In the mid-1970s paper was the primary medium for storing and providing access to bibliographic information. That changed in notable increments over the next two decades, gentle precursors to the revolutionary change brought about by the World Wide Web’s possibilities and its ever-growing ubiquity. Technology transformed the process of searching for bibliographic information more frequently than any other aspect of reference. Reference service’s responses to and accommodation of those changes in searching make an ideal lens through which to examine broader trends.
Paper gave way initially to unseen people operating computers on behalf of reference patrons. At some universities it was possible to submit a request for a search of a database such as ERIC,Dissertation Abstracts, andPsychological Abstracts to a specially trained librarian or by mail to a commercial service provider. The librarian would then interact through a console with a main- frame computer which would search computer tapes and generate a result set of citations to documents. Rarely did this process occur in real time. There were time lags between submission of the written request, its transmittal to the specialist, and return to the requestor of results—as likely as not printed out in all capital letters on wide green bar tractor feed paper. One test of three services revealed turnaround time ranging from 14 to 21 days (Langley, 1976). Results were inconclusive when the quality of these services’ searches conducted by distant mediators was compared to the quality of the results of manual searches of printed indexes. Time lag was a negative; savings in labor was a positive. However in two of the three cases, a manual search of the same sources retrieved more citations than the machine searches. Cost models varied. Some companies based charges ‘‘on the number of citations retrieved, such asDATRIX IIwhich charges $15.00 for the first 150 citations plus $10 for each additional citation. Others base[d] charges on the complexity of the search, as is the case withPASAR’’ (ibid., p. 229).
In the 1976 reference scenario above, the student was responsible for searching for relevant citations. The processes of searching for citations and evaluating each citation’s relevance were conducted simultaneously by the person who would use the information in the cited sources. The advent of electronic databases changed that. The process of searching belonged exclu- sively to the librarian; the process of evaluating citations was usually shared by the librarian and the patron. Here in a kernel is the most important trend in reference service over the past 30 years. The respective roles and, in some ways, the responsibilities of the reference librarian and the user of reference service changed. If in 1977 Vavrek considered it suicidal for the reference librarian to allow usurpation by paraprofessionals of the reference librarian’s role, what, one wonders, would he have said then about the degree to which key elements of that role have transferred to library users.
A. Real-Time Online Searching
Services allowing librarians to conduct database searches online in real time emerged in the late 1970s. BRS and DIALOG dominated the library da- tabase search market. The H.W. Wilson company’s Wilsonline service was a latecomer to online searching. These services acted as wholesalers, making multiple databases available to librarians. This offered the advantage of
going to one or two or three providers, each using a consistent interface to all of its databases. Their systems employed arcane command structures using, for example, ‘‘..’’ or ‘‘/’’ prefixes to commands that compacted multiple factors into the fewest possible number of characters. This required training through classes these companies conducted throughout the country. Specialized jour- nals supplemented this training. Information Today began publication of Onlinein 1977 to meet the needs of librarians and businesses working to use online databases effectively in their organizations. Database began publica- tion the following year. Learned Information began publication ofSearcherin 1993. All three continue to publish today, a testament to the ongoing vitality of the online search business. These magazines offer industry news, trend analysis, and practical information for searching particular databases or for types of information. Information Today’s annual Computers in Libraries conference also thrives; the 21st was held in Washington, DC in 2006.
Once trained to use DIALOG or another service, librarians could place a telephone handset into the acoustic coupler of a portable ‘‘briefcase’’ terminal, and transmit and receive data at a 300 baud rate. Results emerged from these terminals on a slow scroll of thermal paper. A librarian’s preparation before a search consisted of database selection; consultation of a manual providing information on each database’s scope and structure (librarians of that time recall DIALOG’s ‘‘blue sheets’’); consultation of a thesaurus of controlled indexing terms for each database; and a reference interview of the end user.
