Interacting with Work-Sets


Technicalities 42(1) Jan/Feb 2022: 12-15

Welcome to the third in a series of columns about the work, or “the work set” as I called it, with specific reference to FRBRand RDA. I previously claimed that there were four tasks when we consider a group of texts (books, films, etc.) as a work:

·     What are the members of the set (i.e., what is included in the set and what is excluded)?
·     How do we describe the set as a collectivity?
·     How do we organize the members of the set?
·     How do we connect the set externally to other works or entities?

The last two columns discussed those first two issues, and bottom line, I think FRBR gets it wrong, and that RDA leads to a catalog that has a lot more in common with Cutter and the “printed dictionary catalog” than the active reimagination of the catalog made possible by really embracing the work-basis for its design. And all of this is political because I do not yet see evidence that we can build firm rules for the identification and representation of works. I think, instead, identifying the work as the proper basis of the catalog brings the historical and cultural bases of our bibliographic projects into greater focus, where design methods are more likely to succeed than scientific processes, and that our designs should be critically informed as well as technically informed. That is a lot of Latin, or at least Latinate words, so let me provide a concrete example: how do we organize the set of things called “The Wizard of Oz”, including which specific items to include and how to describe and arrange them?" I almost used that mixed-up word “metadata”, one of those words like “television” with both Greek and Latin roots. Really, we have got to do better.

Thinking Beyond Antiquated Technology

The description and arrangement of a set of publications as a single conglomerate unit should depend as much on who will use those descriptions as it does on the technologies and standards we use for our work. For over a century we have proceeded on the idea—false in my opinion—that there is basically a single way to describe these works. In a period where the technology and the standards were limited, such as a single card catalog along with a predominant descriptive standard, we did not have a lot of choice. The emphasis was in developing terse (to fit on a card) standardized representations that could be entered in a computer-based system to be shared across networks. Those descriptions were also positioned within a limited number of subject representation systems such as LCSH, the DDC and LCC. We were limited by a lack of choice and a lack of technological alternatives.

But today we are limited mostly by a lack of imagination regarding our bibliographic projects. There was a lot of stress starting twenty years or so ago about the lack of ranked retrieval in the library catalogs; Google had it, our catalogs did not, and libraries were going to be obsolete. Looking back, it was not that we could not do ranked retrieval, it is that we never made it a priority. I always enjoyed Ray Larson’s work on ranked retrieval in catalogs[1]and in retrospect, while our catalogs remain relatively unsophisticated technologically, Google has not proven to be our equal in its commitments to privacy, its opposition to commercialization and privatization of knowledge, or to equity of access. Most disastrously, Google has translated a series of naive assumptions regarding knowledge that has reinforced forms of bias into the heart of information access mechanisms. Crudely, Google has promoted knowledge deemed popular by adolescent males and major corporations at the expense of diversity and the promulgation of authoritative sources.

A more honest characterization about the technology of our catalogs would admit that we were less concerned about ranking and retrieval, and more concerned about interactivity, under the concept of “Web 2.0”, where

users participate and add content and value. The first decade of this millennium will probably be known for the growth of sophisticated electronic social activity. Users have become accustomed to creating content on the Web, whether it is writing a review of a book at an online bookstore or creating an identity for themselves on MySpace. They are also accustomed to having their say by posting comments to blogs or adding their ideas on a topic to Wikipedia.
[2]

What Do We Mean by Interactivity?

Too often, by interactivity we mean things like user reviews of books and articles in our catalog. I think we have seen enough of those on Amazon to not be too encouraged about the overall quality of commentary regarding books. I try to avoid the reductio ad hitlerum but Mein Kampf enjoys a 4.4 score based on 2,127 ratings on Amazon (Nov. 15, 2021), and most commenters strive to let you know they are not Nazis but rather as “amy” said on March 27, 2018, “Bought this for my son. He’s obsessed with all things WW2 related. I haven’t read it, but he has. He enjoys reading other view points.” Thank you, amy. And really I don’t think I am being unfair in my characterization of the reviews—almost all the other comments are about as helpful.

Lankes and others had a broader view of interaction on the catalog to include how the catalog integrated with other tools and platforms. Thus, for example, they describe downloading data into bibliographic citation software, or into your own note-taking system. They also include using a catalog record to look up a title in Amazon.[3]Ultimately Lankes and others are interested in participation and community, but we still need to consider what kinds of interactions we would like to support. As amy proves, not every form of interaction builds community, much less knowledge. Linking published book reviews to the record of the reviewed book in the catalog? Yes. More linking to eBay and Amazon? That would be a hard pass. I think not only do we want to improve interactivity but also social discourse and perhaps also the quality of our services. That work would also improve our own credibility as sources of reliable information and authentic community.

A more substantive form of interaction would be to weigh how our bibliographic representations could be formed to meet the needs of different communities. I am all for finding ways to facilitate conversations around books and ideas, and the catalog can be a place for that, carefully and under the right conditions. But what about something that is a little less of an add-on to catalogs and a little more at the heart of our service? That could include describing a book differently for different communities. Certainly we have some notion of that when we use different kinds of subject access systems to represent a book, but a better idea would be the development of more specialized forms of access rather than separate positioning within broad general systems like the DDC and the LCC. Perhaps a better example would be the description of a book for teenagers compared to a description of the same book for an adult audience. I am looking at you again, Oz, or perhaps the Alice books. It feels like different audiences are getting different things from those books. Maybe we could change our descriptions in ways to meet those needs.

