Tech

EXCLUSIVE: Documents Detailing Google’s ‘News Blacklist’ Show Manual Manipulation Of Special Search Results

REUTERS/Jim Young

J. Arthur Bloom Deputy Editor

Google does manipulate its search results manually, contrary to the company’s official denials, documents obtained exclusively by The Daily Caller indicate.

Two official policies dubbed the “misrepresentation policy” and the “good neighbor policy” inform the company’s “XPA news blacklist,” which is maintained by Google’s Trust & Safety team. “T&S will be in charge of updating the blacklist as when there is a demand,” reads one of the documents shared with The Daily Caller.

“The deceptive_news domain blacklist is going to be used by many search features to filter problematic sites that violate the good neighbor and misrepresentation policies,” the policy document says. (RELATED: Meet The Five Google Staffers Who Circulated The Petition To Drop Kay Coles James)

That document reads that it was, “approved by gomes@, nayak@, haahr@ as of 8/13/2018.” Ben Gomes is Google’s head of search, who reports directly to CEO Sundar Pichai. Pandu Nayak is a Google Fellow, and Paul Haahr is a software engineer, whose bio on Google’s internal network Moma indicates that he is also involved in, “fringe ranking: not showing fake news, hate speech, conspiracy theories, or science/medical/history denial unless we’re sure that’s what the user wants.”

“The purpose of the blacklist will be to bar the sites from surfacing in any Search feature or news product. It will not cause a demotion in the organic search results or de-index them altogether,” reads the policy document obtained by the Caller. What that means is that targeted sites will not be removed from the “ten blue links” portion of search results, but the blacklist applies to most of the other search features, like “top news,” “videos” or the various sidebars that are returned as search results.

Google CEO Sundar Pichai speaks on-stage during the annual Game Developers Conference at Moscone Center in San Francisco, California on March 19, 2019. (Photo by Josh Edelson / AFP / Getty Images)

In a section of the memo entitled “Eligibility for GNP [Good Neighbor Policy] enforcement,” the types of search results impacted by the policy are described:

“If your product shows any of the following, Misrep and GNP would apply to your PA.

  • Shows content from users and news publishers (percieved 3P voice). Ex: UGC, News corpus, etc.
  • Outputs single answers (perceived to come from the open web). Ex: Web answers, Video answers, etc.
  • Shows content owned, licensed, or edited by Google (perceived to come directly from Google). Ex: Knowledge panels, News summaries, Oneboxes, Munin carousels, etc.”

The “ten blue links” may not be impacted by the blacklist, but virtually every other kind of Google search result is. While hard numbers are not available for how much traffic is directed through the 10 links versus the other search blocks, since the latter appear so high on the results page, the impact could be significant. (RELATED: Applause At Google’s All-Hands Meeting As Company Drops Heritage Foundation President)

“Focus on the user,” said a source at Google who described the program to the Caller. “Users need to trust any content that Google shows them, whether it’s the 10 blue links or other special search results.”

BERLIN, GERMANY – JANUARY 22: Workers push a Google sign from the stage before the festive opening of the Berlin representation of Google Germany on January 22, 2019 in Berlin, Germany. The official opening will take place tonight with Berlin Mayor Michael Muller. (Photo by Carsten Koall/Getty Images)

Sundar Pichai testified before the House Judiciary Committee on Dec. 11 of last year. Democratic California Rep. Zoe Lofgren asked why a search for the term “idiot” returned a photo of President Trump. In response, Pichai said, “This is working at scale, we don’t manually intervene on any particular search result.”

A memo about the deceptive news blacklist was also obtained by the Caller, showing its last edit as Dec. 3, 2018, a week before Pichai’s congressional testimony. This document, which describes the process by which a site can be blacklisted for deceptive news, clearly shows that there is a manual component:

“The beginning of the workflow starts when a website is placed on a watchlist which is used for monitoring of sites to determine if they violate the Good Neighbor Policy. This watchlist is maintained and stored by Ares with access restricted to policy & enforcement specialists working on the Good Neighbor Policy. Access to the listing can also be shared at the discretion of pcounsel and legal investigations on a need to know basis to enforce or enrich the policy violations. The investigation of the watchlist is done in the tool Athena, the Ares manual review tool, and intakes signals from Search, Webspan, and Ares in order to complete reviews. … Once a domain is determined to be violating the misrepresentation policy or the Good Neighbor Policy, such patterns are then added to deceptive_news_blacklist_domains.txt by the Trust & Safety team.”

The document indicates that there is, among other things, a “manual review tool” involved in maintaining the blacklist.

On the blacklist are a number of conservative sites, including Gateway Pundit, Matt Walsh’s blog, Gary North’s blog “teapartyeconomist.com,” Caroline Glick’s website, Conservative Tribune, a property of The Western Journal and the website of the American Spectator.

“You can’t trust the human judgment of Google’s Trust and Safety team,” said the source at Google with knowledge of their practices.

Requests for comment from Google’s press team have gone unreturned as of publication time.

Update: A Google spokesperson provided the following statement to the Daily Caller after this story was published: “We do not manually determine the order of any search result, nor do our algorithms or policies attempt to make any judgement on the political leanings of a website. Our Google News inclusion policies are publicly available online. They provide guidelines on content and behaviors for matters like sponsored content, deceptive practices, and more. Sites that do not adhere to these policies are not eligible to appear on news surfaces or in information boxes in Search. These policies do not impact the way these sites appear in organic blue-link Google Search results.”

Follow J. Arthur Bloom on Twitter