That person typically sat beside the librarian during a scheduled appointment for the search. The end user helped the librarian evaluate results as they appeared. Using this counsel the librarian could modify the search to increase or decrease the size of the set of records retrieved or to achieve more pertinent results. Potential weaknesses were inherent in this collaborative arrangement between librarian and end user. The reference interview might not have identified the user’s need; the user’s expectations of the search might have been unrealistic; or the search might not have employed the most pertinent terms.
The cost structure of database searching, as well as the training required to craft and conduct successful searches, precipitated the division of labor between the librarian (search and discovery) and the patron (evaluation and selection). Typically these consisted of timed charges for connection to a telecommunications network, timed charges for connection to the database aggregator’s system, and charges for every printed or downloaded citation.
The last charge varied from database to database; the format and complete- ness of each citation (e.g., with or without abstract, descriptors, etc.) also modulated cost.
Left to a self-service user, the cost of a single search would have been astronomical and the quality of the results highly questionable. Some
libraries absorbed in whole or part the cost of searches; others passed them on in full to the end user. End users were generally satisfied with the results of mediated searches. One study showed that the majority of users were satisfied with the results of searches and the cost (Hilchey and Hurych, 1985). Seeing opportunities to clarify mystery citations and to provide other information to answer reference questions quickly, librarians incorporated, at no cost to the end user, ready reference searches into their desk services (Brownmilleret al., 1985). They used the Online Computer Library Center (OCLC) and Research Libraries Information Network (RLIN) databases in similar ways (Gould, 1984). Despite their name, ready reference searches were generally conducted away from the desk. Because of the pressures of timed costs, searches were generally conducted in offices where interruptions were minimal. Some li- brarians worried that electronic databases shortchanged researchers ‘‘since the computer supplies only references containing the stated parameters, giving a limited output. Important citations may be missing because the requestor forgot to include a term, but perhaps more important, the search does not give the surrounding or overlapping fields that could have been surveyed in the printed index’’ (Champlin, 1985, p. 216). Despite such reservations, users demanded more of these searches. The introduction of online circu- lation systems and first generation integrated library systems in the 1980s undoubtedly contributed to user’s acceptance of machine searching of other bibliographic databases (Bates, 1984).
B. Early End-User Searching
Given the cost considerations, mediated searching of electronic databases was the norm in the 1970s and into the 1980s, even as microcomputers replaced dumb terminals and telecommunication speeds increased. The Personal Computer (PC) allowed librarians to quickly download large sets of results for the end user to examine and evaluate at leisure. Once library users adopted the IBM PC, introduced in 1981, and the Apple Macintosh, introduced in 1984, for their other academic work, they became interested in using their computers for their bibliographic searches. An experiment conducted at a Boston teaching hospital demonstrated an appetite for self-service searching.
In 1979 the hospital’s library installed a computer terminal that enabled
‘‘casual, walk-in users to access directly a specially prepared file of 400,000 citations in clinical medicine’’ (Dalrymple, 1984). This system, known as Paper-Chase, demonstrated user acceptance of unmediated searching.
The companies that made librarians’ searching of a wide range of dat- abases practical also worked to offer end users the same opportunity. In the early 1980s BRS created its BRS/After Dark service, DIALOG created its
Knowledge Index, and Dow Jones created its Dow Jones News/Retrieval. As the menu-driven BRS/After Dark system implied, these complemented li- brarian-mediated searches. Indeed, After Dark was available only in ‘‘off’’
hours evenings and weekends, times when few libraries offered mediated searches. These end user systems offered access to a subset of the ever- growing number of databases available to librarians through the parent systems. Even with simplified interfaces, these systems required some train- ing—either that or librarians decided on behalf of users that they required training. One library’s test of BRS/After Dark that included a training component concluded that the training was indeed necessary (Trzebiatowski, 1984). It also concluded that the student participants were satisfied with the results of the searches they conducted for themselves (ibid.).
