A new book explores the racial implications of the algorithms that structure our lives.
When Dylann Roof murdered nine African-American churchgoers at the Emanuel African Methodist Church in Charleston, South Carolina on June 15th, 2017, the country was left wondering what factors conspired to lead a someone to commit such an atrocious act of violence. Some looked to his white supremacist roots. Others studied his psychology and his upbringing. One factor that didn’t come up, at least not in most circles: his Internet search history.
Safiya Umoja Noble, author of the new book Algorithms of Oppression: How Search Engines Reinforce Racism, believes it should have. Noble offers a compelling look into the structure of digitized information—most of it driven by advertising revenue—and how it perpetuates racist assumptions and ideologies. Whereas the information age’s most common social justice ambition has been to bridge “the digital divide” between Internet access and marginalized groups, Noble, who teaches at the University of Southern California, argues that access is less important than operation—that is, how search engines (read: Google) rank the pages that respond to our queries. Dylann Roof represents perhaps her most compelling case study.*
During Roof’s trial, his “manifesto,” which he posted on the white supremacist website www.lastrhodesian.com, proved pivotal to exploring Roof’s motivation for doing what he did. In it, Roof explains that his interest in the Trayvon Martin case—the 2012 incident in which George Zimmerman, often described as a “white Hispanic” shot and killed Martin, an unarmed 17-year-old black man—sparked him to wonder how often blacks killed whites. He wrote, “this prompted me to type in ‘black on White crime’ into Google, and I have never been the same since that day.” He continued, “The first website I came to was the Council of Conservative Citizens (CCC). There were pages upon pages of these brutal black on White murders. I was in disbelief.”
Obviously, one must be careful about presuming to identify the tipping point into radicalization. But it’s hard not to invest what Roof next wrote with significant meaning: “At this moment I realized that something was very wrong. How could the news be blowing up the Trayvon Martin case while hundreds of these black on White murders got ignored?”
Noble’s main question in response to Roof’s manifesto is one that did not come up during the trial, but may provide insight into the evolution of Roof’s murderous mentality: Why, when he asked Google his question, was he directed to a white nationalist website trading in vicious propaganda, rather than to accurate statistics provided by the Federal Bureau of Investigation or the Southern Poverty Law Center? Had he gone to the FBI’s webpage, he would have learned that the vast majority of violence against white Americans is committed by white Americans, and that the vast majority of violence against black Americans is committed by black Americans. Had his first hit been the SPLC, he’d have read that the Council of Conservative Citizens is “the modern reincarnation of the old White Citizens Councils,” an organization that has “evolved into a crudely white supremacist group.”
Of course, it may not have mattered. Noble does not necessarily assume that the Charleston massacre would have been averted if these legitimate websites had ranked ahead of white supremacist ones. Roof was quite possibly looking for what his sick predisposition wanted to find, and would have scrolled for pages to find it.
But she does rightly hypothesize that Roof’s fated search touches upon the larger cultural relationship between search engine optimization and the perpetuation of racial ideologies rooted in a legacy of prejudice dating back to Jim Crow and chattel slavery. Noble, to wit, was motivated to write her book when, in 2011, she searched Google for “Black girls” and the first hit was “Sugary Black Pussy,” followed by a list of equally offensive links playing on the stereotype of the hyper-sexualized African-American woman. As long as we continue to believe that accurate information is essential to democratic citizenship, Noble’s essential claim cannot be dismissed as irrelevant to the death of those nine churchgoers: “What Roof found was information that confirmed a patently false notion that Black violence on White Americans is an American crisis.”
Two related objections typically arise in response to Noble’s targeting of Google as responsible for the racism embedded in its algorithms. The first is that the information that emerges on search engines is ultimately a neutral reflection of the “hive mind,” or the popular will, and therefore not the platform’s fault, but rather the collective perspective of Internet users. The second is that it’s beyond Google’s control to determine what link ranks above or below other links.
Noble convincingly debunks both assumptions. “Recall,” she explains, “that what shows up on the first page of search is typically highly optimized advertising-related content. Indeed, Google thrives when “clients are paying Google for placement on the first page.” The farrago of factors that contributed to the CCC allegedly appearing on Roof’s computer before the FBI or the SPLC is impossible to decipher, but given the fact that advertisers influence Google’s algorithms, rather than informational accuracy or attentiveness to prejudicial portrayals, it’s safe to say that it was not a neutral expression of American taste. And as for the claim that Google cannot control what ranks where, Noble cites a number of cases whereby Google freely admits “to engaging in filtering practices.” She also notes “content moderation” from leaked Facebook reports indicating that “people and policies are in place to navigate and moderate content on the web.”
Raymond Kurzweil, the American computer scientist and futurist, has prophesied that “within several decades information based technologies will encompass all human knowledge and proficiency, including ultimately the pattern recognition powers, problem solving skills, and emotional and moral intelligence of the human brain itself.” Imagine, as Noble does, if this prediction is accurate, and then imagine if the algorithms driving AI’s thought processes were, intentionally or not, racist. It’s a prospect every bit as terrifying as Roof’s act of domestic terrorism.
*Update—July 2nd, 2018: This article has been updated with the correct spelling of Safiya Umoja Noble’s name.