ed 380 135 ir 055 439 author michelson, avra; rothenberg, jeff title report … · 2013-08-02 ·...

DOCUMENT RESUME

ED 380 135 IR 055 439

AUTHOR Michelson, Avra; Rothenberg, JeffTITLE Scholarly Communication and Information Technology:

Exploring the Impact of Changes in the ResearchProcess on Archives. Rand Reprints.

INSTITUTION Rand Corp., Santa Monica, Calif.REPORT NO RAND/RP-187PUB DATE 93NOTE 85p.

PUB TYPE Reports Evaluative/Feasibility (142) Journal

Articles (080)JOURNAL CIT American Archivist; v55 n2 p236-315 1992

EDRS PRICE MF01/PC04 Plus Postage.DESCRIPTORS Access to Information; *Archives; Artificial

Intelligence; Change; *Communication (ThoughtTransfer); Curriculum Development; ElectronicPublishing; *Futures (of Society); Human Resources;Hypermedia; Information Dissemination; InformationPolicy; *Information Technology; ProfessionalDevelopment; *Research Methodology; StrategicPlanning; *Trend Analysis; Users (Information)

IDENTIFIERS Connectivity; End User Statistical Computing; VirtualReality

ABSTRACTThe report considers the interaction of trends in

information technology and trends in research practices and thepolicy implications for archives. The information is divided into 4sections. The first section, an "Overview of Information TechnologyTrends," discusses end-user computing, which includes ubiquitouscomputing, end-user interfaces, and "online transition." Connectivityis examined in terms of access to computational and human resources;the trend toward interchange standards; and distributed versuscentralized control. The following technology trends affectingscholarly communication are examined: artificial intelligence;end-user publication and distribution; hypertext and hypermedia;visualization and virtual reality; and caveats. The second section,"Scholarly Communication and the Use of Current InformationTechnology," discusses a number of issues. Identification of sourcesand communication with colleagues is covered. Interpretation andanalysis of sources is discussed, including computer-assistedanalysis through conversion and computer-assisted analysis withartificial intelligence. Electronic publishing and hypermedia areidentified as findings for dissemination of research. Curriculumdevelopment and instruction is also discussed. Section 3, "Responsesby the Library Profession to Changing Research Practices," includespromoting connectivity; conversion; software engineering; andtransformations in professional roles. A fourth section, "Conclusionand Recommendations," is divided into the following parts:establishing a network-mediated archival practice; establishing astrategy for the future usability of electronic records; andrecognizing and rewarding leadership. (Contains 224 references.)(AEF)

,RAND.

UAL DEPARTMENT Of EOCCATOONOffice of &Map Natal RfanCh and anotovernan/

EDUCATIONAL RESOURCES INFORMATIONCENTER IERICI

O This document has been reproduced asrecerved from the person or organizationoriginating it

0 MclOr changes have been made to imotovereproduction duality

Pomtil ot wow Or opinions stat edmthisdocument do not 1100111311fily represent othciaiOERI apeman or policy

Scholarly Communication.and InformationTechnology

Exploring the Impact of Changesin the Research Process onArchives

Avra Michelson and Jeff Rothenberg

2BEST COPY AVAILABLE

"PERMISSION TO REPRODUCE THISMATERIAL HAS BEEN GRANTED BY

E.D. Gi11

TO THE EDUCATIONAL RESOURCESINFORMATION CENTER (ERIC1."

RAND is a nonprofit institution that seeks to improve publicpolicy through research and analysis. The RAND Reprint Seriescontains offprints of RAND research originally published injournals or books. The text is reproduced here with permission ofthe publisher. Publications of RAND do not necessarily reflect theopinion or policies of the sponsors of RAND research.

Published 1992 by RAND1700 Main Street. P.O. Box 2138, Santa Monica, CA 90407-2138

IIIMINIMIMONNIMINNNO RAND/RP-187

Scholarly Communicationand informationTechnology

Exploring the Impact of Changesin the Research Process onArchives

Agra Michelson and Jeff Rothenberg

Reprinted from The American Archivist

236 American Archivist / Vol. 55 / Spring 1992

Research Article

Scholarly Communication andInformation Technology:Exploring the Impact of Changesin the Research Process onArchivesAVRA MICHELSON AND JEFF ROTHENBERG

About the authors: Avra Michelson is a member of the Technology Research Staff at the NationalArchives and Records Administration. Jeff Rothenberg is a Senior Computer Scientist in the SocialPolicy Department of The RAND Corporation.

Avra Michelson would like to express her gratitude to William Holmes and Charles Dollar forthe opportunity to do this research. Jeff Rothenberg would like to thank Tora Bikson for establishingthe connection with NARA that led to this research and Bob Anderson for insightful comments onthe technology issues discussed in the report. The authors also would like to extend a special thanksto Paul Conway for serving as a wise sounding board throughout the research and writing of thereport; Anne Kenney, for taking the time to explain and consider strategies for preservation micro-filming and digital imaging; Richard Cox for his instructive readings of many drafts; Tora Biksonfor her thoughtful review of the final version; and Lida Churchville for extraordinary assistancewith database searching. The recommendations that accompany the report were written as a productof the authors' participation in the 1991 Research Fellowship Program for the Study of Mode aArchives administered by the Bentley Historical Library, University of Michigan, and funded by theAndrew W. Mellon Foundation and the University of Michigan. The authors are deeply grateful toDavid Bearmal, Richard Cox, Charles Dollar, Anne Kenney, Paul Evan Peters, Michael Sperberg-McQueen, Joan Warnow-Blewett, and Jerome Yavarkovsky for their review of drafts of the report,including an early version circulated prior to the Bentley symposium, and for their assistance indeveloping policy recommendations for the report. Also, the ignitors would like to acknowledge thevaluable comments received from many people mentioned in the report who aided the research orwe.e kind enough to review sections.

Reprinted by pemtission from The American Archivist, Vol. 55, No. 2., Spring 1992, pp. 236-315. Copyright m 1992 by The

Society of American Archivists.

5IEST COPY AVAILABLE

Scholarly Communication and Information Technology 237

ContentsExecutive Summary 237 Communication with Colleagues 262Introduction 238 Interpretation and Analysis ofBackground 241 Sources 263Overview of Information Technology Computer-Assisted Analysis Achieved

Trends 244 Through Conversion 264End-User Computing 245 Computer-Assisted Analysis with

Ubiquitous Computing '46 Artificial Intelligence 268End-User Interfaces 247 Dissemination of ResearchThe "Online Transition" 247 Findings 273Summary 248 Electronic Publishing 273

Connectivity 248 Hypermedia 276Access to Computational and Curriculum Development and

Human Resources 249 Instruction 278The Trend Toward Interchange Summary 281

Standards 251 Responses by the Library Profession toA False Dichotomy: Distributed Changing Research Practices 282

Versus Centralized Control 252 Promoting Connectivity 283Summary 253 Conversion 284

Specific Technology Trends Affecting Software Engineering 286Scholarly Communication 253 Transformations in ProfessionalArtificial Intelligence 253 Roles 288End-User Publication and Conclusion and Recommendations 289

Distribution 256 Part I: Establishing a Network-MediatedHypertext and Hypermedia 257 Archival Practice 289Visualization and Virtual Part II: Establishing a Strategy for the

Reality 259 Future Usability of ElectronicCaveats 260 Records 297

Summary 260 Part III: Recognizing and RewardingScholarly Communication and the Use of Leadership 303

Current Information Technology .... 260 Bibliography 303Identification of Sources 261

EXECUTIVE SUMMARYThe emergence and use of information

technology is this century's most signifi-cant development affecting archival prac-tice. In response to this development,members of the archival profession haveexplored both the ways in which new tech-nology can improve the management of ar-chives, and the most appropriate methodsfor managing electronic records that resultfrom the use of automated systems. Butthese two issues only partially address theimpact of technology on archives. A thirdbut indirect influence also deserves exam-ination: technology's impact on scholarlyresearch methods, which has consequencesfor the use and management of archives.

This article considers the policy implica-tions for archives of trends resulting fromthe infusion of information technology intothe scholarly research process.

The article considers the interaction oftwo distinct kinds of trends: trends in in-formation technology and trends in re-search practices, particularly among socialscientists and humanists. Although muchof the rapid growth and evolution of infor-mation technology may be unrelated toscholarly research, and aspects of scholarlyresearch may be evolving in ways that havelittle connection with information technol-ogy, there is nevertheless a strong and im-portant interaction occurring between thesetwo evolutions. The possibilities created by

238 American Archivist / Spring 1992

new technology are prompting transfor-mations in scholarly practice, and thesetransformations are in turn stimulating newneeds among researchers and are inspiringfurther technological breakthroughs. Un-derstanding the nature. of this interaction isnecessary for forecasting the most likelyways in which new scholarly methods willdemand innovative services and responsesfrom the archival community.

This article explores two fundamentaltrends in information technology affectingscholarship: end-user computing and con-nectivity. Several other technologies of rel-evance to scholarship are also considered,including artificial intelligence, end-userpublication and distribution, hypertext andhypermedia, and visualization and virtualreality. Changes in the research process re-sulting from scholarly use of informationtechnology are considered within the broadframework of scholarly communication. Thescholar's use of currently available tech-nology to search for sources, communicatewith colleagues, interpret and analyze sourcematerials, disseminate research findings, andprepare curriculums and instructional applications is examined. Our key finding isthe exploding use among researchers of in-formation technology on research and ed-ucation networks to advance scholarship.Far from being visionary, this suture is al-ready present: It is currently being experi-enced by significant and increasing numbersof scholars from many disciplines. The li-brary profession is responding to the emer-gence of network-mediated scholarship bypromoting global connectivity, performingconversions of print sources to machine-readable form, undertaking the softwareengineering of full-text delivery systems foronline materials, and collaborating withtechnologists in the use of computing andcommunication technology to meet spe-cialized researcher needs.

The report recommends that the archivalprofession:

7

1. establish a presence on the Internet/NREN.

2. make source materials available forresearch use over the Internet.

3. create documentation strategies todocument network-mediated schol-arship and the development of re-search and education networks.

4. develop archival methods suitat'e foroperation with NREN.

5. take user methods arid future com-putational capacity into account in es-tablishing policies on the managementof software-dependent records.

6. recognize and reward initiatives that(a) advance archival management ofelectronic records, (b) respond toscholarly use of information technol-ogy, or (c) promote a network-me-diated archival practice.

This article is the result of nearly twoyears of collaboration between Avra Mich-elson and Jeff Rothenberg. Earlier versionsor derivative presentations of the article werereported at annual meetings of the Societyof American Archivists, National Associ-ation of Government Archivists and Rec-ords Administrators (NAGARA), NationalNet '92, and the Library of CongressWorkshop on Electronic Texts. The articleis available electronically on the file serveroperated by the Coalition for NetworkedInformation. (Contact [email protected] forinstructions.)

INTRODUCTION

The emergence and use of informationtechnology is this century's most signifi-cant development affecting archival prac-tice.' In response to this development,

The term archives refers broadly to historic sourcesof enduring value that document the activities of gov-ernments, organizations, or individuals; it also refersto the rcpositorics respoi,ible for preserving and ;flak-ing available the historic record.

BEST COPv ^mil ABLE


members of the archival profession haveexplored both the ways in which new tech-nology can improve the management ofarchives2 and the most appropriate methodsfor managing electronic records that resultfrom the use of automated systems.' But

'Sec Marion Matters, ed., Automated Records andTechniques in Archives: A Resource Dire.-tory (Chi-cago: Society of American Archivists. lT;3), 12-37,for a bibliography on the topic. A selection of theseminal literature includes: Thomas H. Bickel-son,.4rchives and Manuscripts: An Introduction to Auto-mated Access, SAA Basic Manual Series (Chicago:Society of American Archivists, 1981); Richard H.Lytle, "An Analysis of the Work of the National In-formation Systems Task Force," American Archivist47 (Fall 1984): 357-65 (see also other articles in thisissue of AA, especially Thomas E. Brown, "The So-ciety of American Archivists Confronts the Com-puter," pp. 366-82); David Bearman, TowardsNational Information Systems for Archives and Man-uscrip. Repositories: The NISTF Papers (Chicago:Society of American Archivists, 1987), as well asBearman's Archives and Museum Informatics tech-nical reports and quarterly newsletter; and two specialissues of the American Archivist devoted to "Stan-dards for Archival Description" (Fall 1989 and Win-ter 1990). More recently, archivists have begun toexplore the use of specific technologies to supportarchival functions. See, for instance: Optical DigitalImage Storage Systtm: Project Report (Washington,D.C.: National Archives and Records Administration,Archival Research and Evaluation Staff, March 1991);Avra Michelson, Expert Systems Technology and ItsImplications for Archives, National Archives Tech-nical Information Paper no. 9 (Washington, D.C.:National Archives and Records Administration, Ar-chival Research and Evaluation Staff, March 1991);and Anne R. Kenney and Lynne K. Petsonius, "TheFuture of Digital Preservation," Advances in Pres-ervation and Access, vol. 1 (Westport, Conn.: Meek -Icr Press, forthcoming).

'Charles Dollar identifies salient literature on thistopic in his work, The Impact of Information Tech-nologies on Archival Principles and Methods (Ma-ccrata, Italy: University of Maccrata Press, 1992). Aselection of seminal publications includes: CharlesDollar, "Appraising Machine-Readable Records," inA Modern Archives Reader: Basic Readings on Ar-chival Theory and Practice, edited by Maygene F.Daniels and Timothy Walch (Washington, D.C.: Na-tional Archives and Records Service, 1984); MargaretL. Hedstrom, Archives and Manuscripts: MachineReadable Records, SAA Basic Manual Series (Chi-cago: Society of American Archivists, 1984); HaroldNaugler, The Archival Appraisal of Machine-Readable Records: A RAMP Study with Guidelines (Paris:

these two issues a idress only a portion ofthe impact of technology on archives. Athird though indirect influence that de-serves examination is technology's impacton scholarly research methods, which hasconsequences for the use and managementof archives. This article considers the pol-icy implications for archives of trends re-sulting from the infusion of informationtechnology into the scholarly researchprocess.

The term information technology refersto the computing and communicationstechnology used to obtain, store, organize,manipulate, and exchange information. Thedefinition includes computer hardware andsoftware, as well as the telecommunica-tions devices and computer-based networksthat connect ;nem.' The influence of infor-mation techn 'logy on the research process,already evident, promises to deeply pene-trate scholarly practice as we enter thetwenty-first century. This technology is en-abling academics to change significantly theway they communicate and collaborate,identify and analyze sources, store and re-trieve data, and disseminate the productsof their research. Although technology af-fects the research process across a spectrumof disciplines and professions, this articlefocuses on changes in the social sciences

General Information Programme and UNISIST,UNESCO, 1984); United Nations, AdministrativeCommittee for the Coordination of Information Sys-tems, Technical Panel on Records Management, Elec-tronic Records Guidelines: A Manual for PolicyDevelopment (New York: United Nations, 1989); andResearch Issues in Electronic Records. (St. Paul,Minn.: Published for the National Historical Publi-cations and Records Commission, Washington, D.C.,by the Minnesota Historical Society, 1991). Sec alsoTom Rullcr, "Managing and Appraising GIS Data:Issues and Strategies," unpublished paper presentedat the 1991 annual meeting of the Society of AmericanArchivists, Philadelphia.

'John R. B. Clement, "Increasing Research Pro-ductivity Through Information Technology: A User-Centered Viewpoint," unpublished paper, 19 October1989, p. 3.


and humanities because scholarly patronsof archives tend to be drawn most heavilyfrom these fields.5

Undertaking the research for this reportwas motivated, in part, by efforts in thearchival profession to provide answers toquestions related to the use of source ma-terials, such as the following: Who are the(potential) users of primary sources? Whatare the characteristics of the modern re-search process? How do patrons frame re-search questions ?6 In the past few years,several empirical studies have been con-ducted on patterns of research use withinor across repositories or specific disci-plines.' Although these studies provide

'The terms scholar and researcher generally arcused throughout the paper to refer to social scientistsand humanists unless specified otherwise. Neverthe-less, we believe that the research trends identified inthis report apply to a broader range of the researchcommunity.

6A selection of key literature that has advanced thearchival profession's conceptual framework includes:Mary Jo Pugh, "The Illusion of Omniscience: SubjectAccess and the Reference Archivist," American Ar-chivist 45 (Winter 1982): 33-44; Elsie T. Freeman,"In the Eye of the. Beholder: Archives Administrationfrom the User's Point of View," American Archivist47 (Spring 1984): 111-23; Paul Conway, "Facts andFrameworks: An Approach to Studying the Users ofArchives," American Archivist 49 (Fall 1986): 393-407; and Lawrenc; Dow ler, "Availability and Use ofRecords: A Research Agenda," American Archivist51 (Winter/Spring 1988): 74-86.

'A selection of the key studies includes: MajorFindings, Conclusions and Recommendations of theResearcher and Public Service Component EvaluationStudy (Ottawa: Public Archives of Canada, 1985);Paul Conway, "Research in Presidential Libraries: AUser Study," Midwestern Archivist 11 (1986): 35-56; William J. Maher, "The Use of User Studies,"Midwestern Archivist 11 (1986): 15-26; David Bear-man, "User Presentation Language in Archives, Ar-chives and Museum Informatics 3 (Winter 1989-90):3-7; Paul Conway, Partners in Research.- TowardsEnhanced Access to the Nation's Archives (Washing-ton, D.C.: National Archives and Records Adminis-tration, forthcoming); and Ann D. Gordon, Using theNation's Documentary Heritage: The Report of theHistorical Documents Study, supported by the Na-tional Historical Publications and Records Commis-sion in cooperation with the American Council ofLearned Societies (Washington, D.C.: National His-torical Publications and Records Commission, 1992).

valuable insights on users and patterns ofuse for the period of study, they typicallyfail to consider their findings within thecontext of a broader analysis of scholarlyresearch trends.

Archivists need more than snapshots asa basis for policy formulation. An accuratedepiction of current research practices isnecessary, but archival strategic planningmust also involve an analysis of significanttrends. This article addresses the interac-tion of two distinct sets of trends. Elec-tronic information technology as aphenomenon is experiencing rai id growthand evolution, much of which ; .ty be un-related to scholarly research. lit the sametime, aspects of scholarly research may beevolving in ways that have little connectionwith information technology. Neverthe-less, a strong and important interaction isoccurring between these two movements.The possibilities created by new technol-ogy are prompting transformations inscholarly practice, and these transforma-tions are in turn stimulating new needsamong researchers and further inspiringtechnological breakthroughs. Understand-ing the nature of this interaction is neces-sary for forecasting the most likely ways inwhich new scholarly methods will demandinnovative services and responses from thearchival community.

Trends analysis is inherently somewhatcircular, since technological changes"drive" changes in scholarly practice onlyto the extent that the new technology pro-vides capabilities that scholarly researcherscan use in meaningful and productive ways.It invoives more than the description of ar-bitrary technological trends: Their rele-vance must be derived from the perspectiveof scholarly research. It also involves morethan the description of current trends inscholarship: To the extent that scholarshipuses information technology, it is neces-sarily constrained by what is currently pos-sible. Only by considering the joint evolutionof technology and scholarly methods can a

BEST COPY AVAILABLE


convincing picture of the future be con-structed. The remainder of this article at-tempts to create such a picture in order toexamine its implications for archives dur-ing this decade and beyond the turn of themillennium.8

This article presents a conceptual frame-work for understanding long-term trendsrelevant to the scholarly research process.The topic is introduced by a discussion ofscholarly communication and the early useof computers among academics. An analy-sis of information technology trends mostpertinent to the conduct of research fol-lows. The third section explores, throughcase examples, trends in the use of cur-rently available ormation technology bysocial scientists and humanists. The fourthsection considers model efforts by those inthe library profession to respond to changesin the research process. The article con-cludes with policy recommendations thataddress key changes needed in archivalpractices and methods to respond to trans-formations in scholarly research methods,and the growing prominence of a new elec-tronic communication mediumresearchand education networks.

BACKGROUND

Scholarly inquiry represents a timelesshuman quest to understand the world around

'Because this paper examines the interaction of twodistinct trends, differing frameworks arc used to or-ganize the key sections (Overview of InformationInchnology Trends and Scholarly Communication andthe Use of Current Information Technology). The for-mer uses information technology trends as the organ-izing framework, whereas the last uses the elementsof scholarly communication as a structuring frame-work. The relationship between technology and schol-arship is both dynamic and complex, and ourunderstanding of it continues to evolve. Although itwas suggested to us that the framework used to ex-plore information technology trends should he usedas the organizing framework for the section on currentscholarly practices as well (e.g., a more technologicaldeterminist approach), we consider the dual frame-works, and the analysis of the relationships betweenthem, one of the paper's key virtues.

us. Although this quest for understandingis a sustaining element of human culture,the techniques of the scholar have changedover time. No longer characterized by oraltradition and forum dialogues, the modernresearch process is commonly understoodto entail five processes: (1) identificationof sources, (2) communication with col-leagues, (3) interpretation and analysis ofdata, (4) dissemination of research find-ings, and (5) curriculum development andinstruction for preparing the next genera-tion of scholars. Refinement of the schol-ar's original idea or hypothesis occursthroughout these more tangible processes.The impact of information technology onthese processes is resulting in unprece-dented transformations in scholarly com-munication.

Scholarly communication is the term usedto refer to the interrelationship of the fiveprocesses of modern scholarship.' The termimplies both a dynamic exchange of infor-mation and ideas and an interdependenceamong publishers, librarians and others inthe support of scholarship and the advance-ment of knowledge. Scholarly communi-cation is generally understood to involvethe social exchange of intellectual and cre-ative activity from one scholar to another.'As a concept, it denotes a recognition ofthe mutual reliance of researchers, publish-ers, professional associations, and librariesand archives in fostering intellectual pur-

°The American Council of Learned Societies pop-ularized the term scholarly communication among ac-ademics as a result of their mid-1980s survey on theexperience of more than five thousand humanists asauthors using scholarly publications, libraries, andcomputers. The findings of the report appear in Her-bert C. Morton and Anne J. Price, The ACLS Surveyof Scholars: Final Report of Views on Publications,Computers, and Libraries (Washington, D.C.: Officeof Scholarly Communication and Technology, Amer-ican Council of Learned Societies, 1989).

"Thomas W. Shaughnessy, "Scholarly Commu-nication: The Need for an Agenda for Action ASymposium," Journal of Academic Librarianship 15

(May 1989): 69.

242 American Archivist I Spring 1992

suits. This interdependence implies that achange in the practice of any one of theseagents is capable of inspiring changes inthe entire paradigm. In transforming the wayin which academics learn of primary sourcematerials, search and gather data, interpretand analyze sources, and report findings tothe scholarly community, information tech-nology is influencing significant aspects ofscholarly communication. Consequently,changes in scholarly research patterns haveramifications for archives and libraries."

The influence of modern technology onscholarly communication began with thebirth of computers. More than forty yearsago, the scientific community was the firstof the academic disciplines to introducecomputers into the research process. Ascomputing power expanded, geographi-cally dispersed scientists began collaborat-ing on research questions requiringcomputers. In 1969, in response to the needsof this community, the U.S. Defense De-partment's Advanced Research ProjectsAgency (ARPA) d 'eloped the ARPA-NET, a telecommunications network de-signed to allow the sharing of expensivecomputer resources among govei ament andacademic research laboratories." Scientificcomputing has evolved to include the useof electronic networks for electronic mail(e-mail) and for access to supercomputingprocessing power and to software that fa-cilitates group work."

"For a historical consideration of the relationshipbetween scholarly communication and libraries, seePhyllis Dain and John C. Cole, cds., Libraries andScholarly Communication in the United States: TheHistorical Dimension, Beta Phi Mu Monograph, no.2 (New York: Greenwood Press, 1990).

"Clifford A. Lynch and Cecilia M. Preston, "Ev-olution of Networked Information Resources," Pro-ceedings of tire Twelfth National Online Meeting May7-9, 1991 N.Y., N.Y., Martha E. Williams, ed. (Med-ford, N.J.: Learned Information, 1991): 221-30.

"See, for instance, a recent book of readings, ed-ited by Irene Grcif, Computer-Supported CooperativeWork (San Mateo, ('alif.: Morgan Kaufmann, 1988);for an assessment of the information needs of thc sci-

BEST. COPY AVAILABLE

Since the 1970s, a large and complexarray of networks has emerged to supportcollaborative scientific research. As thescientific need for connectivity increased,network infrastructures at institutions, or-ganizations, commercial enterprises and re-gions expanded. Today, more than threethousand regional, federal, commercial, andorganizational networks connect an esti-mated 5 million scholars in seventy coun-tries:4 The Internet, the existing networkof research and education networks, com-prises thousands of trunk lines that cur-rently carry fron; 1.5 to 45 million bits persecond.'' The National Research and Ed-ucation Network (NREN), authorized in1991 and due to be operational by 1995,will be capable of transmitting 1 billion bitsof datathe equivalent of fifty thousandtypewritten pagesevery second.'`'

In recent years, the global expansion ofelectronic networks has allowed for world-wide collaboration among scientists. Fur-ther, the connectivity provided by greaterbandwidth lets scientists process previouslyunimaginable amounts of data. Expandingthe volume of data able to travel acrossnetworks permits scientists to explore newtypes of questions because greater amountsof data are available with less time requiredfor analysis. Equally important, the prom-

cntia scholar, sec Communications in Support of Sci-ence and Engineering: A Report to the National ScienceFoundation from the Council on Library Resources(Washington, D.C.: The Council, August 1990); fora discussion of state-of-the-art collaboration-orientedsoftware, see Daniel Williams, "New Technologiesfor Coordinating Work," animation 36 (15 May1990): 92-96.

"Clifford Lynch, "Tclecomrunications and Net-working: A Tutorial," presentation made at theAmerican Society for Information Science 54th An-nual Meeting, Washington, D.C. (29 October 1991).

"Lynch and Preston, "Evolution of Networked In-formation Resources."

'From a presentation by Paul Peters, executive di-rector of the Coalition for Networked Information, tothe National Archives and Records Administration on7 May 1991; sec also Ralph Alberico, "The Devel-opment of an 'Information Superhighway'," Com-puters in Libraries 10 (January 1990): 34.


ise of increased computing power and ad-vances in telecommunications will allowscientists to expand the graphical displayof research results, alleviating many prob-lems associated with interpreting very largedata sets." The trends characteristic ofmodern scientific inquirygreater collab-oration, increased use of computer-assistedanalysis of machine-readable sources, andexpanded use of global research and edu-cation networksincreasingly representtrends in the social sciences and humanitiesas well.

In the humanities, scholars initially usedcomputers simply to store and retrieve data.In what is commonly believed to be theearliest project of its kind, Father RobertoBusa in 1949 began his effort to compilean index and concordance to the work ofSt. Thomas Aquinas.'8 But apart from thehard sciences, the field of political scienceis typically regarded as the discipline mostresponsible for transforming computerprocessing into an accepted scholarlymethod. What began as a simple use ofcomputers by political scientists forprocessing survey data and analyzing na-tional opinion polls became a standard so-cial science methodology: quantitativeanalysis. During the past four decades, fol-lowing the lead of survey researchers, arange of scholars within academic disci-plines began to use computer technology toprocess large sets of numeric data."

"Clement, "Increasing Research Productivity," 3;for a discussion of the role of imagery in human un-derstanding, scc Mary Alice white, "Imagery in Mul-timedia," Multimedia Review (Fall 1990): 5-8.

'"Sec David S. Miall, ed., Humanities and theComputer: New Directions (Oxford: Clarendon Press,1990), 2.

'"As their numbers grew, quantitative scholars Ac-ccssfully campaigned for the establishment of dataarchives, special repositories designed to preserve andprovide access to machine-readable collections of sur-vey, census, polling, and legislative data. Sec Kath-ken M. Heim, "Social Scientific Needs for NumericData: The Evolution of the International Data ArchiveInfrastructure," Collection Afnnagement 9 (Spring19871: 1-53.

The advance of information technologyover the past several decades has aston-ished even the most visionary technolo-gists. Although certain predictions haveproved too optimistic, the overall rate ofadvance has matched or surpassed the pro-phesies of most experts, and it shows everysign of continuing unabated during the nextfew decades. Indeed, from 1980 to 1985,the period that marked the birth of personalcomputers, their use among scholars soaredfrom nonexistent to more than 50 per-cent.20 Today, the scholarly use of personalcomputers extends beyond storage and re-trieval of data and includes text editing,formatting, and text analysis. Increasinglyscholars are turning to technology to do sta-tistical analysis, create databases, producespreadsheets, and compile graphical im-ages of data. Many scholars consider tech-nology an essential instructional tool forgenerating simulations, capturing data, and

="Morton and Price, ACLS Survey of Scholars, 33.The ACLS study represents the only currently avail-able direct survey of scholars on their use of com-puters. But the survey polled only scholars who aremembers of professional associations. For the pastfew ycars, EDUCOM and the University of SouthernCalifornia have conducted an annual survey of aca-demic computing directors on campus planning, pol-icies, and procedures affecting the use of desktopcomputers. According to reports by academic com-puting centers, 39.5' of faculty at two year publicand four year public and private colleges and univer-sities have access to or own computers. This figure,however, is considered unreliable, as it is based onestimates by academic computing staff, rather than ondirect counts. Furthermore, no one believes that actualusage has dropped from 1985 to 1991, as implied bythe discrepancy between the ACLS and EDUCONI?USC figures. According to Kenneth C. Green, theEDUCOM/USC survey developer and author of thereport on the findings, "our limited knowledge aboutstudent and faculty access to and use of technology isappalling." Green argues that a direct survey of schol-ars is needed to identify actual computer usage. SccUSC Center for Scholarly Technology Newsletter,"Despite Budget Cuts, Campuses Attempt to Main-tain Computing Services," (October 1991); KennethC. Green, "A Technology Agenda for the 1990s,"Change 23 (January/February 1991): 6-7; and Ken-neth C. Green and Skip Eastman, Campus Computing/990 (Los Angeles: University of Southern ('alifor-nia, Center for Scholarly Technology, 1990).

12


providing individualized assistance to stu-dents.2'

The driving force behind the advance ofinformation technology has been the de-velopment of faster, smaller, and cheaperelectronic devices, which can be used toproduce machines with greater capabilitiesfor manipulating and processing informa-tion. These machines have in turn inspiredthe production of more powerful and im-aginative programs and solution techniques(computational methods or algorithms) forsolving problems that would be intractablewithout this new computational power. Theavailability of increased computationalpower, in turn, has enabled the design ofnew computer hardware and software, pro-ducing a snowball effect in which each newgeneration of system facilitates the designof its successor. This process can be ex-pected to continue until designers reach thefundamental limitations of physics and ex-haust all technological alternatives, whichnoes not appear imminent. An improve-ment in computational power of six ordersof magnitude (a factor of a million) overthe past two decades can be attributed toroughly equal improvements (three ordersof magnitude each) in hardware and soft-ware.22 It is not unreasonable to expect acomparable improvement to occur over thenext two or three decades. As a result, inthe next few decades an unimaginableamount of computational power will beavailable to scholars. This capacity com-pels the archival profession to determinethe implications of the use of information

''Sec Miall, Humanities and the Computer, 4; andJean-Claude Gardin, "The Future Influence of Com-puters on the Interplay li?.tr, ;en Research and Teach-ing in the Humanities," Humanities. CommunicationNewsletter 9 (1987): 17-18.

"Grand Challenges: High Performance Computingand Communications, The FY 1992 U.S. Researchand Development Program, A Report by the Co,,.mittee on Physical, Mathematical, and EngineeringSciences, Federal Coordinating Council for Science,Engineering and Technology, Office of Science andTechnology Policy (1991), 14-15.

technology by scholars for conventional ar-chival practices.

Although the future evolution of infor-mation technology is fairly predictable inbroad outline, predicting precise details ofhow the technology will evolve is more dif-ficult. For our purposes, however, it is thebroad outline of these trends that is mostimportant. Our discussion of technology,therefore, avoids mentioning specific de-vices, techniques, or research results. In-stead, the next section examines trends ofinformation technology that are likely tohave the greatest impact on scholarly com-munication and, by implication, on ar-c.iives management. The focus here is onbroad descriptions and projections mostrelevant to the future of scholarly research.Later in this paper we examine how schol-ars ate actually using information technol-ogy in their current work.

OVERVIEW OF INFORMATIONTECHNOLOGY TRENDS

The two most obviousand for the pur-pose of this paper, the most importantinformation technology trends that pertainto scholarly communication are end-usercomputing and connectivity. These trendsare distinct and separable, and each is dis-cussed in detail below. Ultimately, how-ever, it is the integration of the two thatwill have the greatest impact on scholarlycommunication. End-user computing en-hances the autonomy of the researcher, i.e.,the researcher's ability to use the power ofcomputation to conceptualize and executeresearch without sacrificing intellectualcontrol by delegating computational tasksto specialists. Connectivity enhances theresearcher's abilities to access data, collab-orate, seek input and feedback, and dis-seminate :deas and results. The confluenceof these trends produces a rich interplay ofsynergistic effects, which are explored be-low.

