Mt. Rushmore, Theodore Roosevelt closeup.
Image via Wikipedia

It’s all well and good to say that libraries should acquire books based on usefulness rather than focusing on quality. But in practice, there are several real problems with such an approach.

First of all, librarians are in a bad position to judge usefulness. A book may be useful to a patron for any number of reasons that can’t be anticipated by someone else, even if that someone else has a library degree. Any book could turn out to be useful, depending on the task at hand. The traditional library is made up almost entirely of predictions made by librarians; the books in the collection represent librarians’ collective attempts to guess what their patrons will need in the future. But since actual usefulness is completely unpredictable, we’ve been forced historically to treat quality as a proxy for usefulness. “We’ll buy this good book rather than that bad book,” we’ve implicitly said, “because a good book is going to serve our patrons’ needs better than a bad one”—this despite the obvious fact that a high-quality book on a topic irrelevant to the patron’s needs may well be of far less utility than a relevant and mediocre one, or even (as I argued in the first part of this piece) a very bad one that is bad in useful and instructive ways.

Second, although quality is a relatively stable property, usefulness is not. A very good biography of Theodore Roosevelt may be superseded by a later one, but it won’t become a worse book than it was just because the new one is better. The usefulness of a book, however, will not only vary greatly from user to user, but is  also likely to vary over time for an individual user. A software manual, for example, will be very useful to some patrons and completely useless to others — and when a new version of the software is released, the old manual may become effectively useless to the ones for whom it was very useful before. Similarly, the Roosevelt biography may be an essential resource for a graduate student today, but become much less useful to her next year after her thesis is accepted. And if it’s superseded by a later biography, it will become less useful to some patrons than it was beforehand, despite its continued high quality.

Third is the problem of rationing access. Historically, research librarians have rationed access by deciding how much of what kind of material should be in the collection, and we discriminated according to reasonably consistent criteria: relevance to the curriculum, price, quality, etc. As a result, our libraries now boast excellent and well-crafted collections — large chunks of which are never used, partly because despite their quality as collections, they don’t contain the very specific materials particular students and researchers need in order to do their work. This is the inevitable result of trying to guess what other people will want. But suppose, for the sake of argument, that librarians actually were able to figure out a way to discriminate based on utility rather than quality. How then would we allocate our acquisition budgets? Do the needs of a junior faculty member take precedence over the needs of an undergraduate student? How about the needs of a tenured faculty member versus those of an untenured one? Or of a Nobel laureate as compared with those of a post-doc? Libraries have never provided everything that everyone needed—to do so would be impossible, even with a generous budget.

Possible solutions to the first two problems have begun emerging in recent years. First, given the impossibility of predicting the usefulness of particular books to library patrons, valuing usefulness over quality will tend to lead us to let patrons select the books we purchase rather than to continue selecting them on behalf of patrons. This obviously implies a program of patron-driven access. While I don’t advocate the wholesale abandonment of speculative and librarian-driven acquisition for all libraries, I do think that this reality—the impossibility of guessing what will and won’t be useful—suggests the importance for most research libraries of exploring patron-driven access models, and even doing so quite aggressively. Given the option of knowing for certain that a book will be used, why choose instead to buy one that runs a strong likelihood of not being used? Patron-driven models are enormously disruptive to the traditional library, of course. But as librarians, our responsibility is to serve our patrons well, not to preserve the structures in which we’ve become comfortable.

Second, as for the unevenness (across patrons) and instability (over time) of usefulness: the implications of this reality are fairly clear, but alarming. They suggest an argument against building permanent collections at all. If usefulness is of central importance, then permanence of access should take a back seat to fluidity—the point becomes not so much to build a defined and carefully-crafted collection as to create a space in which patrons can gain access to a much larger and more loosely-crafted “collection,” the contents of which might be to a great degree shifting and dynamic over time. Permanent access would be established (if at all) only for those materials that get used. Such an approach would mark another truly radical and fundamental change in the traditional understanding of what constitutes a research library; it would signal a shift from the model of the library as collection to one of the library as conduit—again, an extremely disruptive and even threatening structural shift for librarians.

