ACRL News Issue (B) of College & Research Libraries 1052 I C&RL News The future o f reference III: A response By Cynthia A. K ehoe Librarian, Balcones L ibrary Service C enter University o f Texas at A ustin I w ould like to focus on two types o f problem s we m ust deal w ith in moving towards th e library as concept, with th e electronic inform ation system as a prim ary m edium and artificial intelligence as a m ajor tool in aiding u ser access. W e face both pragm atic and conceptual p ro b lems. T he prim ary conceptual problem has to do with th e n atu re o f reference. O ne characteristic o f th e reference process is th at patrons are asking questions about som ething unknow n to th em , making it difficult to express a question clearly. T he patron m ust often talk around th e problem , ju st as th e description o f a do n u t hole relies on th e d e scription o f th e donut surrounding it. T he identifi cation o f an inform ation n eed m ust often be in te r active; problem s are clarified through dialogue. Users frequently leave out th e context o f a qu es tion, or phrase th e ir questions based on assum p tions about possible sources or a likely solution. R obert Taylor, in his classic article o f 25 years ago on inform ation seeking and question negotia tion, spoke o f four levels o f inform ation n eed — th e visceral or unconscious, th e conscious b u t vaguely defined, th e formalized, and th e com prom ised n e e d .1 T he com prom ised n e e d is th e one often brought to an inform ation system or expressed to a librarian. It has b een m olded to fit th e inform ation system, and may b ear little resem blance to the original inform ation need. O ne task o f th e re fe r ence librarian is to help determ ine w hat th e con scious inform ation n eed o f th e user is. E xpert systems are usually based on a major assum ption th a t may be false in m any reference situations— th at the question posed by th e user is a close approxim ation o f th e actual inform ation need. T he system may have a p ro ced u re for n ar rowing a request, b u t rarely is th e re a m eans o f com pletely reshaping th e question. T h ere is little means for aiding the u ser to move from a com pro m ised query to the conscious inform ation need. R eference librarians all have favorite anecdotes o f users whose initial questions bore little resem ’R o b e rt S. T aylor, “T h e P ro cess o f Asking Questions,” American Documentation 13 (1962): 391-396. blance to th e actual n e e d th at was eventually id e n tified. E xpert systems may eventually be able to cope w ith such situations, b u t they are not th e re yet and I do not expect th em to be for some time. P art o f th e new vision o f th e library as concept includes interaction betw een patrons and re fe r ence librarians th a t is m uch less often face-to-face. This prospect rem inds m e o f an anecdote about Xerox PARC, th e research c e n te r which does a lot o f w ork in office autom ation. A n u m b e r o f years ago, th e Xerox PARC co m p u ter scientists designed a glitzy new co m p u ter system for th e office, incor porating all th e latest co m p u ter technology and d e veloping some new ones, th a t could do all kinds o f w onderful things. T he Xerox PARC staff was very excited, and eager to d em onstrate it in another m eeting. T he system was placed in a Xerox corpo rate office, and it failed. N obody used it. W hen th e research staff exam ined why, they found th at it was because th e system, while exciting to technologists, bore little relation to th e way people work. Xerox, being smart, h ired psychologists and an thropologists and specialists in organizational b e havior as p art o f th e ir research team s, and th e next generation o f products was m uch m ore successful. W hat does this have to do with libraries? The possibility o f electronic reference work is exciting to many, b u t it doesn’t yet appear to fit in with how m any o f ou r patrons use libraries. Part o f th e b arrier is simply lack o f knowledge am ong patrons a b o u t th e possibilities. It still surprises m any people th a t they can call a library and ask a re fe r ence question over th e phone. B ut p a rt o f the b arrier is th a t use o f these systems requires a change in behavior for many people th a t hasn’t b e e n d e m o n stra te d to th e m to have a strong enough payback to w arrant th e change. T he library is still, for many people, a place to go because o f th e im portance patrons place on in ter action with th e staff and im m ediate access to a wide range o f materials. E ven in a library service unit, such as th e one in which I work, with lots of technology and almost no collection, most patrons still com e to th e “library” rath e r than contact us by o th e r m eans in o rd er to req u est m aterials available elsewhere. Patrons p re fe r to talk with a person, prim arily for reassurance, I suspect. T hey w ant to December 1 9 9 0 / 1053 be sure that they’ve given the staff all the necessary information for a request, to ask questions, and to allow us to ask questions. The process is highly interactive, personalized, and immediately respon sive. We may eventually design avenues o f access to the library