- Facebook plans to ban misinformation about the 2020 Census as well as a range of disruptive information about the elections.
- The company is also making a public, searchable database of all jobs and credit ads listed on its platform.
- The announcements come as part of a civil rights audit into Facebook's practices.
Facebook is banning misinformation about the 2020 Census and elections as it updates it policies to deal with online trolls and other bad actors.
In a report released over the weekend, Facebook COO Sheryl Sandberg detailed a number of steps it's taking to secure the 2020 elections. The company is banning ads that tell people not to vote and last year banned a range of "misinformation" on voting, including posts that misrepresent when, where or how to vote or threaten violence relating to voting or an election, it said.
The company also plans to introduce a "census interference policy" later this year, in the lead-up to the once-a-decade count of all U.S. residents. The policy will "prohibit misrepresentations of census requirements, methods, or logistics, and census-related misinformation," the company said in its report.
The report's lead author, former American Civil Liberties Union executive Laura Murphy, was hired by Facebook in May 2018 to assess its performance on vital social issues.
Facebook has been widely criticized for allowing trolling and misinformation on its sites in the lead-up to the 2016 elections, with various actors using the platform to target vulnerable populations, discourage voting and stir white nationalism. Federal officials havethat similar tactics could wreak havoc on the 2020 Census.
The census ultimately determines the allocation of hundreds of billions in federal funding as well as the assignment of electoral districts. Certain groups of people — including black and Hispanic residents, Native Americans on reservations, renters and poor people — have been under-counted in prior counts, while homeowners and white residents were overcounted. The Trump administration's attempt to add a last-minute citizenship question to the census, though it has by the Supreme Court, sets the stage for further confusion.
Searchable database for jobs and loans
Facebook will make a searchable database for jobs, loans and credit-card offers available for all U.S. users, the company said in its report. It's taking those actions as part of a Marchthat accused the company of encouraging discrimination in its targeted advertising.
The groups had claimed Facebook violated anti-discrimination laws by preventing audiences, like single mothers and the disabled, from seeing many housing ads, or excluding women and older workers from job ads.
The searchable housing ads database will roll out by the end of 2019, Facebook said, and Murphy said she expects the employment and financial service offerings databases to be available within the next year.
Murphy said she's "very excited" about the move and believes they could help boost the social mobility of millions of people in the United States.
Targeted ads tailored to individuals are Facebook's bread and butter — accounting for the vast majority of its more than $50 billion in annual revenues last year. It's unlikely that making the ads searchable would have a significant effect on Facebook's business. But analysts have cautioned that any restrictions on Facebook's ability to target ads could scare off advertisers.
Galen Sherwin, senior staff attorney at the ACLU and the group's lead attorney in the case against Facebook, said making the three databases searchable by anyone "definitely creates greater access to information about economic opportunities."
However, civil rights groups are concerned that the secretive, proprietary algorithms that govern how the company steers ads— even when not consciously targeting specific groups — could still be discriminatory.
"I wish we could see into the black box," Sherwin said.
Facebook still faces a U.S. Department of Housing and Urban Development complaint over the targeting and delivery of housing ads. Murphy, the auditor, said she thinks the company understands it's "going to have to look at the algorithms" behind them.
The company also faces privacy and anti-trust investigations in the U.S. and Europe over data-collection practices that critics call invasive and struggles to police speech globally with sometimes lethal repercussions.
Hate speech monitors
Facebook continues to walk a fine line in its attempts to shed "harmful content," including hate speech and fascist speech. To be more effective, the company will dedicate some content moderators on hate speech alone. A few dozen are involved so far, the company said in its Saturday report. All come from the more than 20,000 outsourced content moderators who screen the platform.
Two days after the report was released, ProPublica reported on a 9,500-member secret Facebook group for border control agents in which users joked about migrant deaths, discussed throwing burritos at Latino congress members and posted fake pornography of a congressional representative. The group has been in existence for three years, according to ProPublica.
Recommendations from the Facebook audit include ending a carve-out for humor as an exception in hate speech and devising better mechanisms for blocking harassment, which can be especially overwhelming for targets when automated.