1 of 17

Falsehoods Librarians Believe About Names

Islandoracon 2019

Vancouver, BC

October 9, 2019

2 of 17

Charlie Tillay (they/them) -- Cataloging and Metadata Librarian at Tulane University; undergoing NACO training; involved as a MIG co-convener; please call me Tillay

Rosie Le Faive (she/her or they/them) -- Systems and Metadata Librarian at the University of Prince Edward Island; Islandora longish-timer, Angry giver of fucks

Tillay:My name is Tillay. I have personal stakes in many of the questions we’re going to ask and answer about names and identity today. I'm also a NACO contributor through my work at Tulane University as Cataloging and Metadata Librarian. I also have some experience with local authorities in digital libraries, and I've been attempting to map those to Linked Data as a co-convener of the Islandora Metadata Interest Group.

Rosie: Im Rosie. I'm sort of a metadata librarian at UPEI. I want to do things the right way. I also have been doing Islandora since… July 2012.

R: It’s been over 9 years since Patrick McKenzie wrote “Falsehoods Programmers believe about names”, and computer systems still mangle my last name because there’s a space in it. I get it, migrations are hard. At Islandoracon two years ago, Alan and I gave a presentation to this extent.. I still haven’t finished those migrations, by the way. So I get it. Migrations are hard. But I get frustrated, and angry. Why is it so hard to encode a space? I know it wasn’t done out of malice or deliberate exclusion, but because the people in the room making decisions about how it would work, didn’t realize that they were making decisions.

3 of 17

Foundational Information

Authority File is to Authority Record as

Drupal Vocabulary is to Taxonomy Term

4 of 17

Foundational Information

RDA superseded AACR2.

It was an attempt to answer these questions about modeling library data in Linked Data.

5 of 17

It’s tempting to think “let’s mint objects!”

WORK

Author

dc:creator

“Rosie Le Faive”

WORK

Author

dc:creator

“Rosie Le Faive”

AGENT�(minted object)

Name

foaf:name

foaf:person

a

6 of 17

Doing a MODS to RDF mapping will mean...

“Rosie Le Faive”

“Rosie A. Le Faive”

“Rosie La Favie”

“R. A. Le Faive”

“Le Faive, Rosie”

“Le Faive, Rosemary”

“Rosie Le Faive”

“Faive, Rosie Le”

normalize

“LeFaive, Rosemary”

7 of 17

Doing a MODS to RDF mapping will mean...

“Rosie Le Faive”

“Mint an object”

(normalized form)

Name: Rosie Le Faive

(Drupal taxonomy term)

8 of 17

Doing a MODS to RDF mapping will mean...

“Rosie Le Faive”

“Mint an object”

(normalized form)

Name: Rosie Le Faive

ORCID: 0000-0002-0116-0984

LCNAF:

a.k.a.: Rosemary Le Faive

d.o.b.: [redacted]

(Drupal taxonomy term)

R: Now that it’s a “thing”, we could give it more properties. Like other forms of the name! Or existing identifiers! (I have an ORCID, not an LCNAF record)...

T: Do you have a standard about which unique existing identifiers you’ll link to, and when, and prove those are the right ones?

R: Reconciliation! Yeah, we’ll search known authority files… I mean, for ORCID they make it explicit that institutions aren’t allowed to search by name, and then has to be taken from the work in hand, we’re not allowed to search and say “yeah, that’s the right name, and the right institution” and assume from that that we can link up the ORCID, but if they published the thing in our repo, and they put their ORCID on it, the “real published authors” who are in LC, we’ll string search the name authority file, and assume “Montgomery, L.M. (Lucy Maud) is the one we want.

T: You know that’s harder than you think, right? Especially if your name isn’t formed the same way as in the authority file. And what about when you don’t have enough information to tell if there’s a match, like, if you just had initials?

R: eeehhhh… we’d leave it blank? The identifiers are optional.

9 of 17

Doing a MODS to RDF mapping will mean...

Name: J. Smith

ORCID:

LCNAF:

a.k.a.:

d.o.b.:

(Drupal taxonomy term)

10 of 17

What is Authority Work?

Choosing a preferred form of name is authority work.

Adding links to other identifiers (incl. LCNAF) is authority work.

Disambiguating between different people with similar names is authority work.

T: You've probably been doing more authority work than you think. If your Islandora 7 gathers names into links or faceting based on character strings, this implies authority work even if you didn’t do authority work. In Islandora 8 you will definitely have to make decisions about authorities. In Islandora 8, names autocomplete in the metadata forms, which is de facto authority work. In order to get your records into Islandora 8, you will have to make a decision about whether you’re creating an object for each string, pulling in strings with no connection to objects at all, or something in between. These are decisions about authorities.