and its contents in addition to face-to- face that patrons will regularly use, but this will be a slow process. W e will need to utilize the research we already have about information-seeking behav ior, as well as gather new data on how that Behavior changes as new technologies becom e available, in order to design successful access points. W e also need to examine what patrons think will improve the library. They often seem much less interested in new means o f access than in w hether we are effective at our present services. How fast can you put w hat they want in their hands? Many visions of the library’s future are prim arily technological, and do not give us any time fram e for their realization. Some o f the technologies dis cussed are available now (for a price); others may not be available for 20 years. The prim ary prag matic problem s we face have to do with the re sources needed to develop our electronic libraries. Project costs to develop a small expert system (of only 50 to a few hu n d red rules) start at $25,000, and take the equivalent o f more than .25 person years of the program m er alone.2 To design an expert system to help answer ready reference questions, an expe rienced AI program m er would need to spend at least ten hours p e r week for a year, working with a reference librarian who would need to find th re e to five hours p er week for that same period. In addi tion, to support this system, we would have to purchase workstations on which to run it. If we want to provide dial-up access, we may n eed to work with the campus com puter center to expand networking capabilities— again for an additional cost. And the program and equipm ent would n eed to be m aintained and updated. In an era in which library budgets are shrinking, and personnel bu d g ets often seem to us particularly hard hit, we m ust seek large amounts o f additional funding to build our new information systems. Expert systems in business can rarely be justified on the basis of savings in the num ber of employees. The savings come from elsewhere— often m ore ac curacy in decision making saves th e com pany money. The benefit for libraries will not be as easily dem onstrated. These systems are not likely to re sult in fewer staff, though there may be some shifting of duties. F o ra reference librarian to spend 200 hours over the course o f a year as part o f an AI 2Paul H arm on et al., Expert Systems Tools and Applications (New York: John Wiley & Sons, Inc., 1988): 184. design team is 200 hours away from som ething else. Academic libraries have been retrenching in vari ous ways for a n um ber o f years, and it is becoming increasingly difficult to find trade-offs and tasks to give up in o rd er to allow staff to take on new projects. A major role o f librarians in a reference room is instruction in th e use o f particular tools. It has been suggested that we should make m ore o f this in struction com puter-based. Aside from the issue that such instruction is rarely in a vacuum, and th ere is often a broader information need to be addressed, the cost o f designing a one-hour com puter-based training program is high. The prim ary expense again is personnel. It takes about 200-250 hours o f developm ent to produce one hour of com puter-based instruction.3 This is not to advocate refusing to design these systems because we cannot afford them , b u t it is a rem inder th at we will need to work hard to justify our projects to funders, to increase the level of funding, and to find it from new sources if we are to succeed. Smaller academic libraries in particular will face serious difficulties in providing these tools. In the transition towards a m ore electronic li brary, we may find it increasingly hard to cope with patron dem ands— particularly as we provide in creased access to all kinds o f materials in citation form, b u t do not provide electronic delivery o f the full text as quickly. Faculty are not necessarily willing to face trade-offs betw een buying materials and providing electronic access during this period o f change. Libraries have already found that having an online catalog often leads to increased circula tion— especially in areas that were not as accessible through card catalogs, such as governm ent docu m ents and special collections. Adding end-user searching capabilities, such as CD -ROM in the reference room, increases the dem ands on refer ence staff, and increases the num ber o f items re quested through interlibrary loan. T here is also a greater need for instruction in th e various new com puter-based information systems. One o f the side effects o f electronic information systems is the greater potential for information overload. Many filters (good and bad) disappear in the electronic environm ent. Librarians will need to work with faculty to teach undergraduates how to b e tte r evaluate and filter information. The proliferation of com puter-based inform a tion systems continues to be m uch m ore rapid than th e im provem ent in search interfaces. Standardi zation of interfaces is slow. AI will help out here eventually. But again, this is going to come slowly, 3Estim ate by Dr. Patricia Smith, Associate Pro fessor, D epartm ent o f Curriculum and Instruction, University of Texas at Austin. 