It didn’t take long for Todd Lockwood to realize that a hierarchical book classification system would not work for the Brautigan Library. He was, after all, following through on Richard Brautigan’s vision to archive and curate unpublished manuscripts by unknown but inspiring writers. It was 1990, and people were traveling to Burlington, Vermont, from all over the world to experience what this strange and tender library had to offer.
But the source material, Brautigan’s first novel The Abortion: An Historical Romance 1966, didn’t cover book classification systems. By this point, Lockwood had far surpassed Brautigan’s original vision of a library that only sought to collect and preserve “the unwanted, the lyrical and haunted volumes of American writing” by opening the library up to the public at any capacity.
So Lockwood and early supporters of the Brautigan Library put their heads together and whipped up the Mayonnaise System: a thirteen-category classification system as easy to shelve as category and date in the order each manuscript was received. Some categories like “Adventure” and “Poetry” were seemingly more straightforward than “Street Life” or “Social/Political/Cultural.” Writers would choose which category they felt captured the essence of their works. The Mayonnaise System is more loyal to what the works mean than what they say. It’s a classification system where nothing must be as it seems.
This is in contrast to the Library of Congress Classification, the most widely used library classification system among academic libraries in the U.S. The decade the Brautigan Library implemented the Mayonnaise System was the decade Library of Congress Classification was rounding the corner toward its centennial. Melissa Adler, a professor at the University of Kentucky’s School of Library and Information Science, calls 1876 one of the most important years in library classification history.
As Adler points out, this was the year the American Library Association (ALA) and Library Journal was founded, the year Melvil Dewey published the Dewey Decimal Classification and Charles Ammi Cutter issued the foundational Rules for a Dictionary Catalog as part of the U.S. Bureau of Education’s “Special Report on Public Libraries.” Rules for a Dictionary Catalog introduced librarians and information scientists to the concept of evolutionary order. Rules for a Dictionary Catalog would come to evolve into Expansive Classification. In Expansive Classification, Cutter establishes twenty-one classes to organize all knowledge, followed by subclasses and sub-subclasses. The Library of Congress adopted these top classification elements in 1898, making Cutter’s classificatory principles among the most highly influential and lasting in the field.
The Mayonnaise System is more loyal to what the works mean than what they say.
Cutter’s life’s work revolved around the belief that a classification system informed by evolutionary order would evolve naturally and standardize collection growth across all libraries. And while no one can deny the need for a high-functioning information schema, the theory of evolutionary order is rooted in scientific racism. Thomas Dousa, a metadata analyst librarian at the University of Chicago Library theorizes in his research that Cutter may have been influenced by the men in his social circles, such as John Fiske—a librarian also passionate about introducing evolutionary-based classification into library catalogs—and Richard Bliss—an anthropologist who defended Cutter’s “natural” system for how it categorized social science topics were “not usually considered susceptible to a natural and systematic arrangement” in Library Journal.
It seems Cutter was in good company with the early librarian leadership establishment, receiving support for introducing scientific racism, or the now-debunked belief that empirical evidence exists to support or justify racial inferiority and racial superiority into his book classification system. We see the outgrowth of scientific racism, as well as scientific sexism and scientific heterosexism in Library of Congress Subject Headings, which are rooted in what Amanda Ros, a cataloger at Texas A&M University calls “the straight white American male assumption.” Research from Ros shows that about one-tenth of Library of Congress Subject Headings contain the word “men” in the title, compared to the number of subject headings that include “women.”
Ros finds that without gender, race or geographic qualifications subjects containing the word “astronauts” in Library of Congress Subject Headings can be assumed to mean white American men. As Ros illustrates in a 2019 article for The Conversation, “Women are designated with ‘Women astronauts’ and ‘African American women astronauts,’ but there is no subject heading for male astronauts. A book about astronauts who are men would have the general subject ‘Astronauts,’ unless the racial identity prompted the use of a subject like ‘Hispanic American astronauts’ or ‘Indian astronauts.’ Likewise, a book about Russian astronauts would have a geographic subdivision added: ‘Astronauts – Soviet Union’ instead of ‘Russian astronauts.’
As of March 2021, the Library of Congress reported nearly 300,000 Library of Congress Subject Headings, making it the “most comprehensive, non-specialized controlled vocabulary in the English language,” and it’s evolved into organized chaos. With Library of Congress Subject Headings, each heading breaks down the original subject but it becomes chaotic when most subtopics are interconnected to other topics. A single physical book can only be shelved in one place, and the more specific these subjects are, the less likely it is that lesser-known authors will be discovered and that canonized authors will appear. This can make it difficult to find works by various and historically under-represented or misrepresented people unless they look directly for them.
The results for library patrons and researchers is a gamifying of information discovery that comes from trial and error of the right combination of keywords and operators that classify and label diverse knowledges and cultures through the process of evolutionary othering. Imagine you are looking for information about astronauts, but don’t know where to start. So you type in the keyword “Astronauts” and the system spits out 10,000 results. You scroll through the top five, and they’re all about Neil Armstrong. You would have had to go through dozens of pages before finding anything on Mae C. Jemison. In order for books about Jemison to appear, you would have needed to type in “African American women astronauts.”
In this scenario, without an advanced search, locating anything outside of the white American male assumption makes locating and using works from diverse and historically underrepresented or misrepresented people intentionally difficult to find. Or, your search is hyper-specific and does include the qualifier “African American women,” but the results are still sparse to none because books on Jemison do not account for the ways identity intersects. In this scenario, the subject heading does little more than offer dead ends.