Rationing presents perhaps the stickiest problem of the three. In the past, libraries rationed in a highly impersonal manner: we subscribed to Journal X but not Journal Y, and bought Monograph A but not Monograph B, and for the most part no one took such choices personally. Everyone (well, most everyone) understood that the library’s resources were limited and that hard choices had to be made for the good of the collection and its users. But switching the focus from quality to usefulness implies opening up many more options to patrons without any promise of new money. How can scarce resources be allocated rationally within such a structure? Will patrons themselves be evaluated and found more or less deserving of support depending on their status and the relevance of their work to institutional priorities? If not, then what will be the means of discrimination, given that discrimination remains essential wherever needs outstrip resources (which is to say everywhere)? A new system of resource rationing might be less disruptive to librarians, but much more disruptive to library patrons.

I confess that I have no real answer to that last question. For one thing, the complexity of a more rational allocation mechanism—one that accounts not only for the differing needs of different classes of user but also for the priorities of the institution served by the library — would be mind-boggling. For another, I can think of no strategic and need-based allocation mechanism that wouldn’t ignite a firestorm of controversy among library patrons, many of whom would be outraged at the apparent unfairness of an allocation structure that explicitly privileges some disciplines and some individuals over others. Such a structure would bring to the surface a reality that has always existed, but which we generally prefer not to discuss in the context of library service: the fact that in every institution of higher education, some disciplines matter more than others, and that different classes of student and employee can expect different levels of collections support. I honestly have no idea how to resolve this particular problem, and I solicit the wisdom of the commenting community.

To sum up: it’s human nature to emphasize the importance of what you can do well (such as judging quality) and minimize the importance of what you do poorly (judging usefulness). In the days when building speculative collections was the only way to meet patrons’ needs, librarians could get away with that approach better. As budgets tighten, prices rise, and information resources keep moving into a digital environment in which patron-driven access options naturally proliferate, it’s quickly becoming harder for research libraries to justify the speculative building of just-in-case collections. This, as much as anything else, is what I believe is currently driving libraries away from time-honored models of traditional librarianship, and the repercussions of that shift are going to be as strong as they are unpredictable.

Enhanced by Zemanta
Rick Anderson

Rick Anderson

Rick Anderson is University Librarian at Brigham Young University. He has worked previously as a bibliographer for YBP, Inc., as Head Acquisitions Librarian for the University of North Carolina, Greensboro, as Director of Resource Acquisition at the University of Nevada, Reno, and as Associate Dean for Collections & Scholarly Communication at the University of Utah.


24 Thoughts on "No Such Thing As a Bad Book? Part 2 — Implications and Problems"

Interestingly, you wind up providing a seemingly powerful argument against PDA. Given the necessity of rationing, PDA requires discrimination, which is unacceptable, hence PDA is unacceptable.

But what you are calling discrimination is really just allocation, which is the problem that budget systems are designed to solve. That is, under PDA there needs to be specific acquisition budget allocations for the different classes of patrons. These allocations probably have to be made by the University, not the Library. Does this help?

BTW, back when I was a faculty member (Carnegie Mellon) I was under the impression that the faculty picked the books and journals. Silly me.

> Interestingly, you wind up providing a seemingly powerful
> argument against PDA.

You say “powerful argument against,” I say “problem with.” Potato, potahto. It seems to me that the price of intellectual honesty is the ability to acknowledge and discuss the downsides of one’s position on an issue. (It’s also the price of coming up with an actual working solution to any problem.)

> Given the necessity of rationing,
> PDA requires discrimination, which is unacceptable, hence
> PDA is unacceptable.

No, David, that syllogism doesn’t work. PDA isn’t about eliminating discrimination; it’s about moving the locus of discrimination. It means having patrons discriminate based on their manifest needs instead of having librarians discriminate based on their speculation about patrons’ needs.