Even when you’re not creating authority records, if you’re adding links to an authority record in your bibliographic record, you may be impacted by these issues if you’re:

Using the wrong record, especially if you’re matching identities programmatically
Using a record that’s for the right person, but that has ethical issues

11 of 17

Authority work:

Includes a constellation of standards

There are many standards - RDA, DCM Z1, and LC-PCC PS

They instruct on how to:

Form names
Assign names
Link multiple names for a single person
Describe a person who has a complicated identity
Cite sources for your data
Enhance the record with personal information

RDA and DCM Z1 (Descriptive Cataloging Manual, Z1: Name and series authority records), LC-PCC PS (Library of Congress-Program for Cooperative Cataloging Policy Statements) are the three I most often use, pulling info about punctuation from one, how to encode the data in the schema from another, and figuring out what data to include from another.

There are more rules than anyone can ever master. There are dozens of special cases: people who have a name phrase (think Will.i.am), people who have a single preferred name (Cher), persons who are a pseudonym for multiple people (Carolyn Keen), people who wrote under multiple pseudonyms (Mark Twain), Queens and Bishops best known by their titles, etc. There are detailed instructions on when to include fields, what exactly should go in each field and subfield, how to justify those elements with citations, what sources to use for citations. I’ve been training on NACO for months and I still know I’ve only touched the surface.

12 of 17

Sample Fake Authority Record

Citations/670 fields:

Suave, M. L., How to catalog cats, 1904: $b t.p. (M.L. Johnson) introduction (cataloger, librarian, and humorist) pg. 4 of cover (M.L. Johnson, they/them, born Marquisse Liane in Alexandria, Louisiana, lives in Washington D.C., Principal Cataloger at the Libraries of Congressional Mishaps, regular contributor to the humor and book review sections of the New Onion News, identifies as a cataloger)
New Onion News, 31 February 1918, viewed online 31 February 2022: $b Being Non-binary (M.L. Johnson, assigned female at birth, announced their gender as non-binary when they were 34 years old, identified as non-binary privately from the age of 5)
New Onion News, 1 March 1929, viewed online 31 February 2022: $b Queer Marriage Complications (M.L. Suave chose to take their wife’s last name, wife is a trans woman, Suave rejects the label of lesbian)

T: Here’s a record I created to demonstrate the conflicts involved in personal name authority work. I’m not going to explain each element or the precise RDA instruction I applied each time, we’re just going to act out a pro-and-con discussion of the decisions made in this record so that you can get a sense of the extent of human thought, care, and research required to create a record and the ethical ambiguity involved.

T: The place to start when you create an authority record is always the source(s) of the information about the person.

R: Why did you cite these random-seeming sources?

T: Our instructions are to create the first citation from “the piece in hand” and that’s the book I had to catalog. We then have a list of preferred sources within each item (title page vs. cover vs. author’s blurb) and a set of pieces of information you ideally note down (such as every unique way you find the name represented). Then I repeated this process for two articles the author was discussed in or for which I could tell they wrote. Next I may have looked at bibliographies or encyclopedias. The last resort, and ideally only to get data to break a conflict with the normalized version of another authority record’s authorized access point, is to contact the agent themselves.

R: Why wouldn’t the first thing to do be to contact the person and get this information directly from them?

T: I don’t entirely know the full history of why catalogers do this this way. In many cases it may be that it was assumed that we couldn’t contact the person. Often the subject of an authority record is dead, or you have no contact information for them. It may have been that we were “staying in our lane”...gathering data available in the library or the catalog of record made sense as our first resort because that was the pond we swim in. Certainly there’s some concern now that contacting people to ask them about including information in an authority record might in and of itself be a violation of privacy.

R: Similar to Facebook employees emailing you to ask you if they can add your gender to your profile since you haven’t yet?

T: Essentially, that’s the ethos we’re trying to avoid.

13 of 17

Choosing one string to represent an identity, making it unique, and linking it to other strings used for that person is the core of authority work

The 100 field

R: Are you even sure you should write their name in indirect order (ie. Last, First)?

T: I’m fairly confident because I did follow the RDA instruction 9.2.2.4 and “chose the part or combination of parts to record as the first element” based on “reference sources published in their language/place of residence”. I didn’t have evidence that their “preference is known to be different from normal usage”, so I didn’t follow that exception. And before you ask, I also used this language and this transcription of their name because it’s the form that “corresponds to the language of most...” of their writing (RDA 9.2.2.5.2).

R: But why do this for their work anyway? Why not just include their name as they would have written it, as a string?

