Beyond ‘Ditch Dewey’ in the academic library

An old 90s television displaying a blue screen, set in front of a red brick wall

‘Why don’t we just ditch Dewey?’

Well, why don’t we just. After all, it’s a fair question, provocatively asked by a client services librarian at a session of our week-long end-of-year zoom gathering. My heart sank. Of course somebody asked this. A fellow colleague cheekily posted in the zoom chat ‘Your time to shine, Alissa!’ I can’t recall ever meeting him properly, but I guess he knew who I really was? I could see the University Librarian about to respond, as the Q&A portion of the hour was intended for the library executive, when suddenly I found myself interrupting, unmuted, to over a hundred people:

‘Uh, could I just say something here?’

With approximately zero seconds notice I ad-libbed five or so minutes of explanation around this idea, having entirely forgotten to put my camera on. My hands were shaking by the end of it. Surprise public speaking is really not my thing.

Here’s a much fuller version of what I think I said, with some added points that I only thought of after a strong peppermint tea and some chillout time with my fern collection. I know my comments were recorded for an internal audience but I’m deliberately not going back and listening to them! While I’m not a classification expert, I have spent many years agitating for more critical attitudes to this work, and I am the metadata team leader here, after all. If anyone was going to give a (decent) answer to this question, it was gonna be me.

Smart move, loser. Now you’re gonna have to say something!

Shut up, brain.


These days, ditching Dewey is no longer an outrageous, unthinkable suggestion. Only 19% of American academic libraries were found to be using DDC in 2018, with that number steadily dropping. The biggest problem now is what to replace it with. If there existed a better classification system for generalist academic libraries like ours, chances are someone would be using it already. I could spend my entire career devising something better and still never be finished – and it’s not like OCLC are resourcing this work for DDC anymore. The only other system in wide use in Australia is Library of Congress Classification (LCC), which isn’t exactly a better option. It encodes the biases, perspectives and priorities of the United States government, just like LCSH does, and why would we want that in our library? The nexus of classificatory power is located far, far away from us, when it needs to be right here. LCC has a tendency to historicise Indigenous peoples—that is, place everything about them in the history section, as if they have ceased to exist entirely—and encodes archaic and offensive perspectives on topics as diverse as Arabic literature and the geography of Cold War-era Eastern Europe.

Genrefication isn’t really an option for academic libraries. Our collections are designed for serious research and study, not recreational reading or other types of lifelong learning; whimsical genres are largely inappropriate for an academic setting. Besides, even public library genrefication projects, such as the one at the flagship branch a stone’s throw from my office, are often built on Dewey’s foundations; the books are shelved in distinct genres, yet continue to use a DDC number as a shelfmark, meaning the substrate logic remains the same but is made impenetrable to a casual browser. It makes the browsing experience more frustrating, because you no longer have the granularity of Dewey to guide you, only a broad category. Academic library users are typically quite focused in their browsing. We couldn’t just say ‘here’s the economics section’ and leave them to it—we need the kind of granularity only a formal classification system can provide.

Our print collections have been largely unavailable for browsing for the best part of two years. We’ve been doing distance education for decades and have a large and growing cohort of exclusively online students. It’s not like a lot of people are actively browsing our physical collections right now. Also, reclassifying an entire print collection fills me with dread! My team of 1.6 FTE are nowhere near resourced enough for such a enormous undertaking – physically retrieving, reclassifying, restickering and reshelving every single print book in our branches would take months and involve huge amounts of work. We just don’t have that kind of capacity.

Besides, is large-scale reclassification truly the best use of my team’s limited time and considerable talents? I feel like there are more immediate and more focused things my team and I can do to improve the cultural safety of our metadata. We could be adding AUSTLANG codes and AIATSIS subject headings to our First Nations materials, overlaying records from Libraries Australia that already feature this data. We could work to contextualise offensive and culturally unsafe depictions of First Nations topics, adding content warnings where necessary. We could work with our Indigenous knowledges institute and our colleagues in the Archives to apply Traditional Knowledge and Biocultural Labels for materials we collectively hold, ensuring their provenance, protocols and permissions are clearly documented.