Nor is discrimination “just allocation.” Allocation decisions follow from discrimination, but they are not the same thing. You can allocate without discrimination by, for example, applying a budget cut (or increase) evenly across all disciplines rather than imposing the change unevenly according to mission-based criteria.

Nor are “budget systems… designed to solve” allocation problems — they are designed to carry out allocation decisions, which have to be made prior to to the application of a budgeting system. You make the allocation decision before you feed money into the budget.

Nor are allocation decisions at this level typically made by libraries’ sponsoring institutions. The university does indeed allocate budget funds to the library, but those allocations are usually broadly defined, and further allocation decisions always have to be made by the library. The university doesn’t typically tell the library how to distribute its materials budget, to use the example most relevant to this discussion.

Finally, your understanding of how library budget allocation worked when you were at Carnegie Mellon may have been mistaken, and it may not have been. At most research libraries, selection decisions have traditionally been made by librarians. However, at a small number of college and research libraries, faculty members do all or most of the selection for the library collections themselves. That may have been the case at Carnegie Mellon when you were there (and may still be now, for that matter).

Rich, you seem to be objecting more to my language than to my point. I understand you to be looking for “…a more rational allocation mechanism—one that accounts not only for the differing needs of different classes of user but also for the priorities of the institution served by the library…” in order to implement PDA.

I am proposing that the university allocate something like “PDA drawing rights” to the various user classes. My basic point is that while the library cannot unilaterally make such an allocation, for the political “firestorm” reason you give, but the university can. They do that sort of thing all the time, in other contexts, such as building space. The library loses a degree of control this way, but that seems to be implicit in PDA.

OK, I see what you’re saying in regard to university allocations. Yes, it would definitely give the library political cover if the university were to create a resource fund for PDA purchases, allocated accross disciplines according to the university’s priorities. Not only would that give the library cover in regard to the (inevitably) uneven distribution of those resources, but it would also prevent the exhausting of those funds being seen as a library failure.

My initial response, of course, is to say “Such a program would be especially welcome if it involved _additional_ resources rather than a redirection of money that would normally come to the library budget.” But if I’m honest with myself, I have to admit that even a redirection of existing budget flows would represent a step in the right direction according to the philosophy I’ve outlined above.

Your second solution to the problem (the fluid library) hints at a deeper solution. Perhaps you mentioned it in Part I or in a previous post. Clearly, electronic production and delivery, combined with some different pricing model, cuts through the Gordian knot of judiciousness, by enabling libraries to acquire “everything”. I would think this would allow librarians to focus more on providing better and better subject indexing.

Of course, all the problems you cite would re-emerge in another form, but with less impact on the actual dissemination of information: do I buy by the megabyte? there will of course be different sizes and grades of “everything”. Shall I buy 1-license to the Journal of Physical Anthropology or 2? Or should I rent such and such a History book only for the summer of the topics sesquicentennial? And so forth.

Now, can libraries band together and accept no other medium than electronic? Can publishers design the new pricing models? Of course! How fast can it all happen? I will not spare you the obvious truism: too fast for some, not fast enough for others.

Nick, I think I have to disagree that “electronic production and delivery, combined with some different pricing model… (enables) libraries to acquire ‘everything’,” even if we put “everything” in quotes (which I assume you’re doing to indicate that you mean “more or less everything,” which is the way I’ll use the quotes as well).

What electronic production and delivery do quite well is enable libraries to _show_ our patrons “everything” — to help make “everything” more discoverable and then immediately usable. But it’s hard to imagine a pricing model that would make the preemptive _acquisition_ of “everything” possible. And even if it were possible, I struggle with the idea that we should deliberately buy the wrong content, which is what most of “everything” includes. In a perfect world we would never buy a document (or access to a document) that isn’t needed. In the real world, of course, things are imperfect, and pricing models sometimes require us to buy batches of content despite the fact that much of the content isn’t wanted, because that’s the only way we can afford access to the content that is wanted. I’d like to see things change in such a way that acquisition can be made less wasteful than that.

Thank you for your thoughtful reply.