A number of more specific technology

13 LLST COPY AVAILABLE


trends are also likely to affect scholarlycommunication. Most of these are exam-ples of end-user computing or connectivity(or the integration of the two), but eachwarrants attention in its own right. The mostrelevant of these appear to be artificial in-telligence, end-user publication and distri-bution, hypermedia, and visualization andvirtual reality.

End-User Computing

In the current context, end-user comput-ing refers to the direct use of computers byresearchers.' The general trend toward theincreased use of computers is understand-able. Computers continue to become bet-ter, cheaper, more accessible, and moreusable. Software continues to become mornapplication-oriented, and user interfacescontinue to improve. Databases continue tobecome larger and more relevant. As theuse of computers becomes more common,,users continue to increase in number andsophistication, generating greater and greaterdemand for computation while driving priceseven lower by expanding the size of themarket. But the increasingly direct use ofcomputers by their end-users is a more re-cent and more interesting trend, and its im-plications for research are profound.

The term end-user refers to someone whophysically uses a computerthe person whotouches the keyboard and reads the screen.

"For most users, the trend toward direct accessbegan with personal computers (PCs), but it actuallybegan soon after the advent of the modern computer.The very first computers of the early 1950s were es-sentially single user machines and, since users had tobe very aware of their machines' foibles (and typicallyhad to be present while running their programs inorder to deal with problems), they necessarily becameintimate end-users. Later, more reliable mainframecomputers often ran jobs in batch mode (batches ofwork were run together instead of individually) to im-prove their utilization, which tended to distance usersfrom their machines. In the early 1960s, however,timesharing reintroduced direct access by allowingmultiple users to share a mainframe machine remotelyfrom their terminals.

The end-user may or may not initiate orconsume the results of the computation. Itis useful to distinguish the end-user fromthe "ultimate user" of a computer: some-one who initiates and consumes the resultsof a computation, without necessarilytouching or seeing the machine. The ulti-mate user is the person who causes a com-putation to be performed and who uses theresults of the computation, i.e., the personwhose work involves computation, whetheror not it involves using a computer directly.

End-user computing occurs when the end-user and the ultimate user are the same.The crux of end-user computing is that theend-user is able to initiate computations andget results without going through an inter-mediary. To some extent, this is a detail:What difference does it make if a compu-tation is performed by a researcher or aprogrammer? But the distinction is an im-portant one, since it bears on how centralthe computation is to the researcher's thoughtprocess. If a researcher is the ultimate userof a database, for example, but is not theend-user, then some intermediary (librar-ian, data archivist, programmer, secretary,or assistant) is interposed between the re-searcher and the database, limiting the re-searcher's ability to interact directly withthe data, to browse through it, to exploreits idiosyncrasies, and to become intimatewith it. Similarly, if a researcher askssomeone else to write a program to com-pute summary statistics, the researcher willbe unaware of the decisions embedded inthat program or the problems encounteredin writing it.24 This kind of insulation fromthe computational process may free the re-searcher from menial tasks, but it also lim-its his or her ability to define the computation

"Although writing a program does not guaranteethat one will becomelet alone remainaware of itsimplications and limitations, using a program writtenby someone else virtually guarantees that the user willnot be aware of then,


correctly, use it appropriately, and under-stand the implications of its results.

From a practical standpoint, end-usercomputing is attractive because of its con-venience. An end-user need not find a pro-grammer or data processing specialist (andan available machine) to get an answer toa computational problem. This reduces thethreshold of effort required to performcomputation, allowing users to consider ita more integral part of their work style.

The ramifications of end-user computingin the research process are deeper and moresubtle than they may first appear. Only bybecoming intimate with the computationalprocess can a researcher fully realize thepotential of computation in performing re-search. Only when the researcher is an end-user does computing become familiarenough and convenient enough to be a nat-ural part of the research process. This isnot an end in itself, but it is important be-cause it allows the researcher to conceiveof new kinds of research that become pos-sible only when computation becomes anintegral part of research. End-user com-puting is an important trend because theactivity of computation allows researchersto reconceive the nature of research itself,i.e., the kinds of questions posed, themethodologies used, the type and extent ofsources analyzed, and the form of presen-tation of the findings. (Examples are dis-cussed in a later section.)

To summarize: End-user computingmeans direct access to computational ca-pability; the key implication of this in thecurrent context is that it allows computa-tion to become an integral part of a re-searcher's thought processand thereforeof the research itself.

Ubiquitous computing. One trend thatis still relatively new is the advent of port-able computing, using laptop, notebook, oreven pocket-sized ("palmtop") computers.This portability means more than just beingable to carry a computer from one locationto another. It implies the ability to carry a

BEST COPY AVAILApLE

part of one's working context (database,text, notes, and correspondence) in a ma-chine that can be used on location, in meet-ings, or while traveling. This context maybe "downloaded" to a portable machinefrom a researcher's home machine and usedfor on-site research or during interactionswith other researchers to modify data, re-cord notes, work on evolving documents,and many other tasks. The results of thiswork can thea be "uploaded" to the re-searcher's home machine, by a telecom-munications link from the remote locationor by a direct transfer of data after the re-searcher returns home.

In addition to portable machines them-selves, cellular modems (modulator/de-modulators) allow computers tocommunicate over cellular telephone links.This allows the user to link computers whiletraveling anywhere that cellular telephonecoverage is provided; it is already possibleto connect to a remote computer or data-base from a portable computer while ridingin a taxi in any major city in the UnitedStates. Whether this kind of remote com-puting will ultimately become a commonactivity depends on tradeoffs between thesize, cost, and capacity of portable versusremote computers and the attendant tele-communications costs.

The important point is not the size andcapability of portable machines, but ratherthe freedom they give the user to performcomputations and to access data from anylocation. For example, another way ofachieving the same result would be to pro-vide computer terminals in public places;this would be analogous to the use of stan-dard (noncellular) telephones, which areubiquitously available anywhere in the de-veloped world. The French government hasimplemented just such an approach to com-puting in its Minitel system, which is avail-able in homes and post offices throughoutFrance.' Because of these alternatives, it

'`David L. Margulies, "C'est la France, C"est Min-


is useful to think of this as a trend toward"ubiquitous computing" rather than "port-able computing." This is discussed furtherunder Connectivity below.

End-user interfaces. The design of soft-ware for end-users has also had a tremen-dous impact on the growth of end-usercomputing. For end-users who are notcomputer specialists, "access" to compu-tation means more than simply having acomputer or communicating with one. Touse . ,..omputer effectively, such users needsoftware that allows them to work in waysthat are natural to them, without having tolearn the intricacies of an arcane computersystem. Software for end-user computingmust have two key attributes: It must pro-vide functionality that is of use to the end-user, and it must present an interface thatis usable by an end-user.

Appropriate functionality requires thatsoftware be either generically useful (suchas word processors, electronic mail, data-bases, spreadsheets, and mathematical pro-grams) or designed for some specific task.that the user performs. Task-specific pro-grams (or applications) tend to be writtenfor users in a given industry or type ofwork.26 But if its interface makes it diffi-cult to use, neither generic nor task-specificsoftware is of much value to any but themost dedicated and tenacious of end-users.

The trend toward improving end-user in-terfaces began in the early 1960s.27 Many

itcl," PC Computing 2 (January 1989): 194; EllisBooker, Vive Ir. Minitel," Telephony 215 (8 August1988): 24; and S. Nora and A. Minc, The Comput-erization of Society: A Report to the President of France(Cambridge, Mass.: MIT Press, 1980).

"Both general-purpose and task - specific programsbecome more useful when they can be tailored to theneeds of a particular end-user. Examples of this areword processors that allow users to define their owndocument formats, function keys, "macros," etc. Theultimate general-purpose program is a programmingsystem (or language) that allows end-users to definenew computations at will (i.c., to write programs);end-users may become programmers to a limited ex-tent by tailoring software to their own needs.

"For example, Cliff Shaw's JOSS system is widely

of the principles of current user interfaceswere developed by Engelbart's group at theStanford Research Institute (SRI) in the1960s and early 1970s.28 This led to thedevelopment of a number of systems at Xe-rox Corporation's Palo Alto Research Cen-ter (PARC) in the late 1970s, culminatingin the introduction of the Star in 1981.29The Xerox Star pioneered the point-and-click, window- and menu-blsed "desktopmetaphor" that is currently in vogue. Thistrend toward better user interfaces gainedmomentum with the development of per-sonal computers, and it has now reached apoint where many systems can be learnedand used effectively by most users withoutany formal computer training. Although theterm user friendly has become such an ad-vertising cliché that it is now all but mean-ingless, its overuse is a measure of the extentto which the computer industry recognizesthe importance of user interface design forend-user computing.

The "online transition." One of the keyfactors that facilitates end-user computingis an "online transition"3° in which com-

regarded as one of the earliest successful timesharedsystems designed for direct access by researchers. SeeJ. C. Shaw, JOSS: Conversations with the JohnniacOpen-Shop System (Santa Barbara, Calif.: RANDCorporation, P-3146, 1965); J. C. Shaw, "JOSS: ADesigner's View of an Experimental On-Line Com-puting System," in American Federation of Infor-mation Processing Societies Conference Proceedings(Fall Joint Computer Conference), Vol. 26 (Bal:i-more, Md.: Spartan Books, I964): 455-64.

"In additQn to inventing the mouse, this visionarygroup developed many of the concepts that form thefoundation of modern user interface design, as wellas producing one of the first hypertext systems. Foran early description of this work, sec D. C. Engelbartand W. K. English, "A Research Center for Aug-menting Human Intellect," American Federation ofInformation Processing Societies Conference Pro.ceedings (Fall Joint Computer Conference) vol. 33.(May 1974), 395-410.

29J. Johnson, T. L. Roberts, W. Verplank, D. C.Smith, C. H. Irby, M. Beard, and K. Mackey, "TheXerox Star: A Retrospective," IEEE Computer 22(September 1989): 11-26.

"'The term online originated in the electric powerindustry. Generating equipment is said to be "online"



puting becomes more useful the more it isused. If a user is still Lund to the tele-phone, paper mail, paper documents, paperfiles, and paper memos, then computationremains an infrequently used tool that doesnot integrate with the rest of the work en-vironment. When electronic mail (e-mail)begins to replace telephone and paper mes-sages and when machine-readable elec-tronic documents and files begin to replacepaper, the user's working context is inte-grated in new ways.

The online transition produces a newphenomenon: Many previously separateforms of communication become integratedby being stored in electronic form. For ex-ample, if telephone messages and tele-phone directories are both electronic, userscan forward information from a phone mes-sage in e-mail and can use telephone num-bers or other information from a phonemessage to search their phone directoriesfor information about callers. Many mes-sages that traditionally have come by tele-phone will in the future be sent by e-mailinstead, since e-mail is asynchronous (therecipient does not have to be present to re-ceive an e-mail message) and provides amore legible and reliable medium for mes-sages containing text or data. Similarly,users can easily copy text from letters,memos, and informal messages into newdocuments and search their contents elec-tronically, rather than visually scanning vo-luminous printed material.

when it is connected to a power distribution grid. Theterm is used in information science to refer to infor-mation and other resources being electronically ac-cessible to users by means of computers andcommunication devices. Similarly, it refers to usersbeing able to access their work resources electroni-cally, i.c., having terminals, communication facili-ties, computer accounts, etc., as needed to work inthis way. (Information that is not accessible in thisway, or users who do not have access to their workin this way are referred to as being "offline.") Theterm online as used in the database and lihrary do-mains is derivative and analogous but considerablynarrower. It is used here in its more general sense.

In the early stages of the online transi-tion, computation does not fully realize itspotential because it is not yet integrated intothe user's work style. This creates a chicken-and-egg problem. Users are not motivatedto use computation until its benefits out-weigh the cost of learning to use it (andchanging one's work style to make use ofit); but its benefits are realized only afterit becomes an integral part of one's workstyle. This problem produces a learningcurve in which progress initially is slow,but it accelerates as the online transitionproceeds. This curve rises steeply above acertain point, when a critical mass of theuser's context becomes integrated online.

Summary. The exact ways in whichcomputation will be delivered to end-usersin the future will be determined by factorsthat involve trade -offs among the costs ofcomputers, various kinds of memory andcommunicati,m, and issues of privacy,convenience, and control. The form in whichcomputation is delivered will continue toevolve as the relative costs and benefits ofvarious alternatives change. Ultimately, theend-user may not even knowand shouldnot carewhether the response to a requestis generated locally by the machine sittingon the user's desk, remotely by a special -purpose processor, or by some combinationof the two. The importance of the trendtoward end -user computing for researcherslies not in the details of its implementationbut rather in its potential to transformscholarly communication by making com-putation an integral part of the researcher'sthought process and work style.

Connectivity

The trend toward end-user computing isintimately related to the equally importanttrend of connectivity. This term describesthe researcher's ability to access data,processing capabilities, and other research-ers electronically in ways that facilitate theresearch process. Connectivity is a broader

17


concept than communication. Like com-munication, connectivity includes the abil-ity of computers to talk to each other andto access remote databases, but it also in-cludes the ability of researchers to worktogether in useful ways, to solicit feedbackfrom each other, to disseminate their ideasand results, and to integrate their researchsources and products. Connectivity re-quires communication, but it further as-sumes that information is in a usable formthat facilitates interchange and integration.

Many aspects of end-user computing relyon connectivity. The online transition re-quires that a sufficient critical mass of theuser's context be available online. That is,the various categories of data that comprisethis context (such as telephone messages,e-mail, memos, and documents) must allbe accessible electronically and must bestored in a common, interchangeable form,so that data can be shared and exchangedamong these different categories. Conven-tional wisdom recognizes that a critical massof users must be online before they willtruly benefit from their connectivity, but itis at least as important that a critical massof information and tools be online if usersare to reap the benefits of connectivity.Furthermore, convenient and effective in-terchange must be available across this crit-ical mass of information and tools before auser can profitably make the online tran-sition.

Access to databases also requires con-nectivity, especially if the user needs to seethe most up-to-date version of dynamic data.Access to dynamic data is particularly im-portant for research, where the most recentadditions to a database (representing newpublications, ideas, data, or research) areoften the most valuable, even though theymay change only a small fraction of theoverall database. If a database is static (i.e.,does not change very ofte), it can be cop-ied onto local systems, either by physicallysending disks to different sites or by down-loading data over a network (which again

requires connf-tivity). However, if a re-mote database is dynamic, a user can seethe most up-to-date version of the data onlyby either viewing the updated database overa network (relying on connectivity) or byupdating a local copy of the database ondemand (again, over a network) and view-ing the copy. Access to dynamic data there-fore depends on connectivity.

An infrastructure of connectivity allowscomputation to be performed and data tobe stored wherever it is most cost-effective,given that the relative costs of memory,computation, and communication are con-tinually changing. Connectivity allowscomputation and data to be reallocated fromlocal to remote resources (computers, disks,etc.) as these costs change. This realloca-tion has traditionally required physicalchanges to system configurations (such asmoving disk drives or rewiring buildingswith cables), but in principle this can bedone without physical intervention, re-sponding automatically to changing costsor shifting demands. Connectivity there-fore facilitates end-user computing by al-lowing it to take advantage of evolving costfactors.

The trend toward ubiquitous comput-ingwhether provided by portable com-puters, publicly available terminals, or otheralternativesrelies on a similar form ofconnectivity to link users to their working"office" contexts by remote or portableaccess. Ultimately, it will become irrele-vant whether a user's working context ex-ists in a single place or is distributed overa number of sites and machines. Connec-tivity will allow users to access their com-putational and informational contextswherever and whenever they need them.

Access to computational and humanresources. Although access to data andone's working context is the most obviousaspect of connectivity, it has other impli-cations as well. In general, connectivity al-lows users to access resources. These maybe data resources, but they may also be


specialized computational or human. re-sources. Two related initiatives'intended toencourage such interactions by providingwidely available, high-capacity networkingare the National Research and EducationNetwork (NREN) and the High Perform-ance Computing and Communications(HPCC) efforts. The capacity of a networkis measured by its bandwidth, which is thenumber of bits of information it can trans-mit per second.3' The NREN and HPCCefforts are targeted to produce gigabit (bil-lion-bit per second) transmission capacitiesduring the next decade.32 In addition toproviding high-capacity "backbone" com-munications, related initiatives include ef-forts aimed at integrating the communicationof text, images, voice, video, and othermedia. The NREN is intended to supportthe transmission of other media as well astext, although it should be noted that non-textual media require much greater trans-mission capacity. When fully implemented,NREN should greatly facilitate collabora-tion and resource sharing among research-ers.

Efforts such as NREN also are importantbecause, despite the evolution towardcheaper computers, there may always bestate-of-the-art computing facilities that re-main too costly for individual researchersto own. For example, large parallel com-puters may allow searching through hugedatabases for complex patterns, but the mostpowerful of such machines may always betoo expensive for any one researcher or evenany one research facility to justify theirpurchase. Connectivity will allow research-ers to share such facilities through remoteaccess.

Beyond access to machines, connectivityallows researchers to communicate and col-laborate with each other and with special-

"An average page of text consists of approximately20,000 bits, although this volume can he reduced(compressed) for transmittal.

'=Grant! Challenges, 17-19, 54.

ists in other fields. The vast web ofinterconnected networks (sometimes re-ferred to informally as "World Net") al-ready allows researchers to broadcast ordirect queries and requests by e-mail to alarge proportion of the researchers in a givenfield, regardless of their nationality or lo-cation. This process is not always directlycontrolled by the initiator of a request:Queries may be forwarded by their initialrecipients across networks and gatewaysbetween networks to individuals, electronicmailing lists, and electronic bulletinboards,33 eliciting responses from distantand unlikely places. Integrated networkingis greatly facilitated by an open systemsapproach, allowing multivendor softwareand hardware to communicate using stan-dard protocols. The International StandardsOrganization's Open Systems Interconnec-tion (OSI) reference model serves as a stan-dard for interconnection of this kind.34 Thesedevelopments are producing a truly globalcommunication capability, which is ex-panding rapidly and spontaneously.

The communication aspect of connectiv-ity goes beyond the use of e-mail for askingquestions or broadcasting general infor-mation. It is causing a major shift in theway many researchers collaborate and in-teract.35 The use of e-mail allows arbitrary

"Electronic bulletin boards arc analogues of theirphysical counterparts. They allow online users to re-motely view notices posted electronically by other users.

."The OSI reference model is discussed in detail inA. S. Tanenbaum, Computer Networks, 2d cd. (En-glewood Cliffs, N.J.: Prentice-Hall, 1988), 14-34.

"We are unaware of any research on c-mail useamong scholars, but for recent studies on the use ofe-mail and other collaborative electronic media in in-ternational organizations, see T. K. Bikson and S. A.Law, Electronic Mail Use at the Bank: A Survey andRecommendations (Washington, D.C.: Information,Technology, and Facilities Department, World Bank,September 1991); and Tora K. Bikson and Sally AnnLaw, "Electronic Information Media and RecordsManagement Methods: A Survey of Practices in UnitedNations Organizations," ACCTS Electronic Informa-tion Media and Records Management Survey Report,A RAND Note (N-3453-RC) (Santa Monica, Calif.:RAND Corporation, 1991).


text and data files to be transmitted in sim-ple, linear text formats, without concernfor machine compatibility or knowledge ofremote file systems. Researchers can gen-erally transform any relevant informationinto text and send it as the body of a mes-sage. Transforming formatted information(such as structured documents or page lay-outs) into linear text so that it can be ex-changed in this way requires that the senderand recipient have software capable of per-forming the appropriate transformations.Standards for transforming such informa-tion into linear text are evolving in re-sponse to this need. For example, theStandard Generalized Markup Languageoffers a standard textual representation forstructured documents, whereas Post-Script' offers a widely used de facto stan-dard textual representation for formattedpage images. Such standards already allowusers to send textually encoded documents,pictures, or formatted page layouts by e-mail instead of on paper. The e-mail recip-ient can view or print the transmitted in-formation after transforming it back to itsoriginal form. This capability will continueto improve as standards for graphics andother media evolve.

Connectivity also promises to' erase thegeography" that separates students fromteachers, classes, or other resources of in-terest. The educational notion of "distancelearning" has evolved from the correspon-dence course to the use of televised instruc-tion, but networking allows a much richerform of educational interaction. Particu-larly in upper-level scholarly subjects, it isnow possible to envision geographicallydistributed seminars that bring together in-terested scholars and students without re-gard to their physical locations.

The use of e-mail, teleconferencing, and

"'Adobe Systems, Inc., PostScript Language l?ef-e'en«. Manual Weadmg, Mass.: Addison WesIcy.1990),

remote windowing is producing a new phe-nomenon: computer-supported cooperativework (CSCW).37 Through CSCW, groupsof researchers can work together, sharingtheir context and coordinating their work,regardless of their locations, schedules, andwork styles. Connectivity allows coopera-tion in all phases of research, includingconcept formation, literature and back-ground search, analysis, publication, peerreview, and dissemination. This trend hasthe potential to both reduce the time re-quired to perform and publish research andimprove its quality through earlier and widerreview. CSCW also facilitates interdisci-plinary research through online discussionforums that are open to all interested par-ties, not just credentialed members of aparticular discipline. This openness makesit easier for researchers from different fieldsand institutions to collaborate, which maybroaden the perspective of scholarly com-munication. Finally, the trend toward shar-ing the research process may well changethe conception of the research product itselfinto something more multidimensional thana traditional document, allowing it to re-flect multiple views and opinions. (See thesection on hypertext and hypermedia laterin this paper.) Note that the implicationsexplored here are not derived from tech-nological determinism: The technology it-self does not produce such changes. Rather,the changes result from the trend towardsharing and collaborating, which the tech-nology facilitates.

The trend toward interchange stan-dards. True connectivity involves the abil-ity to interchange information, whichrequires that information be represented ina standard form. The relative youth of in-formation science as a field and the rapidevolution of computers and communication

"For an excellent annotated bibliography of currentwork in CSCW, see Saul Greenberg, "An AnnotatedBibliography of Computer Supported CooperativeWork," Sla:111 Bulletin 23 (July 1991): 29-62.

2 0


technology have produced chaotic alterna-tives for representing and communicatinginformation. This may be unavoidable in afield in which technology and paradigmsare still evolving. By their very nature, novelideas do not always fit into previous pat-terns. Similarly, new computational capa-bilities often produce new informationstructures that do not easily translate intoexisting standard forms. Furthermore, thedevelopment of new standards is a slowprocess because it requires compromise andconsensus. The development of standardsis therefore a difficult undertaking, and theytend to lag behind the latest technologicaladvances. Nevertheless, the growing em-phasis on interchange standards is a vitaland worthy trend, without which the prom-ise of connectivity cannot be realized.

Standards are beginning to evolve for text(as discussed in the section on Computer-Assisted Analysis Achieved Through Con-version), and ultimately they will extend tographics, voice, three-dimensional model-ing, animation, video, and other media aswell. In the early stages of this process, thegoal is to develop usable initial standardsquickly, without precluding their extensionand modification in the future. This trendtoward extensible standards is motivated bya recognition of the inevitable lag bet Neenstandards and technological advance. De-veloping such extensible standards is a ma-jor technical challenge, involving asignificant effort to translate among differ-ent standards and different versions ofevolving standards. Ideally, such transla-tion will minimize the need for the user tobe aware of the underlying standards, andinexpensive computation will providetransparent translation among standardswithout user intervention.

In addition to interchange standards, atrend is developing toward defining stan-dards and policies for privacy and author-ization of access. As collaboration becomesmore common, it will become increasinglyimportant for researchers to be able to pro-

BEST COPY AVAILABLE21

tect their data, analysis, and results. Pla-giarism, theft, tampering, and sabotage willundermine the advances of connectivity iftechnical, administrative, and legal solu-tions to these problems are not imple-mented. Even the computation andcollaboration processes themselves must beprotected from unauthorized auditing andanalysis. Various agencies or individualscould easily misuse or abuse knowledge ofthe kinds of questions a researcher asks andthe thought processes involved in formu-lating research. The trend toward increas-ing interest in privacy and security issuesis evidenced in a number of recent confer-ences and publications.38

A false dichotomy: distributed versuscentralized control. One of the most in-triguing implications of the trend towardconnectivity is its potential to redefine themeaning of control over intellectual arti-facts. In particular, the traditional dichot-omy between distributed and centralizedcontrol may no longer be appropriate. Thisdichotomy is based on the natural but out-dated assumption that control is a functionof location in the physical world. Tradi-tionally, a resource has been considered tobe under centralized control if it exists inonly one physical location and is main-tained by agents residing at that location.Conversely, a resource is considered to beunder distributed (decentralized) control ifit consists of multiple copies or parts thatare dispersed among multiple locations and

"Computers, Freedom and Privacy Conference,sponsored by Computer Professionals for Social Re-sponsibility, San Francisco Marriott, Burlingame, Calif.25-28 March 1991; The National Conference onComputing and Values (NCCV), held at ResearchCenter on Computing and Society, Southern Con-necticut State University, New Haven, Conn. 12-16August 1991; and the seventh Annual Computer Se-curity Applications Conference, sponsored by Aero-space Computer Security Associates and AmericanSociety for Industrial Security, and the Associationfor Computer Machinery and the Institute of Electricaland Electronics Engineers, St. Anthony's Hotel, SanAntonio, Tex., 2-6 December 1991.


maintained by agents dispersed among thoselocations. This dichotomy applies reason-ably well to physical resources, but it failsto work for resources created by electronicconnectivity.

The physical location of a resource haslittle meaning in the electronic domain.Connectivity allows resources to be repli-cated and distributed among numerousphysical locations while behaving as thoughthey existed in only one location (and viceversa). The key to this phenomenon is theseparation between an electronic resource'sphysical location and its availability: A da-tabase may reside on a storage device inone location while being viewed or modi-fied via a terminal in another location. Sim-ilarly, a database that appears to exist inonly one location may actually consist ofpieces distributed and replicated among nu-merous locations and may he viewed ormodified by numerous agents via com-puters at different locations. This charac-teristic is the definition of connectivity:Ac -ctss becomes independent of location.The notions of centralized and decentral-ized (distributed) control simply do not ap-ply in this context. New forms of controland policies for when to employ themarelikely to evolve as connectivity replacesphysical access to resources.

Summary. End-user computing andconnectivity have been discussed sepa-rately here for expository reasons, but theirfull impact lies in their mutual synergy.Connectivity elevates end-user computingabove simple word processing or calcula-tion by allowing end-users to access remotedatabases, share information in many dif-ferent media and forms, connect to theirworking contexts wherever they are, com-municate with their peers, and collaboratein all phases of research. End-user com-puting in turn provides one of the main mo-tivations for improving connectivity:Networks do not connect machines, theyconnect people. The combined trends of end-user computing and increasing connectivity

will shape the evolution of research (alongwith many other endeavors) well into thenext century.

Specific Technology Trends AffectingScholarly Communication

The major trends of end-user computingand connectivity will manifest themselvesin many ways. This section identifies anumber of specific technology trends thatwill superimpose themselves over thisbackground. Each subsection discusses anarea of technology that is expected to havea particular impact on research. Althoughnot exhaustive, this examination includessome of the technology that are likely tohave the greatest influence over the nextdecade, i.e., artificial intelligence, end-userpublication and distribution, hypermedia,visualization, and virtual reality.

Artificial intelligence. Current trends inartificial intelligence (Al) have the poten-tial to affect scholarly research in a numberof ways. Al may provide intelligent aidsfor analyzing and interpreting sources; au-tomated "agents" that can help researchersstay abreast of new findings; and tools tohelp formulate research coi,:-.:pts. AI mayalso enable researchers to model their sub-ject areas to test hypotheses. Finally, Alhas the capacity to produce intelligent tu-tors that may help researchers leverage theirteaching skills.

The recent commercial success of expertsystems (and more generally, knowledge-based systems) has brought AI out of theivory tower where it had evolved since theearly days of computing. A number of gen-eral-purpose programming languages andenvironments (expert system shells) forbuilding expert systems have appeared onthe market, allowing users with little or noformal training in Al to take advantage ofsome of the most common AI techniques.Yet AI encompasses much more than justexpert or knowledge-based systems. As oneof the frontiers of computing, it attempts

22


to find ways of using computers to solveproblems they cannot now solve. AI isdriven by dual motivations that sometimesconflict with and sometimes enhance eachother. The first of these, which can bethought of as a "modeling" motivation,seeks to use computers to model and un-derstand intelligence. The second, whichcan be thought of as an "engineering" mo-tivation, simply seeks to solve difficultproblems, by whatever means. AI effortsthat are motivated by modeling tend to fo-cus on defining intelligence, understandingcognitive processes, and addressing prob-lems whose solutions are acknowledged torequire intelligence. AI efforts motivatedby engineering simply try to solve difficult,worthwhile problems, using any availabletechniques, regardless of whether the tech-niques simulate human intelligence.

Because of these dual motivations andbecause AI is a frontier (and therefore nec-essarily dynamic and evolving), it tends toinclude many disparate activities and tech-nology, ranging from the automation offormal mathematical logic to the design ofartificial neural networks. Several themesrun through AI, such as representingknowledge, language, and meaning andfinding relevant patterns or solutions amonglarge, complex sets of alternatives. The pri-mary influences of AI on scholarly com-munication are l'kely to be its ability toanalyze linguistic and pictorial informa-tion, its ability to find patterns, its abilityto create automated "agents" that act on auser's behalf, and its ability to model real-ity and formulate concepts.

The bulk of scholarly data is currently intextual form, and text will undoubtedlycontinue to be the major target of scholarlyresearch for some time. Other forms of data,such as visual imagery (including draw-ings, paintings, photographs of sites or ar-tifacts, holograms, and film and video),spoken language, sounds, and music may,however, play greater roles as the technol-ogy for their encoding and analysis im-

23

proves. AI software's growing ability tounderstand the semantics (and eventuallythe pragmatics) of language and to analyzerelationships and identify patterns will makeit an increasingly attractive tool for per-forming scholarly analysis. In addition, AIhas developed a number of techniques fordealing with beliefs and uncertain, contra-dictory, or hypothetical information, whichmay help researchers who must often gen-erate hypotheses and rely on contradictoryor uncertain conclusions and beliefs in or-der to find patterns and relationships. Cou-pled with growing databases of encoded textand fast processing, these techniques willenable researchers to look for new, unex-pected patterns across a wide range of sub-ject areas. Similar capabilities eventuallywill extend to visual imagery and sound,allowing integrated analyses of text, speech,music, and pictorial data. Although it willprobably be some time before AI will becapable of truly understanding literarytext39and even longer before it will becapable of understanding spoken languageor visual imageryit is already capable offiltering large bodies of text to find literaryaspects or relationships that are of partic-ular interest to a researcher. In this role,AI will not replace the analytic insight ofthe researcher, but it will enhance the re-searcher's ability to scan large collectionsof information and find patterns worthy ofanalysis.

One of the major emphases of AI re-search has been to develop intelligent agentsthat can behave autonomously on behalf oftheir users. Robots (which are still largelyexperimental) are the most dramatic ex-amples of such agents, but another class ofagents is more relevant to scholarly re-search. These are informational agents, suchas literature-search or SDI (selective dis-

"See Nancy NI. Ide and Jean Veronis, "ArtificialIntelligence and the Study of Literary Narrative," Po-elks 19 (1990): 37-63.


semination of information) agents, whichcan search for information of interest to aresearcher, using criteria specified in a formsimilar to a database query. Such agentsultimately may perform a number of serv-ices, such as translating a researcher's queryinto the form required by particular data-bases; periodically repeating a query orsearch; monitoring activity on a network orin a database and alerting the user when"interesting" events occur; soliciting, col-lecting, and filtering information from manysources; responding to routine requests fromother researchers for information or to othercorrespondence; and coordinating theschedules and activities of a collection ofresearchers engaged in collaborative effort.Such agents will eventually take over manyof the traditional activities of a secretary:They will make up for their relative lackof initiative and creativity by being tireless,dedicated, and inexpensive.4°

In addition to its role in the analytic phaseof research, AI may have an impact on theconcept formation that leads to research. Inthis earliest conceptualization phase, re-searchers often generate informal hy-potheses about a subject area, in an attemptto define interesting research thrusts. Anumber of tools currently emerging from"knowledge acquisition" efforts in AI havethe potential to help identify viable hy-potheses and useful concepts. These con-cept-formation tools help the user formconcepts by asking questions that can dis-criminate between examples and counter-examples of an evolving concept, based onattributes that the user declares as definingthe concept. For example, a researcher mightattempt to define a concept such as "ado-lescent imagery" in a body of text in terms

"For research on intelligent agents, see Robert E.Kahn and Vinton G. Cerf, Ari Open Architecture fora Digital Library System and A Plan for its PetriNnnent, The Digital Library Project, Volume I: TheWorld of Knowbots (Washington, D.C.: Corporationfor National Research Initiatives, March 1988).

of attributes such as age, immaturity, andsexual embarrassment. A concept forma-tion tool might attempt to find examples ofsuch images, asking the user to rate eachcandidate passage according to each attrib-ute. Based on these ratings, the tool mightthen show which of these passages appearto be examples of the concept and whichones appear to be counterexamples, therebyhelping the user form a consistent and use-ful definition of the desired concept.

