Reference Annealing: let’s stop reinventing the answer

RQ [Reference Quarterly], Summer 1995 v34 n4 p459(5) “Reference Annealing: let’s stop reinventing the answer” by Nancy Humphreys

Abstract: Reference librarians should cooperate to avoid duplication of effort and create a networked online reference catalog. Hypertext links could make a virtual reference library especially valuable for reference work.

“Public Library Services Continue to Deteriorate” was the lead headline in the California State Library Newsletter as 1994 began. While California may be affected by some unique state politics, other areas of the United States and other types of libraries besides public libraries are experiencing a similar crunch. Across the country, schools of library and information science are being dismantled, and administrators are considering the “benefits” of doing away with the reference desk. This is a proposal for dealing with this crisis.

Librarians believe that providing equal services to patrons will result in making society more equal, but in a society that is unequal to start with, this doesn’t happen. Many more people need library services than are now being served, and these people need information desperately. As a result, a vital issue, especially in times of declining budgets, is how to get more results for each dollar spent. Rather than watch as reference services are cut, librarians need to see how reference service can be improved. In particular, we need a painless way to get more results for the time we spend at the reference desk. To do this, we need to reduce the redundancy of the work we do.

Every day, many reference librarians across the country are asked to answer the same questions. Many sources can be used to answer the same question. However, the source chosen by a librarian may not be the best one or the most complete one. Sometimes different sources will each give a partial answer to the same question, partial answers that could be combined1 to produce the complete answer. Reference librarians working alone duplicate each other’s work and fail to completely answer questions because we do not collaborate. Once in awhile, a librarian may go to colleagues for advice on a particular reference query. This may be done in person at the desk, at meetings where “favorite tools” are shown, through a bulletin board where problems are discussed online, or through a library network referral service. However, most reference questions are answered by librarians working alone.

What is the problem with the reference librarian, like the Lone Ranger, usually riding alone? The problem is that we are creating an enormously expensive service that, contrary to our desire to be democratic, is available to only a small part of the public or clientele we serve. In addition, we lose a priceless resource every time a librarian leaves the field. The bulk of the connections built up over years of experience in any librarian’s brain remain knowledge unique to that person. These “connections,” i.e., memories of where to go to obtain certain pieces of information, may not even be the librarian’s own unique knowledge at all times. How many of us have racked our brains for that elusive memory that just won’t come to the surface? The fact is, we aren’t giving our best service by constantly working in isolation.

The irony of this situation is that catalogers, often viewed by the public and their colleagues as the less people-oriented group, are far more collaborative in their work than their colleagues in reference departments. Catalogers do not reinvent a monographic entry if they do not have to. OCLC, RLIN, and regional catalogs are superhighways where super-catalogers are able to both share their thoughts and save themselves work. Reference librarians, on the other hand, seem to be content to trek through the wilderness on foot or horseback. We cannot afford to keep doing this.

Just as catalogers use computers to avoid partially or totally cataloging a title that has already been cataloged by someone else, reference librarians need to construct cooperative, computer-linked consortia to eliminate the need to partially or totally research questions that have already been answered by someone else. This will speed up the process when patrons are three abreast in line at the reference desk, and will allow us to help more people. I would call the method for doing this “reference annealing.”

In the 1980s, Neil Larson, a Berkeley hypertext software designer, wrote in his newsletter MaxThink about the concept of “information annealing.” Larson took the concept of annealing from metallurgy and applied it to social interaction. Larson defined “information annealing” as the use of a form of computer-assisted indexing, such as hypertext links, by people who wish to combine and fuse together their thoughts about an issue. Carried one step further, we could coin the term “reference annealing” as a term that covers librarians using computers to collectively combine our wisdom about how to answer reference questions.

Currently, reference librarians tend to use ready reference or I&R files to hold their collective wisdom, but these files are very cumbersome to set up and maintain. The solution I propose instead uses the online catalog as a place for collecting and remembering our wisdom. Reference annealing would require libraries to implement a networked online reference catalog using hypertext software. The advantages of using a catalog over an I&R system to communicate among ourselves are clear: (1) catalog entries are already standardized, (2) catalog entries are already typed, (3) catalog entries are regularly weeded whenever books are weeded. Lastly, catalog records contain a lot of information. This means there are lots of places in a catalog record to insert hypertext links.