I had a conversation with one of our liaison librarians and her boss a few weeks ago about her campus library shelving First Nations Dreaming stories in 398.2, the myths and fairy tales section. The juxtaposition of core cultural and religious texts with lightweight children’s stories is manifestly inappropriate; the former’s continued placement here, in spite of clear DDC instructions to the contrary since 2003, is a damning indictment of the cultural incompetence of library cataloguers.1 My liaison librarian colleague mentioned this situation in the zoom chat—I suspect it was a surprise to many of our colleagues, including the University Librarian. I responded after I had finished speaking with ‘Absolutely – I look forward to fixing this next year’. I was going to do it anyway but it’s nice to have senior management buy-in for these things 🙂

The Dewey Decimal Classification was designed by a particular man, at a particular time, in a particular place, with a particular collection and for a particular audience. His notoriously questionable values and those of the classification system that bears his name are, by and large, not those we share today. But who is ‘we’? Who gets to put books where? Whose values are encoded and embodied in the placement of books on shelves? What values would our institution like us to project in our physical and digital spaces? What about the multitude of value systems that our students and researchers bring to those spaces? How do we represent such multitudes in a linear shelf arrangement? Should we even try?

Don’t get me wrong, colleagues: I am all in favour of doing classification differently. But please don’t underestimate the difficulty and the sheer amount of work involved. It’s not as simple as just ‘ditching Dewey’.


At this point I ran out of courage and trailed off. The University Librarian, who I get the impression chooses their words carefully, said nothing for a few agonising seconds before inviting a question from another audience member. I had noticed them listening intently as I spoke. I hoped they didn’t mind me interrupting.

Two days later, in another session of our week-long end-of-year zoom gathering, the conversation turned to the ethics of AI, and how systems reflect the biases and perspectives of the people who build them. The UL remarked that the discussion really went to the heart of critical librarianship—recognising that the library profession also has a long history of perpetuating all sorts of biases and harms in our work. And I just about fell off my chair.

Was this real? Did I really work in a library where the University Librarian not only uses the phrase ‘critical librarianship’ in front of the entire staff but makes a point of actively living those values? Did I really hear the directors echo those comments and agree that perhaps it’s time to reconsider our classifiation practices? Did I really hear the UL namecheck me twice in the closing session, for both my colleague’s presentation on our project to fix our batch file loading processes and also for my impromptu Dewey comments? Did I really say all that in front of everybody?!

Truly I feel like I’m working in paradise. It’s one thing to blurt out my Big Metadata Feels in response to a question that wasn’t even directed at me, but it’s quite another for my senior management to embrace these ideals, making a point of publicly supporting the work I do and the things I am so passionate about. Hearing ‘critical librarianship’ out loud at work has just about made my year. Every day I am so grateful to be here. I can’t wait to make good on this promise.


  1. I did not say this part out loud. 

Classifying autism spectrum disorders in DDC, LCC and NLM

Everyone, justifiably, wants to see themselves reflected in their library’s classifications. But the two major classification systems used in Anglophone libraries—Dewey Decimal Classification (DDC) and Library of Congress Classification (LCC)—as well as the (American) National Library of Medicine classification (NLM) have a long history of reflecting the biases, perspectives and limits to knowledge of the times and spaces they were first devised. Sometimes small aspects are updated, but structural biases are baked in, and far harder to fix.

The following outlines the treatment of autism spectrum disorders in these three major classification systems. None of them are sympathetic to the neurodiversity movement, and range from the benign to the downright offensive. It’s an insight into the history of social and medical attitudes toward autism, but a classification system is not the right place to be storing that history. I wish we could move with the times.

Dewey Decimal (DDC)

Works on the medical aspects of autism are classed at 616.85882, under ‘Intellectual disabilities; developmental and learning disorders’. This is how the medical establishment sees us, so therefore this is how Dewey sees us. The broader number 616.8588 sits between factitious disorders (including Munchausen syndrome) and ADHD, and is itself part of a grab-bag of socially-marginalised disorders at 616.858 that also include personality disorders, gender-identity disorders and ‘disorders of impulse control’. Can’t say I love this particularly pathologised perspective—and that’s even after looking the other way at ‘Diseases’!

The scope note reads: ‘Class here comprehensive works on pervasive development disorders’, with a note for PDDs other than autism to be classed at 616.85883. This echoes the DSM-IV and ICD-10 (that is, a previous) approach to autism, which classed autism as one of five pervasive developmental disorders. The DSM-5 and ICD-11 moved to using the term ‘autism spectrum disorder’, encompassing a range of autistic traits and severities, including those previously categorised as Asperger’s syndrome. Asperger’s is classed at 616.858832, but as this term is no longer used, I imagine the call number will eventually fall out of use as well.