Much of AI research focuses on model-ing. In order to act intelligently or solvecomplex problems, Al systems often createmodels of reality about which they can rea-son or which they can manipulate in orderto decide how to act in the real world. Tra-ditional simulation and mathematical mod-eling techniques are severely limited in thetypes of questions they can answer. Sim-ulation users, for example, typically spec-ify the initial state of a simulated world andthen run the simulation to see what hap-pens. This "toy duck" view of modeling("wind it up and see where it goes") cor-responds to asking questions of the form"what if . . . ?" (i.e., what would happenif the world were to proceed from this giveninitial state?). This ability to ask "what if. . . ?" questions is often touted as the ul-timate analytic capability, but many otherkinds of questions are at least as importantin many situations.'" These include suchquestions as: Why did some agent take aparticular action? Why did a given eventhappen? Can a particular event ever hap-pen? Under what conditions will a givenevent happen? Which events might lead toa particular event? How can a desired resultbe achieved? Ongoing AI research in this

"M. Davis, S. Roscnschein, and N. Shapiro, Pros-pects and Problems for a General Modeling Meth-odology (Santa Monica, Calif.: The RANDCorporation, N-1801-RC, June 1982); and J. Roth-enberg, "The Nature of Modeling," in Artificial In-telligence, Simulation, and Modeling, edited by L.Widman, K. Loparo, and N. Nielsen, 75-92 (NewYork: John Wiley & Sons, August 1989).

24


area is producing powerful new techniquesfor modeling intentions, causality, goals,beliefs, and other phenomena to allow an-swering questions that go beyond "what if. . .

This trend toward model-based systemswill provide researchers with techniques forconducting experiments, evaluating hy-potheses, and exploring alternative inter-pretations of reality with minimal cost andrisk (since they are carried out within acomputer). As a simple example, sociolog-ical or cultural models could be built toexplore alternative hypotheses about an an-cient civilization, using the model to makepredictions that can be compared with his-torical evidence. Al techniques such as thesemay help researchers conceptualize re-search as well as perform analyses.

The modeling capabilities of AI are alsothe key to its use in education. Intelligenttutors are an outgrowth of joint research ineducation ancrAl; typically; they involve amodel of the subject matter to be taught (adomain model) and a model of the student.The domain model elevates an intelligenttutor above the level of simple programmedinstruction because it enables the tutor toanswer unanticipated questions about thesubject matter. Students can therefore aska much wider range of questions and pur-sue many alternative paths of instruction.Similarly, the student model helps the tutordetermine which concepts the student ishaving trouble understanding. This helpsthe tutor address the student's underlyingproblem rather than simply repeating newmaterial or backing up blindly to reviewprevious material. Although intelligent tu-tors are still largely experimental, they ap-pear to hold great promise for improving

"See J. Rothenberg, "Using Causality as the Basisfor Dynamic Models," in Proceedings of the Thirdinternational Working Conference on Dynamic Mo-delling of Information Systems (DYNMOD-3) (Delft,The Netherlands: Delft University of Technology,1992), 277-92.

the educational process, particularly forstudents who are self-motivated and self-paced. Ultimately, this should allow schol-ars to leverage their teaching skills by de-veloping tutors that embody their expertise.

In summary, current trends in artificialintelligence may affect scholarly researchby

providing analysis aids that can helpfind and interpret relevant source data,text, and other media.creating informational agents that canperform some of the routine tasks ofkeeping abreast of new findings, act-ing as tireless monitors of develop-ments in a field.providing tools to help researchers ex-plore, formulate, and refine researchconcepts and hypotheses.enabling researchers to model theirsubject areas to try out hypotheses andpredict where to find confirming (orfalsifying) evidence.facilitating the development of intel-ligent tutors that can help researchersdisseminate their knowledge andteaching skills to wider audiences.

Since AI is one of the frontiers of infor-mation science, it is also not unlikely thatadditional developments in this field willhave unforeseen consequences for the ev-olution of scholarly research.

End-user publication and distribu-tion. An equally important though less ex-otic computing trend is the grow .ng abilityof end-users to publish and distribute theirown work. This is already creating alter-natives to traditional publication in schol-arly journals, not only reducing the time ittakes to publish research but, more impor-tantly, changing the channels of distribu-tion, redefining the review process, andtransforming dissemination by means ofelectronic connectivity.

The most prosaic form of end-user pub-lication is the production of camera-readyprinted documents, suitable for publicationor reproduction and dissemination without


further typesetting or layout work (some-times referred to as "desktop publishing").Even this simple modernization of the tra-ditional publication process has profoundimplications. As with all forms of end-usercomputing, end-user publication involvesthe author of a document much more di-rectly in its production. Because of thisavailability of layout and production toolsduring the draft phase, a document ap-proaches its final form at an earlier stageof development. For example, figures,footnotes, and final formatting can be in-corporated into early drafts, giving review-ers a more readable product and helping toeliminate errors and, in general, to improvethe product. Ideally, the author's controlover questions of typography, graphics, andlayout means that the final document rep-resents a more accurate and integrated rt -flection of the author's overall intent. Thecorresponding disadvantage is that authorsmust learn new publication skills, for whichthey may have little inclination, patience,or talent. Of course, end-user publicationdoes not preclude the use of secretaries,graphic artists, or publication specialists toreintroduce traditional expertise in the pub-lication process, but this intervention tendsto subvert the advantages of end-user pub-lication by slowing the process and reduc-ing the author's control.

Beyond modernizing the traditional pub-lication process, end-user publication al-lows authors to publish their workelectronically, bypassing the production anddistribution of paper documents entirely.Electronic documents can easily reproducemost of the desirable attributes of paper,and they provide increased flexibility forcorrection, revision, access, and dissemi-nation. During the production phase of adocument, these features facilitate remotecollaboration and early review and theygreatly simplify the revision process. End-user publication also facilitates a radicallydifferent view of the research process, inwhich ideas are disseminated for review and

feedback in the earliest stages of research,i.e., prior to documenting or even perform-ing he research. (Examples of this are dis-cus:Jed later in this paper.)

Electronic dissemination makes use ofincreasing connectivity to bypass tradi-tional distribution channels, reduce the costof reproduction and mailing, and enable re-cipients of a document to redistribute it byforwarding it in electronic form.43 Thecopyright and other legal implications ofelectronic dissemination are only beginningto be explored. Similarly, direct, online az-cess to the source of a documen*. makes iteasier than ever to plagiarize ideas, text,and even complex graphics without leavingany trace. These problems must be ad-dressed by technical, legal, administrative,and, ultimately, cultural policies. Such pol-icies are likely to evolve more slowly thanthe technology they seek to civilize, leav-ing a gap between practice and policy forat least the next decade or two; this gap ispart of the cost of the technological revo-lution of scholarly research.

Hypertext and hypermedia. All re-search studies must explicitly or implicitlyaddress a number of questions that rep-resent different dimensions of inquiry, suchas What is the problem? What assumptionswere made about the problem? What re-lated research exists? What is original aboutthe study? What methodologies were con-sidered? What approach or method waschosen, and why? What sources and datawere used? What analysis was performed?What were the results? How she lid the re-sults be interpreted? 'What other interpre-

°Computers and networks arc being used in thecommercial sector as well, both to help automate theprocess of publishiag traditional books and journalsand to develop novel electronic products. This elec-tronic publishing industry has so far had little impacton end-user publication, but it may be too soon to tellwhether this industry will ultimately attempt (or man-age) tin appropriate aIii commercialize the new than-nels distribution and dissemination th. end-usersare cui critly developing for themselves.

26


tations are worth considering? How do theresults and interpretations depend on theresearcher's assumptions? What are the im-plications of the research? It is difficult toanswer all such questions without inundat-ing and confusing the reader.

Similarly, presenting complex subjectmatter to students requires answering anal-ogous questions about context, backgroundhistory, alternative approaches or formu-lations, and relationships to other disci-plines. Traditional textbooks and otherinstructional materials seldom address theseissues adequately.

Such questions are inherently interre-lated and multidimensional. Answering themin a strictly sequential, linear fashion is oftenconstraining and unrevealing. Yet writtendocuments necessarily present their argu-ments linearly. In addition, an expositorysequence that provides insight to one readeror audience may not be enlightening to an-other. Cross-references, references to otherdocuments, repetition, overviews, andsummaries can ameliorate these problems,but only at the cost of redundancy and addedwork for the reader (flipping pages to findcross-references or consulting other docu-ments). Furthermore, documents, which arcinherently static, are hard-pressed to por-tray processes or other dynamic phenom-ena. The effectiveness of graphics issimilarly limited by the static nature of theprinted image. Oral presentations can beless linear than documents, can be tailoredto specific audiences, and are better suitedto presenting dynamic phenomena, but theyare ephemeral and cannot provide the depthof the printed word.

Electronic information technology prom-ises to transcend these limitations by deliv-ering research results in an interactive,electronic form that is nonlinear and mul-tidimensional and that integrates written,spoken, and graphic media in a permanent,dynamic, customizable presentation. Theterms hypertext and hypermedia suggest thenovel characteristics of this new approach:

27

1. It provides rich, dynamic linkagesamong the elements of a presenta-tion. For example, using electronicretrieval and display, a reference fromone item of text to another (whethera cross-reference, a bibliographic en-try, or a citation in another work) canbe viewed instantly in a windowwithout the user's having to turn pagesor find another document. Such linkscan be used to present different di-mensions of analysis, alternative se-quences of exposition, optionaldegrees of elaboration or depth, sup-porting evidence, references, data, orcontextual background. The multidi-mensional nature of such structures isdenoted by the prefix "hyper." Au-thors can use this linking to presentdifferent kinds of information or todefine alternative paths that generatedifferent presentations or variants froma single master document.

2. Hypermedia combines several mediathat currently can be presented elec-tronically, such as text, color graph-ics, and sound (including voice).These can all be linked together aseasily as text, producing presenta-tions that combine the features ofdocuments and oral presentations.

3 These media can be presented dy-namically. This allows animatinggraphics, synchronizing voice withanimation to describe processes, andcontrolling the pace of a presentation,as in an oral briefing.

4. This approach is interactive, allowingthe reader to control the sequence,speed, cicpth, and focus of the pre-sentation, within limits set by the au-thor.

The concept of a nonlinear document"

"Although hypertext and hypermedia products arcvery different front Itaditional documents, they arcgenerally referred to as "documents" for want of abetter word.


can be traced bael-. at least as far as theseminal paper "As We May Think," byVannevar Bush, in 1945." The electronicimplementation of this concept is begin-ning to transform the traditional notion ofa document into a multimedia, nonlinearform of presentation. The publication of re-search results in hypermedia form may makethem more accessible and more captivat-ing, thereby greatly increasing the impactand influence of research, particularly out-side the traditional scholarly community.The result may be greater public recogni-tion of policy issues identified by re-searchsuch as the need to preserve historicsites or artifactsin much the same waythat popular televised documentaries haveincreased public awareness of myriad sci-entific, cultural, and environmental issues.Furthermore, the use of hypermedia maytransform the research process itself byproviding a natural way to represent andkeep track of interrelated facts, references,hypotheses, and arguments, as well as re-actions, revisions, and annotations to sup-port collaboration. Finally, hypermedia maytransform educational material into a new,multidimensional experience that will cap-italize on the exploratory tendencies ofscholarly students.

Visualization and virtual reality. Re-cent trends in visualization and virtual real-ity have the potential to transform the wayscholarly researchers interact with their dataand perform their analyses. The world ofscientific computing has begun to developtechniques that allow scientists to visualizethe results of complex computations.Graphic techniques and animation are beingused to display complex data in ways thatattempt to make significant patterns leapout at the user. Abstract relationships areoften easier to grasp if they are translatedinto graphical presentations, such as false-

"Vannevar Rush, As We May Think," AtlanticMonthly 176 (July 1945): 101-08.

color maps, cluster plots, or adjacencygraphs. These techniques apply equally toany field in which complex data, patterns,and relationships must be understood. Manyareas of scholarly communication may profitfrom this technology by visualizing quan-titative or qualitative data to gain insightinto its meaning or to present complex re-sults in a perspicuous form.

Though it is typically viewed as a verydifferent trend, the technology of virtualreality is closely related to visualization. Avirtual reality is a simulated world createdin a computer, using traditional simulationor AI modeling techniques such as thosediscussed above. The user "enters" a vir-tual reality by wearing a display helmet orgoggles to create the visual illusion of beingin the simulated world (e.g., showing dif-ferent views as the user's head turns). Theuser interacts with the virtual reality bywearing devices such as instrumented glovesor suits that sense the user's hand or bodyposition, thereby allowing the simulatedworld to react. The result is something likean intensified video game, in which the userperceives the virtual reality and interactswith it for some purpose.

The power of virtual reality is that it har-nesses the user's full sensory and motorcapabilities in exploring an abstract world,rather than relying on more limited facul-ties such as reading and typing. Coupledwith modeling and visualization, this hasthe potential to allow a researcher to inter-act intimately with a virtual world createdout of data or analytic results and to explorethis world in a much more direct, exper-iential way than would be possible by read-ing numbers or even by viewing a graphicaldisplay. In addition to its potential fortransforming certain aspects of the analyticprocess, virtual reality technology might alsobe of use during concept formation (allow-ing researchers to explore abstract spacesof concepts, represented as visual worlds)or for bringing the education of scholarlysubjects to life (allowing students to ex-

26


perience subject matter as a virtual world).Virtual reality may also be viewed as a log-ical extension of hypermedia, in which re-search results may be presented as a virtualworld to be explored, rather than as a doc-ument to be seen or heard.

Caveats. The trends described herein arenot without their dangers. 1 he legal issuessurrounding electronic dissemination andconnectivity have been pointed out above,as have some of the possible violations ofprivacy that result from working in an open,networked environmen'. Every technolog-ical advance has its own risk for misuse,whether this risk is legal, ethical, or merelya matter of lost productivity and quality.For example, the indiscriminate use of end-user publication and distribution may by-pass carefully established mechanisms foreditorial and peer review, leading to a pro-liferation of low quality, unprofessionalpublications. Similarly, the use of hyper-media by authors who are not trained ingraphic design or media presentation mayproduce a flood of incoherent researchproducts whose complexity makes them in-accessible to their intended audiences. Thenaive use of modeling toois, visualizationtechniques, and virtual reality may seduceresearchers into believing results that seemcompelling despite the fact that they havenot been validated. Researchers and audi-ences alike may tend to accept conclusionsbased on state-of-the-art computations, suchas AI, with less than the required skepti-cism, especially if these computations ex-hibit a veneer of intelligence.

These dangers are real and may well pla-gue scholarly researchers for decades tocome, as they adopt new methods empow-ered by technology. Nevertheless, thesetrends appear inevitable and are likely tochange the form and substance of scholarlycommunication in fundamental ways.Whether this change will ultimately im-prove the quality of that research is a ver-dict that only the future can deliver.

Summary

The availability of quantitative data andnumerical techniques for analyzing themhave had a marked effect on scholarly com-munication over the past several decades.The technology trends discussed here, aswell as others that may prove to be impor-tant, are likely to have an even more pro-found impact. This impact will do morethan simply change the work styles ofscholarly researchers: It will affect theirthought processes as well, suggesting newkinds of research questions and new kindsof answers. It will change the way re-searchers collaborate and interact with theirpeers and the way they produce their re-sults. It will change the form of these re-sults, the way they are distributed anddisseminated, their audiences, and the im-pact they have on the research communityand the public. These changes, already un-der way, will have profound implicationsfor the information services, libraries, andarchives that serve the research process.

SCHOLARLY COMMUNICATIONAND THE USE OF CURRENTINFORMATION TECHNOLOGY

The previous section explored key trendsin information technology most relevant toscholarly communication. This sectionconsiders the use of currently available in-formation technology by social science andhumanities scholars to advance scholarshipand intellectual productivity. The use oftechnology across the full spectrum ofscholarly communication is considered byexamining how researchers rely on tech-nology to: (1) identify sources, (2) com-municate with colleagues, (3) interpret andanalyze data, (4) disseminate research find-ings, and (5) develop curriculums and aidinstruction. Case examples of scholarlypractices illustrate broader tendencies within

29


the field." For analytical purposes, the im-plications of these practices for archivaladministration are discussed in the finalsection (Conclusion and Recommenda-tions) of this report. Although the discus-sion focuses primarily on practices in thesocial sciences and humanities, the emerg-ing patterns exhibited in these professionsmirror those found in a broad range of dis-ciplines and occupations.47

The old assumptions commonly sharedby archivists and librarians about the re-search process characterize a decreasingsegment of the scholarly community." In-stead, a paradigm shift is occurring in theresearch styles of social scientists and hu-manists, as in the scientific community,where: electronic communication is gain-ing prominence; direct online searching isreplacing intermediary searching; researchcollaborations are becoming more com-mon; electronic sources available in homesand offices are becoming an alternative toreading room visits; source materials orig-

' ^The authors do not endorse particular techniques,uses of technology, or the validity of results reportedin the case examples. Because empirical data onscholarly use of information technology does not ex-ist, this section relies on case examples intended tobe illustrative of broader trends within the social sci-ences and humanities. Academic computing officers,however, arc beginning to recognize the need for suchdata, and some have expressed interest in conductingcampuswide or intcrcampus surveys on scholarly useof technology.

"For a discussion of the impact of informationtechnology on the research process of intelligence an-alysts see Michael R. Leavitt, The Analyst and Tech -nology -2000, prepared for the U.S. IntelligenceResearch and Development Council (January 1991).

"'new ling such assumptions as: patrons discoversource materials essentially through word of mouthand through supplemental assistance by intermedi-aries; humanities and social science scholars conductresearch basically as individuals; primary sources, bynature, require viewing in reading rooms fortified withprofessional assistance; primary sources are best storedand viewed in their original form or on microfilm; thequalitative methods used in the analysis o: sourcestypically preclude computation; and the standardscholarly products (e.g., publications) arc linear doc-uments distributed in print form.

inally created in print are being convertedto machine-readable form; standard schol-arly research practices are extending to theuse of artificial intelligence to interpret andanalyze materials; and electronic publish-ing and nonlinear technology, such as hy-permedia, are prompting the developmentof new forms of scholarly research prod-ucts. The following explores how scholarlycommunication practices among social sci-entists and humanists are changing as a re-sult of the use of currently availableinformation technology.

Identification of Sources

According to the professional literature,the key way scholars learn about relevantresearch materials is through their col-leagues. But in the last few years, word ofmouth has been supplemented by new formsof electronic searching through online pub-lic access catalogs (OPACs). For instance,most campuses provide academics with di-rect access via personal computers to theinstitution's online library catalog.49 In-stead of visiting the library, researchers cannow explore descriptions of the library'sholdings from their offices. Furthermore,if the institution's catalog proves insuffi-cient, scholars can access more than twohundred major American library catalogs,including those of the universities of Cali-fornia, Michigan, Pennsylvania, and Wis-consin, via the Internet.5° For a

comprehensive search of this nation's li-brary holdings, the Research Libraries In-

"See Clifford A. Lynch, "Library Automation andthe National Research Network," EDUCOM Review24 (Fall 1989): 22; and Communications in Supportof Science and Engineering, 1-7.

"Conversation between Avra Michelson and PaulPeters, executive director of the Coalition for Net-worked Information, 7 May 1991. Clifford Lynch,director of library automation for the University ofCalifornia, reports that as many as 30 percent of thelog-ons to the universitywide MELVYL library cata-log arc from remote sites.

30


formation Network (RLIN) is available overthe Internet, and plans are under way ton e OCLC available on the Internet aswe_, .51 Humanist scholars report that, byproviding a comprehensive means to browsethrough libraries in their homes and officeswhen convenient, direct access to biblio-graphic databasef. represents a source of in-tellectual empowerment.52 The use of onlinecatalogs probably represents the mostwidespread example of scholarly practicesin the social sciences and humanities thatir.volve end-user computing and connectiv-ity.

Communication with Colleagues

The search for sources and the need torefine intellectual ideas motivate academicsto communicate with their colleagues. In-deed, communication of this sort is fun-damental to the advancement of scholarship.Beyond the most common methods of com-munication (such as face-to-face discus-sions, telephone conversations, writtencorrespondence, or public presentations)scholars are using e-mail and a variety ofnew electronic communication formats de-rived from it for academic interchange.Scholars naturally still talk to one another,but many information exchanges occurthrough network communications rather thanthrough oral discourse.53 E-mail exchangesare growing at an astonishing rate, and cur-

"See Clifford A. Lynch, "The Growth of Com-puter Networks: A Status Report," Bulletin of theAmerican Society for Information Science 16 (June/July 1990): 10; and Robert Webcr, "Libraries With-out Walls?" Publishers Weekly 237 (8 June 1990):S20S22.

"Stephen Lehmann and Patricia Renfro, "Human-ists and Electronic Information Services: Acceptanceand Resistance," College and Research Libraries 52(September 1991): 411.

"Many argue that although less interactive, for briefexchanges e-mail is a far deeper medium for com-munication than oral discourse. Unfortunately, therearc currently no studies on the nature and extent ofthe use of e-mail among scholars, but its significanceas a new communication medium is indisputable.

rently constitute approximately half thetraffic on research and education net-works.54 The global spread of e-mail hasbeen rapid, and it is now possible forAmerican scholars to communicate via e-mail with colleagues in close to 140 othercountries. The popularity of e-mail amongscholars emphasizes the increasing impor-tance of network connectivity in the dailylife of academics.55

As an outcome of e-mail, scholars arecreating new formats for substantive ex-change to supplement conventional com-munication. For example, nearly thirtythousand public-access electronic bulletinboards are currently available through re-search and education networks. This is upfrom fourteen thousand such applicationscounted one year earlier.56 BITNET, a net-work developed during the mid-1980s toprovide rapid communication among re-searchers, educational institutions, andfunding agencies, reports more than twothousand listservs. Listservs are discussiongroups that allow people with common in-terests to communicate with one another bysending to a special network address mailthat is automatically distributed to eachperson who has subscribed to a particularlist.57

"Presentation by Paul Peters to a joint meeting ofthe National Association of Govrnment Archives andRecords Administrators Comi....tee on InformationTechnology, and the SAA Committee on AutomatedRecords and Techniques, Washington, D.C., 22 April1992.

"The EDUCOM/USC Survey of Desktop Comput-ing in Higher Education estimates that 25 percent (ex-trapolated figure) of faculty at four year public andprivate universities and colleges use e-mail. See Greenand Eastman, Campus Computing 1990, 23. See thetwo studies by Tora K. Bikson and Sally Ann Lawcited in the previous section for studies on the use ofe-mail within an office environment.

"Christopher Lindquist, "Ferret Lovers Unite andDownload," Computetworld 25 (12 August 1991): 1.

"Theodore J. Hull, SNX.4 Reference Report, Cen-ter for Electronic Records, National Archives andRecords Administration (June 1991 draft), 2; and EricThomas, Revised List Processor (Listserv(t frecpl 1),Release 1.5d, Ecolc Centrale de Paris, from Lis-



Of the thousands of electronic discussiongroups, or conferences, operating on theInternet, close to 600 are devoted to schol-arly topics in the social sciences and hu-manities." The rate of growth of thesescholarly electronic conferences is aston-ishing. From 1990 to 1991, 200 new con-ferences were identified on the Internet. Forthe eight months prior to March 1992, anadditional 150 conferences in the social sci-ences and humanities were added to the ex-isting directory of listings." Scholars haveestablished conferences in virtually everyfield within every discipline. For example,there are conferences on topics such as Hel-lenic culture, folklore, modern British andIrish literature, the Vietnam War, and eigh-teenth-century world history. There areconferences devoted to the study 'f coun-tries or regions, such as Peru, Iberia, LatinAmerica, and the Baltic states. There areconferences on the works of single authors,such as James Joyce, John Milton, ThomasPynchon, and Hegel, and there are confer-ences devoted to concepts such as libertar-ianism, intuition in decision making, ethics,and fraud in science.°

The Humanist, an electronic conferenceestablished several years ago serves as afocal point for discussions of humanitiescomputing techniques and research meth-

[email protected] (13 June 1991 17:55:18),1.

"Diane Kovacs, Directory of Scholarly ElectronicConferences, 3rd ed. (Kent State, Ohio: Kent StateUniversity Libraries, August 1991), [available on Bit-net/Internet at ListserveLkentvm or FTP fromksuvxa.kent.cdu. The directory, an indispensable,growing resource, is also available in print as Direc-tory of Electronic Journals, Newsletters and Aca-demic Discussion Lists by Michael Strangelove andDiane Kovacs, edited by Ann Okerson, 2nd cd.(Washington, D.C.: Association of Research Librar-ies, March 1992). The conference figures cited reflectupdated information that Diane Kovacs was kindenough to share with Avra Michelson.

"'These figures may undcrrepresent actual scholarlyactivity, as Kovacs warns that the directory's cover-age of Usenet is less than comprehensive.

"For a description of these conferences, sec theKovacs directory cited earlier.

ods. It also broadcasts announcements andincludes a column for ongoing queries andresponses that cover a broad range of issuesof interest to humanist scholars. The Hu-manist is transmitted daily to about twothousand readers, including subscribers inEurope and the Near East.61 A Britishcounterpart, Humanities Online Bulletin,operates as a forum for humanists to ex-change experience, solicit advice and in-formation, notify one another of projects,review publications, and make announce-ments. The almost thirteen hundred regis-tered readers are mostly members ofhumanities departments in British univer-sities.62 These electronic discussion groupsserve a unique role in scholarly communi-cation in that they permit the rapid inter-change of current information, ideas, andperspectives. No other medium has per-mitted scholars to communicate with an in-ternational group of peers quickly andeffortlessly at the front end of the researchprocess. (The scholarly implications of thenew exchange mediums are examined fur-ther in the section Dissemination of Re-search Findings later in this report.)

Interpretation and Analysis of Sources

The use of information technology to as-sist in interpreting and analyzing data rep-resents one of the most important paradigmshifts toward end-user computing in schol-arly research practices. Scholars are bothconverting primary textual sources to ma-chine-readable form to allow for conven-tional computational processing and usingartificial intelligence to do new types ofmachine-assisted interpretation and analy-

6,Elaine Brennan and Allen Rencar are the currentco-editors of the Humanist. Information on the Hu-manist from a telephone conversation between AvraMichelson and Allen Renear, 17 December 1990, andfrom th.: description of the conference that appears inthe Kovacs directory.

6213rendan Loughridge, "Information Technology,the Humanities and the Library," Journal of Infor-mation Science 15 (JulySeptember 1989): 280.


sis. The use of computing to perform inter-pretation and analysis is a developmentaltrend with broad implications for scholar-ship. These practices suggest fundamentalchanges in scholarly methods, and each willbe examined in depth.

Computer-assisted analysis achievedthrough conversion. Social scientists andhumanities scholars use both quantitativeand qualitative methods to analyze and in-terpret sources. Typically, the search forand evaluation of evidence involves bothtypes of methods. At one end of the con-tinuum, quantitative analysis involves theuse of mathematical processes such as acount of frequencies and distributions ofoccurrences, or higher level statistical tech-niques. At the other end of the continuum,qualitative analysis typically involves non-mathematical processes oriented towardlanguage, interpretation, or the building oftheory.63

Scholarly analysis often involves theprocessing of large and sometimes massiveamounts of textual sources.6" But research-ers have discovered that many of the meth-ods of interpretation and analysis related toboth quantitative and qualitative methodsare processes that can be performed bycomputers. For example, computers cancount (e.g., they can count words, births,deaths, marriages, commercial activity, andeven brush strokes used in a Rembrandtpainting). Computers can perform regres-sion analysis to suggest cause and effectrelationships. Through the use of advancedtechnology, computers can perform patternrecognition, do semantic analysis, analyzetext, and model concepts. And computerscan perform these processes faster, overmore sources, and with greater precision

"Nigel G. Fielding and Raymond M. Lcc, eds.,Using Computers in Qualitative Research (London:Sage Publications, 1991), 4.

"The use of nontextual sources of evidence, suchas photographs, film footage, artifacts, and sound re-cordings is significant as well.

than scholars who must rely on manualinterpretation of data.

But if computers are to be used for thesepurposes, source materials must be in ma-chine-readable form. For this reason, manyscholars, once they have identified the keysources for their research, are convertingthem to machine-readable form so that theyare in a form amenable to computer-as-sisted analysis.°

Scholarly conversion of sources to ma-chine-readable form has been occurring forat least forty years. At first the practice wasgenerally limited to numeric data. But inmore recent years, the scholarly appetitefor machine-readable data has extended totext as well. Textual conversion projectsundertaken by individual scholars or underthe auspices of academic institutions are farmore prevalent than one might expect, es-pecially in the fields of linguistics, classics,religion, and even history. The Center forElectronic Text in the Humanities estimatesthat there are currently eight thousand se-ries of converted electronic text.66 The con-version efforts among scholars are anexample of the manifestation of end-usercomputing, in an effort to store, retrieve,manipulate, and analyze large amounts ofsources in electronic form. The availability

"See Avra Michelson, "Forecasting the Use ofNREN by Humanities Scholars," paper presented atthe panel "New Constituencies for the NREN," 27March 1992, National NET '92, Washington, D.C.Available electronically on the Coalition for Net-worked Information filescrver.

Contact [email protected] for transfer information."'Conversations between Avra Michelson and Mar-

ianne Gaunt, Center for Electronic Texts in the Hu-manities, Rutgers University, on 30 October 1990,and 14 May 1991. Rutgers and Princeton universitiesrecently announced the creation of the jointly spon-sored Center for Electronic Texts in the Humanitiesto respond to the information needs of a new gener-ation of scholars. The center will develop an inter-national inventory of machine-readable textual sourcematerials, provide catalog entries through the Rc-scarch Libraries Information Network (RLIN), and ul-timately make electronic textual source materialsavailable to researchers on research and education net-works.

33


of an electronic corpus of sources on a topicencourages new types of questions to beasked and hypotheses to be explored.

The conversion of paper-based textualsource materials to machine-readable formoccurs worldwide. The earliest Americanconversion project, the Thesaurus LinguaeGraecae (TLG), was founded in 1972 byTheodore F. Brunner at the University ofCalifornia at Irvine to create an electronicdata bank of extani ancient Greek texts fromthe period of Homer (ca. 750 B.C.) throughabout A.D. 600. The massive electronic fileis used by researchers in Greek languageand literature, linguistics, ancient history,philosophy, and religious studies to accessGreek texts and related documents in fulltext. In conjunction with the AmericanPhilological Association, many membersof the classicist profession participate in theongoing compilation. Today the TLG is animmense, growing database of more thaneight thousand works of classical Greek lit-erature stored on CD-ROM, copies of whichare available at two hundred locations inthis country and abroad.°

Another conversion effort, the Americanand French Research on the Treasury of theFrench Language (ARTFL) draws on thework of the French government since 1957to create a new dictionary of the Frenchlanguage. In conjunction wi'h the devel-opment of the dictionary, the French de-veloped an electronic database ofapproximately 150 million words derivedfrom major literary and philosophical worksand scientific and technical texts. For in-stance, the auxiliary database contains thenovels of prominent and popular authors,correspondence, literary criticism, an ex-tensive collection of poetry and theater,

"For information on TLG, see Theodore F. Brun-ncr, "Data Banks for the Humanities: Learning fromThesaurus Linguae Graceac," Scholarly Communi-cation 7 (Winter 1987): 1, 6-9; and David S. Miall's"Introduction," in Humanities and the Computer: NewDirections (Oxford: Clarendon Press, 1990), 5.

travelogues, biographies, historical works,political documents, biblical commentary,philosophical and economic essays, andwritings on biology.

In 1979, the National Endowment for theHumanities (NEH) granted funds to theUniversity of Chicago to conduct a survey'of North American French literary scholarsand historians whose work focused on theeighteenth to twentieth centuries. The pur-pose of the survey was to evaluate the po-tential usefulness of the ARTFL databaseto their work. Based on the scholars' en-dorsement, France deposited the corpus offifteen hundred machine-readable texts atthe University of Chicago in 1982. Afterthe database was restructured to allow fortext analysis, the electronic materials weremade available to researchers. As an on-going project at the University of Chicago,scholars continue to augment the databasewith, for example, a collection of trouba-dour poetry estimated to include 65 percentof the genre's extant poems; a collectionof texts from the 1848 revolution, includ-ing radical newspaper articles, pamphlets,posters, speeches, and manifestos by pro-letarian leaders; and a collection of sev-enteenth-century French theater pieces.