I agree, it is hard to imagine the pricing model for “everything”. I have fleeting thoughts of buying content by the megabyte, or of buying tranches of “everything” as Satellite radio offers, but I really don’t know how it would work. Or the ‘market’ will deliver it, with its usual benevolence. And I too share the same dislike for buying a book that isn’t needed.

But as I understand it, the option of PDA arises in part because acquisition librarians also cannot choose or, what was the number, 40% of their choices are wrong. And that’s after a prior winnowing by publishers who choose what to publish: how often are they “wrong”?

Bear in mind, I am making this up as I go along. What is being destroyed is fairly clear: paper and shelves. But what will replace them is up for grabs. One thing that seems fairly real on the horizon, however, is this: In a world where the energy and money of triaging which books to acquire could be invested in the analytic indexing of their information, the role of librarians and publishers would shift and merge, and a premier library would be characterized less by the extent and “quality” of its collection than by the intelligence and vitality of its index, and its integration with the information it represents. Or maybe Google’s Page ranking really is enough?

Nick, I think you’re right that the centrality of paper and shelves is being destroyed, but I also think that this development is only a symptom of a deeper one, which is that the concept of the collection itself is being obviated. Our information environment is now one in which you don’t necessarily have to purchase access ahead of time (and therefore guess what will be wanted) in order to ensure that your patrons will have access; this represents a challenge to our most fundamental practices in libraries. I also think you’re right to point out that this newly emerging information context puts a very great premium on some kind of indexing — though what that will end up looking like, I’m not sure. Many of my colleagues pooh-pooh the effectiveness of full-text searching, and they’re not always wrong. On the other hand, I think we librarians have a tendency to underplay the power of full-text searching, in part because we just don’t like it — it allows our patron to bypass all of the mediating layers that we’ve learned so painstakingly to apply.

How interesting. I argue that indexing is pointless since there are dozens of ways to index the same body of information, all equally valid. This result flows from my general theory of the structure of information content. See

This is related to what are called faceted taxonomies, which are usually just alternative taxonomies. If there are just 10 different was to index a body of information then any one is highly unlikely to be the right one for a given user. This is why my work centers on full text search. The ultimate goal is for the computer to read the text and answer our questions.

I think it’s a fallacy to infer that if a book does not circulate in any given library, or even in all libraries, somehow the decision by the publisher to bring it out was “wrong.” Remember that university presses are making their decisions based not just on editors’ judgments but the judgments collectively of external expert reviewers and members of the faculty editorial board that has the final say about what gets published. This board can sometimes include as many as twenty or more faculty members from diverse fields. The combined wisdom that goes into each decision is, of course, not infallible, and hindsight will lead publishers themselves to think they have made a mistake in publishing a certain book occasionally. But these decisions do reflect a considered judgment of senior people in a field that the book will contribute to the advancement of knowledge in that field in a significant way. Any given book may have more or less “use” in a library (do usage statistics cover in-library use?), but it may turn out to be influential in a field regardless of how much library use it has. Think of a book that it so crucial to a field that everyone buys it instead of relying on the library. Its minimal library use would hardly support the conclusion that a “wrong” publishing decision had been made.

> I think it’s a fallacy to infer that if a book does not
> circulate in any given library, or even in all libraries,
> somehow the decision by the publisher to bring it
> out was “wrong.”

Sandy, if you can give me an example of a book that has gotten broad usage and readership everywhere except in libraries, then I’ll happily concede that in that case, it was not the wrong book to publish. However, if that book never gets used in a library, then it was certainly the wrong book for that library to buy, regardless of its objective quality. (Again, I’m not going to take up SK space to rehash an argument we’ve thoroughly explored elsewhere; anyone interested can refer to our exchange in Against the Grain earlier this year.)

How about Harry Frankfurt’s “On Bullshit,” which was so small in format as not to be capable of having a library slip placed in it and so cheap (under $10) as to encourage purchase rather than borrowing? It has sold over 350,000 to individuals, by the way. I wonder how many academic libraries ordered it (in contrast with, say, public libraries)?