1054 / C&RL News and th e transition may be rocky. These inform ation systems are not easy to use. In a study in which Stanford undergraduates w ere briefly train ed to use an online catalog w ith Boolean capabilities, one-third could not use it.4 And full-text databases are m ore difficult to use than such catalogs. O ne o f th e most com m on problem s is in how to form ulate a search query— th e point at which AI has th e least help to offer at this tim e. P art o f th e difficulty lies in th e fact th at m ost patrons use these systems infre quently, and th erefo re are not able to rem e m b er all th e necessary com m ands and strategies o f th e m ore complex systems from session to session. This tends to discourage casual users. O nline vendors have rep o rted increasing n u m bers o f passwords allotted to end-users. But th e libraries which keep track o f th e searching (and th e bills) rep o rt th a t these systems generate initial enthusiasm , b u t th a t only a small n u m b e r o f p a trons continue to use them . Some o f th e new inform ation technologies, such as online catalogs, do generate im m ediate and continued in terest am ong patrons, b u t not all inform ation systems have b een as successful. In o rd er to im prove th e search interfaces o f these systems, a great am ount o f research still needs to be done. T h ere are not a lot o f prototypes resulting from inform ation retrieval research th at are obviously b e tte r th an c u rre n t systems, esp e cially for a range o f u ser system and subject exper tise, a n d ju st waiting to b e im plem ented. Librarians (and o th e r knowledgeable library staff) should be involved in research and developm ent projects, b u t many o f th e efforts th a t are n ee d e d will req u ire a significant investm ent on th e p art o f libraries and o th e r agencies, in staff tim e and o th e r resources. T he research needs open up many possibilities for cooperative efforts in applied research am ong li braries, with o th er d epartm ents in th e university, and with o th er agencies such as publishers. Such cooperative efforts im pose th e ir own constraints. As we move towards a g reater use o f technology, we also m ust invest m ore strongly in ou r staffs. Staff m em bers n eed training in th e use o f com puters and inform ation technologies. Are all staff able to use m icrocom puters and th e online catalog? Is th e reference staff expected to try each new C D -R O M for an hour or so w hen it first arrives? O r do reference librarians feel guilty if they’re trying a new C D and a patron wants to use it? In addition to investing in our staff b o th through in-house training and providing release tim e and funding for training outside th e library, we also have to invest in technology for th e staff to use. It is difficult to help patrons with equipm ent we rarely 4Christine Borgman, The Users Mental Model o f an Information Retrieval System: Effects on Per form ance (dissertation, Stanford University, 1984). have access to ourselves. P art o f w hat we know from studies o f th e diffusion o f innovations is th at an im portant p art o f getting patrons to accept a new technology is to make th e staff com fortable with it— including th e less obvious staff m em bers. In an academ ic library, one m eans o f ensuring accep tance o f new co m p u ter tools is to teach your stu d en t staff how to use th em (when they’re not teaching you). T h eir enthusiasm in o th er settings is a great benefit. As we incorporate inform ation technologies, we m ust not forget th a t it will be years yet before everyone has a co m p u ter at hom e, w ith telecom m unications capabilities, and before nearly all p a trons are co m p u ter literate. T he n u m b e r o f nontra- ditional college students has b een on th e increase for many years. In addition, th e frequently quoted statistics showing th at m ore and m ore elem entary and high schools have com puters are misleading. T he n u m b e r o f schools owning m icrocom puters is quite high; th e n u m b e r o f com puters in each school is still low. O ne-third o f all public schools have less than te n m icrocom puters.5 T he students graduat ing from high school in th e next few years will have vast differences in th e ir co m p u ter experiences. Libraries may have to play several roles in m ain taining equity providing adequate co m p u ter access to th e ir inform ation tools in th e library, and helping to train students who do not yet have th e necessary skills to use th e inform ation technologies. T he o th e r m ajor aspect o f equity is cost. C an we ask undergraduates to help pay for these systems and, if so, in w hat form should th e paym ent be? As we attem p t to utilize m ore and m ore com p u ter-b ased technologies, libraries are going to have to becom e m ore aggressive in seeking fund ing. E q u ip m en t is expensive, and adding it does not necessarily result in great savings elsewhere. M ore libraries will n e e d to have a grants specialist on th e ir staff, who is given both adequate tim e and training to pu rsu e funding. G rant money, however, is finite and project-oriented. Library adm inistra tors will have