A variety of scholars use the ARTFL,including Keith Baker, a University of Chi-cago historian of ideas. Baker's researchconcerns the attempt to redefine traditionalterms during the Enlightenment to conformwith the new political and social order. Ac-cording to Baker, the "advantage of theARTFL Project is that it provides a broadbasis for systematic analysis of . . . keyterms,"68 such as the occurrence of im-portant political phrases like `!opinion pub-lique" in eighteenth-century texts.69 Other

'Alice Musick McLean, Robert Morrissey, andDonald A. Ziff, "ARTFL: A New Tool for FrenchStudies," Scholarly Communication 8 (Spring 1987):8.

"For another example of research devoted to thehistorical analysis of language, sec Mark Olsen and


instructors have used ARTFL to teachFrench nationalism or to study the literarymyth of Charlemagne as recorded from themiddle ages to the nineteenth century.ARTFL is available online to scholars andstudents at institutions that participate in aninter-university fee-based consortium.'

A third large file, the Medieval andModern Data Bank (MEMDB) was foundedin 1982 at Rutgers University by RudolphM. Bell and Martha C. Howell to establishan electronic library for medieval and earlymodern historians. The data bank .consistsof text descriptions of currency exchangerates, including a master data set of tabularworks concerning medieval and early mod-ern history. More than thirteen thousandmedieval currency exchange quotations fromthe mid-twelfth century to 1500 A.D. areavailable, covering Europe, Byzantium, theLevant, and North Africa. MEMDB is agrowing database; plans for expansion in-clude adding taxation records, wills and in-ventories, parish records, vital statistics,comnany records, import/export records,household/estate accounts, palacopathol-ogy studies, and such reference aids asglossaries of weights and measures, gaz-etteers of Latin and vernacular place names,and calendars of dates. The Research Li-braries Group (RLG) is pc-Taring a CD-ROM version of MEMDB for release.'

Louis-Georges Harvey, "Computers in IntellectualHistory: Lexical Statistics and the Analysis of Polit-ical Discourse," Journal of Interdisciplinary History18 (Winter 1988): 449-64.

"'McLean, et.al., "ARTFL," 1, 6-9."Information reported in a phone conversation by

Marianne Gaunt, the Center for Electronic Texts inthe Humanities, Rutgers University to Avra Michel-son 14 August 1991. Sec also Loughridgc, "Infor-mation Technology," 281; and Rudolph M. Bell(Rutgers University), "User Perspectives and Re-quirements: Creator of Non-bibliographic DatabasesHas to Share with Others," unpublished paper pre-sented to the Library of Congress Network AdvisoryCommittee Meeting, 29-31 March 1989, Washing-ton, D.C.: also information reported by Rudolph Bellto Avra Michelson in a phone conversation 13 No-vember 1991.

The TLG, ARTFL, and MEMDB rep-resent discipline-specific electronic com-pilations, but many smaller and often morediverse humanities conversion projects alsoexist.72 For instance, under the direction ofRobert Hollander at Princeton University,the Dante Project converted to electronicform the complete text of sixty commen-taries on Dante in Italian, Latin, and Eng-lish. Before the Dante conversion project,many of these works were unavailable inthe United States.73 The purpose of Vic-toria Kirkham's Penn Boccaccio Project atthe University of Pennsylvania is to de-velop an electronic archives that establisheslinks between the author's writings and theseven thousand illustrations of his work thatwere created contemporary to his lifetimein the fourteenth century through the six-teenth century.74

Several archival conversion projects areunder way in England. For example, theBrotherton Library is compiling a completedatabase of its seventeenth and eighteenthcentury manuscript verse. The Universityof York History Department initiated a jointeffort with the York Archaeological Trustboth to develop a computerized database ofthe town's title deeds and to create a re-construction of the region's topographicalevolution between the twelfth and sixteenthcenturies. At the University of Southamp-ton, scholars are developing an online da-tabase of the papers of the first Duke ofWellington."

At Bar-Ilan University in Israel, YaacovChoueka is constructing a Jewish culture

"Georgetown University's Center for Text andTechnology has compiled a database of descriptionsof more than three hundred conversion projects, manyof which comprise hundreds of series.

"Constance Gould, Information Needs in the Hu.inanities: An Assessment, Prepared for the Programfor Research Information Management of the Re-search Libraries Group, Inc., Stanford, Calif.: 1988,27.

"Ibid., 27."Loughridgc, "Information Technology," 281.

35

BEST COPY AVAILABLE


database, the Global Jewish Database/Res-ponsa Project. This database includes aboutfifty thousand rabbinical answers to ques-tions about Jewish life and culture, the Ba-bylonian Talmud, Midrash literature,medieval commentaries, Maimonides Code,and the full text of the Hebrew Bible. Whencompleted, the database will contain fulltext of nearly all written Hebrew works upto the tenth century as well as about onethousand major sources on Jewish cul-ture.76 At the University of Pennsylvania,Robert Kraft and John Abercrombie, work-ing in conjunction with the Packard Hu-manities Institute, issued a CD-ROMcontaining at least ten versions of the Bibleas well as a dictionary of New TestamentGreek, classical Latin texts, Greek inscrip-tions, and various texts including Sanskritsources.''

England's Oxford Text Archive (OTA)is a large repository of machine-readabletext and includes text bases in more thantwenty-five languages. Recently it servedas a key source for a dissertation on JaneAusten's novels.78 In Pisa, Italy, the Isti-tuto di Linguistica Computazionale, one ofthe oldest and largest repositories of ma-chine-readable classical and modern texts,has converted an extensive variety of ma-terials, including Italian newspapers andperiodicals, modern novels and poetry, andworks of nonfiction." Similarly, the Brit-ish Domesday project assembles a varietyof textual and visual information on con-temporary Great Britain."

"Gould, Information Needs in the Humanities,39,and Loughridge,"Information Technology," 280.

"Gould, Information Needs in the Humanities, 38."The Center for Electronic Texts is cataloging the

records of the OTA. The catalog is made possiblethrough an NEIL grant, and the center is describingthe approximately eight hundred records that comprisethe OTA and making the description: available throughRLIN. For information on the OTA, sec Miall, ed.,"Introduction," in Humanities and the Computer, 5,and Gould, Information Needs in the Humanities, 27.

"Gould, Information Needs in the Humanitin, 27."'Mall, "Introduction," in Humanities and the

Computer. New Directions, 5.

Other conversion efforts involve do-mains such as italian Renaissance musicand lyric poetry, Spanish texts, medievalmedicine-related drawings and illustra-tions, the papers of Charles Sanders Peirce,81and the works of literary greats such asShakespeare, Shelley, Faulkner, and Mil-ton.82 Besides these institution-based con-version efforts, hundreds of smaller projectsare addressing the needs of particular teamsof researchers. Humanities scholars predictthat the millions of words of text alreadyavailable in machine-readable form rep-resent only a minute fraction of source ma-terials to be converted in the next ten tofifteen years.83 Scholars contend that thereuse of textual databases by those otherthan the original converters will soonifit does not alreadyconstitute the predom-inant use.

In an effort to compile a massive elec-tronic text corpus that will serve as a com-prehensive research resource, languagescholars have initiated the Data CollectionInitiative (DCI). Sponsored by the Asso-ciation for Computational Linguistics, the

"This new effort involves a consortium that in-cludes the current documentary editing project on Peircecentered at Indiana University-Purdue University,working with the philosophy departments at Harvardand Texas Tech universities, Georgetown Universi-ty's Center for Text and Technology, Brown Univer-sity's Computing and Information Services, and GeorgeWashington Universit's Department of Communi-cation. The consortium plans to convert Peirce's largeprint manuscript collection housed at the HoughtonLibrary, along with secondary commentaries on Peirce'swork, to machine-readable form. The database wouldalso include provisions for electronic scholarly com-munication on the vastly interdisciplinary work ofPeirce.

"References appear on the database of electronictexts compiled by Georgetown University's Center forText and Technology.

"Association for Computers and the Humanities,the Association for Computational Linguistics, andthe Association for Literary and Linguistic Comput-ing, "Proposal for Funding for an Initiative to For-mulate Guidelines for the Encoding and Interchangeof Machine-Readable Text," unpublished proposalprepared for the National Endowment for the Human-ities, 1988, 12.

36


DCI is the most extensive international col-laboration of its kind. The ultimate goal ofthe project is to develop a global electroniclibrary of text available for online research,primarily to serve the needs of computa-tional linguists. Coordinated by Mark Lib-erman at the University of Pennsylvania'sDepartment of Linguistics, the DCI in-cludes a broad sample of materials, suchas the archives of the Challenger investi-gation commission, which constitutes about2.5 million words of deposition and hear-ings transcripts; portions of the Library ofAmerica volumes; 200,000 U.S. Department of Energy scientific abstracts; U.S.Department of Agriculture Extension Ser-vice fact sheets; the Federalist Papers; theKing James Bible; computing journals; ansample correspondence and dictionaries."

Besides acquiring a large corpus of elec-tronic text, scholars are developing encod-ing standards for documents, to ensure thatconverted files can be read on a variety ofcomputers and software. The Text Encod-ing Initiative (TEI) is a collaboration amongthe Association for Computers and the Hu-manities, the Association for Computa-tional Linguistics, and the Association forLiterary and Linguistic Computing, whichreceived funding from the National Endow-ment for the Humanities (NEH), the Eu-ropean Economic Community, and theMellon Foundation to determine the ele-ments and the methods for encoding ma-chine-readable text for electronicexchange." The first phase of funding is

"Information on the DCI from phone conversationsbetween Avra Michelson and Don Walker, Bellcore(14 May 1990), and Mark Liberman, AT&T Bell Lab-oratories (5 June 1990); sec also Mark Liberman,"Report to the ACL Executive Committee on the ACl/DCI," (5 June 1990).

"Some scholars consider questions of what to en-code as serious a concern as how to encode. For in-stance, should encoding indicate the physical conditionof a document by marking the presence of ink spots,water stains, brittleness of paper, etc.?

BEST COPY AVAILABLE

devoted to the needs of literary, linguistic,and text-oriented historical research."

The TEI encoding standards closely fol-low the International Standards Organiza-tion's standard ISO 8879, the StandardGeneralized Markup Language (SGML).This interchange format specifies how toencode (or mark up) texts so that they canbe shared in a machine- and software-in-dependent form by different research proj-ects for different purposes. The TEIencoding standards use delimiters and tagsto distinguish markup from text and to ex-press specific information about the formatof a document.87 A draft version of the TEIstandards is circulating to scholars and in-dustry for review.88

The extraordinary projects under way byscholars to convert source materials to ma-chine-readable form, assemble an elec-tronic corpus of textual data, and establishdata format standards for the interchangeof text are in essence efforts aimed at fa-cilitating end-user computer-assisted analy-sis of sources within the social sciences andhumanities.

Computer-assisted analysis with arti-ficial intelligence. Some scholars are con-

"Association for Computers and the Humanities etal., Proposal for Funding, 59.

"SGML is a standard set of instructions for com-posing machine-readable tag sets and grammars. SGMLapplications, such as the TEI guidelines, establish tagsand delimiters for the interchange of all types of text,including rules for encoding many types of documentstructures and data elements. The encoding allowscomputers, using appropriate software, to "read" thestructure of a document (e.g., to know that an an-thology of poems contains individual poems and thateach possesses a title, stanzas, and lines), and to pres-ent it as such to the user; for further explanation ofSGML, see C. M. Sperberg-McQueen and Lou Bur-nard, eds. Guidelines for the Encoding and Inter-change ofMachine-Readable Texts (Chicago, Oxford:Text Encoding Initiative Version 1.1, October 1990).

"Scholars in Europe have formed the HistoryWorking Party, a subgroup of the Text Encoding In-itiative to ensure that TEl encoding guidelines addressthe needs of historians (e-mail via Internet from Don-ald A. Spaeth to Avra Michelson, 9 August 1991).

37


verting records to machine-readable formso that artificial intelligence (AI) can beused to assist in data interpretation andanalysis. The use of AI in scholarly re-search signals a new phase in social scienceand humanities end-user computing.89 Asearly as 1986, a panel of specialists broughttogether by the National Science Founda-tion reported that AI methods held greatpromise for research in the social sciences,especially in relation to the analysis andinterpretation of complex situations, re-search design, and theory formation."Within the humanities, scholars contend thatthe ability to process incomplete and in-consistent data with software that supportsuncertainty and changes in beliefs makesAI uniquely suitable for many research ef-forts.91

In the area of Al, political scientists cur-rently are the most sophisticated experi-menters outside the hard sciences. Theirprominence with AI calls to mind their ear-lier role as the pioneer users of computa-tional processing with electronic numericdata. They are using artificial intelligence,especially in the area of international rela-tions, to model decision making for the studyof "deterrence, escalation control and wartermination."92 The applications involve thechoices defense programs and military op-erations confront during peace, as well asmethods for evaluating choices during a

"See, for instance, Miall, Humanities and theComputer, 2; or for an earlier discussion, E. Casettiet. al., "Regarding the Feasibility and Desirability ofConferences on 'The Methodological Research Fron-tiers and the Social Sciences,' " Final Report to theNational Science Foundation (NSF Award No.: OIR8406230), 10 September 1986, 13.

9`)Casetti, et al., "Regarding the Feasibility and De-sirability of Conferences," 13.

"Miall, Humanities and the Computer, 6."'See Paul K. Davis, "A New Analytic Technique

for the Study of Deterrence. Escalation Control, andWar Termination," in Artificial Intelligence and Na-tumal Security, edited by Stephen J. Cimbala (Lex-

inn, Mass.: Lexington Books, 1987), 35-60.

conflict. They explicitly address compli-cations that decision riikers face, such asconflicting principles and objectives, ill-defined alternatives, the complexity ofproblems, and the pervasive uncertainty ofassumptions. As a result of intensive work,some existing prototypes are evolving intomore advanced applications.93

Besides the "conflict-oriented" proj-ects, other examples include AI prototypesthat interpret Sino-Soviet negotiating ses-sions," recognize patterns over a large,complex data set of historical events forpurposes of prediction,95 and generate hy-potheses by exploring data to induce rules.One application of this last type analyzesthe factors that influence different satisfac-tion levels of state legislators with legisla-tive outcomes. The developers contend thatthe existing application can be adapted foruse with similar research questions.96

The discipline's innovators argue that AItechniques should be considered "standardcomponents in every political scientist's toolkit."" In making their case, they arguethat many foreign policy questions rep-resent suitable AI applications, such as thedegree to which the Soviet economy de-clined under Brezhnev or the impact of thedevelopment of a navy on China's foreignpolicy.98 One of the discipline's journals,

"Ibid., 35-55."Sec William dcB. Mills, "Rule-Based Analysis

of Sino-Sovict Negotiations," Social Science Com-puter Review c 8 (Summer 1990): 181-95.

c5Philip A. Schrodt, "Pattern- Matching. Set Pre-diction, and Foreign Policy Analysis," in ArtificialIntelligence and Nationai Security, 89-107.

9'G. David Garson, "The Role of Inductive ExpertSystems Generators in the Social Science ResearchProcess," Social Science Microcomputer Review 5

(Spring 1987): 11-18.dcB. Mills, "Rule-Based Analysis," 182;

and Paul A. Anderson, "Using Artificial Intelligenceto Understand Decision Making in Foreign Affairs:The Problem of Finding An Appropriate Technol-ogy," in Artificial Intelligence and National Security,133.

"Sec also, for instance, the t:.1 or so articles inArtificial Intelligence and National Security.


Social Science Computing Review, keepsreaders current on AI software with regularreviews of expert systems shells.

Unlike-political scientists, most histori-ans using AI tend to apply it to a narrowerrange of research questions. The chief useof AI in historical research is in applica-tions designed to build nominal record link-ages to reconstruct the population historyof past societies." This technique is usu-ally used with family and community re-construction, an area of study already quitecomputer-oriented. Nominal record linkageinvolves the analysis of parish and census-like records to reconstruct individual ident-ities and relationships among individuals.It is a complex process that requires muchinterpretation because of the prevalence ofhomonic names (multiple names, with thesame sound and often the same spelling,which refer to different people), name var-iations, and the need to link evidence re-lated to the same individual from separaterecords. Historians typically consider anindividual's vital dates, residence, profes-sion, filiation, and other available data todecide whether several pieces of evidencerefer to the same person.

Nominal record linkages typically in-volve analysis of a large and diverse set ofrecords. Once the records of an individualhave been linked, then a similar processmust be performed to link the records offamilies and, ultimately, of communities.Historians are finding, however, that AI canbe used to perform some interpretations as-sociated with the task. For example, in

"See Kevin Schurer, "Artificial Intelligence andthe Historian, Prospects and Possibilities" in Inter-pretation in the Humanities: Perspectives from Arti-ficial Intelligence, Library and Information ResearchReport no. 71, edited by Richard Ennals and Jean-Claude Gardin, 169-95 (Cambridge: Cambridge Uni-versity Press, 1990); and Joaquim Carvaiho, "ExpertSystems and Community Reconstruction Studies,"History and Computing II, edited by Peter Denley ct.al. (Manchester: Manchester University Press, 1989),97-102.

France, historians at the Institut de Re-cherche et d'Histoire des Textes are usingexpert systems technology to identify un-ambiguously individuals, based on thir-teenth and fourteenth century parishregisters.'°° Similarly, the Cambridge Groupfor the History of Population and SocialStructures has been using artificial intelli-gence for both nominal records linkage andto disambiguate household relationships. Inthe Cambridge project, AI performs someof the rudimentary aspects of analysis butstill leaves the hard questions of interpre-tation to the historians. Kevin Schurer, amember of the group, describes it this way:

The study of history should bedriven by theory rather than fact. AItechniques may help historians to ex-amine the relationship between factsmore closely, and may add to the un-derstanding upon which interpreta-tions are made, yet they can neveract as a substitute. In the examplesgiven, expert systems may help us todetermine the degree of householdcomplexity in the past, or the levelsof fertility. They may "positively"identify that females married on av-erage at age 24 and had a completedfamily size of between five and sixat the beginning of the 19th century,compared to an average age of 26 anda completed family size of aroundthree at the end of the century, yet itis the task of the historian to theorizewhy this transition occurred.1°'

Although the primary use of AI amonghistorians has been to reconstruct kinship

"Caroline Bourlet and Jean -Luc Mincl, "A Dec-larative System for Setting Up a Prosopographical Da-tabase," in History and Computing, edited by PeterDenley and Dcian Hopkin (M,Inchester, England:Manchester University Press, 1987), 190.

"Schurer, "Artificial Intelligence and the Histo-rian," 190.

39


and community relationships, other uses alsoare being explored. For instance, Frenchsocial historian Beatrice Henin developeda computer file of leasehold documents cre-ated by notaries and property inventoriestaken at the time of death to study seven-teenth-century Marseilles. Toward the endof her research, Henin became interestedin the interior decor of houses from differ-ent social classes. Her use of artificial in-telligence to analyze textual descriptions ofpictures on the walls of rooms, largely withreligious themes, led he develop a newmodel for understai. .Ag Protestant andCatholic families in seventeenth-centuryEngland. 1°2

Another European effort, the RESEDAProject, uses AI to respond to historicalquestions from a biographical database ofFrench public and private figures during thefourteenth and fifteenth centuries. In ad-dition to biographical information, the da-tabase contains abstract data aboutindividuals, such as their beliefs, inten-tions, opinions, and mental attitudes. Usinga hypotheses template, the system sorts theinformation to discover relevant facts, andinfers information from the data to answerquestions involving conjecture.m

In yet another type of project a scholaris using AI to extend the findings of Tzve-tan Todorov's The Conquest of America:The Question of the Other (1985). Todo-rov's work concerns the use of "signs andcommunication (and failed communica-tions) within the cultural encounter.sl04 Jim

"Richard Ennals, Artificial Intelligence: Applica-tions to Logical Reasoning and Historical Research(Chichester, England: Ellis Horwood Limited, 1985),125.

"Gian Piero Zarri, "Artificial Intelligence and In-formation Retrieval: A Look at the RESEDA Proj-ect," in The Analysis of Meaning: Informatics 5, editedby Maxine MacCafferty and Kathleen Gray (London:Queens College Oxford, 1979), 166-72.

"Jim Doran, "A Distributed Artificial IntelligenceReading of Todorov's The Conquest of America: TheQuestion of the Other, by Tzvctan Todorov, 1985,"in Ennals and Gardin, Interpretation in the Humani-ties, 166.

Doran at the University of Essex uses AIto add another dimension to Todorov'sanalysis by analyzing the belief systems andtheir impact on the behavior of the key per-sons and cultural groups examined in To-dorov's book. Using evidence for beliefsalready embedded in Todorov's work, Doranfurthers the analysis by systematically ex-amining the relationship between the be-liefs and the conquest of America. This effortsuggests one way in which scholars are ex-ploring the use of artificial intelligence toextend an existing analysis of source ma-terials. It uses AI to examile the relation-ship between reasoning and beliefs, tocategorize "faulty" belief systems, and toconsider metabeliefsbeliefs about be-liefs.1°5

In the field of history, the principal in-vestigators using AI in their research tendto be credentialed as historians, not as com-puter scientists. There is, however, an in-teresting exception. Kenneth L. Jones is anavid avocational genealogist who works withthe Cartographics Application Group at theJet Propulsion Laboratory (JPL) in Pasa-dena, California. As a hobby, Jones beganusing his AI background to unravel hisfamily genealogy. The system he devel-oped was fairly comparable to those al-ready described: It provides records linkagesby disambiguating individuals, families, andgeopolitical boundaries. But the applica-tion's level of sophistication caught the at-tention of the American intelligencecommunity. In developing the system, Jonesproduced a form of knowledge represen-tation (the depiction of knowledge as sym-bols in a form that a computer canmanipulate), which he refers to as "knowl-edge visualization." Knowledge visuali-zation entails the use of graphics to clarifyor make more intelligible the relationshipsamong interrelated fragments of knowl-edge. Conferring with colleagues at the JPL,

"Ibid.

40

272 American Archivist / Spring 199'

Jor.es realized that the intricate matrices hewas developing to assist in family researchcould be applied to any problem that in-volves the conceptualization of complex in-terrelationships among objects, such astracking money-laundering or counter-ter-rorism activities. Jones's work on this sys-tem continues, with funding from the U.S.Army's Joint Tactical Fusion Office.'

Besides research-oriented applications,serious efforts are under way to use soft-ware engineering to develop a scholarlyworkstation devoted to the needs of histo-rians. Manfred Thal ler is a historian andkey participant in the Historical Worksta-tion Project sponsored by the Max-Planck-Institut fur Geschichte in Gottingen, Ger-many, an institute dedicated to fundamen-tal research in the humanities.1 °' Since 1978,the institute's research has been designedto improve software for historians. Theworkstation project focuses on the devel-opment of three components: software thatcan access information from both currentand historical sources, databases that arc asavailable and easy to use as books, andknowledge bases that allow the other com-ponents to draw upon information in his-torical reference works. The developers planto use artificial intelligence to providetransparent interaction between subsys-tems, to create new rules in the knowledgebases when new facts are inferred, and toguide users to relevant information. Var-ious elements of a production prototype ofthe workstation are being tested. Some arestill under development, and some of themore difficult aspects of context-sensitiveinterpretation are still in the design phase.

Among sociologists, Edward Brent re-

"'From a presentation made by Kenneth L. Jonesat the Eighth Annual Intelligence Community Al'Ad-vanced Computing Symposium, Greenbelt, Mary-land, 12 March 1991.

"See Manfred Thalia, "The I listorical Workstalion Project," unpublished paper delivered at the sev-enteenth Imemational Congress of Historical Sciences,Madrid 29 August 1990.

fers to the current era as "the first hint ofwhat it might be like to have computers thatact less like clerks and more like col-leagues."108 His remarks pertain to the earlybenefits sociologists report in using AI fortheory development, especially to differ-entiate dependent variables from indepen-dent variables, to develop theories based oncausal models, and to extend sociologicaltheory by transforming theoretical asser-tions into logical ones. Sociologists are de-veloping applications using artificialintelligence for these purposes.")" In the fieldof literature, scholars are using natural lan-guage understanding for the rapid disam-biguation of words stored in machine-readable dictionaries and, within limiteddomains, to comprehend the "meaning" ofa story. in other literary uses, expert sys-tems have been developed that log and ana-lyze differing interpretations of text amongreaders.'m

Scholars are beginning to use artificialintelligence as a tool to assist in the inter-pretation and analysis of sources in nearlyevery corner of the social sciences and thehumanities."' In addition to those men-tioned, researchers in the fields of archae-ology, linguistics, music, art history, anddesign are exploring the value of "intelli-

""Edward Brent, "Is There a Role for ArtificialIntelligence in Sociological Theorizing?" AmericanSociologist 19 (Summer 1988): 164.

""'Ibid., 160-64." "'See for instan e, Nancy M. Ide and Jean Ve-

ronis, "Very Large Neural Networks for Word SenseDisambiguation," paper presented at European Con-ference on Artificial Intelligence, Stockholm, August1990; Nancy M. Ide and Jean Veronis, "ArtificialIntelligence and the Study of Literary Narrative," Po-etics 19 (1990): 37-63; and David !Alan, "An ExpertSystem Approach to the Interpretation of LiteraryStructure," in Ennals and Cardin, Interpretation inthe Humanities, 196-214.

"'The Foundation for Intelligent Systems in theSocial Sciences, Arts and Humanities is a new organ-ization that publishes a quarterly newsletter, Mte lhgent Systems. devoted to applications in thesedisciplines. For further information, contact the foun-dation's directot, Stephen Toney, at 2205 GabriclDrive, Las Vegas, Nevada 89119.

41


gent" tools, such as expert systems shellsand specialized software, capable of per-forming functions attractive to a variety ofdisciplines."' During this decade, as pri-mary sources become more available inmachine-readable form and as commercialAI software becomes more sophisticated andprevalent, it is likely that scholars will turnincreasingly to AI for research assistance.

Dissemination of Research Findings

The scholarly obligation to report re-search findings is typically fulfilled throughthe publication of articles in peer-reviewedprint journals or monographs. Until re-cently, the defining feature of a publicationwas its linear and printed format. But theemergence of electronic publishing and hy-permedia are challenging this definition ofa document. The scholarly use of electronicpublishing and hypermedia is a result of thedual trends toward end-user computing andgreater connectivity. Considered together,these new dissemination and presentationformats are beginning to transform themanner in which findings arc shared in thescholarly community.

Electronic publishing. Introduced lessthan a decade ago, electronic publishing al-ready represents a $6.5 billion business ac-

"For further information on shells, see AvraMichelson, Expert Systems Technology and Its Impli-cations for Archives, National Archives Technical In-formation Paper no. 9 (Washington, D.C.: NationalArchives and Records Administration, March 1991),9-10. An example of specialized software is Ex-Samplc which helps researchers determine an appropriatesample size for a study. Ex-Sample is reviewed inEdwin H. Carpenter and Rick D. Axelson, "Statis-tical and Graphical Research Methods: State of theArt," In Social Science Computer Review 7 (Winter1989), 508. Another example, IXL's Discovery Ma-chine, performs patternmatching over large amountsof data that typically would go undetected throughmanual analysis. For a report on its use, see KarenD. Schwartz, "Agencies Use Software to Dig Up LinksAmong Data," Government Computer News 19 (15October 1990): 60.

cording to current estimates.113 The mostviable commercial electronic publishing ef-forts involve indexing and abstracting textsand electronic versions of full-text printjournals. Through electronic publishing, itis increasingly possible for researchers toaccess on their computers full-text versionsof "newspapers and newswires, popularmagazines and scholarly journals, financialand directory sources, and referencebooks. 114 For example, electronic ver-sions of more than forty medical journalsare available in full text, as are some of themost important scientific and technicaljournals 115 More than three hundred full-text newsletters can be accessed througheither NewsNet or Dialog files. Businessand industry periodicals enjoy wide cov-erage in electronic form, as do specializedtitles like marketing reports."6 Unlike bib-liographic databases developed primarily foruse by information specialists, full-text da-tabases generally are designed for the end-user. Researchers, enthusiastic about theconvenience of these databases, also findelectronic publishing attractive because itpromises to increase the pace of publicationand expand opportunities for dialogue amongscholars.

An electronic resource directory createdby Bibliofile, Fulltext Sources Online,identifies more than fifteen hundred full-text and information sources available on-

"'Council on Library Resources, Communicationsin Support of Seim, . and Engineering, Report to theNational Science Foundation. Washington, D.C.:Council on Library Resources, August 1990. 11-8.

"'Sec Ruth A. Pagel!, "Primary FTDBs for theEnd User: New Roles for the Information Profes-sional," Onlini, Review 13 (April 1989): 143.

"'Ibid., 146. The hunt Library at Carnegie MellonUniversity is compiling an electronic full-text corpusof extended runs of computer science journals on ar-tificial intelligence. The specific journals and runs arccited in a subsequent section of this paper (sec "TheLibrary Profession's Response to New Forms ofScholarship! Software Engineering" section).

w'Pagell, "Primary ETD% for the End User," 143--46.


line."7 The trend watchers in the industryestimate that by the year 2000 much ofscholarly and profe';sional publishing willoccur electronically, involving the trans-mittal of journals and books over high-speednetworks by authors to the publishers, andthen from publishers readers."8

Further, publishers are discovering thatthe electronic versions of certain printedproducts are beginning to turn a profit. In-deed, Harry Boyle of Chemical AbstractsService (CAS), one of the world's largestindexing and abstracting companies, de-scribes the shift occurring in his companyin this way:

The revenue base for the printedproduct is shrinking. The revenue basefor the electronic product is growing.Fifteen years ago the printed productwas paying the bills. In the next fiveyears, the electronic form of theproduct will be the dominant way thatthe database is used and the printedwill become secondary. We are rap-idly approaching the point where theelectronic use of the product is in factgenerating a lot of the revenue neededto build the database, and the printedproduct is becoming the secondaryconcern. I don't think we will stopthe printed product. But if you lookat the economies inside the company,you'll know that electronic use is

paying the bills and it is subsidizingthe printed product which is an exactreverse of what we saw fifteen yearsago."'

'''Richard Van Orden, "Content-Enriched Accessto Electronic Information: Summaries of Selected Re-search," Library Hi Tech 31 (1990): 28.

" "Robert Weber, "The Clouded Future of Elec-tronic Publishing," Publishers Weekly 237 (20 June19901: 76.

" "Jeffrey K. Pemberton, "Online s I LaryBoyle on CAS's New License Policy . . . Effects onSearching Prices,'' On/in; 12 (March 19881: 21,

On the surface, electronic publishingseems to imply only a change in the formof distributing publications. But scholars inthe social sciences and the humanities havebegun to use the existing research and ed-ucation networks to engineer a new formof publication distinct from commercial ef-forts. These publications are academic-based, scholarly created and controlled,(often) refereed, electronic-only, network-delivered journals. Although scholarlyelectronic journals were invented only sev-eral years ago, already about three dozenhave sprung up in an array of disciplines,along with sixty newsletters and the thou-sands of electronic conferences used for lessformal communications.m

PSYCOLOQUY is one of the best ex-amples of the innovative genre of elec-tronic journals.121 The journal's editor,Stevan Hamad, a cognitive psychologist atPrinceton University, has edited an influ-ential nonelectronic journal (Behavioral andBrain Sciences) for more than fifteen years.Hamad decided to edit a scholarly elec-tronic journal as a result of his experienceparticipating in an early electronic confer-ence. He characterized early users of net-works as primarily computer enthusiasts andgraduate students. These two audiencespossessed enough time and motivation toventure into the new medium of conferenc-ing, a unique form of communication thatallows people, dispersed in time and place,to share ideas, ask questions, comment onwork, and sustain narrative discussions. As

12"Nliehael Strangelove, Directory of ElectronicJournals and Newsletters, ed. 1, July 1991. (To re-trieve electronically, contact the author at<441495(il'uottawa>; the directory is also availablein print through the Association of Research Libraries,Washington, D.C). Ann Okerson, of the Associationof Research Libraries, provided updated infotmationon current journal numbers to Avra Michelson in March1992.

'''The PSYCOLOQUY discussion is from notes ona presentation by larnad at the "RefereedJournals" session on 21 March 1991, at the NationalNct'91 Conference in Washington, D.C.


an early participant in an AI conference,Harnad decided to transmit work in a formmore polished than customary, as if he werewriting for a peer-reviewed journal. To hisgreat surprise, he found the exchange tre-mendously helpful to his intellectual work.Instead of waiting several years to receivepeer responses, he received instantaneousreactions to his work over the networks.Further, the responses arrived at the begin-ning of his intellectual process rather thanat the end, as happens with conventionalpublishing. Inspired by his conference ex-perience, Hamad wondered what it wouldbe like to experience with the best mindsin his field the same kind of instantaneousdialogue he had established with computerenthusiasts and graduate students. Thisprompted him to create PSYCOLOQUY, afully refereed, scholarly, electronic-onlyjournal, sponsored by the American Psy-chological Association.

