Serendipity, Algorithms, and Managing Down the Collective Print Collection

by Frank Winter, Librarian Emeritus
University of Saskatchewan

My first contribution to Brain-Work introduced the conceit of the Gone-Away World, and explored the mindset exhibited by Traditionalist librarians who believe that the traditional rationale for developing a local collection and the traditional collection-based services of reference, information literacy and so on are still in existence (Winter, 2014). The core belief of this mindset is the overarching importance of the local print collection. I asked, “[i]f the local collection of printed scholarly books by-and-large just Goes-Away as a useful and important service, what choice will we have as university libraries and university librarians but to adapt? Evidence-based solutions will be increasingly necessary for good decisions in this environment.” In the paragraphs below I explore in a bit more detail the use of serendipity by Traditionalist librarians as one specific example of resistance against any evidence-based reduction of the local collection.

The concept of the collective print collection (where print = books, rather than journals which have long-since migrated almost completely to digital format) has been a very significant development in the options available to research libraries as they consider how they might allocate their resources to meet the existing, evolving, and competing demands of their users within the overarching expectations of what the library must contribute to the mission of the university. Specifically, this concept offers options to “manage down” local print collections so that some of the substantial resources required to maintain them can be reallocated to address other priorities.

Originating with Lorcan Dempsey, the concept has been developed by OCLC as it investigates how and why some library services might now be more efficiently and effectively organized and delivered at supra-institutional or network levels (Dempsey, 2013). As part of its research, OCLC began to more systematically investigate the characteristics of the print holdings of its members with respect to overlap, uniqueness, and geographical distribution. What has emerged is a deeper and more nuanced understanding of how non-local print book collections could be organized and managed including the importance of proximity as it relates to delivery options. OCLC has identified as many as twelve North American mega-regions as well as other, smaller regions. One of the smaller regions has been named Canada Extra-Regional, which includes libraries in the prairie provinces (Demspey and Malpas, 2015). Careful reading of the research reveals many caveats such as the awareness that implementing various actions that might work in larger mega-regions might not work as well in smaller ones such as the Canadian prairies, which have smaller aggregate print collections distributed across relatively larger geographic areas.[1]

Proposals to manage down the local collection – even ones to shift materials to an on-campus repository let alone into a regional print storage facility – typically encounter resistance from local users and librarians based on any number of factors. One factor that is often invoked is that removing materials from the open stacks will reduce opportunities for serendipitous discovery by users. Preserving the opportunity for serendipity is advanced by some librarians and some users as a positive value to be privileged when making decisions with respect to how and where collections are housed.

Although I once heard a very senior and accomplished professor tartly dismiss serendipity as “the tool of the lazy scholar,” we all have our serendipity stories. The most amusing serendipity story I ever heard was told by Dan Cohen, Founding Executive Director of the Digital Public Library of America (DPLA). A colleague went into the stacks and reached up to the top shelf to pull off his desired book. The book beside it dropped on his head and led to a lifelong program of research (Cohen, 2014). Well, who among us would not want to preserve the conditions for such fruitful accidents? In the same blog posting, Cohen described some interesting work at the DPLA to determine whether they could “engineer serendipity” into its user interface although, regrettably, without any capacity for virtual book bonking.

It always has struck me how contingent serendipity is, depending as it does on a chain of events such as a physical book being published, being collected, being on the shelf in a particular spot, having a user actually go into the stacks and coming across the item, as well as the innumerable, unknowable factors taking place inside the mind of the user.[2] The deliberate incorporation into the local collection of hundreds of thousands of e-books let alone hundreds of thousands of other digital objects over the past decade seems to me to only further attenuate serendipity in the stacks as a factor to be given much weight.[3]

Librarians mask the complexity that underlies their operations so that their users can get their hands on the desired object in as friction-free a manner as possible. The invocation of serendipity when making decisions about the disposition of local collections in research libraries conceals and even, in some cases, denies the decades- and even centuries-long application of a long chain of information labour and expertise that brought about these local collections in the first place. Those books did not just happen to be there. The librarian-related actions that got them there such as collection building, describing, organizing, and preserving, have been termed “epistemological engineering” (Guédon, 2001).

It is but a short step from engineering to algorithms if one considers the algorithm to be a codification of the innumerable operations involved in such engineering. Currently there is considerable discussion concerned the algorithms that permeate our lives, from Google’s 200+ “signals,” through Facebook, Amazon, and eHarmony, to the relevance ranking in online catalogs and discovery systems. I have found the work of Tarleton Gillespie especially helpful here with his careful unpacking of the many parts of algorithmic culture (Gillespie, 2014). And Ian Bogost has written recently that, “Concepts like ‘algorithm’ have become sloppy shorthands, slang terms for the act of mistaking multipart complex systems for simple, singular ones. Of treating computation theologically rather than scientifically or culturally.” (Bogost, 2015). For him, it is imperative to recognize all the other inputs (even preceding data and many of them human) that are required in order for an algorithm to produce its answer. One of these multipart complex systems is the library collection and the services that surround it.

