Abstract
Safiya Noble’s Algorithms of Oppression: How Search Engines Reinforce Racism argues that current approaches to evaluating, studying, and contextualizing the social effects of algorithms do little to address how they promote the expansion and reproduction of existing racist and sexist modes of power. Noble uses black feminist theory and critical race studies to orient information sciences and algorithmic studies toward an agenda that centers how algorithms affect the ability of marginalized communities to fight race and gender-based oppression. In doing so, Noble’s work lifts up the voices and experiences of communities who are systematically excluded or misrepresented in how search engines manage society’s access to, and engagement with, information.
Noble’s study focuses mainly on Google’s monopoly over our information infrastructure and economy. While she also addresses corporate users of search like Yelp, and public users like the Library of Congress, her singular focus on Google allows readers a clear picture of the risks posed in giving corporations and institutions the power to manage how we search and find information on and off the Internet. Noble’s commitment to changing our relationship with the companies and institutions that provide access to the world’s information is a call for scholars, the tech industry, policy experts, lawmakers, and the general public to trouble the connection between commercial interests and the algorithms that help us search for information that contextualizes our social experiences. In presenting algorithms and search as integral to how we acquire knowledge, Noble aims to make readers more conscientious about how capital influences people’s access to reliable, relevant, and responsible sources of information. Although her primary aim is to demonstrate how commercial search engines reinforce systematic oppression against black Americans (what she calls “technological redlining” [p. 1]), she often expands her framework to include other marginalized communities such as non–black American women, other communities of color, and people who practice religions other than Christianity. By doing so, she demonstrates the wide-reaching costs of erasing the social power of algorithms for communities who rarely have a say in how they are categorized and indexed through search.
Noble begins her work by challenging the belief that algorithms are neutral and unbiased managers of information. For her, seeing algorithms in this manner obscures the programmers responsible for writing the algorithm’s code, as well as the commercial interests using algorithms to their advantage. Noble is not the first scholar to write about the biased nature of search engines (Itrona and Nissenbaum 2000). Nor is she the first author to use Google as a primary case study for the dangers that information monopolies pose to the free and equal exchange of information (Vaidhyanathan 2011). However, her perspective doesn’t just bring awareness to how Google’s control manipulates our information economy. She also draws attention to how search results are indicators for how we unequally come to value, care, and know about different groups of people. Specifically, Noble disputes Google’s assertion that racist and sexist search results stem from glitches in the algorithm’s code. Instead, she contests that “glitches” in Google searches for terms such as “black girls,” “professor,” or “criminal” provide evidence that Google’s code is embedded within the same human epistemologies for race and gender that devalue the experiences of communities not represented in the company’s mainly white and male workforce. While Google invest in projects dedicated to increasing the number of minorities in tech, Noble resists such measures as a panacea for how programmers unconsciously encode bias into search. She sees investing in such future-oriented projects as indicative of Google’s failure to address their present hiring practices that bypass pools of qualified programmers from marginalized communities. Moreover, as the mostly white and male programmers at Google are not expected to learn to create algorithms that consider the importance of representation for marginalized communities, the company passes the responsibility of addressing bias in algorithms on to future generations of marginalized programmers, while asking little or nothing of their current workforce.
Noble spends a great deal of time getting readers to question what it means to acquire knowledge on the Internet. She writes, when it comes to online commercial search engines, it is no longer enough to simply share news and education on the web. We must ask ourselves how the things we want to share are found and how the things we find have appeared. (p. 155)
This quote represents the criticality she wants to develop in her readers. For Noble, what structures our path toward information is just as important as critiquing the types of information we find on the Internet. However, the proprietary nature of Google’s search algorithm means Noble cannot directly analyze its code to understand how the algorithm facilitates the appearance and discovery of information on the web. This forces her to think about the architecture and structure of algorithms not in terms of their specific code but as the social and economic conditions that influence Google’s webpage rankings. Her work identifies advertising and SEOs (search engine optimization) as two ways commercial interests are able to influence Google’s PageRank system. Noble’s work asserts that the majority of people searching the web believe their search results are credible and reflect the search habits of all Internet users. By exposing the work SEOs and advertisements play in Google’s PageRank system, Noble dispels the notion that top search results signify popular search habits. Instead, she shows that what often floats to the top of result pages are paid advertisements and misinformation that possesses the potential to misrepresent and harm marginalized communities.