PSYCOLOQUY is an interdisciplinaryjournal that publishes articles and reviewsconcerning psychology, neuroscience, cog-nitive science, behavioral biology, linguis-tics, and philosophy. Its editorial board offifty scholars reflects the range of disci-plines published by the journal. Journalsubmissions, refereeing, editorial work, anddistribution are handled entirely electroni-cally. There are currently more than twothousand individual subscribers on Bitnet.A large number of institutional subscribersalso receive PSYCOLOQUY through Use-net, a network connected to most of theuniversities ao.d research institutions of theworld, allowing all individuals at these sitesto access the journal. In 1990, LibraryJournal named PSYCOLOQUY one of theyear's best journals.

Harnad contends that the most importantdifference between electronic journals andprint publication is not the form of distri-bution but the medium's potentially revo-lutionary contribution to the furthering ofscholarship and the creation of knowledge.The real contribution of the electronic me-

dium is that it does what no other mediumcan do. Instead of waiting a year or twofor peer feedback (the typical amount oftime it takes to publish and then respond inprint), and instead of receiving the feed-back when already strongly invested in thenext research project, scholars enjoy rig-orous intellectual dialogue with one an-other, freed from the constraints of timeand place, at the front end of the researchprocess.122 The instantaneous distributionof ideas among peers permits a new andcritically important type of interaction thatfurthers scholarly inquiry in a way not pos-sible previously. The electronic medium isunique in its capacity to support interactiveimprovement of scholarship at a speed muchmore commensurate with the speed ofthought.

Other examples of scholarly, electronic-only journals include Post Modern Culture(North Carolina State University), an in-terdisciplinary journal of literary theory,culture, and creative writing; Artcom, de-voted to the interface of art and commu-nication technology; Quanta (CarnegieMellon University), an electronic journalof science fiction and fantasy; the BrynMawr Classical Review, a review journalof books on Greek and Latin classics; On-line Journal of Distance Education andCommunication (University of Alaska), de-voted to the development and practice ofdistance education; Ejournal (State Uni-versity of New York at Albany), an inter-disciplinary journal on the theory andpractice of electronic communication; New

'"Stevan Hamad, "Scholarly Skywriting and thePrepublication Continuum of Scientific Inquiry,"Psychological Science I (November 1990): 342. Asimilar point is made by Cliff McKnight in his article,"Using the Electronic Journal," in Scholarly Com-munication and Serials Prices: Proceedings of a Con-ference Sponsored by the Standing Conference ofNational and University libraries and the British Li-bran Research and Development Department 11-13June 1990, edited by Karen Brookfield (New York:Bov,,ker-Saur, 1991).

44


Horizons in Adult Education (SyracuseUniversity Kellogg Project), a refereedjournal for the field; Journal of the Inter-national Academy of Hospitality (VirginiaPolytechnic Institute and Blacksburg StateUniversity), publishing refereed articles onbasic and applied research on hospitalityand tourism; and the Public-Access Com-puter Systems Review, exploring electronicaccess to library materials.123 At a meetingrecently convened by the Association ofResearch Libraries (ARL), electronic jour-nal editors established the Association ofScholarly Journal Editors, a "closed" elec-tronic communications list for discussingcommon concerns and new publishing ef-forts.'

Aside from scholarly controlled elec-tronic journals, commercial publishers arebeginning to explore the profitability ofpublishing electronic-only academic jour-nals. The American Association for theAdvancement of Science (AAAS), in con-junction with the Online Computer LibraryCenter, Inc. (OCLC), announced the pub-lication of its first electronic-only journalin 1992. The publishers expect The OnlineJournal of Current Clinical Trials, a newpeer-reviewed medical journal, to distrib-ute the findings of original research severalmonths faster than its print counterparts.The journal represents the first commercialelectronic effort to display typeset-qualitygraphs, tables, and equations. The editorsare Edward J. Huth (former chief editor fornineteen years of the Annals of Internal

"'Information on electronic journals from Stran-gelove, Directory of Electronic Journals and News-letters; for a discussion on publishing a scholarlyelectronic journal, see Charles W. Bailey, Jr., "Elec-tronic (Online) Publishing in Action . . . The Public-Access Computer Systems Review and Other Elec-tronic Serials," Online 15 (January 1991): 28-35.

'"Informal presentation made by Ann Okerson(Associatior of Research Libraries) to a session on"Non-Commercial Publishing" at the Spring meetingof the Coalition for Networked Information, 19 March1991, Washington, D.C.; also, letter from Ann Okerson to Avra Michelson dated 8 July 1991.

Medicine), Curtis Meinert (of the JohnsHopkins Center for Clinical Trials), andThomas C. Chalmers (associate director ofthe Technology Assessment Group, Har-vard School of Public Heaith).125 If thecommercial publication of scientific jour-nals proves successful, it is likely that theircounterparts will emerge in the social sci-ences and humanities.

Hypermedia. During the last ten years,hypermedia has developed into a maturetool that supports electronic browsing byallowing users to follow links through text,images, and audio and visual records. Theelectronic links that characterize the tech-nology also make it possible to composeand deliver research products in newways.126 As hypermedia becomes a main-stream technology during this decade,scholars are encountering the prospect ofredefining the modern product of research.Should a hypermedia document provide au-tomatic links that take a reader from a foot-note to the actual cited work? Shouldhypermedia documents chronicle throughlinks the intellectual process of discovery?What new typ, of authoring guidelines arenecessary for research products developedin hypermedia? Scholars are beginning totackle some of the hard questions raised bythe availability of a technology that allowsfor a more complex organization of ideas.The scholarly creation and consumption ofhypermedia documents is another exampleof the trend toward end-user computing,further stimulated in this case by the onlinetransition.

One historian argues that the power ofhypertext (hypermedia restricted to text) isthat "it produces documents not intended

11iThe Online Journal of Current Clinical Trials,brochure published by the American Association forthe Advancement of Science and OCLC, ca. 1991.

"-Tor an introduction to hypermedia, sec JeffConklin, "Hypertext: An Introduction and Survey,"in Computer-Supported Cooperative Work: A Book ofReadings, edited by Irene Greif (San Mateo, Calif.:Morgan Kaufmann Publishers, 1988), 423-75.


to exist in printed form."'" He describesthe contrast between a standard historytextbook and a hypertevr product throughthis example:

Imagine a computerized book ofdocuments. As you open it to theMonroe Doctrine, you see the severalparagraphs of the President's addresswhich make up the statement of for-eign policy. Gliding a mouse-di-rected cursor over the words, an iconpops up next to the words, "RussianImperial Government." By clickingthe mouse, you reveal a brief essayon the Russian Czar's interest in

Alaska. The word "Czar" in thatsubtext can bring up the Czar's actualstatements on the subject, and"Alaska" can trigger a map of thePacific Northwest. After folding theseasides back into the original docu-ment, you reach the phrase, "Withthe existing colonies or dependenciesof any European power we have notinterfered and shall not interfere," andclicking the mouse reveals an anno-tated list of interventions prior to 1823.That screen will activate a map ofCentral and South America showingthe new revolutionary governmentsand the dates of their independencefrom Spain.'28

Scholars already have begun to produceresearch projects in nonlinear formats. AtStanford University, for example, a hyper-media Shakespeare application created byLarry Friedlander allows users to view ona video monitor filmed versions of Shak-espearean plays, while viewing on anotherscreen a synchronized presentation of theplay's text and stage blocking material. At

'James B. M. Schick, Teaching Iltsuny with aComputer: A Complete Guide (Chicago: LyceumBooks, 1990), 63.

1281bid .

any point, users can refer to dictionariesand historical notes to increase their un-derstanding of the performance. The sys-tem also allows users to create animatedversions of plays, provides interactive tu-torial instruction on theater topics, and sup-ports note taking.129

Another hypermedia application, de-signed for use in an undergraduate poetrycourse, uses software to convey the ideasthat poems are related to other poems, thatthey may be related to other art forms, andthat they may be related to both other poemsand other forms of art simultaneously.'3°Since poems often refer to lines from otherpoems, use a painting to develop an anal-ogy, quote a piece of literature, or alludeto a music score, a hypermedia documentcan make poetry truly come alive by usinglinks to demonstrate concretely the culturalattachments among art forms.

Two other projects are representative ofefforts to use hypermedia as a new author-ing medium. The Faculty of Art and De-sign at Coventry Polytechnic in Englandconsidered the possibilities of using hyper-media as an authoring medium for fouryears. As a result of their deliberations, thefaculty decided to allow students to submitthe curriculum's required thesis in hyper-text. The thesis is a research product on thehistorical and theoretical portions of thecurriculum. Hypermedia enables the art anddesign students to incorporate their designand visualization aptitudes into the organ-ization and presentation of a theoreticalwork. After this experiment with studenttheses, the faculty will evaluate the effec-tiveness of hypertext as an authoring me-

129Charles W. Bailey, Jr., "Intelligent MultimediaComputer Systems: Emerging Information Resourcesin the Network Environment," Library Hi Tech, 8

(1990): 31."'John M. Slatin, "Text and Hypertext: Reflec-

tions on ihe Role of the Computer in Teaching Mod-ern American Poetry," in Humanities and theComputer: New Directions, edited by Dav:d S.(Oxford: Clarendon Press, 1990), 129-31.


dium and will consider the most appropriatecontexts for its use."'

Finally, in anticipation of the widespreaduse of hypermedia as an authoring tool, theBritish Library Research and DevelopmentDepartment is funding Project Quartet, aresearch effort to develop a standard set ofguidelines for creating hypermedia docu-ments. The principals on the project arguethat researchers authoring in hypertext needguidelines for establishing nodes and linksto provide the necessary hooks for readers.They contend that the skills used for writ-ing in paper media do not adequately servethe needs of scholars authoring in the elec-tronic age. The project hopes to establishglobal taxonomies for hypertext authoringthat can be used across systems.'32

Curriculum Development andInstruction

The enormous amount of literature oncomputer-aided instruction makes it appearthat faculty in the social sciences and hu-manities use computer technology to im-prove their teaching to an even greater extentthan for research. This is not surprising,

"'Alan Dyer and Kate Milner, "An Examinationof Hypertext as an Authoring Tool in Art and DesignEducation," in Humanities and the Computer; NewDirections, 137-48.

'"Cliff McKnight, John Richardson, and AndrewDillon, "Hypertext Authoring: Some Basic Issues,"Humanities Communication Newsletter 11 (1989): 25-29. See also other publications issued by this group,such as Hypertext in Context, The Cambridge Serieson Electronic Publishing (Cambridge: CambridgeUniversity Press, 1991); "Human Factors of JournalUsage and Design of Electronic Texts," Interactingwith Computers 1 (1989): 183-89; "The Effects ofDisplay Size and Text Splitting on Reading LengthyText From Screen," Behaviour & Information Tech.nology 9. no. 3 (1990): 215-27; and Bill Tuck, CliffMcKnight, Marie Hayet and David Archer, ProjectQuartet, Library and Information Research Report no.76, (Wctherby, England: British Library, 1990). Cor-nell University is also experimenting with the usabil-ity of an online hypermedia presentation of thousandsof articles published : the Journal of the AmericanChemical Society. Sec Michael Alexander, "But CanYou Read It Likc A Book?" Computetworld 24 (19November 1990): 18.

since the fundamental aspect of the tech-nological revolution is that faster, smartermachines affect the ways we think and leans.According to Maty Alice White, directorof the Electronic Learning Laboratory atColumbia University's Teachers College,information technologies change "how werepresent information, and therefore howwe view a problem . . . how we analyzeproblems, and because they change that viewand that analysis, they can change how wemake decisions. These are intellectual tools,the very stuff and excitement of educa-tion."133 The scholarly use of computersto develop instructional applications is an-other example of the trend toward end-usercomputing, while connectivity representsthe key trend that allows for new styles ofdistance education.

Teachers at every educational level arerevising curriculums to include computer-supported instruction, such as simulations,cognitive modeling, and individual-ori-ented learning. The trend for an increasingportion of academia is toward "computercampuses" where students are required topurchase a specific set of computer equip-ment upon enrolling. Some universities havebegun to fund positions devoted exclu-sively to helping faculty develop instruc-tional software or incorporate informationtechnology into the classroom.13` Corn-

'"Mary Alice White, "The Third Learning Revo-lution," Electronic Learning 7 (January 1988): 6.

"'See Schick, Teaching History with a Computer:A Complete Guide, 207-08. Schick cites Drexel, NorthCarolina/Chapel Hill, North Carolina State, the Uni-versity of Southern California, and Stanford amongothers, as campuses that have hired staff to stimulatecomputer-oriented curriculum. Richard Kesner, chiefinformation officer at Babson College, reports that hiscampus is hiring staff for this purpose as well. TheEDUCOM/USC Survey of Desktop Computing inHigher Education estimates that more than 40 percentof two and four year public and private colleges anduniversities provide support for faculty developingcomputer-based instructional courseware. Sec Ken-neth C. Green and Skip Eastman, Campus Computing1990 (Los Angeles: University of Southern Califor-nia, Center for Scholarly Technology, 1990), 15.


puter simulations are especially popular inmany disciplines because they submergestudents in a different social context, al-lowing thenito consider "what-if" scena-rios. The simulations tend to be particularlyeffective in promoting an understanding ofhistory because the first-person experienceof time and circumstance helps students ap-preciate that the past is shaped by indivirluals reacting to social events and forces.135

An extensive array of simulation soft-ware is available for instructional purposes,including hundreds of applications in thefield of history alone. For example, the ex-perience of the Constitutional Convention,complete with delegate selection and theratification process, is available to studentsusing simulation software. Other simula-tions allow students to experience U.S.congressional committee debates on read-mitting southern states to the union afterthe Civil War, or to participate in the pres-idential decision on whether to take actionin the Pullman Strike of 1894. A NationalGeographic Society product simulates theconstruction of the Transcontinental Rail-road, challenging students to decide aboutsuch issues as construction, labor, and re-lations with Native Americans. Anothersimulation focuses on the military tacticsused by the Soviet Union with Nazi forces.At Stanford University, a French Historyprofessor developed a simulation that es-tablishes a seventeenth century bourgeoiscontext for students to negotiate a strategicmarriage, consider proper investments, andmanage the family's inheritance to promotetheir stature. Some simulations employ ar-tificial intelligence techniques to demon-strate more fully the meaning of historicalcontext. For instance, AI-enhanced simu-lations are available for such events as theRussian Revolution and the development ofthe European Economic Community.'

"Schick, Teaching liivtoty till a Computer, 1111-02.

"For on extensive (laical bibliogiapl'y of current

Although simulations are one of the mostpopular forms of computer technology foundin the classroom, other types of applica-tions are also in use. In an effort to com-puterize a full discipline's curriculum,Gregory Crane, of Harvard University'sClassics Department, established the Per-seus Project. The Perseus database at-tempts to provide an interactive multimediacurriculum on classical Greek civilization.It contains a vast corpus of the discipline'ssources, including translations of majorGreek texts, introductory materials de-signed for novice students, Greek languagetexts for more advanced students, color im-ages and line drawings of archaeologicalartifacts and maps, essays and themes onkey facets of Greek literature, a chronol-ogy, and a classical encyclopedia. The hy-permedia application provides coursematerials for such disciplines as art, ar-chaeology, classics, history, law, philoso-phy, and political science. In early 1992,Yale University Press released version 1.0of the Perseus database, which runs onMacintosh computers with the HyperCardprogram.'"

Anticipating the availability of large vol-umes of humanities source materials on-line, the Stevens Institute of Technology,with support from the Humanities GrantProgram of the New Jersey Department ofHigher Education, is exploring how accessto electronic source materials is apt to re-structure humanities education. In particu-lar, they are interested in learning howelectronic texts, such as those compiled for

simulations available for the study of history, secSchick, Teaching History with a Computer, 122-45,and for a description of the Stanford University sim-ulation see pages 100-01. The Social Science Com-puter Review regul.aly reviews commercial simulationsoftware designed for educational purposes. The Alsimulations arc mentioned in Enna Is, Artificial Intel-/is:owe, 125.

" -Sec a brief review of the project in Social ScienceCompiler Review 7 (Summer 1989): 211; also,1.oughridge, "Information Technology," 281.


specific disciplines, can best be integratedinto undergraduate course work. As part oftheir research effort, they plan to evaluatestudent learning patterns and actual studentperformance with electronic curricu-lums.'38

In a different approach, a group of fac-ulty at Manchester Polytechnic in Englandis developing "viewbooks" for use in his-tory curriculums. These disk-based bookstake the form either of annotated historicaldocuments with introductions and conclu-sions cr of texts and tables. Approximatelytwenty-five different books are available,and they permit information to be retrievedthrough various techniques from the data-bases. The developer is currently designinga viewbook shell that will allow instructorsto insert the text of their choice into thedatabase. This type of application will ben-efit from advances in document-conversionscanning technology.13"

Commercial software specially designedfor particular disciplines is becoming avail-able, which will facilitate the use of elec-tronic source materials in classrooms. Onerecently released package, for example,displays nineteenth-century statistical cross-tabulations and regressions on France, En-gland, and Wales)4° Faculty also are mak-ing use of AI shells; a history instructorfound a particular software shell enhancedwith artificial intelligence well-suited foran application on the Norman Invasion. Thissame instructor also chose an expert sys-tems shell to construct a learning tool for

'"Sec Edward A. Friedman, James E. McClellanIII, and Arthur Shapiro, "Introducing UndergraduateStudents to Automated Text Retrieval in HumanitiesCourses," in humanities and the Computer, 103-12.

19See Richard H. Trainor, "History, Computingand Higher Education," in History and Computing II,38-39; for a discussion of scanning technologies, seeTimothy C. Wciskcl, "University Libraries, Inte-grated Scholarly Information Systems (ISIS), and theChanging Character of Academic Research." LibraryIli Tech 6 (1988): 15.

" "Sec review in Social Science Computer Review7 (Summer 1989): 211.

the study of the Middle Ages."' The useof information technology by the British inhistory curriculums is so great that thecountry's academics have established a for-mal organization to support the exchangeof technical resources. Headquartered at theUniversity of Bath, the National Informa-tion for Software and Services organizationcoordinates the sharing of historically ori-ented software and data files among collegeprofessors."2

Apart from computer-assisted curricu-lums, teachers in the United States are usinginformation technology to support a newstyle of education. Distance learning, in es-sence an improved successor to correspon-dence course work, interactively linksteachers and students in scattered locations.During the past few years, a majority ofstates have become active proponents ofdistance learning. The findings of a recentsurvey show that thirty-two states "cur-rently have at least one statewide networkfor distance learning, and nearly half havemore than one."'" Enthusiasm for dis-tance learning seems to emanate as muchfrom advances in storage and retrieval tech-nology as from telecommunication's net-works that expand the ability to useinformation at distant locations. Indeed, astudy conducted by the U.S. Office ofTechnology Assessment found that dis-tance learning no longer serves only iso-lated rural schools. Rather, it has becomethe vehicle for bringing advanced, special-ized course work and an array of expertsto many classrooms. Existing programsmake it possible for a high-school studentin Mississippi to study Japanese, and forWashington State to provide advanced

"'See Martyn Wild, "History and Ncw Technol-ogy in Schools: Problems, Possibilities and the WayForward," in History and Computing 1!, 30.

12See Trainor, "History, Computing and HigherEducation," in History and Computing II, 40.

"Barbara Kurshan and Marcia Harrington, State.wide Education Networks: Survey Results (Roanoke,Va.: Educorp Consultants, April 1991), 2.


placement English courses to all who qual-ify. In Maine, teachers enrolled in a mas-ters' program attend after-hours graduatecourses in their classroom via distancelearning, instead of undertaking a four- tofive-hour commute.'" The feasibility ofusing distance learning to maximize uni-versity students' control over the time, place,and pace of education is being evaluatedthrough experimental courses. The flexi-bility of a distance-learning program is aptto be particularly attractive to full-time em-ployed students enrolled in advanced de-gree programs."5

The infusion of technology into educa-tional programs is occurring rapidly. Ex-amining the effectiveness of technology asan educational tool represents a popular areaof research, though findings are still some-what preliminary. One study on the impactof the use of Al tutors in high-school ge-ometry classes found that the individuallypaced applications fostered a healthy com-petitiveness among students.146 In a tradi-tional classroom, the students never had theopportunity either to get ahead of or fallbehind one another. With the AI tutor,however, self-paced learning stimulatedstudents to rival one another, as they wouldcall out in class the "page" on the monitorthey had advanced to through correct an-swers.

The study also observed that the majorityof students enjoyed the AI tutoring morethan conventional classroom instruction andthat the enjoyment translated into increased

"'See U.S. Congress, Office of Technology As-sessment, Linking for Learning: A New Course forEducation, OTA-SET-430 (Washington, D.C.: U.S.Government Printing Office, November 1989), 2-3,54.

"See Gil Rogers, "Teaching a Psychology Courseby Electronic Mail," Social Science Computer Re-view 7 (Spring 1989): 60-64.

'"Sce Janet Ward Schofield, Debra Evans-Rhodes,and Brad R. Huber, "Artificial Intelligence in theClassroom: The Impact of a Computer-Based Tutoron Teachers and Students," Social Science ComputerReview 8 (Spring 1990): 24-41.

motivation. In addition, the study found thatstudents appreciated the independence fromadult control and that with the computerthey were free to vent anger and frustrationunacceptable with teachers. But probablymost important, the research discovered thatthe students experienced the tutor as a gameand thus associated it with play. The elec-tronic games popular among youth, com-bined with computer-assisted learning, inessence are preparing the next generationfor a new era. As a result of changes oc-curring in education and play, young peo-ple are being thoroughly indoctrinated intothe computer culture. The use of informa-tion technology and electronic communi-cation will be deeply ingrained in the nextgeneration of researchers, who will havebeen computer veterans since elementaryschool. The current demands for electronicinformation available through networks inhomes and offices can only escalate anddeepen among tomorrow's scholars.

Summary

As the preceding section indicates, theclear trend in the modern research processis toward scholarly identification, use,interpretation, and analysis of sources inelectronic form, and the gaining promi-nence of new forms of computer-assistedcommunication and instruction. The re-search process is already changing, and thischange is accelerating and spreading acrossa wide range of disciplines. Because a keyfactor promoting this change is the availa-bility of new information technology, ana-lyzing how trends in information technologyinteract with current trends in scholarlypractice can help predict the future evolu-tion of the research process.

The analysis of information technologyundertaken above points to two major tech-nology trends that are likely to transformscholarly practice: increased end-user com-puting and increased connectivity.. Thisanalysis also implies that a number of more


specific technology, including artificial in-telligence, end-user publication and distri-bution, hypermedia, and visualization andvirtual reality, are likely to have a signifi-cant impact on the research process. Theeffects of these trends, along with changesin scholarly practice that are already underway, point to a future in which researchersuse computation and electronic communi-cation to help formulate ideas, accesssources, perform research, collaborate withcolleagues in their own and other disci-plines, seek peer review, publish and dis-seminate results, and engage in manyprofessional and educational activities. Farfrom being visionary, this future is alreadypresent: It is currently being experiencedby significant and increasing numbers ofresearchers from many disciplines.

How should the archival profession re-spond to these changes in scholarly prac-tice? Are the techniques and functionsdeveloped by the archival profession tomanage printed media adequate for the needsof researchers who operate in a global elec-tronic networking environment? Should es-tablished archives convert printed materialto machine-readable form? If so, what se-lection criteria should be used? What con-stitutes the "reference function" in the ageof research and education networks andelectronic communication? These issues firstare addressed through case examples drawnfrom the experience of the library com-munity, and then by a set of recommen-dations specifically designed for the archivalprofession.

RESPONSES BY THE LIBRARYPROFESSION TO CHANGINGRESEARCH PRACTICES

On several occasions in the recent past,libraries and professional associations havesponsored inquiries into scholarly use oftechnology. For example, the AmericanCouncil of Learned Societies conducted asurvey in 1985 to 1986 that noted the rapid

increase in the use of technology by thescholarly community.147 In a more recentstudy sponsored by the Harvard College Li-brary and the American Council of LearnedSocieties, the Conference on ResearchTrends and Library Resources brought so-cial science and humanities scholars to-gether to explore new trends in researchmethods. Scholars spent several days con-sidering the impact of new technology, in-terdisciplinary research, and the use ofinnovative formats of materials on theirwork.148 In another effort, the AmericanAcademy for Arts and Sciences sponsoredan exchange between scholars and librar-ians to develop policy recommendations toimprove access to library materials. A keyobservation shared by these inquiries is thatscholars increasingly want online access toelectronic source materials available throughpersonal computers in their homes or of-fices.

Visionary leaders within the library com-munity are beginning to implement pilotprojects designed to improve the library'srole in advancing scholarship and its re-sponse to changing research methods. Theseprojects hold particular interest for archi-vists as the key distinction between theprinted form of archival and library mate-rials is disappearing. Indeed, in an elec-tronic environment, concepts, such as"unique" and "multiple," which have beenused to distinguish archival sources fromlibrary materials, are less meaningful. It isnot surprising that librarians hold differingopinions regarding the most appropriate rolefor libraries in the electronic environment.Some librarians argue for continuitythecontinued commitment to collection devel-

"'Morton and Price, The ACLS Survey of Scholars:Final Report of Views on Publications, Computers,and Libraries, 33.

'"Lawrence Dowler, "Conference on ResearchTrends and Library Resources," 22-23, February 1990,unpublished draft report (Cambridge, Mass.: HarvardUniversity, Widener Library, n.d.)

51


opment. Those who hold this position ar-gue for consolidating library resources inthe activities of selection and collectionmanagement and for relinquishing a rolefor libraries in converting source materialsto electronic form. In contrast, the propo-nents of change claim that the continuityapproach could mark the end of the era offree access to information because com-mercial vendors would step in to convertlibrary materials and make them availablefor a fee in electronic form. The advocatesof information-based institutions championa new vision of the library without wallsan enterprise comprising many electronic

including commercially producedproducts) that provide network access topatrons. Regardless of their perspective, bothsides agree that patron demands for elec-tronic access to library materials will bemet by someone.'" This section examinesseveral leading projects and programs un-dertaken by the library community to ad-dress changes in the research environment,focusing on four new trends in professionalactivity: (1) promoting high-performanceconnectivity, (2) conversion of printed ma-terials to machine-readable form, (3) soft-ware engineering for next-generationsystems, and (4) transformations in profes-sional roles.

Promoting Connectivity

In the last few years, library leaders haveforged a new political alliance with aca-demic computing centers and tne telecom-munications industry to support thedevelopment of high-performance comput-ing networks capable of rapidly transmit-ting huge amounts of data and high-resolution graphics. A high-performancecomputing network is needed because the

"'For two perspectives on the topic sec: StephenE. Ostrow and Robert Zich. in Research Collectionsin the Information Age: The Library of Congress I.coksto the Future, edited by John Y. Cole (Washington,D.C.: Library of Congress, 1990).

several thousand academic, governmental,regional, and private networks that alreadyoperate worldwide cannot transmit data andimages fast enough or in large enoughchunks to keep pace with the needs of sci-entific research. Furthermore, faster net-works with higher bandwidths will expandinfrastructure support for scholarly ex-change of visually-oriented material (suchas that required for medical research), on-line electronic publishing, and high-speedinterchanges of text and graphics in the artsand social sciences.

Recognizing the need for infrastructures(or "highways") to disseminate materialselectronically, the Association of ResearchLibraries (ARL) in 1990 joined with aca-demic and administrative computing cen-ters to form the Coalition for NetworkedInformation (CNI). CNI is a collaborationamong three distinct groupsEDUCOM,CAUSE, and the ARLwho have unitedto "promote the creation of and access toinformation resources in networked envi-ronments in order to enrich scholarship andenhance intellectual productivity."Is° Themost immediate focus of the coalition's workis to establish the National Research andEducation Network (NREN), a federallysupported high-performance computingnetwork. In the interim, NSFNet (a net-work administered by the National ScienceFoundation), in conjunction with the thou-sands of other existing networks, serves asthe precursor for the future operationalNREN.

The coalition is optimistic about imple-menting NREN as a gigabit-per-secondnetwork. In 1991, Congress passed the HighPerformance Computing Program that es-tablishes the mandate for NREN. Althoughthe original motivation for NREN emergedfrom the scientific community's require-ments, the broader constituency rep-

" "From Coalition for Networked Information, Mis-sion Statement, March 1990.

52


resented by CNI envisions a network devotedto kindergarten through high school (K-12)programming, as well as leading-edge re-search. Indeed, EDUCOM recently desig-nated a full-time staff position for thedevelopment of network K-12 programs.CNI's commitment is to :lie developmentof a network available to all the nation'steachers, students, and researchers.'5'

When fully implemented, NREN will al-low researchers at universities, nationallaboratories, nonprofit institutions, govern-ment research centers, and private industryto exchange sources, communicate in realtime, share preliminary fin.itngs, and dis-seminate publications electronically. In-deed, the dramatic changes in the waysresearch is conducted and infonnation isexchanged are key factors driving the de-velopment of NREN. Through remote ac-cess hookups, NREN will provide thenation's researchers and students, regard-less of the type and size of their college,with the same computing tools, data files,supercomputers, electronic libraries, spe-cialized research facilities, and educationaltechnology.152 It is anticipated that NRENwill support the transmittal oc at least 1

billion bits of data every seconc' by 1995.Recognizing the impact a network with

such unprecedented speed and capacity willhave on their institutions, librarians havejoined with other information professionalsto support the development of NREN. Ascoalition members, librarians are partici-pating in a range of NREN-related activi-ties, including CNI's seven working groupson: (1) encouragement of academic pub-lishing; (2) expansion of commercial elec-tronic publishing; (3) development of

"'From Kenneth King, president, EDUCOM, un-published paper presented at the "NREN Governanceand Policy" session at National Net'91 Conference,22 March 1991, Washington, D.C.

"'See NREN: The Aational Research and Educa-tion N ,.,tork (Washington, D.C.: Coalition for theNational Rese,Irch and Fducation Network, 1989).

network architectures and standards; (4)formation of proposals for legislative codes,policies, and practices; (5) organization ofdirectories and resource information serv-ices; (6) creation of teaching and learningprograms; and (7) improvement of networkmanagement and user education.

Through the activity of building a high-performance network, a new vision of thelibrary is emerging. No longer simply a placeto visit, libraries are becoming "virtual en-terprises" of electronic information.153

Conversion

As a concrete step toward the realizationof networked electronic libraries, some re-positories have begun to convert to ma-chine-readable form records originallycreated on paper. The American MemoryProject at the Library of Congress (LC)represents a leading example of this typeof effort.''' Over the next five years, theLibrary of Congress, with nearly $1 millionper year in congressionally appropriatedfunds along with private donations, willconvert into electronic form large archivalcollections from their holdings relating to

'"For additional information on NREN see mostrecent issues of EDUCOM Review; also, Jean Loop,National Research and Education Network: Overviewand Sionnuoy (Washington, D.C.: Association of Re-search Libraries, July 1990); Charles E. Catlett, "TheNSFNet: Beginnings of a National Research Inter-net," Academic Computing 3 (January 1989): 18-21,59-64; Stephen B. Gould, "Computing and Telecom-munications in the Federal Government," CRS Re-view 11 (July/August 1990): 12-15; for informationon CNI, see organizational papers available from PaulPeters, CNI, 1527 New Hampshire Avenue, N.W.,Washington, D.C. 20036.

1s4Additional programs now under way include theHunt Library at Carnegie Mellon University and theImage Transmission Program at the National Agri-cultural Library. Other libraries are creating CD-ROMson specialized subject areas. The Marine Corps hasannounced that it is compiling an online version ofthe Marine Corps University warfighting collectionthat will allow marines to "fight smart" wherm.ithey are stationed; sec Kevin M. Bacrson, "MarinesPut Library On-Line," Federal Computer Week,5 (2September 1901): 1, 4.

53


American culture and history.155 The pur-pose of the project is to use advanced tech-nology to make electronic versions ofcollections available to libraries across thecountry.

The collections chosen for the initial roundof conversion primarily document aspectsof turn-of-the-century life in America. Theyare drawn from a cross-section of originalformats, including rare pamphlets, earlymotion pictures, sound recordings, per-sonal papers, and still photographs. A va-riety of image, text, and audio types willbe linked to catalog information in the stan-dard MARC (MAchine-Readable Catalog-ing) format.

