While the web-indexing process is not the same as classification systems such as DDC, the application of the theoretical model is still valid for thinking about conceptualizing algorithms and indexing models that could actively intervene in the default normativity of racism and sexism in information resources.
Problems in Classifying People
The idea of classification as a social construct is not new. A. C. Foskett suggests that classificationists are the products of their times.11 In the work of Nicholas Hudson of the University of British Columbia on the origins of racial classification in the eighteenth century, he suggests that during the Enlightenment, Europeans began to construct “imagined communities,” citing Benedict Anderson’s term.12 He says, “This mental image of a community of like-minded individuals, sharing a ‘general will’ or a common national ‘soul,’ was made possible by the expansion of print-culture, which stabilized national languages and gave wide access to a common literary tradition.”13 Classification systems, then, are part of the scientific approach to understanding people and societies, and they hold the power biases of those who are able to propagate such systems. The invention of print culture accelerated the need for information classification schemes, which were often developing in tandem with the expansion of popular, scholarly, and scientific works.14 Traces of previous works defining the scientific classification of native peoples as “savage” and claims about Europeans as the “superior race,” based on prior notions of peoples and nations, began to emerge and be codified in the eighteenth century. Extensive histories have been written of how racial classification emerged in the eighteenth and nineteenth centuries in North America as a paradigm of differentiation that would support the exclusion of native and African people from social and political life.
By the nineteenth century, the processes involved in the development of racial classification marked biological rather than cultural difference and were codified to legally deny rights to property ownership and citizenship. These historical practices undergird the formation of racial classification, which is both assumed and legitimated in classification systems. Without an examination of the historical forces at play in the development of such systems, the replication and codification of people of African descent into the margins goes uncritically examined. This process can be seen in knowledge organization that both privileges and subordinates through information hierarchies such as catalogs and classification systems. The field of library science has been implicated in the organization of people and critiqued for practices that perpetuate power by privileging some sectors of society at the expense of others.
Traditional library and information science (LIS) organization systems such as subject cataloging and classification are an important part of understanding the landscape of how information science has inherited and continues biased practices in current system designs, especially on the web.
Opportunities abound for the interdisciplinarity of LIS to extend more deeply into cultural and feminist studies, because these social science fields provide powerful and important social context for information about people that can help frame how that information is organized and made available. To date, much of the attention to information organization, storage, and retrieval processes has been influenced and, more importantly, funded by scientific research needs stemming from World War II and the Cold War.15 The adoption of critical race theory as a stance in the field would mean examining the beliefs about the neutrality and objectivity of the entire field of LIS and moving toward undoing racist classification and knowledge-management practices. Such a stance would be a major contribution that could have impact on the development of new approaches to organizing and accessing knowledge about marginalized groups.
If the information-retrieval priority of making access to recorded information efficient and expedient is the guiding process in the development of technical systems, from databases to web search engines, then what are the distinguishing data markers that define information about racialized people and women in the United States? What have primarily been missing from the field of information science, and to a lesser degree library science, are the issues of representation that are most often researched in the fields of African American studies, gender studies, communications, and increasingly digital media studies. Information organization is a matter of sociopolitical and historical processes that serve particular interests.
A Short History of Misrepresentation in Classifying People
In order to understand how racial and gender representations in Google Search express the same traditional bias that exists in other organizational systems, an overview of how women and non-Whites have been historically represented in information categorization environments is in order. The issue of misrepresentations of women and people of color in classification systems has been significantly critiqued.16 Hope A. Olson, an associate dean and professor at the School of Information Studies at the University of Wisconsin, Milwaukee, has contributed among the most important theories on the social construction of classification that many of us in the field assign to our students as a way of fostering greater awareness about the power that library, museum, and information professionals hold. Those who have the power to design systems—classification or technical—hold the ability to prioritize hierarchical schemes that privilege certain types of information over others. An example of these biases include the cataloging of people as subjects in the Library of Congress Subject Headings (LCSH), which serve as a foundational and authoritative framework for categorizing information in libraries in the United States. The LCSH have been noted to be fraught with bias, and the radical librarian Sanford Berman details the ways that this bias has reflected Western perspectives:
Since the first edition of Library of Congress Subject Headings appeared 60 years ago, American and other libraries have increasingly relied on this list as the chief authority—if not the sole basis—for subject cataloging. There can be no quarrel about the practical necessity for such labor-saving, worry-reducing work, nor—abstractly—about its value as a global standardizing agent, as a means for achieving some uniformity in an area that would otherwise be chaotic. . . . But in the realm of headings that deal with people and cultures—in short, with humanity—the LC list can only “satisfy” parochial, jingoistic Europeans and North Americans, white-hued, at least nominally Christian (and preferably Protestant) in faith, comfortably situated in the middle- and higher-income brackets, largely domiciled in suburbia, fundamentally loyal to the Established Order, and heavily imbued with the transcendent, incomparable glory of Western Civilization.17
Eventually the LCSH abolished labels such as “Yellow Peril” and “Jewish Question” or made substitutions in the catalog, changing “Race Question” or “Negroes” to “Race Relations” and “Afro-Americans,”18 but the establishment of such headings and the subsequent decade-long struggles to undo them underscored Berman’s point about Western racial bias. (In fact, it was Berman who led the field in calling for antiracist interventions into library catalogs in the 1970s.) Patriarchy, like racism, has been the fundamental organizing point of view in the LCSH. The ways in which women were often categorized was not much better, with headings such as “Women as Accountants” in lieu of the now-preferred “Women Accountants”; women were consistently an aberration to the assumed maleness of a subject area.19
Furthermore, efforts at self-identity from the perspective of marginalized and oppressed groups such as the Roma or Romanies cannot escape the stigmatizing categorization of their culture as “Gypsies,” even though their “see also” designation to “rogues and vagabonds” was finally dropped from the LCSH.20 A host of other problematic naming conventions including “Oriental” instead of “Asian” and the location of Christianity at the top of the religious hierarchy, with all deviations moving toward the classification of “Primitive,” suggests that there is still work to be done in properly addressing and classifying groups of people around identity.21 Olson says, �
��the problem of bias in classification can be linked to the nature of classification as a social construct. It reflects the same biases as the culture that creates it.”22 These types of biases are often seen in offline information practices where conquest is a means of erasing the history of one dynasty or culture by the subsequent regime.23 Olson’s research has already shown that classifications reflect the philosophical and ideological presumptions of dominant cultures over subordinate cultures or groups. For example, in traditional Dewey Decimal Classification (DDC), over 80% of its religion section is devoted exclusively to Christianity, even though there are greater numbers of other religious texts and literature.24 Olson points to the Library of Congress Classification (LCC) and its biases toward North American and European countries in volumes on the law, with far fewer allocations of space for Asia, Eurasia, Africa, the Pacific area, and Antarctica, reflecting the discourse of the powerful and the presumption of marginality for all others.25
In this respect, Olson reminds us that the ordering of information provided in classification schemes “tends to reflect the most mainstream version of these relationships” because “classificatory structures are developed by the most powerful discourses in a society. The result is the marginalization of concepts outside the mainstream.”26 In other words, the most mainstream (e.g., White, heterosexual, Christian, middle-class) controlling regimes in society will privilege themselves and diminish or subdue all others in the organization of what constitutes legitimate knowledge. When we inherit privilege, it is based on a massive knowledge regime that foregrounds the structural inequalities of the past, buttressed by vast stores of texts, images, and sounds saved in archives, museums, and libraries. Certainly, classification systems have some boundaries and limits, as they are often defined in whole by what is included and what is excluded.27 In the case of most library databases in the United States, Eurocentrism will dominate the canons of knowledge. Knowledge management reflects the same social biases that exist in society, because human beings are at the epicenter of information curation. These practices of the past are part of the present, and only committed and protracted investments in repairing knowledge stores to reflect and recenter all communities can cause a shift toward equality and inclusion in the future. This includes reconciling our brutal past rather than obscuring or minimizing it. In this way, we have yet to fully confront our histories and reconstitute libraries and museums toward reconciliation and reparation.
Search engines, like other databases of information, are equally bounded, limited to providing only information based on what is indexed within the network. Who has access to provide information in the network certainly impacts whether information can be found and surfaced to anyone looking for it. Olson’s research points to the ways that some discourses are represented with more power, even if their social classifications are relatively small:
In North American society, taking away women, African Americans, Hispanic Americans, French Canadians, Native peoples, Asian Americans, lesbians and gay men, people with disabilities, anyone who is not Christian, working class and poor people, and so forth, one is left with a very small “core.” An image that shows the complexity of these overlapping categories is that of a huge Venn diagram with many sets limited by Boolean ANDs. The white AND male AND straight AND European AND Christian AND middle-class AND able-bodied AND Anglo mainstream becomes a very small minority . . . , and each set implies what it is not. The implication of this image is that not every person, not every discourse, not every concept, has equal weight. Some discourses simply wield more power than others.28
Arguably, if education is based in evidence-based research, and knowledge is a means of liberation in society, then the types of knowledge that widely circulate provide a crucial site of investigation. How oppressed people are represented, or misrepresented, is an important element of engaging in efforts to bring about social, political, and economic justice.
Figure 5.1. Google autocorrects to “himself” rather than “herself.” Search sent to me by a colleague, June 16, 2016.