OCLC also entered the online end user search market in 1991 when it introduced FirstSearch for access to its vast bibliographic database and ad- ditional databases (Benaud and Bordeianu, 1993). To simplify database se- lection, FirstSearch organized databases in broad topical areas such as Arts and Humanities, Education and Social Sciences, and Business/Law/Public Affairs. Searches cost a fixed amount. To control costs most libraries that offered FirstSearch to their users during the service’s first few years required users to purchase cards, each card authorizing a set number of searches. This was awkward to administer. Even though it was simpler to use than the other online end user systems and it offered the option of simple and advanced searches, FirstSearch did not become widely used until simpler, more con- venient access replaced the card system. FirstSearch also went through several major revisions to its user interface, each time becoming friendlier for the untrained, every-now-and-then searcher.
Despite users’ satisfaction with the end user online systems of the 1980s, libraries did not adopt them widely. Some librarians doubted that end users, despite having been left for decades to do their own manual searches in paper indexes, could conduct quality online searches and were thus better served by librarian-mediated searches. These doubts, user training requirements, the need in some places for a dedicated telephone line, the need for an on-call librarian to troubleshoot problems with search and telecommunications sys- tems, limited hours of service, the inability to download results, and cost factors all worked against wide acceptance and implementation of these services (Ankeny, 1989). Furthermore, a powerful competitor soon entered the market.
C. Maturing End User Searching: The Disc
In the mid-1980s the Information Access Corporation introduced its Info- Trac system, ‘‘A laserdisc database system that provides nearly instant access
to the citations of over 1000 periodicals as well as The New York Times and the Wall Street Journal’’ (Walker and Westneat, 1985). To create this user- friendly system, InfoTrac brought together a laser disc that held the search- able bibliographic citations, one or more dedicated IBM PCs, and an equal number of inkjet printers. Special caps replaced some standard caps on the PCs’ keyboards. These bore labels such as ‘‘SEARCH’’ making it easy for the user to issue commands to the system. InfoTrac updated its database monthly by shipping a new laserdisc. A drawback to the system was its shallow temporal coverage; initially it offered indexing back to only 1982. Cost—
typically more than $14,000—was a barrier for some libraries. Everywhere InfoTrac was installed, students took to it immediately, some willing to wait as much as an hour to use it rather than an immediately available paper index providing similar coverage.
The nature of some librarians’ criticism of InfoTrac prompted one commentator to ask, ‘‘one wonders if maybe part of the problem is the librarians not the products’’ (Kleiner, 1987, p. 261). Criticism of the quality of the database, its mix of the scholarly and the popular, and its lack of Boolean search capabilities revealed a difference in values between some reference librarians and many library users. The librarians valued scholarly information, precision in information identification, and powerful search capabilities. The users valued printable citations, acceptably relevant citations, and labor- and time-saving ease of use. That is not to dismiss many librarians’ conviction that they could perform more efficient, more precise, and more relevant searches than end users. It does, however, raise questions about the librarian’s and the end user’s respective roles. A series of studies at Louisiana State University (LSU) showed that ‘‘about one-third of the students needed assistance’’ with InfoTrac (ibid., p. 260).
Unfortunately the study did not report on the type(s) of assistance needed.
Some assistance surely responded to problems with inkjet printers. Reference librarians quickly learned to clear paper jams and to bend a paperclip straight to insert it into an ink cartridge to ‘‘massage its bladder’’ to get the last bit of ink out of it. The LSU study reaffirmed roles of the librarian and the end user. The librarian possesses and can apply expert knowledge of information sources, storage and retrieval systems, and search techniques.
Since reference service’s origins in the 19th century, reference librarians have applied their expert knowledge to help individuals identify and obtain in- formation relevant to their needs. The reference librarian serves as conduit to information, not gatekeeper. Even so, the reference librarian is not the ex- clusive conduit; individuals are free to find their own paths to information and to make their own judgments about its value. Librarians directed users to InfoTrac and provided brief information on how to use it; end users
conducted their own searches and judged the value of citations from the printed results.
In response to InfoTrac’s popularity many libraries placed appointment sheets beside each of the PCs so users could reserve it for a specific time, usually a 15-minute period. ‘‘A trade-off normally exists between the com- plexity and power of a retrieval system and its ease of use; the more ca- pabilities a system offers, the more difficult it is to master’’ (Harter and Jackson, 1988). InfoTrac offered ease of use over power and complexity. As a result it may have attracted new library users who didn’t have the patience to slog through annual printed indexes but who were engaged by computerized information retrieval.