In fiscal year 1991, the Library of Con-gress prepared four collections for elec-tronic dissemination, including about 300broadsides from the Continental Congressand Constitutional Convention; three hoursof sound recordings of speeches (sixty ex-amples) of political leaders during WorldWar I and the presidential election of 1920;two dozen short motion pictures of Presi-dent McKinley at the start of his secondterm and at the 1901 Pan-American Exhi-bition in Buffalo, New York; and about25,000 photographs from a well-knownpostcard and scenic-view company foundedby William Henry Jackson. By the end of1992, the library will supplement these withcollections of Civil War photographs, ap-proximately 350 African-American pam-phlets (11,000 printed pages written between1820 and 1910), local history books fromCalifornia, early films of New York City,and life histories from the Federal Writers'Project.

The library's selection process attempts

'"The American Memory project has received giftsfrom the David and Lucille Packard Foundation, theAnnenberg Fund, Inc., Armand Hammer's OccidentalPetroleum Corporation, and Jones International, Lid.,as well as gifts or loans of equipment from AppleComputer, IBM, and Pioneer. Sec Library of Con-gress, "American Memory," I.(' /*ma'amtin (26 February 1990): 83-87.

to strike a balance between popular, readilyavailable collections and unprocessed col-lections that comprise a backlog arrearage.Selecting an arrearage collection providesan impetus for processing it. As selectionsare made, the planners consult both withLibrary of Congress curators and with out-side scholars. The first set of AmericanMemory collections is being evaluated inforty school, university, public, and speciallibraries to assess patterns of use. The re-sults of this evaluation will provide furtherguidance.

Compared with all the holdings of theLibrary of Congress, American Memory willconvert only a relatively small amount dur-ing the first few years. The program's ex-tent reflects the high cost of conversion,the institution's desire to reduce its arrear-age, and the typical difficulties encoun-tered in the introduction of a newtechnology. To ximize the use of whatit 1,as prepared, however, the library is

placing special emphasis on educationalapplications. Besides providing the collec-tions proper, American Memory's presen-tation also will include introductoryinformation in interactive, computerizedform and in print.

The ultimate goal of the American Mem-ory project is to make materials availablevia telecommunications, but this goal willbe fully realized only in the later 1990s.Until then, the collections will be dissem-inated on disks: CD -RUMS for digital in-formation and analog videodiscs for motionpicture and some still photographic collec-tions. But whether on disk or in a network,every American Memory working proto-type will model what Ricky Erway, anAmerican Memory associate coordinator,describes as a "library without walls."American Memory will he operating as apilot project through 1995.'`"

''''l further nilormanon on Ar»ctiudn Memory,contact the Library of Congress, Special Projects Of-


Software Engineering

Many libraries are considering ways toexpand bibliographic access as part of theirplans to develop next-generation librarysystems.157 But few are taking as ambitiousor comprehensive an approach to the processas the staff at Carnegie Mellon's UniversityLibraries. With a $1.7 million grant fromthe Pew Memorial Trust and several mil-lion dollars of donated hardware from Dig-ital Equipment Corporation, the library isdeveloping a system that will provide theuniversity's faculty, students, and admin-istrators with access to bibliographic data-bases, full-text documents, and networkgateways.'58 Library Information System II(LIS II), implemented in 1991, is designedto improve the quality of retrieval and de-livery of textual information to users. In a

fice, Washington, D.C. 20540, (202) 707-6233. In-formation on the project from discussions by AvraMichelson with Ricky Erway on 28 December 1990and Erway and Carl Fleischhauer on 4 February 1991and from documents supplied by the Library of Con-gress.

'"For the development of enhanced bibliographicrecords, sec, for instance, Van Orden, "Content-En-riched Access to Electronic Information," 27-32; FloWilson, "Article-Level Access in the Online Catalogat Vanderbilt University," Information Technology andLibraries 8 (June 1989): 121-31; and KatharinaKlemperer, "New Dimensions for the Online Cata-log: The Dartmouth College Library Experience," In-formation Technology and Libraries 8 (June 1989):138-45; the Klemperer article also discusses Dart-mouth's approach to the development of an integratedcampuswide information system.

'5"Information for this section is from a site visitby Avra Michelson to the University Libraries thatincluded meetings with Thomas Michalak, Tom Do-pirak, and Denise Troll on 27 March 1991. See alsotwo reports on the work of the project: Denise A.Troll, Library Information System 11: Progress Reportand Technical Plan, Mercury Technical Report Se-ries, no. 3 (Pittsburgh, Pa.: Carnegie Mellon Univer-sity, 1990); and Nancy I-1. Evans et. al., The Visionof the Electronic Library, Mercury Technical ReportSeries, no. 1 (Pittsburgh, Pa.: Carnegie Mellon Uni-versity, 1989). Also sec William Y. Arms and ThomasJ. Michalak, "Carnegie Mellon University," in Cam-pus Strategies for Libraries and Electronic Informa-tion, edited by Caroline Arms (Bedford, Mass.: DigitalEquipment Corporation, 1990), 243-73.

bold departure front the standard approachto library automation, Carnegie Mellonseparated its public catalog from other li-brary administrative functions. As such, LISII is devoted strictly to user-oriented re-trieval, whereas OCLC's LS/2000, an au-tomated system with integrated modules, isin use for .other aspects of library admin-istration.

The technical goal of LIS II is to producefor networked campuses an affordable li-brary retrieval system that adheres to avail-able standards. During the first phase, thesystem will run on University Library in-stalled workstations. Since January 1992,LIS II has been available across campusthrough workstation or VT 100 access. AMacintosh interface is scheduled to be re-leased by the end of 1992. The applicationgoals of the current system are to providethe following:

Online bibliographic access to all uni-versity resourcesBibliographic access at the article levelto journal literatureElectronic access to external data-basesOnline access to a range of campusinformationOnline access to textual information'

The system's distributed architecture hasbeen designed to support further researchand development toward the realization ofan electronic library.

Although the system's software supportsstandard bibliographic retrieval, it also pro-vides enhanced access to select antholo-gies, plays, edited collections, exhibitioncatalogs, and conference proceedings. Sev-eral thousand bibliographic records for these

155The University of California at Berkeley's Officeof Information Systems and Technology also is de-veloping a campus networked information system tosupport bibliographic and nonbibliographic databases,full-text documents, nontextual documents, and hy-permedia links.

BEST COPY AVAILABLE


types of publications have been embel-lished manually or by establishing systemlinks with nearly one dozen commercialproducts that include tables of contents, ti-tle pages, and book reviews. One-page ab-stracts are included in the bibliographicrecords of campus-issued scientific andtechnical reports. The intL nt of this type ofrecord enhancement is to improve the rel-evance of system retrievals.

Besides record enhancement, the staffplans to mount two types of full-text da-tabases, journal articles and campuswideinformation, on the system. Elsevier, Per-gamon, and the Association of ComputingMachinery (ACM) have agreed to providethe University Libraries with machine-readable journals and technical reports inthe subject field of computer science. ACMwill provide extensive runs of four of itspublications: Computing Reviews (tenyears), Collected Algorithms (twenty-fiveyears), Communications (two years), andGuide to Computing Literature (ten years).Carnegie Mellon is also negotiating anagreement to make the publications of theAmerican Association for Artificial Intel-ligence available in machine-readable form,and it is working with academic researchinstitutions to collect machine-readablecomputer science technical reports. Con-centrating the full-text offerings in an areasuch as artificial intelligence and computerscience will allow the University Librariesto further evaluate scholarly informationneeds by studying the use of textual infor-mation in a single di 'line.

The University Libraries also are install-ing a CD-ROM jukebox system from Uni-versity Microfilms, Inc. That systemincludes full-text images of general andbusiness journals linked to bibliographic ci-tations in tape-mounted databases on LISII. In the final phase of the project, theimages will be delivered to workstationsacross campus.

The full-text, campus-oriented docu-

ments require an indexing scheme entirelydifferent from that developed for standardbibliographic data. The new system willprovide campus software licensing andavailability information, career and place-ment resources, the Carnegie Mellon Pol-icies and Procedures Manual, theundergraduate catalog, user help files forother campuswide systems, listings of fac-ulty and staff publications (including re-search profiles), and indexes and full textof campus newspapers. Standard office ref-erence materials, such as phone books, en-cyclopedias, and dictionaries, are alreadyavailable.

Development of the system's user inter-face is based on staff findings on user workhabits and information-seeking behaviors.According to the research, patrons rarelyrefer to documents in isolation from otheractivity. For this reason, the LIS II archi-tecture has been designed to integrate witha larger work environment, supportinglinkages to word processors, databases, e-mail, and parallel applications. Toolkits(special software routines) permit LIS IIusers to make individual databases avail-able across the network. Other features al-low patrons to store searches for reuse, movein one keystroke from a journal article ci-tation to the full text of the article, andimprove queries by browsing indexes thatreveal how often terms are used. The win-dowed screen environment can be custom-ized by each user.

The creation of an electronic library linkedto other electronic libraries requires sus-tained effort. LIS II provides in substantialmeasure an architecture support full-textelectronic delivery of documents in librar-ies. In creating this system, the developersclarified many issues and resolved otherimportant issues in the areas of distributedstorage and retrieval systems, informationcapture and representation, information re-trieval and delivery, and management andeconomic concerns. Carnegie Mellon plans


to make the software developed for LIS IIavailable to other libraries.

Transformations in Professional Roles

Library literature contains many propos-als for new roles for library professionalsin the electronic age.'6° Among these theprogrammatic achievements of the Labo-ratory for Applied Research in AcademicInformation serves as one of the best op-erational models for redefining the librari-an's role on campus. A division of theWilliam H. Welch Medical Library at TheJohns Hopkins University, the laboratoryis a collaboration among academic schol-ars, scientists, and librarians. They shareresponsibility for the creation, structuring,representation, dissemination, and use ofscholarly knowledge through the use ofcomputing and communication technology.Created in 1987 by Nina W. Matheson andRichard E. Lucier, the Laboratory exploresstrategies for integrating the lil,tary morefully into the scholarly communicationprocess.'6' Lucier has developed what heterms the "knowledge management model,"which extends the library's traditional stor-age and retrieval and information transfer

'6°See, for instance, the ideas developed by EldredSmith in his book The Librarian, the Scholar, andthe Future of the Research Library (New York:Greenwood Press, 1990), especially 60-63 and 83-84. Articles by Bert B. Boyce and Kathleen M. Heim,"The Education of Library Systems Analysts for theNineties," and John Corbin, "The Education of Li-brarians in an Age of information Technology," inComputing, Electronic Publishing and InformationTechnology: Their Impact on Academic Libraries, ed-ited by Robin Downes (New York: Haworth Press,1988), 60-63 and 83-84, respectively; and TimothyC. Weiskel, "University Libraries, Integrated Schol-arly Information Systems (ISIS), and the ChangingCharacter of Academic Research," Library Hi Tech6 (1988): 7-27.

'''This section is based on briefings of Avra Mich-elson by Richard Lucier and Valerie Florence, 7 May1991; sec also Richard Lucicr, "Knowledge Manage-ment: Refining Roles in Scientific Communication,"EDLICOAI Review 25 (Fall 1990): 21-27. For infor-mation on particular projects, sec Welch Library Is-sues, vol. 2, nos. 1, 4, and 6.

functions to include a third function,knowledge management.

In the knowledge management model, li-brarians are teamed with content special-ists, software engineers, and social scientiststo identify the specialized information needsof a constituency and then address the needswith the aid of information technology. Inthis model, the laboratory performs threetypes of work: (1) knowledge base andsoftware development; (2) research andscientific support through ongoing needsassessments and quality control of data, ed-ucation and training; and (3) service throughthe management of the computing andcommunications infrastructure. The socialscientists assess information needs by usingstandard methodologies, such as partici-pant observation, formal and unstructuredinterviews, and document analysis.

The laboratory recently received a three-year grant from the Council on Library Re-sources (CLR) to document the knowledgemanagement model and explore the feasi-bility of implementing the model innonmedical environments. The CLR fundsalso support an invitational symposium onknowledge management. The laboratory'skey projects have been the development ofthe Online Mendelian Inheritance in Man(OMIM) and the Genome Data Base, whichare comprehensive scientific sources usedby geneticists worldwide for gene map-ping, genetic disease diagnosis, and patientcare. These online projects allow an inter-national group of scientists to collect, or-ganize, and electronically distribute mappingand disease information on approximately100,000 genes that regulate human healthand development. The constantly evolvingGenome Data Base is maintained by morethan one hundred &dentists around the wor.i.Lucier considers the database to be a formof dynamic, interactive publication that,unlike static print publications, always pro-vides the most current information andanalysis by the most respected scientificauthorities.

57


Through the development of the GenomeData Base, OMIM, and other projects, thelaboratory has demonstrated that knowl-edge management represents a "practicalworking alternative to existing roles and re-lationships in the creation and managementof scholarly knowledge. "162 Lucier willexpand his work in the development of thenew Center for Knowledge Management atthe University of California at San Fran-cisco.

This section reviewed some of the li-brary community's strategies. The nextsection recommends actions that the archi-val profession can take to respond tochanging research methods. These actionsare an important step toward confrontingthe transformation of scholarly practice thatis as imminent as the new millennium.

CONCLUSION ANDRECOMMENDATIONS

The scholarly use of information tech-nology is resulting in dramatic changes inresearch practices. Essentially two trendsarc evident: one toward end-user comput-ing and the other toward connectivity. Toan increasing extent, social scientists andhumanists are performing their own com-putation in the context of ever greater con-nectivity. The scholarly use of computersand communication technology for re-search and information exchange has bothshort-term and long-term ramifications forarchival practice. In the short term, the ar-chival profession needs to address the in-creasing prominence of network-mediatedscholarship. In the long term, the role ofthe archival profession in the developmentof next-generation archives that operate in

162"Knowledge Management: A Collaboration ofAcademic Scholars, Scientists and Librarians," un-published statement on the three-year project spon-sored by the Council on Library Resources, TheWilliam H. Welch Medical Library, Laboratory forApplied Research in Academic Information (15 July1990).

conjunction with global networks needs tobe defined. The following recommenda-tions suggest concrete actions the archivalprofession can take to address both of theseissues during the next decade:

Establish a presence on the Internet;NREN.Make source materials available forresearch use over the Internet.Create documentation strategies todocument network-mediated scholar-ship and the development of researchand education networks as a newcommunications medium.Develop archival methods suitable foroperation with NREN.Take user practices and computationalcapacity into account in establishingpolicies on the management of soft-ware-dependent records.Recognize and reward initiatives thatadvance (a) the archival managementof electronic records; (b) the responseto scholarly use of information tech-nology; and (c) a network-mediatedarchival practice.

These recommendations are considered inthe three-part discussion below.

Part I: Establishing a Network-Mediated Archival Practice

The archival profession, first and fore-most, must respond to the emergence ofnetwork-mediated scholarship. New meth-ods of searching for sources, communicat-ing with colleagues, disseminating researchfindings, and providing instruction suggestthat scholarly communication is increas-ingly mediated through electronic net-works. The existing Internet and the futureNREN represent the new meeting groundwhere scholars turn for bibliographic in-formation, scholarly dialogues and feed-back, the most current publications in theirfields, and high-level educational offer-ings. Increasingly, full-text versions ofjournals, magazines, newsletters, and even

58


primary sources are available through net-works. In response to this new phenome-non, the archival profession needs toestablish a presence on research and edu-cation networks and to evaluate the impli-cations of new forms of scholarlycommunication for standard archival prac-tice.

But before attempting to introduce pol-icy or collaborative action, the archivalprofession must start using the networks.Indeed, the use of networks is the chiefaction archivists can take in response tochanging patterns of scholarly communi-cation. A presence on the Internet is essen-tial if archivists are to establish credibilityas legitimate network collaborators.

Establishing an archival presence on thenetworks is affordable. Telecommunica-tions hookups involve a modem, commu-nication software, and an e-mail addressprovided through a link to an already ex-isting network connection. For archivistswho do not already possess a modem andwho choose not to use public domain com-munication software, the cost entails a one-time expenditure of, at most, several hundreddollars. Ongoing connect charges in theUnited States arc minimal. Most archivistsshould experience little trouble obtainingan electronic mail address because the ma-jority of campuses are already wired fornetwork connections, as are federal and stateagencies and many private orpanizations andcorporations, especially thou affiliated withscientific research and development. In fact,several hundred archivists' already par-ticipate on BITNET in the network list Ar-chives and Archivists. Once hardware andcommunication software are in place, thearchival profession can become an Internetparticipant.

'"As of June 1992, approximately 440 archivistssubscribed to the Mind Archives and Archivists list-scrv.

Recommendation 1: Archivists shouldbegin monitoring and responding toscholars' intellectual activities conductedon networks.

Besides the standard methods for keep-ing current on research trends, archivistsshould participate in scholarly electronicconferences. To participate, one signs up,or "subscribes," to a conference. Becausethousands of conferences exist, archivistsshould use conference lists and compileddirectories to select those that involve sub-ject areas most closely approximating theholdings of their repository. For instance,a repository strong in women's historysources may subscribe to the lists devotedto women's and gender studies. An insti-tution noted for its collection of pre-CivilWar holdings may choose a conference de-voted to eighteenth century America. So-cial welfare archives may sign up forconferences related to social work, socialactivism, and family studies. Those withstrong collections of utopian records mayselect the Shaker conference. Repositoriesnoted for their hold;-Fs on the arts may jointhe many conferences oil theater, film, anddrama.'"

One way scholars use these conferencesis to exchange information about sourcematerials related to research topics. In aneffort to participate in these dialogues,NARA's Center for Electronic Records be-gan monitoring several scholarly confer-ences in 1991. The conferences offer thecenter a forum for responding to severaldozen additional inquiries each month fromscholars and librarians relating to the cen-ter's holdings. One center staff membercurrently spends about thirty minutes eachday monitoring four BITNET Listservs ontopics related to government documents,

"'Examples of electronic conferences are from Ko-vacs, Directory of Scholarly Electronic Conferences,3rd rev.

50


electronic data sets, social science data lists,and the Vietnam War)65

These conferences not only provide ameans for keeping up with trends in schol-arly research but also provide a mechanismfor establishing a presence on the networksby attaching a name and institutional affil-iation to each communication. As simplis-tic as this sounds, a more substantiveinvolvement with networks can occur onlywhen archivists are familiar with the are-na's discourse and techniques and when thearchival profession is established as a net-work participant. We therefore recommendas an initial action that archivists establisha presence on the Internet by participatinri,in network conferences.

Recommendation 2: The archivalprofession should identify and implementarchival methods appropriate to newforms of scholarly communication.

Establishing a presence on the networksis a necessary first step. But in addition toconference participation, the archivalprofession should pursue archival methodsresponsive to changes in scholarly com-munication. These new archival practicesand techniques include: providing accesson the Internet to source materials in ma-chine-readable form, initially as bit-mappedimages; documenting the activities of net-work-mediated scholarship; and establish-ing archives that operate in the Internet!NREN environment.

2 (a): The archival profession shouldmake source materials available on theInternet. The archival profession shouldmake sources directly available to scholarsvia research and education networks. Thesources should include both records that

"'Conversation between Avra Michelson and Ted(lull of NARA's Center for Electronic Records, 26August 1991; also Ted Hull, "NNXA Reference Re-port," NARA Center for Electronic Records, June1091, draft.

originate in electronic form and those cre-ated in nonelectronic forms. Since thetransfer of nonelectronic records to ma-chine-readable form is a formidable under-taking, this discussion focuses primarily onconversion strategies.

Converting nonelectronic sources to ma-chine-readable form is justified for severalreasons. First, the scholarly expectation thatfull-text materials should be available on-line as a research convenience is unmistak-ably evident and growing.t66 Indeed,electronic document delivery represents theundisputed standard for the informationfield. Second, beyond convenience, con-version of source materials to machine-readable form is essential for analyses thatrely on computational processing. Third,with increasing frequency, the types ofquestions posed by researchers require en-tire electronic libraries of sources, insteadof a single collection, available for com-putational processing. From this perspec-tive, the larger the corpus of convertedcollections, the greater the research value.

As further justification, in the absenceof an archival role in the conversion ofsource materials, the commercial sector iscertain to prevail. This is not to suggest thatmany types of conversion projects wouldnot be more suitable as commercial sectorundertakings. But as the transition to theonline era proceeds, archivists have the re-sponsibility to ensure that publicly avail-able records remain so when converted tomachine-readable form and to alert citizensto the danger of losing the right of freeaccess through inaction.

The proposal to convert source materialsto machine-readable form is neither radicalnor original. Many leaders in the libraryprofession argue that conversion is one of

" "Shrinking travel allocations also may spur re-quests for online access, if the cost of geographicallydispersed archival research exceeds academic bud-gets.


the most important actions librarians cantake to establish a comprehensive record ofscholarship.167 As discussed earlier, somelibraries are already performing pilot con-versions. Further, the Commission on Pres-ervation and Access recently released severalreports recommending that preservationmicrofilming include the generation of dig-ital images.168 In an alternative approach,Cornell University Library, in conjunctionwith Xerox Corporation and the Commis-sion on Preservation and Access, demon-strated the feasibility of directly convertingtext to digital form, avoiding the costs as-sociated with microfilming.'69

In arguing that archivists should convertnonelectronic holdings to machine-reada-ble form, we are not suggesting that it iseither feasible or desirable to convert allrecords. The volume of archival holdingsis simply too great, and many holdings do

"'Sec, for instance, Smith, The Libraria,i, theScholar, and the Future of the Research Library, 71-72. Clifford Lynch also recommends conversion ofsource materials to digital form in "Achieving thePromise: A Proposed Strategic Agenda for Librariesand Networked Information Resources in the 1990s,"unpublished paper presented at the Networks for Net-workers II Pre-Conference, Chantilly, Virginia 17-19Dc. -mber 1990, 18 (Also published under that titlein Networks for Networkers: Critical Issues for Li-braries in the National Network Environment, editedby Barbara Evans-Markuson with Elaine W. Woods[New York: Neal-Schuman Publishers, forthcoming]).

"'Sec Donald J. Waters, From Microfilm to DigitalImagery (Washington, D.C.: Commission on Pres-ervation and Access, June 1991), and Michael Lesk,Image Formats for. Preservation and Access (Wash-ington: D.C.: Commission on Preservation and Ac-cess, July 1990). These reports explore microfilmingas a means to achieve digitization.

"'''The Cornell project, co-managed by Anne R.Kenney and Lynne K. Pcrsonius, involves the directconversion of one thousand volumes of brittle booksto digital form. Half of the volumes are mathematicalbooks, some of which are handwritten or contain for-mulas and graphic images. The Cornell project usesXerox hardware that is capable of producing both dig-ital output and enhanced print output from a digitalcopy. This collaborative effort has produced mean-ingful data on costs, procedures, and models associ-ated with digitization programs useful to the archivalprofession. Sec Kenney and Pcrsonius, "The Futureof Digital Preservation."

not warrant the investment. Rather, our pointis that it is time to begin breaking the tiewith the printed past and establishing aconnection with the machine- readable fu-ture.

Converting source materials to machine-readable form entails the resolution of manyissues that are beyond the scope of this pa-per. However, we wc.Ild like to commenton a few basic archival questions related toconversion: What should be converted?What electronic form should conversion re-sult in? What kind of new descriptive de-vices are necessary to facilitate theindependent use of electronic versions ofsource materials?

What should be converted? Most repo-sitorie:i periodically, if not regularly, mi-crofilm deteriorating collections of enduringvalue. Applying current technology, mi-crofilm preservation projects could be ex-panded or transformed to digital conversionprojects through the development of sev-eral funded, model programs. The benefitof establishing digital conversion programsbased on preservation microfilming is thatmany procedures in place for microfilmingare also suitable for imaging. First, mate-rials for preservation microfilming typi-cally are selected because they are in needof preservation attention and are deservingof wider access. These two elements areadequate criteria for the current selectionof collections to be digitized.'

""Other categories of records also may make goodcandidates for conversion even though they are notdeteriorating. In selecting records primarily to providegreater access, other factors should be considered, in-cluding the nature and extent of use of the records,the institutional visibility or impact afforded by thecc..version, the type of image required for use, thevolume and condition of the records requiring con-version, special labor costs, and the extent to whichconversion can be accomplished through scanning,optical character recognition (OCR), or manual input.But we think it would be a mistake for the archivalprofession to expend much effort at this point on re-fining selection criteria until the results of a numberof digital conversion projects can be analyzed. Fur-

61


Second, the document preparationprocesses used with microfilming are largelycompatible with digital conversion."' Thismeans that handling procedures in place forpreservation microfilming can essentiallybe applied to digitization. Third, micro-filming and digitizing can be intertwinedtechnical processes. That is, while it istechnically possible to generate a microfilmcopy as output from a digitized collection,it is also possible to generate a digitizedcopy of a record set from microfilm output.This means that it is possible to create mi-crofilm and then digitize the output, or dig-itize directly and then generate microfilm.As such, repositories concerned with thelongevity of digital storage mediums, or theirability to move digital data from one gen-eration of technology to another, can con-tinue to rely on microfilm for preservationpurposes and still convert records to ma-chine-readable form.

Repositories that plan to microfilm areencouraged to establish pilot digital pro-grams that draw on many structures alreadyin place for preservation microfilming. Thearchival profession needs tested models toestablish the most cost-effective proceduresfor administering ongoing conversion pro-grams. Pilot projects should provide suffi-cient technical and programmatic guidanceand an awareness of how digital sourcesare used, to equip the profession with theability to implement large-scale digital con-versions.

What electronic form should digital con-versions result in? The profession's assess-

thcr, the Commission on Preservation and Access hascontracted with Margaret Child to reconsider currentcriteria used to select source materials for preservationmicrofilming. Presumably the archival profession willfind the results of this study relevant to digital con-version efforts as well.

"See Archival Research and Evaluation Staff, Op-tical Digital Image Storage System: Project Report(Washington, D.C.: National Archives and RecordsAdministration, 1991), 6; and Kenney and Person-ius,"The Future of Digital Preservation," 9.

ment of appropriate electronic forms willprobably change over time. The overridingconcern, however, must he to identify thekinds of representations patrons need. Dothey need a facsimile image of documents?A stream of straight ASCII text that can bemanipulated? ASCII text encoded with tagsthat identify document structures and for-mats? Although the electronic forms thatpatrons need depends on the type of re-search they are conducting, very little isknown about the actual use of electronicdocuments for different types of research.

Trends in the technology suggest that inthe future the archival profession should beable to provide access to electronic sourcesboth as bit-mapped images and encoded text.But current limitations make large-scaleencoding of text an unrealistic undertaking.For many reasons, the existing methods ofperforming ASCII conversions, manual keyentry or automatic optical character rec-ognition (OCR) are inadequate. For ex-ample, the cost of performing key entrywith great volumes of materials is prohib-itive, and OCR processes are unreliable withhandwritten script and unusual type fonts.In contrast, bit-mapped conversions, whichresult in image representations (like fac-similes, but potentially of far greater res-olution), are readily attainable with today'stechnology. Further, automatically con-verting bit-mapped images of modern printeddocuments to ASCII is typically considereda straightforward process (equivalent toOCR). If desired, encoded text can be gen-erated from the ASCII version, providedthe relevant structural information has beenretained.

ASCII and encoded text differ from bit-mapped images in that the latter cannot besearched and computationally processedwithout considerable programming. It is

highly probable, however, that softwaredesigned to encode text automatically willimprove and reduce in cost during this dec-ade. If this happens, it may be feasible tojustify large-scale textual encoding. Until

62


then, sources not amenable to OCR shouldbe converted to bit-mapped images. Butsince bit-mapped images will not satisfy theresearch needs of certain scholars, archi-vists should monitor advances in OCR andstructure-encoding software.

What kind of new descriptive devices arenecessary to facilitate independent use byresearchers of electronic versions of sourcematerials? Digital versions of large archi-val collections will need specialized find-ing aids, descriptors, navigational aids, orinformational hooks to facilitate their in-dependent use.12 Developing these findingaids and navigational tools represents a keychallenge for the information profession.Nonetheless, it would be ill-advised to con-vert unstructured and voluminous collec-tions to machine-readable form, or to makecollections that originate in machine-read-able form available for independent use,without addressing the need for a descrip-tive system suitable to the electronic envi-ronment. As a further complication, standardbibliographic approaches to retrieval areproving an inadequate method for locatingand managing remote electronic text banks.But metadata, data about data that archi-vists typically collect about z body of rec-ords, may serve as the basis for asupplementary descriptive system to com-plement existing bibliographic informa-tion. Administrative histories, accessionrecords, and other contextual data used toestablish the provenance of a collection mayprove very useful in retrieving informationfrom electronic sources in the absence ofhuman intermediaries.'73

"For a justification of the need for new accesstools, see Clifford A. Lynch and Cecilia M. Preston,"Internet Access to Information Resources," in An-nual Review of Information Science and Technology(ARIST), vol. 25, edited by Martha E. Williams (Am-sterdam: Elsevier Science Publishers B.V., 1990); andLynch, "Achieving the Promise," 24-25.

'"Charlcs Robb, at the Kentucky Department forLibraries and Archives, is developing a locator systemfor statewide information using metadata to comple-

It is encouraging to note that contextualinformation accreted to each document inrecords originating in machine-readable formis likely to be greater than in their printcounterparts. For example, e-mail mes-sages interchanged on the Internet identifythe sender and institution, the receiver(s)and institution(s), the date and time oftransmittal, and the subject of the com-munication. Archival intervention into thedesign phase of software could result in theaccumulation of other metadata that wouldbe useful for both accountability and re-trieval purposes. We therefore endorse theNational Historical Publications and Rec-ords Commission's proposal to research theimplications of capturing and retaining data,descriptive information, and contextual in-formation in electronic form, and we spec-ulate that the findings of this research canalso advance the development of descrip-tive systems suitable for independent useby end-users."

2 (b): Archivists should develop andimplement a strategy for documentingnetwork-mediated scholarship as a newphenomenon of scholarly communica-tion. A key finding of this report is thesubstantial level of scholarly activity beingconducted outside the purview of tradi-tional archival practice. Network-mediatedscholarship raises two very different but re-lated documentation issues for the archivalprofession. The first is the need to docu-ment the origin and administration of re-search and education networks themselves.The second is the need to document the

mcnt existing bibliographic information. This ap-proach may be useful in developing descriptive systemsthat provide access to information within a collectionas well. Sec Charles Robb, "Networking Metadata inKentucky," unpublished paper presented to the Na-tional Association of Government Archivists and Rec-ords Administrators, Chicago, July 1990.

'"This recommendation is part of Research Issuesin Electronic Records, 10-11. Charles Dollar also ar-gues that archivists should define metadata elementsin his report, The Impact of Information Technologieson Archival Principles and Methods, 98-100.

63


programmatic use of these networks for theadvancement of scholarship and learning.As an approach and process, documenta-tion strategy' represents a tool that archi-vists can use to address these documentationproblemse.g., to identify the key agentsoperating in the network environment, todetermine the universe of documentation thatexists, and to develop recommendations forpreserving documentation of enduring value.

The large number of agents and the globalscope of activities associated with researchand education networks suggests that ar-chivists may want to collaborate and seekmulti-institutional funding for documenta-tion projects. At least three types of archi-val repositories are well-positioned to initiatesuch projects: (1) college and university ar-chives, because network research and ed-ucation efforts originate largely in academia;(2) government archives, because govern-ment is a key partner in most academic-based collaborative research projects andnetwork-mediated education programs(either as a fonder, research associate, ornetwork administrator); and (3) disciplinehistory centers (such as the American In-stitute for Physics, the Beckman Center forthe History of Chemistry, and the BabbageCenter), as these centers, by definition, ex-plore a universe of documentation and areheavily devoted to science and technology,disciplines in which network-mediatedscholarship is currently the most pervasive.

The documentation effort should identifykey representatives to participate in stra-tegic discussions, such as those from theInternet and scholarly communities, aca-demic computing centers, private industry,

("Two seminal essays that together provide an in-tellectual foundation for the concept of documentationstrategy, as well as an examination of procedures andcase examples arc: Helen Willa Samuels, "Who Con-trols the Past," American Archivist 49 (Spring 1986):109-24; and Larry Hackman and Joan Warnow-Blew-ett, "The Documentation Strategy Process: A Modeland a Case Study," American Archn at 50 (Winter1981): 12-47.

and government research laboratories. Agoal of the effort should be to clarify theprincipal records-creating agents and theactivities that warrant preservation. Theproject report should include a statement onthe nature of electronic archival records andthe relationship of these sources to non-electronic documentation.