Interacting with Work-Sets

But it is the arrangement of works, variously defined, that I would like to consider. I already argued that different communities might consider the work-set for a given work either expansively or narrowly, for example, if scholars might make increasingly refined distinctions amongst textual variants (known as “splitters”) or another group who might group all the versions of A Christmas Carol as a single thing (“lumpers”). Can we think of a way that different kinds users can access different representations in the same catalog, through an interactive design? For example, we could provide a simple overview of a work (or set of works, depending on how the set is bounded) that could then be explored in greater detail. “Here is a set of things called A Christmas Carol, it includes these things…,” providing further refinements and descriptions.

One reason we have never really considered interactive ways of ordering the individual components in a conglomerate work-set is the technology to do it in the catalog is so awful. I complained about Svenonius’ concept of “work languages”[4]last time though the fault is not hers. She is just trying to justify a practice that involves a concatenation of main entry, standard title, and transcribed title proper, such as “Melville, Herman. 1802-1880. Billy Budd, sailor. Billy Budd, foretopman.” The difficulty is not only that a single sequence, arranged alphabetically by standard title, subarranged by year of publication and an alphabetical arrangement by title proper, followed by translations arranged alphabetically by language of translation, subarranged by year of translation… Whew! Experts cannot make sense of that even when they know to look for it, much less users who do not. A single linear arrangement? Wikipedia says the Bible has been translated into 3,415 languages, comprises dozens of books in various arrangments, divided into major sections such as the Old Testament, in various chronological arrangements and origins, along with associated commentaries and derivative products (looking at you, Ben Hur!), and the thought is that we can place those into a single linear sequence?

The consideration of a single linear data structure as the solution to this problem is odd. But even worse, that linear structure was to be achieved through the use of a concatenation of various data elements, derived by instruction from multiple standards each numbering hundreds of pages long, by different experts with mixed levels of training, using separately constituted bodies of evidence… Whew, again!

So let us begin again, anew, with a couple of premises. First, bibliographic objects and data are complex and variable. Billy Budd is relatively simple, and we are not even sure of the title, and we have no good way to resolve problems like that with certainty. Second, users (and communities of users) are also complex, and the various uses and interpretations to which they put bibliographic objects are also complex and varied.

For awhile we considered FRBR or some approximate as the solution to these problems. It looked like we might start using more sophisticated data structures allowed by computing to model bibliographic objects. But RDA is a reiteration of old technologies with some new updated thinking about works. That reiteration to organize works using headings to structure data is a disappointment, and it is clear that the technology has failed.

Better Living, But Not Through Technology

Better technological design will no doubt improve our solutions, and our services. I have positioned FRBR here as deserving particularly scrutiny. Around the same time as it was published I was working with others to think about alternative designs for the control of work-sets, particularly an approach that intended to describe how a work-set evolved over time.[5] We could, in retrospect, call that the geneticapproach, in the sense of the successive generation of texts arising from a common origin with some overly vague notion of causal antecedence. For a period of time, I felt the solution would be to model a work-set by expressing the data in a semantic structure like RDF. To my eyes it would still be an interesting experiment.

But the real problem is not the selection of theappropriate solution, it is the thought that there is a single technical solution to these sets of problems. We have allowed the technology—the standards, our archive of bibliographic data, the systems we use to access those data—to shape and condition the way we think about problems like works and work-sets. Despite the concerted efforts of Svenonious and the thinking behind FRBR, I do not think we will be able to develop a set of detailed instructions that will unambiguously determine the members of any given work and how they should be arranged, along with the confidence that such characterizations are valid for all users in all settings.

Instead of asking “what is a work” maybe we should ask “What is the Wizard of Oz?”, “What is Billy Budd?” and “What is the Bible?” And we should expect that your answer may be different from my answer. The trick for librarians—one reason why we are professionals—is the thought that we can do that work on behalf of others, so the question for us evolves to be “What is the Wizard of Oz from the perspective of young readers?”

Answering these questions, repeatedly, for different work-sets and user communities might, over time, provide us with the answers to the problems that we really face. Those answers will eventually be to provide concrete guidance involving specific technologies. But those problems are not foremost technological, they are firstly deciding on the basic service we want to provide. The answers will be responses to questions like “How do users explore and understand a work-set?” “How do users understand and select individual documents for reading from the work-set?” We should not expect a single solution to these questions, as the answers will vary from work-set to work-set and from user community to user community. Our answers to these questions may provide us with a set of common solutions combining different technologies. Our work may turn from uniform technological solutions to bibliographic problems toward thinking about texts and communities using a broader set of interpretive and creative design methods.

Works Cited

[1]Larson, Ray. Classification Clustering, Probabilistic Information Retrieval, and the Online Catalog. Library Quarterly 61, no. 2 (1991): 133-173, doi:10.1086/602331.

[2]Coyle, Karen. "The Library Catalog in a 2.0 World." The Journal of Academic Librarianship 33, no. 2 (2007): 289-91, doi.org:10.1016/j.acalib.2007.02.003.

[3]Lankes, R David, Joanne Silverstein, and Scott Nicholson. "Participatory Networks: The Library as Conversation." Information Technology and Libraries26, no. 4 (2007): 17-33, doi.org: 10.6017/ital.v26i4.3267.

[4]Svenonius, Elaine. The Intellectual Foundation of Information Organization. Cambridge, MA: MIT Press, 2000, p. 87ff., doi: 10.7551/mitpress/3828.001.0001.

[5]Smiraglia, Richard P., and Gregory H. Leazer. "Derivative Bibliographic Relationships: The Work Relationship in a Global Bibliographic Database." Journal of the American Society for Information Science 50, no. 6 (1999): 493-504, doi: 10.1002/(SICI)1097-4571(1999)50:6<493::AID-ASI4>3.0.CO;2-U.