WebDewey notes that the class number for autism changed with DDC edition 22, published in 2003. Previously autism was classed at 616.8982, as… a subtype of schizophrenia. I gotta admit, this was news to me too. Autism was once considered a form of childhood schizophrenia; while WebDewey doesn’t tell me when a class number was first introduced, I’m guessing this dates from around the 1960s or 1970s. It could be worse, for sure, but it could be a lot better, too.

A class number for the social aspects of autism was harder to find. WebDewey returned no results in the 300s for the search term ‘autism’, but returned two strong suggestions for the search term ‘developmental disabilities’: 305.9085 for works on autistic people ourselves, and 362.1968 for social services to autistic people. The term ‘developmental disabilities’ doesn’t exactly reflect how I see myself, but I’m very aware these schedules were not designed with low-needs autistic people in mind.

Library of Congress (LCC)

Until recently autism had only one LCC call number: RC553.A88, under ‘Internal medicine—Neurosciences. Biological psychiatry. Neuropsychiatry—Psychiatry—Specific pathological states, A-Z—Autism. Asperger’s syndrome’. I must admit, ‘specific pathological states’ is a more polite descriptor than I had expected to see in LCC—I don’t entirely hate it. Being a straight A to Z list it sits between ‘Auditory hallucinations’ and ‘Cognition disorders’.

Library of Congress cataloguer Netanel Ganin recently wrote about his efforts to address this absence, reinterpreting a call number range in the social sciences, HV1570, to include the social aspects of autism spectrum disorders. This accords with the treatment of other disabilities, such as blindness and deafness, whose medical aspects are classed in R and social aspects in HV.

Netanel notes that the full hierarchy of HV1570 (‘Social pathology. Social and public welfare. Criminology—Protection, assistance and relief—Special classes—People with disabilities—Developmentally disabled’) is not without its problems, as LCC can’t help pathologising autistic people as needing ‘protection, assistance and relief’ and most medical literature regards autistic people as being developmentally disabled, which also explains its preponderance in DDC. This class number is, however, an improvement on LCC medicalising the entirety of the autistic experience.

As an autistic cataloguer I applaud Netanel’s work in this area to help books find their most appropriate home in the LCC schedules, and to make the best of a bad system.

National Library of Medicine (NLM)

Sadly, NLM classification is the worst of the lot. Here, autism doesn’t even warrant listing under its own name, instead being lumped under ‘Pervasive child development disorders’ and classified with ‘Schizophrenia Spectrum and Other Psychotic Disorders’, at WM 203.5. The number for autism sits between schizophrenia and neurocognitive and perceptual disorders.

In a rare act of classificatory transparency, NLM schedules are full of ‘[This number not used]’ where a call number has been removed and its subject classified elsewhere – unfortunately, between the Cutter number and the see reference text, one can often surmise which archaic or offensive words or concepts were previously listed.

Unlike DDC, NLM continues to encode the discredited view of autism as a form of childhood schizophrenia by choosing WM 203.5, instead of an unused number in the WM 200s. Yet as medical understandings of autism spectrum disorders have grown and improved, their classification here remains stuck in the 1950s. It’s also very strange that a call number relating to child development disorders, a diagnosis typically made in, you know, childhood, is specified for works relating to adults only.

Adding ‘spectrum’ to the broader category doesn’t change which individual disorders are collocated with each other. Nor does it change the overall message that sends. Am I supposed to be grateful that autism isn’t classed as a mental illness, or an intellectual disability? I would have expected NLM to be more in line with the classification decisions made by the DSM and ICD, but instead they’ve changed a dressing instead of closing an open wound. I hope they will reconsider this classification in future.

Flatland and the limits of fiction

The original cover of Flatland (image courtesy ArchDaily.com)

The other day I decided I’d better get started on the books in my enormous to-read pile, preferably before I have to return half of them to the library I work at. The topmost book just happened to be Flatland: a romance of many dimensions (1884) by Edwin Abbott Abbott (1838-1926). A colleague had recommended it to me as part of a potential display on ‘flat’ books. (We share a building with a public library branch, and I was thinking of doing a book display which, for once, had nothing to do with local history. I seem to recall being in a flat mood at the time.)

I hadn’t even opened the cover when I was distracted by the book’s call number. This happens to me a lot.

530.
11
ABBO

Flatland, as far as I can ascertain, is a work of fiction, and has been since 1884. Why was it classed in non-fiction? Is this a common view? Who made this decision, and why did they make it?

Firstly, let’s examine this number. DDC 530.11 is the home of general relativity, among such physics luminaries as Albert Einstein, Stephen Hawking, Lawrence Knauss and Brian Cox (whose works on the subject are definitely not fiction). Einstein popularised (and drew on others’ efforts in consolidating) the theory of general relativity in 1915.1 That is, over thirty years after Flatland was published. How can this be?