We have to ask ourselves what it means in practical terms to search for concepts about gender, race, and ethnicity only to find information lacking or misrepresentative, whether in the library database or on the open web. Olson’s notion that cultural metaphor is the basis of the construction of classification systems means these cultural metaphors are profoundly represented in the notions of the “Jewish Question” or the “Race Question.” These subject headings suggest both an answer and a point of view from which the problems of Jews and race are presupposed. Simply put, to phrase “Jewish” or “race” as a question or problem to be answered suggests a point of view on the part of the cataloger that is quite different from how a Jewish person or a racialized person might frame themselves. It is here that the context and point of view of library and information science professionals who are responsible for framing people and communities as “problems” and “questions” is important. By examining the ways that Black people specifically have been constructed in the knowledge schemes, the African American studies professor and philosopher Cornel West aptly describes the positionality of how this community is depicted in the West:
Black people as a problem-people rather than people with problems; black people as abstractions and objects rather than individuals and persons; black and white worlds divided by a thick wall (or a “Veil”) . . . ; black rage, anger, and fury concealed in order to assuage white fear and anxiety; and black people rootless and homeless on a perennial journey to discover who they are in a society content to see blacks remain the permanent underdog.29
The library scholar Joan K. Marshall points to the way this idea was expressed in the Library of Congress when “N*ggers” was a legitimate subject category, reflecting the “social backgrounds and intellectual levels” of users, concretizing oppressive race relations.30 Difference, in the case of the Library of Congress, is in direct relation to Whiteness as the norm. No one has made this more clear than Berman, whose groundbreaking work on Library of Congress Subject Headings has forever changed the field. He notes that in the case of both Jews and the representations of race, these depictions are not without social context:
For the image of the Jew to arouse any feelings, pro or con, he [sic] had to be generalized, abstracted, depersonalized. It is always possible for the personal, individual case to contradict a general assertion by providing living, concrete proof to the contrary. For the Jews to become foils of a mass movement, they had to be converted into objectified symbols so as to become other than human beings.31
In the case of Google, because it is a commercial enterprise, the discussions about its similar information practices are situated under the auspices of free speech and protected corporate speech, rather than being posited as an information resource that is working in the public domain, much like a library. An alternative possibility could be that corporate free speech in the interests of advertisers could be reprioritized against the harm that sexist and racist speech on the Internet could have on those who are harmed by it. This is the value of using critical race theory—considering that free speech may in fact not be a neutral notion but, rather, a conception that when implemented in particular ways silences many people in the interests of a few.
The disclaimer by Google for the problem of searching for the word “Jew” leading to White supremacist, Holocaust-denial web results is surprisingly similar to the construction of Jewish identity in the LCSH. Both systems reflect the nature of the relationship between Jewish and non-Jewish Europeans and North Americans. This is no surprise, given that hyperlinking and indexing are directly derived from library science citation-analysis practices. This linkage between the indexing practices of the World Wide Web and the traditional classification systems of knowledge structures such as the Library of Congress is important. Both systems rely on human decisions, whether given over en masse to artificial intelligence and algorithms or left to human beings to catalog. The representation
of people and cultures in information systems clearly reflects the social context within which the subjects exist. In the case of search engines, not unlike cataloging systems, the social context and histories of exploitation or objectification are not taken into explicit consideration—rather, they are disavowed. What can be retrieved by information seekers is mediated by the technological system—be it a catalog or an index of web pages—by the system design that otherizes. In the case of the web, old cataloging and bibliometric practices are brought into the modern systems design.
Library science scholars know that bibliographic and naming controls are central to making knowledge discoverable.32 Part of the issue is trying to understand who the audience is for knowledge and naming and organizing information in ways that can be discovered by the public. Berman cites Joan Marshall’s critiques of the underlying philosophy of the Library of Congress’s subject-cataloging practices and the ways that they constitute an audience through organizational bias, wherein a “majority reader” is established as a norm and, in the case of the Library of Congress, is often “white, Christian (usually Protestant) and male.”33 Indeed, these scholars are taking note of the influence that categorization systems have on knowledge organization and access. What is particularly important in the interrogation of these marginalizing information-management systems is Berman’s reference to the Algerian psychologist Franz Fanon’s articulation of the mechanics of cultural “brain washing” that occurs through racist cataloging practices.34 Berman underscores that the problems of racial representation and racism are deeply connected to words and images and that a racist worldview is embedded in cataloging practices that serve to bolster the image and domination of Western values and people (i.e., White, European, and North Americans over people of African descent). The library practitioner Matthew Reidsma gave a recent gift to the profession when he blogged about library discovery systems, or search interfaces, that are just as troubled as commercial interfaces. In his blog post, he details the limitations of databases, the kinds of gender biases that are present in discovery tools, and how little innovation has been brought to bear in resolving some of the contradictions we know about.35
Algorithms of Oppression Page 15