to work h a rd e r than ever to solicit moneys from university adm inistrators and legisla tures th a t have often b een cutting library funding for years. W hile libraries can som etim es get o n e tim e start-up funds for flashy co m p u ter products for patrons, th e real task ahead may b e to solicit m aintenance and updating funds for these p ro d ucts, and funds for eq u ip m en t for th e staff, m uch less popular causes. I have suggested problem s in two areas— issues related to th e design and use o f inform ation te c h nologies, and problem s caused by th e n eed for additional resources, including support for our 5The Electronic Directory o f Education (M arket D ata Retrieval, 1989). D atabase available through Dialog Inform ation Services, Inc. December 1 9 9 0 / 1055 staffs. I do not intend to suggest that we should not pursue some o f these projects, although we must make our choices carefully. Artificial intelligence is not going to solve many problem s for us in the next five to ten years, and it is difficult to predict its im pact over thirty years. O ur tasks will not be easy, and we will not reach our goals as quickly as we might wish. The future o f reference III: Another response By Dennis Trombatore Librarian, Geology Library The University o f Texas at Austin While I found Pat M olholt’s presentation e n te r taining and challenging, I would like to rem ind everyone that we are talking about the concept of libraries in the context o f universities where, if I can paraphrase, we practice the willing suspension of profit and loss in th e hope of having an effect on people’s lives, to transm it understanding through teaching, and to inquire into th e nature of things. The university is not M cD onald’s, Chevron, or IBM, and though th ere is a mythology o f the uni versity, described by Anne W oodsworth, Pat Mol- holt, et al. in th eir 1989 article as “in mission, character, and organizational structure . . . essen tially a medieval institution,”1 and that mythology may have been deeply altered by big professions, big sports, big research, big government, and big enrollments, I believe it is too soon to replace the library, the so-called heart o f the mythical univer sity, with a Jarvik-7. In spite o f P at M olholt’s subtle efforts to downplay the significance and usefulness o f print collections while skillfully persuading us o f the allure and irresistible vitality of artificial intelli gence (AI) systems, the fact is that no m atter what technological mix we end up being able to afford in university libraries, the key to th e information fu ture is hum an-based services delivered by a suffi cient num ber of people who care and people who hustle to get the job done right the first time. At the 1978 LITA C onference on Closing the Card Catalog, H ugh Atkinson, th en of Ohio State, also spoke about walls— he predicted that online library catalog systems would destroy traditional physical and social work patterns in libraries, in effect allowing workers and work to be distributed in a way that would unify library departm ents at the same tim e that it increased th eir autonomy and 1Anne W oodsworth et al., “T he Model Research Library: Planning for the F u tu re ,” jo u rn a l o f Aca demic Librarianship 15 (July 1989): 132-138. im proved services.2 Atkinson described these work groups as “tribes” of about a dozen people. In effect, this am ounts to a reinvigoration o f the branch library concept (something that has in fact happened) where, rather than splitting off and com partm entalizing print from electronic systems, collection developm ent from technical services, or reference from adm inistration, these necessary segments of an information delivery system are integrated around the mutual online catalog files now available to us. Meanwhile we in branch librar ies, who wear all these hats, can ply our trade w here it counts— footsteps away from our customers. I ’d like to rem ind everyone that all the hullaba loo about access over acquisition is the sad o u t growth o f physical and fiscal exigency, and that the yearning for global interconnectivity is just another run at the same old wish to have everything close at hand. Yet, access without delivery is suicide. To illustrate that, let me ask you to substitute the term “m icroform ” for “A I” in th e access provision model. W e already provide lots of access on micro form, b u t people confound us by refusing to accept it. Why? Because we refuse to pu t muscle behind delivery. In order to save money most libraries offer too few printers, printers that are poorly d e signed, th at cost too much when they do work, and that generally give a lousy product. T he costs o f real AI systems, with serious access and delivery potential, would destroy us, so we will settle for what we can afford— som ething betw een that old magic eight-ball toy and a thought police man— all the while asking ourselves why no one is ever satisfied. 2H ugh C. Atkinson, “The Im pact o f Closing the Catalog on Library Organization,” in Closing the Catalog: Proceedings o f the 1978 and 1979 Library and inform ation Technology Association In sti tutes, ed. by D.K. G apen and B. Juergens (Phoenix: Oryx Press, 1980): 123-133.