At present, use of hypertext links in libraries is usually confined to touch screen terminals used in public access areas of institutions like the National Institute of Medicine Library or the Holocaust Archives. In these places, use of hypertext is passive. The programmer has created links between parts of a database, e.g., a short clip of Hitler speaking, a biography of Hitler, a list of his publications, etc. The viewer touches the screen to go from one of these items to another. Reference annealing would require a more active use of hypertext computer technology. In reference annealing, the viewers (i.e., librarians) would not only follow paths created by others but would add their own paths into the well of information created.

The first step in the process of reference annealing is the creation of a “virtual” reference collection. There are ample funds available for developing shared cataloging databases on the Internet. However, library directors are now focused on putting the catalogs of their entire circulating collections online. While this is certainly beneficial to local communities, the Internet is a global network. Patrons in Iowa don’t need to see the shelflist of a public library in California. Local access is laudable, but we arc getting less and less back for each dollar spent on putting a record of every item of a library’s collection on the Internet. A more cost-effective use of Internet technology would be to create a networked system of reference collections. This virtual reference collection, an imaginary reference shelf constructed by combining catalog records of titles from many libraries’ real reference collections, would be a “place” where reference librarians would go via their computer terminals to lookup sources for their patrons’ questions.

Obviously, an average two or three Library of Congress subject headings per title is not enough to make this virtual reference library very useful. What will make it invaluable is the computer technology created by developers like Neil Larson that allows librarians to insert “notes” about how to use a particular source and to make hypertext links between one catalog record and another. How would these hypertext links work? First, it is important to note that hypertext links are not just subject descriptors. Hypertext links are vehicles that, like Star Trek’s transporter, actually take one from one place to another. In the case of hypertext links in a virtual reference collection, the places are catalog records. When you click a mouse or push a key when the computer’s cursor is on a link, the link takes you from one record to another. When you create a link you are tying one record to another.

Librarians who have done online searching on systems like Westlaw will already be familiar with how hypertext works. In Westlaw, for example, hypertext links let the viewer easily jump from a court case to the full text of any other case cited in the original case. Then the searcher can go to another cited case, or back to the original case first looked at. The links in Westlaw also allow the online readers to look up the meaning of legal terms they come upon in a case.

Hypertext links then can mentally transport the librarian around the shelves of the virtual library in ways that subject descriptors cannot. For example, a directory of private schools would not be listed in the catalog under “youth” or “teenagers” since it is not about that group of people. However, that group of people would probably want to use such a directory. Hypertext links could take a librarian among the records of a library’s directories that young people, in particular, might want to use.

Likewise, hypertext links could make it easier to help patrons who fall into more than one socio-cultural group. For example, the record of a particular alcohol/drug treatment directory might have links from the records of similar directories showing that it alone provides information on which treatment centers both serve women and have facilities for the disabled. Meeting the needs of people of diverse backgrounds is now one of the most difficult problems at the reference desk in every kind of library.

Another question that constantly plagues reference librarians is the issue of what point of view a document takes. Hypertext links allow documents on a particular subject, e.g., capital punishment, abortion, etc., to be grouped by point of view, e.g., (supports) or (refutes). The hypertext links will show that this is a reference librarian’s subjective opinion, not a cataloger’s authoritative “subject heading.”

Hypertext links differ from subject descriptors because they cannot be created by catalogers working in a back room. Only librarians who work with the public will know what ways information in the collection should be linked so that their patrons can be assisted. Another difference from subject headings is that hypertext links might be guided by thesauri (especially an acronyms authority), but would not need a thesaurus. This is because hypertext links are not made from words to documents, but rather from documents to other documents. (Documents in this case are bibliographic records for reference works.) When a librarian inserts a link from one document to others, all the link has to do is make enough sense for another librarian to be able to use it. Computer algorithms can link records containing phrases like (for teenagers) and (for young people). In addition, if a link is too vague, another librarian can add to it to make it more specific. This is the process of “reference annealing.”

