ access on the basis of their sexual orientation or gender identity. It identifies a “tool kit” of practices YouTube has used to do this, including content blocking, advertising restrictions and channel demonetisation. The complaint further alleges that YouTube promotes anti-LGBTQ content by playing anti-LGBTQ advertisements before LGBTQ videos and by permitting unmoderated hateful comments on LGBTQ videos.
According to the plaintiffs’ lawyer, Peter Ob-stler, in addition to forcing YouTube to make its proprietary algorithms available for review, a ruling in his clients’ favour could reveal if YouTube suppresses third-party content in order to reap greater profits by promoting its own content.
Despite efforts by Google/YouTube to dismiss the case, a federal hearing was scheduled for this spring. However, this may be postponed, because the US Department of Justice has informed the plaintiffs that the solicitor general will review Google/YouTube’s claim to immunity under Section 230 of the 1996 Communications Decency Act. This constitutional issue, and the solicitor general’s involvement, suggest that the case may ultimately be decided by the Supreme Court.
As that lawsuit progresses through US courts, advertisers around the world continue to use keyword blacklists that automatically flag terms deemed to be profit-threatening. In order to promote “brand safety”, about 95% of advertisers now use these, according to Jonathan Marciano, director of communications at CHEQ, an advertising verification company.
A September 2019 study conducted by CHEQ found that keyword blacklists, which include terms such as “bisexual” and “same-sex marriage”, demonetise up to 73% of neutral or positive content on LGBTQ news outlets such as The Advocate and PinkNews. Lost advertising revenues may have also contributed to the closure of LGBTQ news outlets including The Pool and INTO.
The advertising industry’s lack of understanding regarding the impact of keyword blacklists will likely lead to what Marciano described as “a slower death” for additional independent news outlets, further shrinking public access to reliable queer-focused resources online.
Sealow, a YouTuber whose research has identified more than 16,000 keywords used to demonetise YouTube channels, argues that researchers need more direct access to the systems used by platforms such as YouTube to ensure that those systems function without any discriminatory behaviour.
Due to these platforms’ global reach, this algorithmic censorship “applies across national boundaries”, according to Patrick Keilty, a University of Toronto professor who researches the politics of digital infrastructures. For example, in December 2018, when Tumblr responded to changes in US law by removing all content it defined as “adult”, Keilty observed that the decision didn’t just affect queer communities in the United States. Those in Mexico, Canada, Germany, the Philippines and elsewhere also felt its impact.
Afsaneh Rigot, a research fellow at Harvard’s Berkman Klein Centre, contends that everyday online tools, including group chat services and online dating apps, should be designed with “at-risk marginalised communities in mind” because doing so results in features and protections for all groups.
Recent studies, including Virginia Eubanks’ book Automating Inequality and Safiya Umoji Noble’s Algorithms of Oppression, have raised public awareness and informed policy debate about how artificial intelligence can reflect and reinforce racist practices. The scope and complexity of content-blocking preclude any single, simple solution to the marginalisation of queer communities online. However, increasingly, algorithm-driven censorship is stifling LGBTQ freedom of expression and amplifying anti-LGBTQ animus, and urgently needs attention.
April Anderson is a librarian at Macalester College, Minnesota, and Andy Lee Roth is associate director of Project Censored, a media watchdog that promotes independent investigative journalism. Anderson and Roth co-authored Stonewalled: Establishment Media’s Silence on the Trump Administration’s Crusade against LGBTQ People