I have contrasted serendipity, a somewhat Latinish neologism originating in the 18th Century, with the Arabic-derived algorithm as a way of foregrounding the information labour that is often concealed in the order of books on the shelves and their discovery by users. For me, however, the word fluke, a pithy 19th century, vaguely Anglo-Saxonish neologism, more accurately describes what happens in cases of serendipity in the stacks of the local print book collection. There are so many variables that determine whether a user stumbles across something relevant that they are almost impossible to identify. It is far less impressive defending flukiness than invoking serendipity as an operating principle when deciding how to manage down print collections


[1] It is important to acknowledge the caution with which university librarians have approached any “managing down” of the local physical collection as a result of the existence of digital surrogates. It took at least ten years for most librarians to become comfortable that appropriate institutional arrangements were place before acting to discard some local print runs of e-journals. It will take at least as long and probably much longer for equivalent actions to take place with respect to print books. And although some commentators invoke initiatives such as Google Books, the Internet Archive, and Project Gutenberg as replacements and supplements for print book collections, these resources are clearly not library collections in any sense recognized by librarians and not acceptable as long term solutions.

[2] Frederick Kilgour’s study examining the sources of user failure incurred in obtaining the desired item in the stacks included factors such as a book never being acquired, being in on order, awaiting cataloguing and processing, circulation, awaiting reshelving, or misshelved, as well as user error (Kilgour, 1989). One could add any number of other contingencies such as being allowed into the stacks in the first place, books that disappeared into in-library graduate study carrels and faculty offices, and so on. Kilgour’s interest was estimating the impact that an automated library information system would have on these factors. It would be interesting to update Kilgour’s work and the literature of user failure generally to determine what effect the incorporation of e-books into the collection would have on those factors.

[3] See, for example, a recent article by David Woolwine (2014). Woolwine writes from the perspective of supporting undergraduate learning in a small to medium size academic library and includes references to studies involving serendipity. I wonder about the relevance of these earlier studies in an age of digital abundance, scholarly e-books, and changing user practices.


Bogost, Ian. 2015. Cathedral of computation. The Atlantic. January 15, 2015.

Cohen, Dan. 2014. Planning for Serendipity.

Dempsey, Lorcan. 2013. The Emergence of the Collective Collection: Analyzing Aggregate Print Library Holdings. In: Dempsey, Lorcan, Brian Lavoie, Constance Malpas, Lynn Silipigni Connaway, Roger C. Schonfeld, JD Shipengrover, and Günter Waibel. 2013. Understanding the Collective Collection: Towards a System-wide Perspective on Library Print Collections. Dublin, Ohio: OCLC Research.

Dempsey, Lorcan and Malpas, Constance. 2015. Evolving Collection Directions. Collection Development Strategies in an Evolving Marketplace: An ALCTS Symposium. Chicago, 30 January 2015.

Gillespie, Tarleton. 2014. The relevance of algorithms, in Media Technologies: Essays on Communication, Materiality, and Society, ed. Tarleton Gillespie, Pablo Boczkowski and Kirsten Foot. Cambridge, MA: MIT Press.

Guédon, Jean-Claude. 2001. In Oldenburg’s Long Shadow: Librarians, Research Scientists, Publishers, and the Control of Scientific Publishing. Washington, D.C., Association of Research Libraries.

Kilgour, Frederick G. 1989. Toward 100 percent availability. Library Journal, 11450-53.

Winter, Frank. 2014. Traditionalists, Progressives and the Gone-Away World. Brain-Work.

Woolwine, David E. 2014. Collection development in the humanities and social sciences in a transitional age: Deaccession of print items. Library Philosophy and Practice.

This article gives the views of the author(s) and not necessarily the views of the Centre for Evidence Based Library and Information Practice or the University Library, University of Saskatchewan.

1 thought on “Serendipity, Algorithms, and Managing Down the Collective Print Collection

  1. Thank you Frank for this thought-provoking blog. I had forgotten the intellectual rigor with which you approach the many challenging professional practice issues that we as librarians are giving thought to these days. A good dose of ‘Winter-speak’ was always a highlight of my week and I appreciate the opportunity to re-engage and get an up-to-date dose of ‘Winter-speak’, especially this week. I am sure many of my colleagues share my view that your retirement from active service has left a void in our collegial ranks.

Leave a Reply

Your email address will not be published. Required fields are marked *