This recommendation involves a certainurgency because existing documentationtends to be transient. In fact, compilers ofseveral network directories report that atleast a half dozen recent scholarly elec-tronic conferences are already defunct, asare more than a dozen electronic newslet-ters and journals." Some argue that theseefforts become inactive when moderatorsswitch jobs and no longer possess theequipment or time to continue in that roleor when the interest in a once-timely topic,such as the Gulf War, dissipates. Insteadof papers removed to an attic for storage,the records of a defunct electronic confer-ence typically take the form of a mass ofbits abandoned on a campus mainframecomputer or file server, awaiting a purgeof the file by a systems administrator in aroutine cleanup. Given this situation, aca-demic computing staff represent key con-tacts for campus archivists concerned withnetwork files. State archivists also shouldbe concerned with the transient nature ofnetwork communication because network-mediated distance education programs areunder way in most state departments of ed-ucation."' In summary, archivists at insti-

("Correspondence via Bitnet on 23 August 1991between Avra Michelson and Diane Kovacs, compilerof Directory of Scholarly Electronic Conferences; also,a list of defunct electronic journals and newslettersappears in Michael Strangelove, Directory of Elec-tronic Journals and Newsletters.

"'See two reports by Barbara Kurshan: StatewideTelecommunication Networks: An Overview of theCurrent State and the Growth Potential (Roanoke,Va.: Educorp Consultants, December 1990), and withMarcia Harrington, Statewide Education Networks:Survey Results (Roanoke, Educorp Consultants,April 1991). Both are available through Bitnct fromthe author (KurshanOt vtvml.bitnct).


tutions that support online scholarlycommunication are urged to seek fundingfor programs to identify and preserve valu-able records related to the administrationof networks comprising the Internet andnetwork-mediated scholarship.

2 (c): The archival profession shouldsupport the development of archives de-signed to operate on global networks. Thegrowth in network-mediated scholarshipsuggests that the archival profession needsto define its role in relation to the devel-opment of archives designed to operate inthe global network environment. The needfor archival operations on research and ed-ucation networks is already widely recog-nized by the network community. Forexample, program planning in the networkcommunity involves archival concerns. Ata biannual meeting of the Coalition for Net-worked Information (CNI), many subcom-mittees reported on work that entailed theresolution of archival functions in a net-work environment.'78 Although separatefrom the archival profession, CNI rep-resents a group that is identifying issuesrelated to the archiving of network re-sources.

Further, the development of electronicnetwork archives is already evident. Mostmoderators of scholarly electronic confer-ences maintain an archives of the confer-ence's transactions accessible via thenetwork.' Others are capturing subject-oriented transactions across research andeducation networks and making the ar-

chives available on the Internet.'''" Still

"'Observation by Avra Michelson at the CNI springmeeting, 18-20 March 1991, Washington, D.C.

"'Correspondence by Avra Michelson with DianeKovacs, compiler of the Directory of Scholarly le( -Ironic Conferences. on 23 August 1991 via BITNET.

''For instance, Edward Vicln.letti at MSEN in AnnArbor, Michigan, collects and makes available de-scriptions of network resources publicized on the net-works femvOt mscn.coml; Nathan Torkingion at theComputing Services Center in Wellington, New Zca-

others are exploring commercial models forpreserving both volume and breadth in net-work transactions.'8' Those involved innetwork archiving communicate with oneanother through electronic conferences aboutsuch issues as data compression algo-rithms, information filtering techniques, andfile transfer protocols.' This means thatseminal models for microarchiving withina network environment are already in place,while those for archiving on a grander scaleare either on the drawing board or beingprototyped, each established apart from thework of the traditional archival profession.

Archivists must not underestimate thesignificance of these actions. The future ofthe archival mission in relation to elec-tronic communication is being defined bya set of agents wholly separate from thework of the traditional archival profession.Further, the scope of the new archival agentsis apt to grow as NR.EN evolves into a pieceof the backbone used in the conduct of of-ficial government business.'83 The appro-priate role for the archival profession in thisarena remains undefined, but the key ques-tions are clear. Can the archival professionestablish the political authority necessary toimprove the archival methods used in con-junction with research and education net-work transactions, and can it rise to the

land, maintains a publicly accessible electronic ar-chives of text on information management capturedfrom network exchanges [gnatf!ikauri.vuw.ac.nzl.

' "'Vielmetti has developed commercial models forarchiving select network transactions.

"'The key electronic conference where these issuesarc discussed is comp.archtvesoadmin moderated byEdward Vielmetti.

1" the U.S. Office of Personnel Management re-cently released guidelines already effective for the ac-ceptance of electronic signatures in the conduct ofofficial government business. With the issue of elec-tronic signatures resolved, the use of networks in of-ficial government business can he expected to increaserapidly. Scc U.S. Office of Personnel Management,Federal Personnel Manual, Chapter 293, Subchapter6, Installment 39, 1 April 1991 (Washington, D.C.:U.S. Government Printing Office, 1991).

BEST COPY AVAILARtE

65


challenge of defining an archival practicesuitable not only for electronic records butalso for a new communication medium?

Part II: Establishing a Strategy for theFuture Usability of Electronic Records

No discussion of information technologytrends can ignore the issues surrounding thestorage and use of electronic records them-selves. Although this subject has been dis-cussed in the archival literature,184 our focushere is on the scholarly research perspec-tive. This article has concentrated on thenear-term effects of information technol-ogy on current scholarly practice and prod-ucts. It is equally important, however, toconsider how new ways of producing rec-ords (whether they are of scholarly originor not) will affect future users of those rec-ords. In particular, how will the creationof electronic records affect future scholarswhen they use such records in their re-search? What current technology trends bearon the ways these future scholars will per-form their research andby implicationon the ways future archives will have toserve them?

One of the main advantages of electronic

'"A selection of the key literature includes DavidBcarman, Archival Methods, Archives and MuseumInformatics Technical Report, no. 21 tSpring 1989);Advisory Committee for the Co-ordination of Infor-mation Systems (ACCIS) Management of ElectronicRecords: Issues and Guidelines (New York: UnitedNations, 1990); U.S. House, Committee on Govern-ment Operations, "Taking a Byte out of History: TheArchival Preservation of Federal Computer Records,"House Report 101-978 (Washington, D.C.: Govern-ment Printing Office, November 1990); Research Is-sues. in Electronic Records (St. Paul, Minn.: publishedfor the National Ilistoricai Publications and RecordsCommission, Washington, D.C., by the MinnesotaHistorical Society, 1991); David Beaman, ed., Ar-chival Management of Electronic Records, Archivesand Museum Informatics Technical Report no. 13(Pittsburgh: Archives and Museum Informatics, 1991);Nihrgarct Iledstrom, "Understanding Electronic In-cunabula: A Framework for Research on ElectronicRecords, American Arrhilivi 54 (Summer 1991): 11454.

information is that it is usually digital, whichensures that it can be copied and transmit-ted without loss or degradation. Yet, iron-ically, the preferred media on which thisdigital information is storeddisk, tape, andeven CD-ROMhave far shorter shelf livesthan acid-free paper or microfilm. More-over, these media tend to become unusablelong before they reach their ultimate agelimits. As technology evolves, it quicklyreaches a point where older media can nolonger be accessed by existing equipment.It is only somewhat facetious to expressthis irony by saying that digital data lastsforeveror five years, whichever comesfirst. There is no theoretical problem withstoring digital information on archival me-dia, including microfilm, but such mediaare not in popular use, nor does evidencesuggest that they will become so. Thisproblem has a straightforward, thoughcumbersome and relatively expensive, so-lution: to "update" or "migrate" data, thatis, to copy the data from one medium toanother as media wear out or become ob-solete. Although various technology trends(including the continued development ofoptical storage devices such as CD-ROM)may improve the longevity of media, theoverall trend of continued improvement andreplacement of media implies that the prob-lem of obsolescence is unlikely to disap-pear in the foreseeable future.

Despite this problem, it is axiomatic thatthe records produced by governments, or-ganizations, individuals, and researchersthemselves will become increasingly"electronic" over the next few decades.This implies that scholars of the not-so-dis-tant future will he confronted increasinglywith electronic records as both the primaryand secondary source materials for their re-search. Moreover, the current first gener-ation of such records will have uniquehistorical significance, representing the mostdrastic change in the form and conceptionof records since the introduction of print-


ing, or even of writing. t85 Yet at the cur-rent rate of technological change, electronicdocuments (and the programs that produceand access them) typically become obsoleteand unusable in a distressingly short time.How can the loss of this unique generationof records be prevented? How will scholarsbe able to understand and analyze thesedocuments decades from 'now? How canarchives hope to preserve such documentsin a form scholars will be able to use?

Furthermore, media longevity is only apartand in many ways the easier partof the problem. Migrating data can keepthem "accessible," but to be usable theymust be more than just accessible: they mustalso be interpretable. The data stored ondigital media are simply binary digits (bits),which cannot he interpreted without atranslation of the codes they represent andan understanding of the structure in whichthey are placed on their media. Migratingdata may solve the media longevity prob-lem, but by itself it does not solve the largerproblem. Like an illiterate monk dutifullycopying text in a lost language, migrationmay save the bits but lose their meaning.Even if we assume that the media longevityproblem can be solved, what technologytrends bear on whether electronic recordswill be interpretable in the future?

This issue is often referred to as that ofsoftware-dependent records, though thereis somewhat more to the problem than thisterm suggests. Software-dependent recordsare electronic documents that can be readonly by using some particular piece ofcomputer software (that is, a program). Ex-amples of software-dependent records in-clude documents created with wordprocessing or electronic publishing pro-grams, spreadsheets, databases, geo-graphic information systems (GISs), and

'"Scc Jay David Bolter, "Text and Technology:Reading and Writing in the Electronic Age," LibraryResources and Mc-limed Services 31 (January/March1987): 12-23.

hypertext/hypermedia. Though a data filefor such a document may be saved on somemedium (such as a disk), the file can beproperly interpreted only by its software;the document itself is accessible (and insome cases may come into existence) onlyby running the software."6 This can bethought of as the problem of "preserving"electronic documents. However, in this case,"preservation" means more than simplypreserving media; unlike printed records,electronic records require software andhardware in order to be accessed and in-terpreted.

The obvious way to access a software-dependent document is to run the softwarethat produced it. However, programs them-selves quickly become obsolete, and run-ning obsolete software is currently verydifficult. Any given program works onlyon certain computers and only with certainsystem software. This means that accessinga document may actually require the userto run this entire hardware and softwareenvironment. In fact, what is typically meantwhen a document is called "software-de-pendent" is that it can be accessed only byrunning the entire hardware and softwareenvironment in which it was created. Theproblem is that such environments becomeobsolete in the blink of an archival eye, andmaintaining them in working condition be-yond that time is a complex, costly, andultimately futile task."' Preserving elec-

'In a very real sense, all electronic documents arcsoftware-dependent. Simple text and numeric files arenot typically referred to as "software-dependent" onlybecause they are encoded and stored in fairly straight-forward ways that currently arc considered obvious(e.g.. simple sequences of ASCII codes representingchismetcrs). Yet even these cannot he accessed or in-terpreted without hardwac and software that can un-derstand their encoding.

"Tot several discussions of this issue. sec DavidBcarman, Collecting Software: A Nese Challenge forArchives and Museums, Archives and Museum Infor-matics Technical Report no. 2 (Pittsburgh: Archivesand Museum Informatics, 1957, reprinted 199(1); andCoalition for Networked Information Director Paul Evan

67


tronic documents in a way that will allowfuture access to their form and meanirg istherefore not straightforward.

There appear to be two general ap-proaches to providing meaningful futureaccess to software-dependent documents.Either they must be transformed in someway that makes them independent of thesoftware that created them, or they must besaved along with some kind of descriptionof their associated software sufficient to al-low accessing them as was originally in-tended. The first approach might befacilitated by the development of standardsfor various kinds of documents, whereasthe second approach might he facilitated bythe development of formal models of com-putation. Several technology trends hear oneach of these approaches.

Software-dependent documents might bepreserved in a usable form by transformingthem so that they become "software-inde-pendent" in some way. For each recog-nized category of program now in use (wordprocessing, database, spreadsheet, etc.) astandard data file format might be defined,along with a standard set of functions thatany such program can perform. For ex-ample, most word processing programsprovide functions for displaying pages oftext, footnotes, and chapter headings. Inprinciple, a data file for a document fromany such program could be transformed intosome standard format, and its behavior couldhe duplicated by some standard pro-gram.'" This transformation process would

Peters, "The Machine Aspects of Preservation, un-published paper (ca. 19(M).

'''SGN1L is an attempt to provide a standard forthis kind of text, though it is generally recognized thateven a standard for text will not magically remove allthe incompatibilities among existing word processingformats. Another example of this approach that hasbeen discussed in the literature involves relational da-tabases. The argument has been made that a databaseproduced by one relational database management sys-tem (ft DBNIS) can be transformed into a standard formthat can be used by any other RDI3MS. Sec the Na-tional Archives and Records Administration's rc

have to be repeated periodically as the stan-dard itself evolved. Standardization trendssuch as those discussed above may helpmake this possible. However, there mayalways be programs whose behavior cannotbe duplicated by any standard or which donot even fit into the recognized categoriesof programs (e.g., word processing or da-tabase). As noted above, standards gener-ally lag behind the advancing technology;until computer science becomes far betterformalized (that is, based on firm, theoret-ical underpinnings), there will always beprograms that defy the most well-con-ceived efforts at standardization. Policiesin various organizations may attempt to forcethe use of programs that conform to stan-dards, but current trends of technologicalinnovation make enforcement difficult be-cause users find it hard to resist new ca-pabilities, whether they are standard or not.

Even aside from standardization efforts,a "natural migration" of documents occursas the programs on which they dependevolve through successive versions. Newversions of programs often provide upwardcompatibility to allow old documents tomigrate into the required updated forms. Itmay be possible, as has been suggested,'"to rely to some extent on this phenomenonto keep documents accessible. The effec-tiveness of this approach, however, is lim-ited by the fact that periodic upheavals occurin software paradigms. Two examples of

spouse to the recommendations in "Taking a lute Outof History, and Kenneth Thibodeau, "To Be Or Nib(to Be: Archives for Electronic Records. in AnimalManagement of Electronic Records, edited by DavidBearman, 1-13. Although this may be true to a largeextent, it is a relatively atypical example; relationaldatabase systems are one of the very few higher levelapplications for which a formal (mathematicalt com-putational model exists. Most other common appli-rations. such as word processing, spreadsheets,hypertext/hypermedia. or GISs arc not nearly this wellformalized.

'""Dollar. The Impact of Information Technologic%on Archival Prourples and Methods. Chapters 1-4,draft version.

300 American Archivist I Spring 1992

such upheavals are the change from simpletextual tables to spreadsheet:; and the changefrom hierarchical databases to relationaldatabases. Such upheavals make it difficultenough to transform documents that arecrucial to the daily functioning of organi-zations; transforming old documents thatare no longer in use may require more ef-

fort than most organizations are willing tospare.

The alternative to transforming software-dependent documents into software-inde-pendent form is to interpret them by some-how using the software that they dependon, despite its being obsolete. Interpreta-tion does not necessarily require actuallyrunning the software. If a complete de-scription existed of how a program inter-prets its data files in accessing a document,it would not be necessary to save the soft-ware itself (or its environment). The doc-ument c^uld be accessed by following thisdescription, effectively recreating the be-havior of the software. In most cases, un-fortunately, such complete descriptions ofsoftware exist only in the form of the soft-ware itself. Computer science is not yet verygood at describing what complex software

Interpreting a software-dependent docu-ment by using the software it depends ontherefore requires either being able to runthe software that has been saved along withthe document (by effectively recreating itsenvironment), or interpreting the softwarewithout running it (effectively recreating,or emulating, its behavior). The former op-tion requires saving vast (though finite)

"'There: arc exceptions to this, such as the rela-tional database case discussed above. In general, how-ever, current formal descriptive techniques cannotcapture the "human level" semantic behavior of pro-grams. What is required is a computational theory,not of how programs work, but of what they do fortheir users; i.e., a theory of human informationprocessing that describes such things as how humanscreate and use documents and how humans interactwith each other to perform research.

documentation for the software and its en-vironment, including detailed technical de-scriptions of any required hardware and allof its components.'" The latter option re-quires a more sophisticated computationaltheory than is currently available, i.e., anunderstanding of the semantics of whatprograms do at the human level of infor-mation processing and how they do it.Without such a theory, it remains impract-ical to interpret software except by runningit in its original hardware and software en-vironment.' Current trends toward im-proving the formal specification of systemsand environments may facilitate the formeroption, whereas trends toward modelinghuman level computational processes mayfacilitate the latter.'93 Finally, it should benoted that the overriding trend toward in-creased computational power may enablethe performance of tasks that now appearunthinkable, just as we now routinely per-form computations that were unthinkable adecade or two ago. Such future tasks mightinclude automatically decoding lost filestructures, transforming obsolete documentformats through successive generations ofstandards, or recreating the behavior of ar-

"'Although this is a huge task, it may not be in-surmountable: These environments could not exist inthe first place if they did not already posses' suchtechnical descriptions. Furthermore, many of thesedescriptions are already in patent or copyright offices,where they might he accessible for this purpose.

"'Recreating the behavior of a program by figuringout what it was intended to do and building a newprogram that does what the original program did issometimes called "reverse engineering." it is widelyrecognized as a difficult task.

'"Advances in computational theory may enablefuture generations of scholars to understand how weviewed and manipulated our documents far better thanwe understand it ourselves. The present is, after all,only the dawn of the information age, and thc organ-izing principles of the new "computation" paradigmare only beginning to mew. Future scholars mayhave a far better formal mathematical) under-standing of computation and human informationprocessing; this would provide them with a theoreticalframework that could explain any kind of software-dependcnce and allow them to reconstruct past capablinks at will.

69

BEST COPY AVAILABLE


chaic computational environments fromimperfect documentation. These computa-tional possibilities may well allow futuregenerations of scholars to derive the equiv-alent standard form of obsolete software-dependent documents in their archives orto reproduce the behavior of the softwarethat produced them at will.

In the context of scholarly research andinformation technology, the issue of soft-ware-dependent records can be phrased interms of two questions: "How can accessto software-dependent documents be pro-vided to future scholars?" and "What tech-nology can help to provide this access?"

To answer the first question, one mustarticulate certain assumptions about whatkinds of access future scholars are likely toneed to such documents and what they willdo with them after they have accessed them.The software used to create a software-de-pendent document determines the capabil-ities available to its author for viewing andmanipulating it. How accurately mustscholars be able to reproduce these capa-bilities? Is it enough to preserve the contentof such a document without its form? Is itenough to preserve its content and formwithout being able to recreate the way itsauthor saw it'?'94 These questions requiremaking assumptions about the kinds of re-search future scholars will perform, whichcan be informed by analyzing trends inscholarly practice, as undertaken above.Given such assumptions, how would alter-

'' Margaret tledstrom suggests that "The solutionto preservation of electronic records lies somewherebetween the present appro .h of preserving only datavalues and the need to retain all of the functionalityof an active records system. There are tremendousadvantages to returning the descriptive, search, re-trieval, and manipulation functions of some auto-mated systems. The ability to retain more complexelectronic records and more of the useful functionalityof automated systems, however, will remain beyondthe control of archivists if they continue to utilize onlythe tactics [that] have been employed in the past.""Archives: To Be or Not to Be: A Commentary," inArchival Management of Electronic Records, 28.

native software-dependent records manage-ment policies constrain or enhance thecapabilities of future scholars in perform-ing their research using software-dependeatdocuments?195

To answer the second question, one mustarticulate other assumptions about the tech-nological future (while recognizing that allsuch assumptions are speculative).'96 Inparticular, what do current technology trendsimply about future capabilities for access-ing software-dependent records?

Saving data files for software-dependentdocuments is a necessary but insufficientstep toward making them usable. As dis-cussed above, data can be migrated to newmedia to keep them readable, but data mustbe more than just readable to be usable:They must also be interpretable. Is theresome way to transform such documents be-fore saving them in archives, so that theycan be used without their software? If so,what would this sacrifice in terms of beingable to recreate the author's original capa-bilities? Alternatively, is there some prac-tical way of saving the software with eachdocument (in particular, without maintain-ing obsolete hardware/software environ-ments) so that the software itself can beused in the future to access the document?If solutions to these problems are not foundand implemented soon, much of the firstgeneration of electronic documentsrep-resenting a unique historical event in theevolution of recordswill be irretrievablylost.

To summarize, there appear to be twogeneral approaches to solving this problem,as discussed in the archival literature:Transform each document and save it insoftware-independent form, or save thesoftware for each document in some way

"'This article raises this question without attempt-ing to ail,.wer it. Our point is that the assumptions thatunderlie any answer must be made explicit.

l'"'The archival literature on this subject has not yetgenerally articulated such assumptions.


that allows it to be used in the future tointerpret the document.

The solutions that have been proposed inthe literature for both approaches (e.g.,translating documents into one of a fewcurrent standard forms or keeping hard-ware/software environments running for aslong as possible) appear to be based on im-plicitly conservative assumptions about fu-ture technology. It seems likely, however,that inevitable advances in computationaltheory and computational power will pro-duce a vastly more capable future, enablingbetter, longer-range solutions to one or bothof these approaches. This analysis has im-plications for the actions that should be takennow to ensure the preservation of these rec-ords. We see the following recommenda-tion as a necessary step toward deciding onsuch actions.

Recommendation 3: The archivalprofession should establish an evolvingpolicy on the management of software-dependent records, informed by anassessment of the kinds of access futureTcholars will require to such records anda realistic assessment of thecomputational capabilities that will beavailable in the future.

Because of the short effective life of mostelectronic media and the rapidity with whichsoftware-dependent documents tend to be-come obsolete and unusable, this recom-mendation has an urgent aspect: Electronicrecords of enduring value that are not ap-propriately preserved will soon be lost toposterity.

The archival profession should take stepsto ensure that its evolving software-depen-dent records management policy considersthe ways that future scholars are likely touse these records and the ways that futuretechnology is likely to facilitate this use.Assessments, such as the one we have 'un-dertaken here, which attempt to analyzetrends in scholarly practice and information

technology should be used to attempt toproject future needs and capabilities thatare realistic, i.e., neither wishfully gran-diose nor unimaginatively chained to thepast. These projections should be used toproduce evolving policies aimed at themoving target that is the future.

Evolving trends in scholarly practiceshould be sought out by the archival profes-sion, in an attempt to coordinate the de-velopment of archival policies with theperceptions and projections of those schol-ars who represent the leading edge of changein scholarly research practice. This coor-dination might be achieved through sched-uling paper sessions or panel discussionson evolving scholarly practice, to be pre-sented at archives and library science con-ferences and at conferences in variousscholarly disciplines. Workshops, journals,or network discussions might also be or-ganized on this subject, soliciting input fromscholars while establishing the archivalprofession as a focal point for this inquiry.

Similarly, archivists should seek outevolving trends in technology, with partic-ular emphasis on formalisms and standardsfor represe sting various kinds of docu-ments ar.a on formal models of computa-tion and human information processing,which ultimately may make it possible todescribe the behavior of software in waysthat will allow it to be emulated in the fu-ture. In this endeavor, archivists should ac-tively engage the computer sciencecommunity as a partner, for example byorganizing sessions or panels on these sub-jects at both computer science and archivesconferences.

Finally, archivists should engage in anongoing effort to understand the most likelyfuture uses of software-dependent records,and they should articulate their assump-tions about future scholarly practice and fu-ture computational capability as a

prerequisite for proposing archival policieson the management of software-dependentrecords.

71


Part III: Recognizing and RewardingLeadership

Recommendation 4: The archivalprofession should reward activities thatadvance archival practice withinformation technology, electronicrecords, and electronic communication.

The archival profession must respond tothe changing patterns of scholarly com-munication and the emergence of a newcommunication medium. Leadership ca-pable of guiding the archival professionshould be cultivated by promoting graduateeducation programs, collaborative projects,and professional coalitions targeted at ad-vancing archival operations in global net-work environments. The Society ofAmerican Archivists ,,nd the field's otherprofessional associations should recognizeand reward excellence in research, pilotprojects, collaborative associations, andprogrammatic implementations related tothe management of electronic records, theuse of information technology to improvearchival practice, and the establishtnent ofarchival methods suitable to modern com-munication mediums.

BIBLIOGRAPHY

Adman, Peter. "Computers and History."In Information Technology in the Hu-manities: Tools, Techniques and Appli-cations. edited by Sebastian Rahtz, 92-103. New York: Halsted Press, 1987.

Advisory Committee for the Coordinationof Information Systems. Management ofElectronic Records: Issues and Guide-lines. New York: United Nations, 1990.

Alberico, Ralph. "The Development of an`In. __nation Superhighway'." Com-puters in Libraries 10 (January 1990):33-35.

Alexander, Michael. "But Can You ReadIt Like a Book?" Computerworld 24 (19November 1990): 18.

Anderson, Paul A. "Using Artificial Intel-

ligence to Understand Decision Makingin Foreign Affairs: The Problem of Find-ing an Appropriate Technology." In Ar-tificial Intelligence and National Security,edited by Stephen J. Cimbala, 133-46.Lexington, Mass.: Lexington Books,1987.

Archival Research and Evaluation Staff.Optical Digital Image Storage System:Project Report. Washington, D.C.: Na-tional Archives and Records Administra-tion, 1991.

Arms, Caroline, ed. Campus Strategies forLibraries and Electronic Information.Bedford, Mass.: Digital Press, 1990.

. "Libraries and Electronic Infor-mation: The Technological Context, PartTwo." EDUCOM Review 24 (Fall 1989):34-43.

Arms, William Y., and Thomas J. Mich-alak. "Carnegie Mellon University." InCampus Strategies for Libraries andElectronic Information, edited by Caro-line Arms, 243-73. Bedford, Mass.:Digital Equipment Corporation, 1990.

Association for Computers and the Human-ities, the Association for ComputationalLinguistics, and the Association for Lit-erary and Linguistic Computing. Pro-posal for Funding for an Initiative toFormulate Guidelines for the Encodingand Interchange of Machine-ReadableText. Prepared for the National Endow-ment for the Humanities, 1988.

Baerson, Kevin M. "Marines Put LibraryOnline." Federal Computer Week 5 (2September 1991): 1, 4.

Bailey, Charles W., Jr. "Electronic (On-line) Publishing in Action . . . The Pub-lic-Access Computer Systems Review andOther Electronic Serials." Online 15(January 1991): 28-35.

"Intelligent Multimedia Com-puter Systems: Emerging InformationResources in the Network Environ-ment." Library Hi Tech 8 (1990): 29-41

Bantz, David A. "The Values of the Hu-


manities and the Values of Computing."In Humanities and the Compute NcwDirections, edited by David S. N it, 27-37. Oxford: Clarendon Press, 1990.

Bearman, David. Archival Methods. Ar-chives and Museum Informatics Tech-nical Report no. 21. Pittsburgh: Archivesand Museum Informatics, 1989.

, ed. Archival Management of Elec-tronic Records. Archives and MuseumInformatics Technical Report no. 13.

Pittsburgh: Archives and Museum Infor-matics, 1991.

. Collecting Software: A New Chal-lenge for Archives and Museums. Ar-chives and Museum Informatics TechnicalReport no. 2. Pittsburgh: Archives andMuseum Informatics, 1987, reprinted1990.

. Towards National InformationSystems for Archives and ManuscriptRepositories: The NISTF Papers. Chi-cago: Society of American Archivists,1987.

. "User Presentation Language inArchives." Archives and Museum Infor-matics 3 (Winter 1989-90): 3-7.

Bell, Rudolph M. 'User Perspectives andRequirements: Creator of Non-Biblio-graphic Databases Has to Share withOthers." Unpublished paper presentedat the Library of Congress Network Ad-visory Committee Meeting, Washing-ton, D.C., 29-31 March 1989.

Bikson, T. K., and S. A. Law. ElectronicMail Use at the Bank: A Survey and Rec-ommendations. Washington, D.C.: WorldBank, Information, Technology, and Fa-cilities Department, September 1991.

. Electronic Information Media andRecords Management Methods: A Sur-vey of Practices in United Nations Or-ganizations. A RAND Note (N -3453-RC). Santa Monica, Calif.: RAND Cor-poration, 1991,

Blake, Virgil L. P., and Rene Tjoumas,eds. Information Literacies for the

Twenty-First Century. Boston: G.K. Hall,1990.

Bolter, Jay David. "Text and Technology:Reading and Writing in the ElectronicAge." Library Resources and TechnicalServices 31 (January/March 1987): 12-23.

Booker, Ellis. "Vive le Minitel." Teleph-ony 215 (8 August 1988): 24.

Bortnick, Jane. "Information TechnologyOverview." CRS Review 11 (July/Au-gust 1990): 1-3.

Bourcier, Danielle. "Interpretation in Le-gal Expert Systems: From Private Con-viction to the Formalisation of Rules."In Interpretation in the Humanities: Per-spectives from Artificial Intelligence, ed-ited by Richard Ennals and Jean-ClaudeGardin, 215-28. Library and Informa-tion Research Report no. 71. Cam-bridge: University Press, 1990.

Bourlet, Caroline, and Jean-Luc Minel. "ADeclarative System for Setting Up a Pro-sopographical Database." In History andComputing, edited by Peter Denley andDeian Hopkin, 186-91. Manchester, En-gland: Manchester University Press,1987.

Boyce, Bert B., and Kathleen M. Heim."The Education of Library Systems An-alysts for the Nineties." In Computing,Electronic Publishing and InformationTechnology: Their Impact on AcademicLibraries, edited by Robin Downes, 69-76. New York: Haworth Press, 1988.

Brent, Edward. "Is There a Role for Ar-tificial Intelligence in Sociological Theo-rizing?" American Sociologist 19(Summer 1988): 158-66.

. "New Approaches to Expert Sys-tems and Artificial Intelligence Program-ming." Social Science Computer Review6 (Winter 1988): 569-78.

Brilliant, Richard. "How an Art HistorianConnects Art Objects and Information."Library Trends 37 (Fall 1988): 120-29.

Brown, Thomas E. "The Society of Amer-

73


ican Archivists Confronts the Com-puter." American Archivist 47 (Fall1984): 366-82.

Brunner, Theodore F. "Data Banks for theHumanities: Learning from ThesaurusLinguae Graecae." Scholarly Commu-nication 7 (Winter 1987): 1, 6-9.

Bush, Vannevar. "As We May Think."Atlantic Monthly 176 (July 1945): 101-08.

Carpenter, Edwin H., and Rick D. Axel-son. "Statistical and Graphical ResearchMethods: State of the Art." Social Sci-ence Computer Review 7 (Winter 1989):503-34.

Carvalho, Joaquim. "Expert Systems andCommunity Reconstruction Studies." InHistory and Computing 11, edited by Pe-ter Den Icy, Stefan Fogelvik, and CharlesHarvey, 97-102. Manchester: Manches-ter University Press, 1989.

Casetti, E., S. Amarcl, A. Blumstein, R.H. Day, A. H. Levis, S. R. Rosenthal,J. Sacks, and C. Werner. Regarding theFeasibility and Desirability of Confer-ences on "The Iv!ethodological ResearchFrontiers and the Social Sciences. "NSFAward No. OJR-8406230. Final Reportto the National Science Foundation, 10September 1986.

Catlett, Charles. "The NSFNet: Begin-nings of a National Research Internet."Academic Computing 3 (January 1989):18-21, 59-64.

Cimbala, Stephen J., ed. Artificial Intelli-gence and National Security. Lexington,Mass.: Lexington Books, 1987.

Clement, John R. B. "Increasing ResearchProductivity Through Information Tech-nology: A User-Centered Viewpoint."Version October 19, 1989.10.17 pre-pared for publication.

Cline, Nancy. "Information Resources andthe National Network." EDUCOM Re-view 25 (Summer 1990): 30-34.

Coalition for Networked Information. Mis-sion Statement (March 1990).

Conklin, Jeff. "Hypertext: An Introduc-tion and Survey." In Computer-Sup-ported Cooperative Work: A Book ofReadings, edited by Irene Greif, 423 -15. San Mateo, Calif.: Morgan Kauf-mann Publishers, 1988.

Conway, Paul. "Facts and Frameworks: AnApproach to Studying the Users of Ar-chives." American Archivist 49 (Fall1986): 393 407.

. Partners in Research: TowardsEnhanced Access to the Nation's Ar-chives. Washington, D.C.: National Ar-chives and Records Administration,forthcoming.

"Research in Presidential Librar-ies: A User Study." Midwestern Archi-vist 11, no. 1 (1986): 35-36.

Corbin, John. "The Education of Librar-ians in an Age of Information Technol-ogy." In Computing, ElectronicPublishing and Information Technology:Their Impact on Academic Libraries, ed-ited by Robin Downes, 77-87. NewYork: Haworth Press, 1988.

Council on Library Resources. "Commu-nications in Support of Science and En-gineering, A Report to the NationalScience Foundation from the Council onLibrary Resources." Washington, D.C.:Council on Library Resources, August1990.

. "Setting Library Policies andPriorities in Research Universities: ASpecial CLR Grant Program," n.d.