T: Theoretically, I only create authorities in a few cases, such where it is necessary to distinguish between multiple people who use the same name, or when I anticipate that this name may show up in records in the future and I have the best chance of creating an accurate record (because my library specializes in materials about them or the location they’re from or I hold their personal papers). In practice, Islandora 8 is rapidly becoming an example of another reason...because the system requires each name to have a unique identifier for validation and I don’t want to create 99 identifiers for one person just to make my system play nice with my records.

R: So, in this case, your sources indicated that this person has a clear preference for their initials, but you included their full name anyway? That’s rather disrespectful.

T: If I had a choice, I would not have included this in the 100 field at all. To show that I know their name would be M. L., I do explicitly leave only those in the delimiter a or personal name. Their fuller form of name is only included to break a conflict with other M. L. Suave’s. That’s why it is written in parentheses after the delimiter q...to acknowledge this is their name, but not what they went by. FYI, sometimes fuller form of name is a nickname they used...it doesn’t imply the colonialist idea of their “real, legal” name. Because I know that this person is trans, I could have left this out and used (humorist) to break conflict instead, but I decided that I would treat their name the same way I would for an apparently cisgender subject.

R: So, assuming that we think that’s a good enough reason to include information that might be a dead name, how do we maintain the meaning implied by the subfields, e.g. subfield q/ in simple strings or even URIs? Because if our access points don’t include those delimiters, how will anyone know that the preferred name is only Suave, M. L., without the “fuller form”?

T: I don’t know, and that’s part of what we’re trying to figure out.

14 of 17

Enhancing Fields?

T: While there are admittedly possible pitfalls about including some of the new fields allowed/encouraged by RDA, there are also tremendous possibilities. Two examples are that we can now choose from a wider variety of characteristics when adding qualifiers to break conflicts (distinguish between two identical name strings), and we may be able to leverage those fields to gather resources based on characteristics Cutter never dreamed of.

R: (worried) There are risks associated with adding personal identifying information like places to these files. Is it worth recording places someone had lived, when it may open them to the risk of, say, identity theft?

T: That’s a worthwhile question, and one that should absolutely be examined for each identity. In this case, I mentioned that I used the information about places associated with this identity to decide which rules to follow about which name to treat as a family name or a last name and whether to record the name in indirect order. By including this data in its own field, it gives us a kind of audit trail(?)...future catalogers can verify what information I based my decisions on. If they have more updated data, they would be able to fix the decisions I made on old or wrong information. That’s the rule-follower answer.

There’s also a rule-tester answer...in AACR2 the options for breaking a conflict were limited. I believe the only options were fuller form of name and/or dates. This created a problem, especially when the verifiable data about a person were limited and especially when dealing with identities for which the available fuller form of name and/or dates are oppressive. By having these other elements I can use, even if I have limited information, I might be able to use an aspect of their identity I know they had control over and I know they chose as part of their identity, rather than something they had no control over. No more describing enslaved people by a name given to them by their captors, their captor’s last name, and the date they were sold...the new RDA allows me to choose to name them with the title they used within their community (possibly a title). No more choosing between deadnaming a trans person by using their “fuller form of name” or putting them at risk of being fired due to ageist retirement policies, I can use their occupation.

R: So… this includes a LOT of judgement for each individual case. How am I going to do this for the huge batch of MODS automatically, so I can ingest it?

T: That’s the point. You can’t. Authority work is work. It involves training. It involves sensitivity and judgement. It can’t be automated; the point of authority work is to have humans doing this work. With care.

R: But even then, there are so many ways to harm a person. If someone uses a pseudonym, including a “legal name” is interesting but might put someone at risk. Including a dead name. Insisting on a “single preferred name” when someone in real life has multiple names, in multiple contexts.

T: The training and instructions NACO cataloguers receive reminds us that we should not include things can that do harm. Cataloguers’ judgement in RDA empowers me to customize the instructions as appropriate.

R: you’d have to be really culturally competent to know what is culturally appropriate.

T: Yes, that’s the point, and also, if someone more culturally competent comes along, they can fix my record through a NACO proposal. Or even the person themselves can contact LC to get their authority record fixed.

R: Yeah, but you have to know where to look, and dig to find the contact info. LC needs to make it a lot more transparent.

15 of 17

Final Thoughts

Foucault, Michel. "The Subject and Power." Critical Inquiry 8, no. 4 (1982): 777-95.

Fox, Violet B. and Kelly Swickard. ""My Zine is My Private Life": Reframing Authority Control from Detective Work to an Ethic of Care." In Ethical Questions in Name Authority Control, edited by Jane Sandberg, 9-24. Sacramento: Library Juice Press, 2019.