For what it’s worth, based on a quick look at WorldCat, I can tell you that five of the six academic libraries in Utah own On Bullshit, as does the state library. I would imagine the pattern is similar in other states; you can check for yourself if you want at

Your discussion raises some interesting questions about how PDA might interact with the new ebook aggregations that Project Muse, JSTOR, and others are beginning to offer. These broader collections respond to your second solution but, to a degree, enable some discrimination to the extent that they offer the option of selecting subsets of the whole collection, topically or by area (humanities, social sciences, etc.).

Usefulness is often so specific that one wonders if a strategy based solely upon that criterion would serve the interests of the university over the long term well. Think, for instance, of how very specialized dissertations often are. The authors of those dissertations may need very specific items relating to their very specific projects, which may have no use to anyone in the university ever again.

And what about self-help books, novels, etc.? How much is the utility of a self-help book supposed to weigh in a decision to acquire it under PDA, or the entertainment value of a popular novel (assuming it is not assigned reading in any course)? Should academic libraries be excluding such works from their PDA programs by fiat, allowing local public librarians to serve those needs?

I have another concern about the changing nature of usefulness. Jonathan Band has controversially argued for a broad extension of fair use by claiming that a book over time is used for different purposes–say, an older biography of TR may be superseded by a newer one based on recently discovered archival sources and the earlier one therefore becomes of only “historiographical” interest and use thereafter. Band claims that this “repurposing” of the original work is grounds for claiming its later use to be “transformative” in such a way as to allow for fair use to apply–e.g., by having it digitized for a collection like HathiTrust. I think this is a very dangerous argument to make, though it has some legal basis in decisions in copyright cases made in the Ninth Circuit.

> Usefulness is often so specific that one wonders if a strategy
> based solely upon that criterion would serve the interests of
> the university over the long term well. Think, for instance, of
> how very specialized dissertations often are. The authors of
> those dissertations may need very specific items relating to
> their very specific projects, which may have no use to anyone
> in the university ever again.

Right — but let’s remember what the alternative to patron-driven selection is: it’s librarian-driven selection, which also runs the very high risk of resulting in the purchase of books that will be rarely used (if ever) and are unlikely to serve the long-term interests of the university. Some PDA models only incur a small charge on the first use; such models are ideal in the kind of situation you describe. Of course, no model (including traditional ones) is ideally suited to every patron need.

> And what about self-help books, novels, etc.? How much is
> the utility of a self-help book supposed to weigh in a decision
> to acquire it under PDA, or the entertainment value of a
> popular novel (assuming it is not assigned reading in any
> course)? Should academic libraries be excluding such works
> from their PDA programs by fiat, allowing local public librarians
> to serve those needs?

It would make obvious sense to do so, except for the fact that people write dissertations on pop-culture topics all the time. That’s part of what makes collection development in research libraries such a complicated thing. Here’s a more practical answer, though: If I had infinite resources, I would exclude nothing from the universe of titles available for PDA. Since I have very finite resources, I do have to make a “rough cut” to keep the pool of possible purchases manageable. I might exclude “For Dummies” books and travel guides from that pool, for example — while leaving the door open to specific requests from patrons who need such titles for their academic work.

> I have another concern about the changing nature of
> usefulness. Jonathan Band has controversially argued for
> a broad extension of fair use

I could advance my own opinion of Jonathan Band’s arguments in favor of broader interpretations of fair use, but my opinion has no weight. His arguments will have a legal impact or they won’t, presumably on their merits. Jonathan Band has done work for ARL, but he is not the mouthpiece of the library community.

“First of all, librarians are in a bad position to judge usefulness.”

When I first read this, I had to check the byline again… This sounds so “Kent”. lol On the contrary, I think that librarians are the ONLY ones in a position to judge usefulness. As you say, the usefulness of an item will change. Librarians are the only ones in a position to take all points of view into consideration, present and future. We may not always do that, but we certainly have the ability to do that with more consistency than faculty, administration, students or other users.

