Future Tense

Facebook Found “Hate Speech” in the Declaration of Independence

A copy of the United States' Declaration of Independence hangs on a wall at the US embassy in Havana.
A copy of the United States’ Declaration of Independence hangs on a wall at the U.S. embassy in Havana. Adalberto Roque/Getty Images

Facebook has come under fire for finding what many Americans have known all along—one of our country’s founding documents contains racist speech.

The company’s hate speech algorithm flagged parts of the Declaration of Independence when the Liberty County Vindicator, a community newspaper in Texas, published excerpts to the social media platform in the days leading up to the Fourth of July. The post containing paragraphs 27–31 was automatically removed, according to editor Casey Stinnett, and the paper received a notification from Facebook saying that the post violated hate speech standards.

It isn’t clear what exact passage triggered the algorithm—but the line about “merciless Indian Savages” seems a likely candidate. The passage, which refers to colonists’ grievances against King George III, reads:

He has excited domestic insurrections amongst us, and has endeavoured to bring on the inhabitants of our frontiers, the merciless Indian Savages, whose known rule of warfare, is an undistinguished destruction of all ages, sexes and conditions.

The post has since been restored, and Facebook has apologized. “It looks like we made a mistake and removed something you posted on Facebook that didn’t go against our community standards,” it told the paper. “We want to apologize and let you know that we’ve restored your content and removed any blocks on your account related to this incorrect action.”

It’s unclear in this case whether Facebook relied on the algorithm or if it was a combination of a flagging, human review, and removal.

“[Facebook’s] apology did not imply that it was algorithmic, though it also did not specify either way,” said Tarleton Gillespie, principal researcher at Microsoft New England, in an email to Slate. “If anything, part of the problem here is that we do not truly understand how these decisions are made, only that they are. This is a bigger problem than getting it wrong once in a while.”

In a blog post in 2017, Facebook defined hate speech as anything that directly attacks people based on “protected characteristics,” including race, ethnicity, gender, disability or religious affiliation—so you can certainly see how that net would have scooped up the phrase “merciless Indian savages.”

Danielle Citron, a law professor at the University of Maryland who works with Facebook’s non-consensual intimate images advisory group in an unpaid capacity, says that the episode could point to the limits of algorithms in flagging hate speech. Hate speech is about context, she says, which algorithms struggle to detect.

“Algorithms may be helpful to flag patterns and groupings of words that in other cases have been appropriately found to constitute hate speech,” said Citron. “But it does a terrible job of adjudicating on its own. So the idea that we would rely on algorithms to flag and to filter and block without human moderation … is a terrible idea.” But even with human moderation, content removal decisions can be very difficult.

Facebook has had hate speech mishaps in the past. Black Lives Matter activist Shaun King was temporarily removed from Facebook in 2016 because he posted hate speech he had received. (That episode stemmed from a team member wrongly removing the post, according to a screenshot King posted.)

“Facebook and the major platforms are being pressured by European member states and the EU Commission to take an overbroad and immediate response to hate speech,” said Citron. Facebook, Twitter, Google, and Microsoft have been pressured “to remove hate speech within 24 hours, which effectively means you have to use algorithms.”

It’s unclear to what extent Facebook is using algorithms to flag hate speech. But the silver lining here could be that it’s forcing Americans, especially white Americans, to confront the racism of founding fathers like Jefferson. “Perhaps had Thomas Jefferson written it as ‘Native Americans at a challenging stage of cultural development’ that would have been better,” Stinnett wrote. “Unfortunately, Jefferson, like most British colonists of his day, did not hold an entirely friendly view of Native Americans.”

The “merciless Indian Savages” phrase was condemned on Twitter in the days leading up to the Fourth of July. Even Twitter’s official Moments account weighed in, “On the #FourthofJuly, a reminder that the Declaration of Independence references ‘merciless Indian Savages.’ ”

“The quote is of clear historical importance, from a treasured document for most Americans—but it does include phrasing, ‘merciless Indian savages,’ that we now no longer use because we recognize the hateful way it paints an entire people,” said Gillespie. “So it is historically important, and it is hate speech.”