Ezli, Erin and Katherine M. Crowe. “This is the Oppressor’s Language Yet I Need It To Talk to You: Native American Name Authorities at the University of Denver.” In Ethical Questions in Name Authority Control, edited by Jane Sandberg, 9-24. Sacramento: Library Juice Press, 2019.

Adolpho, Kalani. “Who Asked You? Consent, Self-Determination, and the Report of the PCC Ad Hoc Task Group on Gender in Name Authority Records.” In Ethical Questions in Name Authority Control, edited by Jane Sandberg, 9-24. Sacramento: Library Juice Press, 2019.

R: Breaking out of Tillay’s head and talking as Rosie:

R: Anyway, it’s still colonialist. It’s still western-centric, assumes the primacy of a written form of name, does not include the names’ cultural context, and naming someone like this, is a symbolic if not conceptual assimilation into a largely white, western, colonialist, heteronormative structure of power where we, potentially outsiders, as Adolpho describes, assume reality can be “transformed into standardized, aggregated, static facts that are capable of consistent documentation”, thus making individuals “subjects”, easier to describe and control.

T: I don’t disagree with you, I just think this is the best we can do for now, and it’s certainly better than assuming all names can be formed last-name-comma-first-name.

16 of 17

Final Thoughts

No perfect solution
Not doing authority work can create problems
Doing authority work carelessly can harm actual people
Clarify the purpose of authorized forms of names for your institution
Choose an authority file for names that meets those purposes
Scrap the Drupal Name module
Consider creating a name module based on a recognized standard. I recommend RDA for now, but I know we can do better eventually.
Create a take-down policy for your metadata. Publicly provide individuals with a way to contact you if they notice errors in your authorities. (LC should probably put one up on id.loc.gov prominently too!)

T: There is no perfect solution. Whether you do authorities or not, information about people is in your Islandora, and you will do harm. However, having trained professionals who know the RDA standard means that you will do less harm. Even when you create records that are problematic and the standard changes, the fact that your names conform to a standard will mean your metadata is more remediatable than if you just stick to strings.

These next steps are all important, but I want to really emphasize one point above all: investment in the people who do this work is vital. Over the last few decades, automation and outsourcing have allowed libraries to shrink their technical services and cataloging departments drastically. The increasing demands of metadata work for our digital libraries show that we have to reverse this trend and quickly. In order for metadata librarians to do their work well, going into the future, they need the grounding and training in traditional cataloging work they can only receive from those knowledgeable in original cataloging and creation of authority records. We have to reverse this trend now...our metadata is already unwieldy and our technical services under-staffed, and it will only get worse if our metadata librarians continue to create records that don’t meet standards and we have fewer catalogers to train them.

17 of 17

Reading List / Sources

Billey, Amber. "Just because we can, Doesn't Mean We Should: An Argument for Simplicity and Data Privacy with Name Authority Work in the Linked Data Environment." Journal of Library Metadata 19, no. 1-2 (2019): 1-17.

Billey, Amber, Emily Drabinski, and K. R. Roberto. "What's Gender Got to do with it? A Critique of RDA 9.7." Cataloging & Classification Quarterly 52, no. 4 (2014): 412-21.

Foucault, Michel. "The Subject and Power." Critical Inquiry 8, no. 4 (1982): 777-95.

Fox, Violet B. and Kelly Swickard. ""My Zine is My Private Life": Reframing Authority Control from Detective Work to an Ethic of Care." In Ethical Questions in Name Authority Control, edited by Jane Sandberg, 9-24. Sacramento: Library Juice Press, 2019.

McKenzie, Patrick. "Falsehoods Programmers Believe about Names." https://www.kalzumeus.com/2010/06/17/falsehoods-programmers-believe-about-names/. (accessed 4 October 2019).

“9: Identifying Persons”. RDA Toolkit. (April 11, 2017) http://access.rdatoolkit.org/rdachp9.html (accessed 4 October 2019).

The Library of Congress Cataloging Distribution Service. “Descriptive Cataloging Manual -- Z1. Name and Series Authority Records.” Cataloger’s Desktop, Issue 3 (August 2019). http://desktop.loc.gov/search (accessed 4 October 2019).

Ezli, Erin and Katherine M. Crowe. “This is the Oppressor’s Language Yet I Need It To Talk to You: Native American Name Authorities at the University of Denver.” In Ethical Questions in Name Authority Control, edited by Jane Sandberg, 9-24. Sacramento: Library Juice Press, 2019.

Adolpho, Kalani. “Who Asked You? Consent, Self-Determination, and the Report of the PCC Ad Hoc Task Group on Gender in Name Authority Records.” In Ethical Questions in Name Authority Control, edited by Jane Sandberg, 9-24. Sacramento: Library Juice Press, 2019.