It can be hard to understand how an algorithm could cause community harm. This is especially true when one could argue that search results are not the same as physically assaulting someone. Noble cuts through this logic by historically contextualizing her concerns about the power and reach of algorithms and search engines. She does so by taking the time to place racist and sexist bias within search against the racist and sexist history of America. This allows Noble to bring in critical race scholars such as bell hooks (1992) and Jennifer C. Nash (2008), to argue that the racism and sexism we see in search is not outside of America’s epistemological framework for understanding black people, specifically black women. These authors help her make connections between the ways we create language around black Americans, and how we come to value black Americans in our everyday lives. By bringing critical race and black feminist thought to bear on the question of algorithms and search, Noble stresses that managing algorithms is not just a technical question, but political. Her primary example is how Google’s search hypersexualizes and commodifies the existence of black women and girls by consistently showing pornographic results at the top of searches for “black girls.” While searching “black girls” may no longer garner pornography as its top result, Noble’s research offers us a chance to wonder what other words and phrases search still fails to equitably represent, and what is the cost of this failure? In chapter 3, Noble shows the consequences of how search categorizes and values marginalized communities through the radicalization of white supremacist mass murderer Dylann Roof. Noble uses Roof’s manifesto to explore how Google’s search aided in the development of Roof’s anti-black racism. While searching for a way to understand the social politics behind Trayvon Martin’s death, Roof searched for “black on white crime.” Instead of websites that would help reframe his concerns, Google returned a list of right-wing and racist websites that supported and heightened Roof’s fears of black people. Roof credits this search with developing his racial consciousness. This resulted in Roof killing nine black parishioners at Emanuel African Methodist Episcopal Church in Charleston, South Carolina. To be clear, Noble is not arguing that search is solely responsible for the radicalization of Dylann Roof or his anti-black racism. However, she does use Roof as an example of how unchecked and unmoderated racism on the Internet has real-life consequences. In this instance, Google’s search had the ability to redirect Roof toward information that would help him understand the social significance of an unarmed black teenager being killed just because he “looked” suspicious. Instead, the company’s search engine led Roof down a path that disregarded data that may have challenged Roof’s beliefs and saved the lives of nine people.
Noble continues to widen her exploration of how algorithms and search are socially and culturally important by questioning how public institutions have historically categorized and searched for marginalized communities. She uses the Library of Congresses subject headings as evidence that the lack of protections afforded by search is not because search exists in digital spaces, but persist because search is based on practices of categorizing information that historically have classified and indexed marginalized people in ways that mark them as culturally and socially inferior. For this reason, Noble also writes about how lawmakers need to take a stronger stance on protecting people’s right to information, as well as their right to be forgotten. Most importantly, she argues that when we lose control over the data that represent who we are, we lose control over how we want to be represented. For marginalized communities that are already struggling for proper representation, Noble’s text highlights the urgency of connecting how we construct a sense of self and others with the work of algorithms and search. If we’re using algorithms to help manage our interactions with information and data that have the power to re/shape how we come to know and recognize each other, then we need to know how algorithms work and influence us daily.
Noble spends her final chapters discussing how we can address the social inequity embedded in algorithms and search. In these chapters, she continues to argue for the expansion of everyone’s “algorithmic literacy” (p. 25). She describes this form of reading and knowing algorithms as going beyond their codes and moving into the social and technological systems that allow algorithms to make decisions that help structure our relationship to knowledge. Building information systems in which everyone has algorithmic literacy means everyone must have access to the tools we need to collect, store, share, and delete information on terms not determined by corporations, but agreed upon by all users. As archivist Bethany Nowviskie (2016) demonstrates in her work on digital community archives, creating such a system is one way to allow communities underserved by our current information structure a chance to write and store their own histories and representations for community or public use. As Noble demonstrates, this is particularly important for black Americans who often lack control over how they are represented in digital spaces and are often falsely rendered as nontechnological subjects.
While Noble offers examples of search engines similar to BlackWebPortal that are designed to deliver more relevant search results to users, she does not extend her critique of representation to the algorithms or individuals that manage these search engines. This is important as search engines built for specific ethnic and racial groups also manage and control how their users search for information. I mention this concern as black Americans have historically routed their struggle for full social and civic inclusion through rights-based claims. These claims have rested on a foundation of a unified recognizable representation of blackness that often pushes people who are not male, middle to upper class, heterosexual, politically moderate, and able-bodied out of our lexicon for what it means to be black, free, and valued. While BlackLivesMatter, BlackOUT Collective, and the #SayHerName campaign are examples of groups and initiatives fighting to change the conditions through which black Americans demand freedom, we are still not at a point where the images and names of black rights leaders like Claudia Jones roll off our tongues with ease (Davies 2008). For these reasons, including the work of critical race scholars like John Jackson (2005) who questions the relationship between authenticity and blackness, or Katherine McKittrick (2006) who demonstrates the importance of how black women are consciously lost to history even when they are glaringly present, would stretch the political imagination of what it means for anyone to design a search engine specifically tasked with producing search results that lessen the micro and macro aggressions black people experience through technology.
Overall, Noble’s work foregrounds a desire to “eliminate social injustice and change the ways in which people are oppressed with the aid of allegedly neutral technologies” (p. 13). It is easy to dismiss this as an impossible goal. Yet, if we take seriously the role Google and other corporations and institutions play in curating our knowledge and data streams, then we must also seriously consider how various methods of curating or intervening in these streams may push us to think differently about the information search allows us to locate and find on the Internet, as well as what and who becomes unsearchable or findable because of the algorithms deployed to manage search. This means opening up “a dialog about radical interventions on socio-technological systems in a more thoughtful way that does not further marginalize people who are already in the margins” (p. 171). In her final chapter, Noble presents the creation of a black feminist technology studies (BFTS) as one potential intervention readers may take up to revolutionize how we interact and study algorithms and technology more generally. Noble positions this work as crucial to recontextualizing black Americans, specifically black women and girls, as demographics that heavily engage with technology and are not late adapters or outside of tech. Therefore, black feminist technology studies would help broaden how we analyze and critique the methods by which politics, value, and culture are embedded into search and tech more broadly. For Noble, creating new epistemologies around science and technology that include historically marginalized people contains the power to help us rethink our relationship to innovation, design, and resistance, in and outside of tech. We need this if we are going to be more intentional about technology’s role in helping us construct a future where we are all free.