Is it because Flatland is primarily about science, and the term ‘science fiction’ didn’t exist when the book was published? Did the cataloguer of this particular item even know it was a work of fiction? I flipped the book over and read the synopsis. “Flatland (1884) is an influentual mathematical fantasy […] A classic of early science fiction.” Hmm. Looks fairly clearcut to me. Maybe they weren’t in a synopsis-reading mood.

I then looked at the book’s CiP details. Which call numbers did they provide?

PR4000              823.
.A22                8
F53
2009

‘Aha!’ I exclaimed, to nobody in particular. So even the CiP cataloguer knew it was fiction, and classified it as such! The LCC call number PR4000 is English literature > 19th century, 1770/1800-1890/1900 > Individual authors (with A22 F53 presumably being the cutter for Abbott, Edwin. Flatland), while DDC 823.8 is Victorian-era English literature. I’m beginning to get a bit cross at our cataloguer by this point, who seemingly hasn’t read either the synopsis or the CiP data.

I’ve now spent over an hour investigating this book. I haven’t even started reading it yet.

Interestingly, the CiP for this work was done by Library and Archives Canada, as the editor of this particular version is Canadian. Hmm. Did Library of Congress treat this book differently? (No offence, Canada)

I had a peek at the Libraries Australia record for this edition. This record is an LC copycat job, using the original Canadian data, but making a few slight changes…

QA699              530.
.A13               11
2010b

Bingo!! So that’s why the cataloguer of my copy of Flatland thought it was non-fiction—because Library of Congress did too! … Or did they?

While DDC 520.11 is for relativity, as discussed above, LCC QA699 includes a fascinating—and critical—scope note: Geometry > Hyperspace > Popular works. Fiction (Including Flatland, fourth dimension)

According to LC, it is so crucial that Flatland be classed with non-fiction works on hyperspace that it’s literally namechecked in the scope note! This is amazing! But why did they do that? And why was this logic reproduced in DDC?

I then decided to browse LC’s collections at QA699 to see what else was there. The most recent work appeared to date from 1971, with the bulk of works (excluding revised and edited editions, of which the book in my hand is one) dating from the late 19th and early 20th century.

A quick scan of works in QA699 suggested almost all of them were indexed with the LCSH $a Fourth dimension. The scope note for this heading reads: Here are entered philosophical and imaginative works.
Mathematical works are entered under Hyperspace.

LC holds 100 items with this heading, but it also holds 5 items indexed $a Fourth dimension $v Fiction, which are not classed in QA699. These items include a book about Flatland: the movie edition (2008) and four novels published in the 21st century.

So what is this telling me? Flatland, according to LC, is a ‘philosophical’ or ‘imaginative work’, which suggests they think it’s too intellectual to be considered merely a work of fiction. But this seems like a load of crap to me. Isn’t all fiction inherently ‘imaginative work’? Is Flatland accorded this kind of respect because it’s old, was written by a white man, and has increased in intellectual stature over time? Did the cataloguer at LC who originally wrote the QA699 scope note (however many years ago that was, and may or may not have been the same cataloguer who processed Flatland) decide that this work was not mere literature, fit for the P class, but a higher-order piece of writing that ought to reside near the subjects it fictionalised?

Hmph. This reeks of classism to me.

But it also explains why the book in my hand has the entirely inappropriate DDC call number 530.11: the cataloguer at LC probably looked up the closest thing to ‘Hyperspace’ in DDC (that being ‘Relativity’), either didn’t notice or didn’t care about the ‘Fiction’ aspect of QA699, classed the book where DDC said to and moved on with their life, with dozens of subsequent copy cataloguers not knowing, not caring, or not being paid enough to reconsider this choice. I’ve stopped being cross at our cataloguer, who clearly saw no reason not to defy the ANBD record. I can understand where they were coming from.

For me, the question now becomes: will I change it locally? Technically I have this ability, but because it’s not a local history book I’m supposed to refer it to our collections team. They have way more important things to do than change the call number of a perfectly findable item, especially because nobody’s yet complained about it.

I’ll think about what to do next. For now, though, after several hours of investigation and writing, I might actually get started on the book.

I think it’s the least I could do.


  1. For a longer explanation of why I didn’t just say ‘invented’ like everyone else, see this piece on ‘Who invented relativity?’ http://www.mathpages.com/rr/s8-08/8-08.htm