Matthew, what you’re describing is not librarians judging usefulness, but librarians judging quality. Although I think you’re giving librarians too much credit (and everyone else in the world too little) when you say that we’re “the only ones in a position to take all points of view into consideration,” you’re right that librarians are very good at judging quality — for all the reasons you gave. The problem is that quality isn’t always a good predictor of a book’s usefulness for a particular research task — for all the reasons I set out in my post.

But wouldn’t a collection development specialist librarian be in a position to know more about the university’s current and future curriculum in the areas the librarian covered than, certainly, students and maybe even some faculty (other than department heads)? Isn’t usefulness a criterion that can be and should be applied at this level at least as much as to any one individual’s very particular needs, which may not reflect any long-term needs of anyone else? You can order a lot of books that relate to only one dissertation project, but how useful will these books be to anyone else later on?

Sandy, I think I answered these questions pretty thoroughly in my two “No Such Thing?” postings, and also in my previous SK posting on the topic of PDA. I’d be happy to discuss further with you offline, but I don’t want to tax the patience of SK’s readers and editors by rehashing the same points too much in this space.

No, I was definitely thinking and talking about usefulness. And although I didn’t really give more than one reason in my comment, it certainly doesn’t apply to our ability to judge quality. Quality doesn’t have to do with perspectives of the potential users but rather the work’s authority, methodology, and positioning in the field.

I’m not sure I understand your problem with librarians being well-positioned to determine usefulness. Most of us are NOT subject experts and therefore can be more objective about a work’s usefulness in general than any given subject expert. We’re also the only ones trained (formally and on-the-job) to deal with and facilitate the use of books, journals, indexes and other traditional (and some not-so-traditional) information. Have I missed the presence of other information specialists that are more common and have as general a professional scope as librarians?

No, not at all — what I think you’re missing are my primary points about usefulness, which are 1) that quality and usefulness are only imperfectly correlated, and 2) that no person can predict the usefulness of any book for any other person. Professional scope isn’t the solution to that problem — it is the problem. Professional scope helps you build a good collection because it gives you a broad perspective on the literature, but it doesn’t help you anticipate what any of the individual researchers you’re serving will need. An extensive collection of Harlequin Romances may be of zero usefulness to most serious literary researchers, and absolutely essential to one researcher. A very high-quality book on Chaucer may be of great use to 200 researchers, and of zero usefulness to 200 others. The problem is that our patrons are not a mass of people who need access to a mass of very good books; they are individuals who need access to particular books in order to perform particular tasks. We librarians are very good at figuring out whether a particular book fits a particular category but very bad at knowing ahead of time which specific book will be needed by any particular patron, and that’s why so many of the things we buy never get used, while at the same time our interlibrary loan and document delivery departments get constant use — it’s not usually because we failed to buy good books, but because we bought good books that weren’t needed and failed to buy books that were. That’s not a criticism of our skill as librarians; it’s a reflection of the impossibility of guessing ahead of time what people will need in order to do their work.

This way of putting the point, Rick, suggests that the best approach might be to rent books that some individuals need for their very specialized purposes but will never be used again by anyone else. Is a book used by only one person once really a good use of library funds? If 60% of library books are checked out, somebody must be guessing right a lot of the time about what will probably be used by more than one person at one point in time. And surely teachers in charge of curricula can construct lists of books that will relate well to courses when offered as supplemental reading, no?

Sandy, the “rental for one person” approach is indeed one of the PDA models that libraries like ours are current exploring. As for whether it represents a good use of library funds to purchase a book that is only used by one person, I think the answer is that it’s clearly a better use of library funds than purchasing a book that is used by zero people, which is what happens a good 40% of the time in a typical research library. Yes, 60% of the time we do guess right — and that was a more or less acceptable success rate when specuative collecting was our only option. But it’s not a very good success rate, and speculation is not our only option anymore.

Can teachers construct lists of relevant books? Of course, and so can librarians. Teachers can go one better, though: they can construct lists of required books. If we know that a book will be used (because we know a teacher will require its use), then we should certainly buy that book.

Comments are closed.