Are the legacies of dead white men the only possible way to organize all knowledge?
This has normalized the practice of information extraction and the pervasive nature of implicit biases that are continually built upon through Library of Congress Subject Headings. This poses the challenge of evolutionary order in classification systems and controlled vocabularies, that evolutionary order evolves to devalue or even propagate discriminatory language toward people as subjects. Despite attempts by the Library of Congress to address the subject of its criticism, concerns about the outdated and discriminatory language embedded in the headlines, the change occur only on a reactionary basis.
Cataloger Sanford Berman introduced a set of guidelines that have influenced Library of Congress subject heading change processes through Prejudices and Antipathies: a Tract on the LC Subject Heads Concerning People. To change a subject heading, the challenger must submit a proposal that proves that the language for the current subject heading is outdated, inaccurate, or offensive—putting the onus on the challenger. The proposal is then posted for public comment (which catalogers, librarians and lawmakers can also comment on). After about a month, Library of Congress staff will either approve, reject or return the proposal for revision. The process becomes elongated when the proposed changes are inherently political.
A recent example of this was when the underlying structural problem with Library of Congress Subject Headings resurfaced again recently when the Policy and Standards Division of the Library of Congress decided to replace the subject headings “Aliens” and “Illegal aliens” with “Noncitizens” and “Illegal immigration.” The prior subject headings have been challenged by catalogers, notably in 2016 by student activists at Dartmouth College who spent years petitioning the Library of Congress to remove “Illegal alien” from its subject heading. In a press statement, the American Library Association noted that the update better reflects its core value of social justice.
However, this victory brings with it yet another oversight. As Kelly Jensen pointed out in a recent Book Riot article, retaining the word “illegal” is problematic when describing immigrants, asylum seekers and undocumented individuals. “By retaining the word ‘illegal,’ the cataloging hierarchy continues to damage and showcases its inability to be as progressive and inclusive—and indeed, socially just—as it should be,” Jensen writes. “Instead, it acts to push back against conservative politicians only without a greater understanding of why a bigger change is necessary.” The language we use to describe people and their experiences impact the way they live their lives, and to group information together based on perceived commonalities of identity creates a barrier of access to knowledge from cultures and perspectives that differ from the positionality of the heteronormative white male in Library of Congress Subject Headings.
Catalogers are right that we need a discovery system that will make libraries useful to users. But are the legacies of dead white men the only possible way to organize all knowledge? If not, what could a uniform library classification system that represents works by the affluent and the marginalized look like? Could we eradicate the white American male assumption by doing away with subject headings entirely using the Mayonnaise System as a guide? It would, after all, transform the discovery process and how we come to experience libraries. It would force us to stop searching with identity qualifiers and instead through themes. It would force researchers to formulate hypotheses based on natural instinct rather than prescribed categorizations, and would allow us to completely rediscover how libraries can serve patrons not through knowledge extraction, but through freedom of exploration and imagination.
Unfortunately, we do not live in a world where mass adoption of the Mayonnaise System is feasible. Libraries worldwide have cataloged and shared with peer institutions billions of materials, many of which use the Library of Congress Subject Headings best practices to convenience patrons. The system built around Cutter’s Expansive Classification is so ingrained in our discovery systems, to undergo a massive database cleanup would mean going into each bibliographic record, identifying the number of Library of Congress Subject Headings and stripping them out.
With libraries focused on creating shared information hubs as part of the 21st-century vision for libraries, a shared controlled vocabulary is essential. The theory of evolutionary order classification has created a systemic mess that can never really be cleaned up.
There are other issues that arise with the widespread adoption of Mayonnaise. Submissions cataloged using the Mayonnaise System often blurred the lines between fiction and nonfiction. While this worked for the Brautigan Library, it would be impractical to blur the distinctions between fiction and nonfiction works with the virulent spreading of disinformation across digital platforms.
(There is also the issue of the mayonnaise jars. While current curator John Barber did away with the physical mayonnaise jars years ago, many bibliophiles may believe that in the early days, the Mayonnaise System may have taken itself too literally. For years, the library used 32-ounce jars of Helmann’s Mayonnaise as bookends to describe the sections for patrons where one category would end and another would start. While Lockwood reports no messes, the risk may not be worth the commitment to visual representation.)
But there is still something to be said for taking a proactive stance against perpetuating the use of controlled vocabularies that contain inaccurate and harmful implicit biases. We must recognize that grouping information together based on perceived commonalities of identity creates a barrier of access to knowledge from cultures and perspectives that differ from the positionality of the heteronormative white male assumption that’s deeply entrenched in Library of Congress Classification and Library of Congress Subject Headings.
The Mayonnaise System may be a pipe dream for library cataloging, but institutionalizing a more thematic approach to how we use libraries is an interesting thought experiment. So I’ll compromise: What if we removed the white male assumption from the Library of Congress Subject Headings by locating all of the subject headings with a white American male assumption attached to them and create a new subheading that divorces the assumption from the reality? For instance, we could add a new subheading called “White male astronaut” under the subject heading “Astronaut,” and make a note about the correction.
With a concerted effort, we could revolutionize the parts of evolutionary order that are based in pseudoscience. While everything remains less detectable, this change could be implemented within a few months and could change the way we think about libraries, research and each other. Then we could finally have a more democratic classification system with a controlled vocabulary that changes the subject from which subject heading is less worse to discussions about implementing a simplified information schema that doesn’t need to distinguish between facets of identity at all. Kind of like the Mayonnaise System.