Cunnyngham, Jon. "Group to Form Foun-dation for Electronic Publishing." Un-published paper. Columbus, Ohio:Foundation for Electronic Publishing, ca.1989.

Dain, Phyllis, and John Y. Cole, eds. Li-braries and Scholarly Communication inthe United States: The Historical Dimen-sion. Beta Phi Mu Monograph, no. 2.New York: Greenwood Press, 1990.

David, Paul A., and W. Edward Stein-mueller. The Impact of Information


Technology Upon Economic Science.CEPR Publication no. 121. Stanford,Calif.: Center for Economic Policy Re-search, December 1987.

Davis, M., S. Rosenschein, and N. Shap-iro. Prospects and Problems for a Gen-eral Modeling Methodology. N-1801-RC.Santa Monica, Calif.: The RAND Cor-poration, June 1982.

Davis, Paul K. "A New Analytic Tech-nique for the Study of Deterrence, Es-calation Control, and War Termination."In Artificial Intelligence and NationalSecurity, edited by Stephen J. Cimbala,35-60. Lexington, Mass.: LexingtonBooks, 1987.

DeLoughry, Thomas J. "Scholarly Jour-nals in Electronic Form Seen as Meansto Speed Pace of Publication and Pro-mote Dialogue." Chronicle of HigherEducation 35 (22 March 1989): Al 1,A16.

Denley, Peter. "The Computer Revolutionand 'Redefining the Humanities'." InHumanities and the Computer: New Di-rections, edited by David S. Miall, 13-25. Oxford, England: Clarendon Press,1990.

Denley, Peter, and Deian Hopkin, eds.History and Computing. Manchester,England: Manchester University Press,1987.

Denley, Peter, Stefan Fogelvik, and CharlesHarvey, eds. History and Computing II.Manchester, England: Manchester Uni-versity Press, 1989.

"Despite Budget Cuts, Campuses Attemptto Maintain Computing Services." USCCenter for Scholarly Technology News-letter. (October 1991).

Dillon, Andrew, John Richardson, and CliffMcKnight. "The Effects of Display Sizeand Text Splitting on Reading LengthyText from Screen." Behaviour and In-formation Technology 9 (MayJune1990): 215-27.

. "Human Factors of Journal Usageand Design of Electronic Texts." Inter-

acting with Computers 1 (1989): 183-89.

Dollar, Charles. The Impact of InformationTechnologies on Archival Principles andMethods. Macerata, Italy: University ofMacerata Press, 1992.

Doran, Jim. "A Distributed Artificial In-telligence Reading of Todorov's TheConquest of America: The Question ofthe Other, by Tzvetan Todorov, 1985."In Interpretation in the Humanities: Per-spectives from Artificial Intelligence, Li-brary and Information Research Reportno. 71, edited by Richard Ennals andJean-Claude Gardin, 143-68. London:British Library Board, 1990.

Doty, Philip, Ann Peterson Bishop, andCharles R. McClure. "The National Re-search and Education Network (NREN):An Empirical Study of Social and Be-havioral Issues." In Information in theYear 2000: From Research to Applica-tions [proceedings of the 53rd AmericanSociety for Information Science AnnualMeeting], edited by Diane Henderson,284-99. Medford, N.J.: Learned Infor-mation, 1990.

Dowler, Lawrence. "Availability and Useof Records: A Research Agenda."American Archivist 51 (Winter/Spring1988): 74-86.

. "Conference on Research Trendsand Library Resources, February 22-23,1990." Unpublished draft report. Cam-bridge, Mass.: Harvard University, Wi-dener Library, n.d.

Downes, Robin, ed. Computing, Elec-tronic Publishing and Information Tech-nology: Their Impact on AcademicLibraries. New York: Haworth Press,1988.

Dyer, Alan, and Kate Milner. "An Ex-amination of Hypertext as an AuthoringTool in Art and Design Education." InHumanities and the Computer: New Di-rections, edited by David S. Miall, 137-48. Oxford: Clarendon Press, 1990.

Ehringhaus, Michael. The Electronic Jour-

75


nal: Promises and Predicaments. Tech-nical Report no. 3, ED316732. Syracuse,N.Y.: Syracuse University, School ofEducation, February 1990.

Engelbart, D. C., and W. K. English. "AResearch Center for Augmenting HumanIntellect." American Federation of In-formation Processing Societies Proceed-ings (Fall Joint Computer Conference)33 (May 1974): 395-410.

Ennals, Richard. Artificial Intelligence:Applications to Logical Reasoning andHistorical Research. New York: HalstedPress, 1985.

Enna ls, Richard, and Jean-Claude Gardin,eds. Interpretation in the Humanities:Perspectives from Artificial Intelligence.Library and Information Research Re-port no. 71. Cambridge: British LibraryBoard, 1990.

Evans, Nancy H., Denise A. Troll, MarkH. Kibbey, Thomas J. Michalak, andWilliam Y. Arms. The Vision of theElectronic Library. Mercury TechnicalReport Series, no. 1, Pittsburgh: Car-negie Mellon University, 1989.

Fargues, Jean. "Remarks on the Interre-lationships Between Artificial Intelli-gence, Mathematical Logic andHumanities." In Interpretation in theHumanitie.s: Perspectives from ArtificialIntelligence, Library and InformationResearch Report 71, edited by RichardEnnals and Jean-Claude Gardin, 265-83.Cambridge: University Press, 1990.

Fayen, Emily Gallup. "Loading Local Ma-chine-Readable Data Files: Issues, Prob-lems, and Answers." InformationTechnology and Libraries 8 (June 1989):132-37.

Freeman, Elsie T. "In the Eye of the Be-holder: Archives Administration from theUser's Point of View." American Ar-chivist 47 (Spring 1984): 111-23.

Friedman, Edward A., James E. McClellanIII, and Arthur Shapiro. "IntroducingUndergraduate Students to AutomatedText Retrieval in Humanities Courses."

In Humanities and the Computer: NewDirections, edited by David S. Miall,103-12. Oxford: Clarendon Press, 1990.

"The Future of Research Libraries." Un-published minutes of conference spon-sored by the American Academy of Artsand Sciences, 8 June 1990.

Gardin, Jean Claude. Artificial Intelligenceand Expert Systems: Case Studies in theDomain of Archaeology. New York:Halsted Press, 1988.

. "The Future Influence of Com-puters on the Interplay Between Re-search and Teaching in the Humanities."Humanities Communication Newsletter9 (1987): 17-21.

Garson, G. David. "The Role of InductiveExpert Systems Generators in the SocialScience Research Process." Social Sci-ence Microcomputer Review 5 (Spring1987): 11-24.

Gaunt, Marianne, and Robert Hollander."A Proposed U.S. National Center forMachine-Readable Texts in the Human-ities." ACH Newsletter 11 (Fall 1989):14-15.

Gillikin, David P. "Document Deliveryfrom Full-Text Online Files: A PilotProject." Online 14 (May 1990): 27-32.

Gordon, Ann D. Using the Nation's Doc-umental), Heritage: The Report of theHistorical Documents Study. Washing-ton, D.C.: National Historical Publica-tions and Records Commission, 1992.

Gould, Constance C. Information Needs inthe Humanities: An Assessment. Stan-ford, Calif.: The Research LibrariesGroup, 1988.

Gould, Constance C., and Mark Handler.Information Needs in the Social Sci-ences: An Assessment. Mountain View,Calif.: Research Libraries Group, 1989.

Gould, Stephen B. "Computing and Te-lecommunications in the Federal Gov-ernment." CRS Review 11 (July/August1990): 12-15.

Graham, Peter S. "Research Patterns andResearch Libraries: What Should


Change?" College and Research Librar-ies 50 (July 1989): 433-40.

Grand Challenges: High PerformanceComputing and Communications, The FY1992 U.S. Research and DevelopmentProgram. Report by the Committee onPhysical, Mathematical, and Engineer-ing Sciences, Federal CoordinatingCouncil for Science, Engineering andTechnology, Office of Science andTechnology Policy, 1991.

Green, Kenneth C. "A Technology Agendafor the 1990s." Change 23 (January/February 1991): 6-7.

Green, Kenneth C., and Skip Eastman.Campus Computing 1990. Los Angeles:University of Southern California, Cen-ter for Scholarly Technology, 1990.

Greenberg, Saul. "An Annotated Bibliog-raphy of Computer Supported Coopera-tive Work." SIGCHI Bulletin 23 (July1991): 29-62.

Greif, Irene, ed. Computer-Supported Co-operative Work: A Book of Readings. SanMateo, Calif.: Morgan Kaufmann Pub-lishers, 1988.

Hackman, Larry and Joan Warnow-Blew-ett. "The Documentation StrategyProcess: A Model and a Case Study."American Archivist 50 (Winter 1987): 12-47-

Hamad, Stevan. "Scholarly Skywriting andthe Prepublication Continuum of Scien-tific Inquiry." Psychological Science 1(November 1990): 342-44.

Hedstrom, Margaret. Archives and Manu-scripts: Machine-Readable Records. SAABasic Manual Series. Chicago: Societyof American Archivists, 1984.

. "Archives: To Be or Not to BeA Commentary." In Archival Manage-ment of Electronic Records, edited byDavid Bearman, 25-30. Archives andMuseum Informatics Technical Reportno. 13. Pittsburgh: Archives and Mu-seum Informatics, 1991.

, "Understanding Electronic Incun-abula: A Framework for Research on

Electronic Records." American Archi-vist 54 (Summer 1991): 334-54.

Heim, Kathleen M. "Social Scientific In-formation Needs for Numeric Data: TheEvolution of the International Data Ar-chive Infrastructure." Collection Man-agement 9 (Spring 1987): 1-53.

Hewitt, Joe A. "Altered States: Evolutionor Revolution in Journal-Based Com-munications?" American Libraries 20(June 1989): 497-500.

Hickerson, Thomas H. Archives and Man-uscripts: An Introduction to AutomatedAccess. SAA Basic Manual Series. Chi-cago: Society of kmerican Archivists,1981.

Hockey, Susan. "An Historical Perspec-tive." In Information Technology in theHumanities: Tools, Techniques and Ap-plications, edited by Sebastian Rahtz, 20--30. Chich,ter, England: Ellis HorwoodLimited, 1987.

Hockey, Susan, Jo Freedman, and JohnCooper. "Computers in the Study of SetTexts." In Humanities and the Com-puter New Directions, edited by DavidS. Miall, 113-22. Oxford: ClarendonPress, 1990.

Hudson, Valerie M. "Using a Rule-BasedProduction System to Estimate ForeignPolicy Behaviour: Conceptual Issues andPractical Concerns." Artificial Intelli-gence and National Security, edited byStephen J. Cimbala, 109-31. Lexington,Mass.: Lexington Books, 1987.

Hunter, Karen. "Academic Librarians andPublishers: Customers Versus Producersor Partners in the Planning of ElectronicPublishing." In Computing, ElectronicPublishing and Information Technology:Their Impact on Academic Libraries, ed-ited by Robin Downes, 35-47. NewYork: Haworth Press, 1988.

Ide, Nancy M., and Jean Veronis. "Arti-ficial Intelligence and the Study of Lit-era :y Narrative." Poetics 19 (1990): 37-63.

. "Very Large Neural Networks for

(1


Word Sense Disambiguation." Unpub-lished paper presented at European Con-ference on Artificial Intelligence,Stockholm, August 1990.

Johnson, J., T. L. Roberts, W. Verplank,D. C. Smith, C. H. Irby, M. Beard, andK. Mackey. "The Xerox Star: A Ret-rospective." IEEE Computer 22 (Sep-tember 1989): 11-26.

Joseph, Earl C. "Twenty-First Century In-formation Literacies and Libraries." InInformation Literacies for the Twenty-First Century, edited by Virgil L. P. Blakeand Renee Tjoumas, 7-16. Boston: G.K.Hall, 1990.

Kahn, Robert E., and Vinton 0. Cerf. AnOpen Architecture for a Digital LibrarySystem and a Plan for Its Development.The Digital Library Project, Volume 1:The World of Knowbots, Reston, Va.:Corporation for National Research Ini-tiatives, March 1988.

Kaufman, Paula, and Angie LeClercq."Archiving Electronic Journals: Who'sResponsible for What." Information Is-sues 2 (Fall 1990).

Kenney, Anne R.; and Lynne K. Person-ius. "The Future of Digital Preserva-tion." In Advances in Preservation andAccess, Volume 1, edited by BarbaraBuckner Higginbotham and Mary E.Jackson. Westport, Conn.: Meck ler Press,1992.

Klemperer, Katharina. "New Dimensionsfor the Online Catalog: The DartmouthCollege Library Experience." Informa-tion Technology and Libraries 8 (June1989): 138-45.

Kovacs, Diane K. Directory of ScholarlyElectronic Conferences. 3rd ed. Kent,Ohio: Kent State University Libraries,August 1991. (Electronic documentavailable via Internet.)

Kurshan, Barbara. Statewide Telecommun-ication Networks: An Overview of theCurrent State and the Growth Potential.Roanoke, Va.: Educorp Consultants,December 1990.

Kurshan, Barbara, and Marcia Harrington.Statewide Education Networks: SurveyResults. Roanoke, Va.: Educorp Con-sultants, April 1991.

Lancaster, F. W. "Electronic Publishing."Library Trends 37 (Winter 1989): 316-25.

Larsen, Ronald L. "Survey and FeasibilityStudy to Develop Pilot Project Proposalfor Distributed Electronic Federal Dep-ository." Unpublished proposal to Coa-lition for Networked Information, 29October 1990.

Leavitt, Michael R. The Analyst and Tech-nology-2000. Prepared for the U.S. In-telligence Research and DevelopmentCouncil, January 1991.

Lehmann, Stephen, and Patricia Renfro."Humanists and Electronic InformationServices: Acceptance and Resistance."College and Research Libraries 52 (Sep-tember 1991): 409-13.

Lesk, Michael. Image Formats for Preservation and Access. Report of the Tech-nology Assessment Advisory Committeeto the Commission on Preservation andAccess. Washington, D.C.: The Com-mission, July 1990.

Lewis, David W. "Economics of theScholarly Journal." College and Re-search Libraries 50 (November 1989):674-88.

Library of Congress. "American Mem-ory." Library of Congress Inform :donBulletin (26 February 1990): 83-87.

Lindquist, Christopher. "Ferret LoversUnite and Download." Computerworhl25 (12 August 1991): 1, 88.

Loughridge, Brendan. "Information Tech-nology, the Humanities and the Li-brary." Journal of Information Science15 (JulySeptember 1989): 277-86.

Loup, Jean. National Research and Edu-cation Network: Overview and Sum-mary. Washington, D.C.: Association ofResearch Libraries, July 1990.

Lucier, Richard E. "Knowledge Manage-ment: Refining Roles in Scientific Corn-


munication." EDUCOM Review 25 (Fall1990): 21-27.

Lynch, Clifford A. "Achieving the Prom-ise: A Proposed Strategic Agenda for Li-braries and Networked InformationResources in the 1990s." In Networksfor Networkers: Critical Issues for Li-braries in the National Network Envi-ronment, edited by Barbara Evans-Markuson with Elaine W. Woods. NewYork: Neal-Schuman Publishers, forth-coming.

. "The Growth of Computer Net-works: A Status Report." Bulletin of theAmerican Society for Information Sci-ence 16 (June/July 1990): 10-11.

. "Library Automation and the Na-tional Research Network." EDUCOMReview 24 (Fall 1989): 21-26.

Lynch, Clifford A., and Cecilia M. Pres-ton. "Evolution of Networked Informa-tion Resources." Proceedings of theTwelfth National Online Meeting, May7-9, 1991, New York, N.Y., edited byMartha E. Williams, 221-30. Medford,N.J.: Learned Information, 1991.

. "Internet Access to InformationResources." In Annual Review of Infor-mation Science and Technology (AR-1ST), vol. 25, edited by Martha E.Williams, chapter 7. Amsterdam: Elsev-ier Science Publishers B.V., 1990.

Lytle, Richard H. "An Analysis of the Workof the National Information Systems TaskForce." American Archivist 47 (Fall1984): 357-65.

MacCafferty, Maxine, and Kathleen Gray,eds. The Analysis of Meaning: Infor-matics 5. London: Queens College Ox-ford, 1979.

Maher, William J. "The Use of User Stud-ies." Midwestern Archivist 11, no. 1

(1986): 15-26.Margulius, David L. "C'est la France, C'est

Minitel." PC-Computing 2 (January1989): 194-96.

Martin, Susan K. "Library Management andEmerging Technology: The Immovable

Force and the Irresistible Object." Li-brary Trends 37 (Winter 1989): 374-82.

McAdams, Alan, Thomas Vietorisz, Wil-liam Dougan, and James Lombardi."Economic Benefits and Public Supportof a National Education and ResearchNetwork." EDUCOM Bulletin 23. (Sum-mer/Fall 1988): 63-71.

McKnight, Cliff. "Using the ElectronicJournal." In Scholarly Communicationand Serials Prices: Proceedings of aConference Sponsored by the StandingConference of National and UniversityLibraries and the British Library Re-search and Development Department, 11-13 June 1990, edited by K. Brookfield.New York: Bowker-Saur, 1991.

McKnight, Cliff, John Richardson, andAndrew Dillon. "Hypertext Authoring:Some Basic Issues." Humanities Com-munication Newsletter 11 (1989): 25-29.

McLean, Alice, Musick, Robert Morris-sey, and Donald A. Ziff. "ARTFL: ANew Tool for French Studies." Schol-arly Communication 8 (Spring 1987): 1,6-9.

Matters, Marion, ed. Automated Recordsand Techniques in Archives: A ResourceDirectory. Chicago: Society of Ameri-can Archivists, 1990.

Miall, David. "An Expert System Ap-proach to the Interpretation of LiteraryStructure." In Interpretation in the Hu-manities: Perspectives from Artificial In-telligence, Library And InformationResearch Report no. 71, edited by Rich-ard Ennals and Jean-Claude Gardin, 196-214. Cambridge: University Press, 1990.

Miall, David S., ed. Humanities and theComputer: New Directions. Oxford:Clarendon Press, 1990.

Michelson, Avra. Expert Systems Technol-ogy and Its Implications for Archives.National Archives Technical Informa-tion Paper, no. 9. Washington, D.C.:National Archives and Records Admin-istration, March 1991.

. "Forecasting the Use of NREN by

73


Humanities Scholars." Paper presentedat the panel "New Constituencies for theNREN," 27 March 1992, National NET'92, Washington, D.C.

Mills, William deB. "Rule-Based Analysisof Sino-Soviet Negotiations." SocialScience Computer Review 8 (Summer1990): 181-95.

"Mission, Themes, and Strategies."Washington, D.C.: Coalition for Net-worked Information, 1 June 1990.

Morris, Dilys E. "Electronic Informationand Technology: Impact and Potential forAcademic Libraries." College and Re-search Libraries 50 (January 1989): 56-64.

Morton, Herbert C., and Anne J. Price.The ACLS Survey of Scholars: Final Re-port of Views on. Publications, Com-puters, and Libraries. Washington, D.C.:American Council of Learned Societies,Office of Scholarly Communication. andTechnology, 1989.

National Archives and Records Adminis-tration. "Management, Preservation andAccess for Electronic Records with En-during Value." [Response to recommen-dations in Taking a Byte Out of History:The Archival Preservation of FederalComputer Records (Twenty-fifth reportby the Committee on Government Op-erations: House Report 101-978)j. Un-published paper 1 July 1991.

. Optical Digital Image StorageSystem: Project Report. Washington,D.C.: National Archives and RecordsAdminis,ration, Archival Research andEvaluation Staff, March 1991.

"The ,National Research and EducationNetwork: A Policy Paper." rev. Wash-ington, D.C.: EDUCOM Networking andTelecommunications Task Force, March1990.

Naugler, Harold. The Archival Appraisalof Machine-Readable Records: A RAMPStudy with Guidelines. Paris: General In-formation Programme and UNISIST,Unesco, 1984.

Nichol, Jon, Jackie Dean, and JonathanBriggs. "Logic Programming and His-torical Research." In History and Com-puting, edited by Peter Denley and DeianHopkin, 198-205. Manchester, En-gland: Manchester University Press,1987.

Nora, Simon, and Alain Mine. The Com-puterization of Society: A Report to thePresident of France. Cambridge, Mass.:MIT Press, 1980.

North Carolina State University, The Li-braries. "The North Carolina State Uni-versity Libraries and the NationalAgricultural Library Joint Project onTransmission of Digitized Images: Im-proving Access to Agricultural Infor-mation." Unpublished paper, 1990.

NREN: The National Research and Edu-cation Network. Washington, D.C.: Co-alition for the National Research andEducation Network, 1989.

Olsen, Mark, an Louis-Georges Harvey."Computers in Intellectual History:Lexical Statistics and Analysis of Polit-ical Discourse." Journal of Interdisci-plinary History 18 (Winter 1988): 449-64.

Osburn, Charles B. "The Structuring of theScholarly Communication System."College and Research Libraries 50 (May1989): 277-86.

Ostrow, Stephen E., and Robert Zich. InResearch Collections in the InformationAge: The Library of Congress Looks tothe Future, edited by John Y. Cole.Washington, D.C.: Library of Congress,1990.

Pagell, Ruth A. "Primary FTDBs for theEnd User: New Roles for the Informa-tion Professional." Online Review 13(April 1989): 143-54.

Pemberton, Jeffrey K. "Online InterviewsHarry Boyle on CAS's New License Pol-icy . . , Effects on Searching/Prices."Online 12 (March 1988): 19-25.

Perritt, Henry H., Jr: "Government Infor-mation Goes On-Line." Technology Re-


view 92 (November/December 1989): 60-67.

Peters, Paul Evan. "The Machine Aspectsof Preservation." Unpublished paper, ca.1990.

Potter, Rosanne G., ed. Literary Comput-ing and Literary Criticism: Theoreticaland Practical Essays on Theme andRhetoric. Philadelphia: University ofPennsylvania Press, 1989.

Pugh, Mary Jo. "The Illusion of Omnis-cience: Subject Access and the Refer-ence Archivist." American Archivist 45(Winter 1982): 33-44.

Rahtz, Sebastian, ed. Information Tech-nology in the Humanities: Tools, Tech-niques and Applications. Chichester,England: Ellis Honvood Limited, 1987.

Research Issues in Electronic Records. St.Paul, Minn.: Published for the NationalHistorical Publications and RecordsCommission, Washington, D.C., by theMinnesota Historical Society, 1991.

Robb, Charles. "Networking Metadata inKentucky." Unpublished paper pre-sented to the National Association ofGovernment Archivists and Records Ad-ministrators, Chicago, July 1990.

Roberts, Michael M. "Implementing theNRENThe Prologue Is Over." EDU-COM Review 25 (Fall 1990): 9-10.

Rogers, Gil. "Teaching a PsychologyCourse by Electronic Mail." Social Sci-ence Computer Review 7 (Spring 1989):60-64.

Rogers, Sharon J., and Charlene S. Hurt."How Scholarly Communication ShouldWork in the 21st Century." Chronicleof Higher Education 36 (18 October1989): A56.

Rothenberg, J. "The Nature of Model-ing." In Artificial Intelligence, Suntila-lion, and Modeling, edited by L.Widman, K. Loparo, and N. Nielsen,75-92. New York: John Wiley and Sons,August 1989.

. "Using Causality as the Basis forDynamic Models." In Proceedings of the

Third International Working Conferenceon Dynamic Modelling of InformationSystems (DYNMOD-3), 277-92. Delft,The Netherlands: Delft University ofTechnology, 1992.

Ruller, Tom. "Managing and AppraisingGIS Data: Issues and Strategies." Un-published paper presented at the 199]Society of American Archivists annualmeeting, Philadelphia.

Samuels, Helen Willa. "Who Controls thePast" American Archivist 49 (Spring1986): 109-24.

Sarin, Sunil, and Irene Greif. "Computer-Based Real-Time Conferencing Sys-tems." Computer-Supported Coopera-tive Work: A Book of Readings, editedby Irene Greif, 397-420. San Mateo,Calif.: Morgan Kaufmann Publishers,1988.

Schick, James B. M. Teaching History witha Computer: A Complete Guide. Chi-cago: Lyceum Books, 1990.

Schofield, Janet Ward, Debra Evans-Rhodes, and Brad R. Huber. "ArtificialIntelligence in the Classroom: The Im-pact of a Computer-Based Tutor onTeachers and Students." Social ScienceComputer Review 8 (Spring 1990): 24-41.

Scholarly Communication: Notes on Pub-lishing, Library Trends, and Research inthe Humanities, nos. 1-8 (June '.985Spring 1987).

Schrodt, Philip A. "Pattern Matching, SetPrediction, and Foreign Policy Analy-sis." In Artificial Intelligence and Na-tional Security, edited by Stephen J.Cimbala, 89-107. Lexington, Mass.:Lexington Books, 1987.

Schurer, Kevin. "Artificial Intelligence andthe HistorianProspects and Possibili-ties." In Interpretation in the Humani-ties: Perspectives from ArtificialIntelligence, Library and InformationResearch Report no. 71, edited by Rich-ard Ennals and Jean-Claude Gardin, 169-95. Cambridge: University Press, 1990.

81


Schurer, Kevin, Jim Oeppen, and RogerSchofield. "Theory and Methodology:An Example from Historical Demogra-phy." 1n History and Computing II, ed-ited by Peter Den ley, Stefan Fogelvik,and Charles Harvey, 130-42. Manches-ter: Manchester University Press, 1989.

Schwartz, Karen D. "Agencies Use Soft-ware to Dig Up Links Among Data."Government Computer News 19 (15 Oc-tober 1990): 60.

Seiler, Lauren H. "The Future of theScholarly Journal." Modern LanguageJournal 74 (1990): 1-9.

Shaughnessy, Thomas W. "ScholarlyCommunication: The Need for an Agendafor ActionA Symposium." Journal ofAcademic Librarianship 15 (May 1989):68-78.

Shaw, J. C. JOSS: Conversations with theJohnniac Open-Shop System. P-3146.RAND Corporation, 1965.

Shaw, J. C. "JOSS: A Designer's View ofan Experimental On -Line ComputingSystem." In American Federation of In-formation Processing Societies Confer-ence Proceedings 26 (1964 Fall JointComputer Conference), 455-64. Balti-more, Md.: Spartan Books, 1964.

Shuman, Bruce A. The Library of the Fu-ture: Alternative Scenarios for the Infor-mation Profession. Englewood, Colo.:Libraries Unlimited, 1989.

Slatin, John M. "Text and Hypertext: Re-flections on the Role of the Computer inTeaching Modern American Poetry." InHumanities and the Computer: New Di-rections, edited by David S. Miall, 123-35. Oxford: Clarendon Press, 1990.

Slatta, Richard. "Historians and Micro-computing, 1989." Social Science Com-puter Review 7 (Winter 1989): 446-57.

Smith, Eldred. The Librarian, the Scholar,and the Future of the Research Library.Contributions in Librarianship and In-formation Science, no. 66. New York:Greenwood Press, 1990.

Speiberg-McQueen, C. M., and Lou Bur-

nard, eds. Guidelines for the Encodingand Interchange of Machine-ReadableTexts. Chicago: Text Encoding Initia-tive, October 1990.

Strangelove, Michael, and Diane Kovacs.Directory of Electronic Journals, News-letters, and Academic Discussion Lists,edited by Ann Okerson. Washington,D.C.: Association of Research Librar-ies, March 1992.

Stutt, Arthur. "Argument Support Pro-grams: Machines for Generating Inter-pretations." In Humanities and theComputer: New Directions, edited byDavid S. Miall, 71-81. Oxford: Clar-endon Press, 1990.

Tanenbaum, A. S. Computer Networks, 2nded. Englewood Cliffs, N.J.: Prentice-Hall, 1988.

Taylor, Betty W., Elizabeth B. Mann, andRobert J. Munro. The Twenty-First Cen-tury: Technology's Impact on AcademicResearch and Law Libraries. Boston:G.K. Hall, 1988.

Tenopir, Carol, and Man Evena Shu."Magazines in Full Text: Uses and SearchStrategies." Onl ne Review 13 (April1989): 107-18.

Thaller, Manfred. "The Historical Work-station 'Project'." Unpublished paperdelivered at the seventeenth InternationalCongress of Historical Sciences, Ma-drid, 29 August 1990.

Thibodeau, Kenneth. "To Be or Not to Be:Archives for Electronic Records." InArchival Management of Electronic Rec-ords, edited by David Bearman, 1-13.Archives and Museum Informatics Tech-nical Report no. 13, Pittsburgh: Ar-chives and Museum Informatics, 1991.

Thorson, Stuart, and Kristi Andersen."Computational Models, Expert Sys-tems, and Foreign Policy." In ArtificialIntelligence and National Security, ed-ited by Stephen J. Cimbala, 147-58.Lexington, Mass.: Lexington Books,1987.

Tiainor, Richard II. "History, Computing

8,


and Higher Education." In History andComputing II, edited by Peter Den ley,Stefan Fogelvik, and Charles Harvey, 35-42. Manchester: Manchester UniversityPress, 1989.

Triebwasser, Marc A. "The Electronic Li-brary: The Student/Scholar Workstation,CD-ROM and Hypertext." Unpublishedpaper presented at the annual meeting ofthe American Political Science Associ-ation, Washington, D.C.. 31 August-4September 1988 (ED 307 896).

Troll, Denise A. Library Information Sys-tem H. Progress Report and TechnicalPlan. Mercury Technical Report Series,no. 3. Pittsburgh: Carnegie Mellon Uni-versity, 1990.

United Nations, Administrative Committeefor the Coordination of Information Sys-tems, Technical Panel on Records Man-agement. Electronic Records Guidelines:A Manual for Policy Development. NewYork: United Nations, 1989.

U.S. Congress, Office of Technology As-sessment. Linking for Learning: A NewCu se fo; Education. OTA-SET-430.Washington. D.C.: U.S. GovernmentPrinting Office, 1989.

U.S. House, Committee on GovernmentOperations. Taking a Byte Out of His-tory,: The Archival Preservation of Fed-eral Computer Records. House Report101-978. Washington, D.C.: Govern-ment Printing Office, November 1990.

Van Orden, Richard. "Content-EnrichedAccess to Electronic Information: Sum-maries of Selected Research." LibraryHi Tech 31 (1990): 27-32.

Waters, Donald J. From Microfilm to Dig-ital Imagery. Report of the Yale Uni-versity Library to the Commission onPreservation and Access. Washington,D.C.: The Commission on Preservationand Access, June 1991.

Weber, Robert. "The Clouded Future ofElectronic Publishing." PublishersWeekly 237 (29 June 1990): 76-80.

. "Libraries Without Walls?" Pub-

fishers Weekly 237 (8 June 1990): S20-S22.

Weiskel, Timothy C. "University Librar-ies, Integrated Scholarly InformationSystems (ISIS), and the Changing Char-acter of Academic Research." LibraryHi Tech 6 (1988): 7-27.

White, Mary Alice. "Imagery in Multi-media." Multimedia Review (Fall 1990):5-8.

. "The Third Learning Revolu-tion." Electronic Learning 7 (January1988): 6-7.

Wild, Martyn. "History and New Tech-nology in Schools: Problems, Possibili-ties and the Way Forward." In Historyand Computing II, edited by Peter Den-ley, Stefan Fogelvik and Charles Harvey,20-34. Manchester: Manchester Univer-sity Press, 1989.

Williams, Daniel. "New Technology forCoordinating Work." Datamation 36(May 15, 1990): 92-96.

Wilson, Flo. "Article-Level Access in theOnline Catalog at Vanderbilt Univer-sity." Information Technology and Li-braries 8 (June 1989): 121-31.

Woodsworth, Anne. "Computing Centersand Libraries as Cohorts: ExploitingMutual Strengths." In Computing, Elec-tronic Publishing and Information Tech-nology: Their Impact on AcademicLibraries, edited by Robin Downes, 21-34. New York: Haworth Press, 1988.

Woodsworth, Anne, Nancy Allen, IreneFload ley, June Lester, Pat Molholt, Dan-uta Nitecki, and Lou Wetherbee. "TheModel Research Library: Planning for theFuture." Journal of Academic Librari-anship 15, no. 3 (1989): 132-38.

Yavarkovsky, Jerome. "A University-BasedElectronic Publishing Network." EDI/-COM Review 25 (Fall 1990): 14-20.

. Technology and the Research En-vironment of the Future: The Impact ofthe Information Science Revolution on theResearch Environment of the Future, Al-bany: New York State Library and The

83


Nelson A. Rockefeller Institute of Gov-ernment/SUNY, February 1989.

Zarri, Gian Pierro. "Artificial Intelligenceand Information Retrieval: A Look at the

RESEDA Project." In The Analysis ofMeaning: Informatics 5, edited by MaxineMacCafferty and Kathleen Gray, 166-72.London: Queens College Oxford, 1979.

RAND /RP -187

85