CD-ROM technology soon eclipsed the laserdisc. In use by the music industry since 1982, the compact disc was adopted for data storage and retrieval later in the decade. Following the success and popularity of Info Trac, database producers complemented their online access with CD-ROM access. SilverPlatter and OVID filled the role that BRS and DIALOG played as online database gateways. Each of these companies developed a proprietary interface for its databases. Database vendors made their content available through one or more of these gateway vendors. During the late 1980s and the early 1990s library administrators demonstrated their willingness to fund the cost of improved reference service. Librarians rearranged reference rooms, often by disposal or reduction in size of the card catalog, a tool superseded by the online catalog and retrospective conversion of the card catalog’s contents. Just ten years after Vavrek decried cost cutting as a threat to reference service, administrators funded tables, chairs, PCs, CD-ROM drives, and even electrical system upgrades to accommodate this popular new information technology. As libraries added more CD-ROM databases they also paid higher licensing fees to allow multiple users simultaneous access.
Mediated online searches became less common as CD-ROM offerings in- creased and as users demonstrated a preference for doing their own searches, even where their library fully subsidized mediated searches.
Even though CD-ROM databases were promoted as end-user products and even though librarians deployed them in public areas, this technology raised the same concerns about the roles of the end user and the librarian that BRS/After Dark, Knowledge Index, and InfoTrac raised. The issue of user education received a great deal of attention, succinctly framed in the ques- tion ‘‘what are the obligations of librarians when aware that users are satisfied even though they are obtaining less than optimal results?’’ (Harter and Jackson, 1988, p. 519) This is a question of authority and control. Does authority belong to the user or to the librarian and who should control a search and its outcome? In a sense, CD-ROM brought the librarian’s and
user’s respective roles full circle. With CD-ROM end user search systems, once again the processes of searching for citations and evaluating each one’s relevance were conducted simultaneously by the person who would use the information in the cited sources.
At the same time, CD-ROM offered ample opportunities for instruction.
When CD-ROM databases became popular in the late 1980s PC ownership among college students and even faculty was far from universal. Much point- and time-of-need instruction dealt with using computer equipment and printers. CD-ROM databases almost universally offered Boolean search, nested search statements, limits, and other capabilities. Librarians had come to expect these features, familiar to them in the online world, in any database and to judge its value in large part on these capabilities. Reference librarians longed for uniformity in search interfaces. It was not uncommon for them to interact in the course of a typical day with two or three CD-ROM interfaces, OCLC, the local online catalog, and several DIALOG online databases not available locally on CD-ROM. All of these raised training issues for librar- ians themselves. Those same training issues were, potentially at least, ap- plicable to the users of the end-user systems. Nonetheless a study conducted towards what turned out to be the end of CD-ROM’s ascendancy revealed that end users ‘‘relied heavily on simple searches’’ and made very little use of Boolean operators and other advanced search techniques (Anderson, 1995).
This prefigured the popularity of Google and its minimalist search interface.
D. Mature End User Searching: The Web
Everything changed and nothing changed in the mid- and late-1990s when the World Wide Web rapidly became the preferred medium for access to bibliographic databases, breaking news stories, prospective dates and mates, books and music CDs, airline tickets, current weather information and fore- casts, government information, sports scores, pornography, and much, much more. In late 1994 the Netscape browser was released. The cliche´ ‘‘spread like wildfire’’ applies to librarians’ and the populace’s acceptance of the Web.
During the latter half of the 1990s database producers and Online Public Access Catalog (OPAC) vendors migrated access to their products to the Web. CD-ROM towers disappeared from reference rooms and PCs connected to campus networks were added. Users clamored for more and more infor- mation online through the Web, especially full-text and full-image journal articles. Reference budgets were reworked to meet user expectations.
By the time the Web debuted, students were accustomed to graphical interfaces and computer mouse manipulation. Instruction in the use of Netscape would have been superfluous. PC use, if not ownership, had become