Hypertext links are usually words enclosed in some sort of brackets and/or highlighted by bold type or italicized font. When the cursor is on those words and the librarian clicks a mouse or uses a key like an arrow key or an enter key, the screen shifts to display a new document. Theodor H. [Ted] Nelson, the person who originated the term, defined hypertext as “nonsequential writing-text that branches and allows choices to the reader, [and is] best read on an interactive screen. As popularly conceived, this is a series of text chunks connected by links which offer the reader different pathways.”[1]

As I imagine a virtual shared reference collection, these pathways will need to wind among different sets of “shelves” in the collection. First there would be the set of shelves that represents the librarians’ own real library reference collection. Then there would be paths among the combined shelves of a consortium of libraries that have similar patrons and collections. Finally there is the entire set of shelves where the paths wind among all the reference books in the whole world. Browsing among any of these sets of shelves could be further limited to specific call number areas and dates of publication. Perhaps function keys would let the librarian choose which “shelves” to browse for an answer. And perhaps different typefaces, fonts, or styles of brackets could delineate in-house links from those created by librarians at other libraries.

In-house hypertext links could be of use even to librarians who work completely alone. In-house links could be mnemonic, assisting a librarian to recall where she or he found an answer before. The single librarian may also index reference works in ways not useful to most other librarians, e.g., a small special library might use hypertext links to find information easily accessed without any links by libraries that own sets of expensive encyclopedias.

These kinds of in-house hypertext links might not need to be shared. Shared hypertext links, on the other hand, will extend every librarian’s ability to answer reference questions quickly. Links embedded in the catalog records of the entire virtual reference library or of a consortium of similar libraries joined together will show a librarian what sources other librarians used previously to answer the question that the librarian is now researching. In the system I envision, the programming is so excellent that the librarian at the terminal can easily and quickly create a hypertext link and add her or his own knowledge to the virtual reference library as she or he uses it to answer reference queries or during a slow period at the desk.

Reference book publishers may have some concerns about such a virtual reference library, fearing that it might reduce demand for real reference materials if the ones available to librarians were used more efficiently. More likely, however, is the scenario where a reference librarian following hypertext paths through an entire virtual collection of reference items on the Internet would see many new sources that should be purchased for use in her or his local collection. And the virtual reference collection on the Internet would raise the funds for these new books because it would manifestly increase the number of satisfied library patrons using real reference collections.

A virtual reference shelf on the Internet could be created by just a few libraries with a small grant to test how it works. Patrons using terminals could only see and not create hypertext links. Librarians would make the links and even if only a few librarians are making links, they will need some coordination to make sure their links are smooth. At some point the virtual reference collection would become large enough that a sysop (system operator) would need to take care of it. If it were very successful, eventually there would need to be an agency that does for virtual reference records what OCLC and RLIN do for all catalog records. This doesn’t mean that hypertext technology will be the answer. Hypertext has its limitations, especially the tendency of its users to become “lost in hyperspace.” Nevertheless, the overwhelming amount of data barreling down the information superhighway demands some device better than a few keywords or subject descriptors to flag down the particular items of data needed.

What I have proposed here is a radical departure from the present way of doing things, but it has great possibilities to benefit libraries. If reference desks can increase the population they serve, they will be increasing their allies in every battle for maintaining or increasing funding. Librarians across the country are obtaining funds to make the Internet available to library patrons. Do we really think that patrons will not continue to need our help in finding the information they need? Of course not! So let’s obtain funds to use the Internet’s capabilities for ourselves so that we can serve our patrons better. Let’s stop reinventing the answer. Let’s start collaborating at the reference desk.


[1.] George P. Landow, Hypertext: The Convergence of Contemporary Critical Theory and Technology (Baltimore: Johns Hopkins Univ. Pr., 1992), 4.

COPYRIGHT American Library Association 1995