Arc Hiv Aria Record Keeping Systems

8/6/2019 Arc Hiv Aria Record Keeping Systems

1/21

Articles

Record-Keeping Systemsby DAVID BEARMAN*

Les cadres d e classement de s dossiers constituent le contexte dans lequel s'Ctablit la valeurprobatoire des dossiers; pour cette raison, leur gestion est cruciale po ur la sauvegarde d ecette signification. La comp rkhension des cadres de classement des dossiers est aussi dela premibe importance pour 1'Ctablissement des besoins archivistiques fonctionnels envue d e la gestion des archives informatiques, pour la dCfinition des normes d e documenta-tion archivistique, et la concep tion des syst2mes de contr6le archivistique. L'auteur soutientque le cadre de classement de s dossiers (plut6t que le fon ds, le record group, ou la sCrie)doit Ctre admis comm e le lieu fondamental de la provenance. Les cadres de classementdes do ssiers doivent avoir la prCfCrence sur les autres concepts parce qu'ils ont des frontikresconnues e t des propriCtCs caractCristiques; ls rCsolvent kgalemen t des difficult6s associkesaux concepts de fonds, de record group, ou de skrie dans les pratiques canadienne,amkricaine, et australienne; et do nnent enfin aux archivistes de nouveaux outils avec lesquelsils peuvent jouer un r6le ac tif ? 1'Lge Clectronique. De plus, mettre I'accen t sur les exi-gences fonctionnelles des cadres de classement des dossiers a pour effet d'integrer lesarchivistes aux vkrificateurs, au personnel de la skcuritk adm inistrative, aux o fficiers d'accbs2 I'information, aux avocats, et a la haute direction, lesquels ont tous u ne responsabilitCenvers la mCmoire collective et sa gestion. L'auteu r soutient que cette intkgration est a lafois vitale au plan stratkgique et intellectuellement souhaitable.AbstractRecord-keeping systems are the locus of the ev idential significance of records; therefore,their management is critical to the preservation of evidential meaning. Understandingrecord-keeping systems is critical to formulating archival functional requirements formanagement of electronic records, defining archival documentation standards, and de-signing archival control systems. The author argues that record-keeping systems-ratherthan fonds, record groups, or record series-should be accepted as the fundamen tal locusof provenance. Record-keeping systems are preferred to these other concepts becausethey have concrete bou ndaries and d efinable properties, solve the problems identified withthe concepts of fonds, record groups, and series in Canadian, American, and Au stralianarchival practices, and give archivists new tools with which to play an active role in theelectronic age. In addition, the focus on functional requirements for record-keeping sys-

"All rights reserved Archivaria 36 (Autumn 1993)


2/21

RECORD-KEEPING S YSTEM S 17tems allies archivists with aud itors, administrative security personn el, freedom of infor-mation and privacy officers, lawyers, and senior managers-all of whom have a responsi-bility for corporate memo ry and its managemen t. Th e author argues that this alliance isboth strategically critical and intellectually desirable.

I . The Place of Record-Keeping Systems in a Model of Archival DataRecord-k eeping system s are a special kind of information system abou t which archivists shouldbe experts. As the nam e sugges ts, record-keeping systems keep and suppo rt retrieval of records,while information systems store and provide access to information. Record-keeping systemsare distinguished from info rmation sy stems within organizations by the role that they play inproviding organizations with evidence of business transactions (by which is meant actionstaken in the course of conducting their business, rather than 'commercial ' t ransact ions).Non -record inform ation systems, on the other hand, store inform ation in discrete chunks thatcan be recom bined and reused w ithout reference to their documentary co ntext. Archivists ou ghtto have a special expertise in record-k eeping system s, because they are the source of archiva lrecords and their context and structure reveal the historical meaning o f archives. Nevertheless,the analysis of record -keeping systems from a theoretical or practical p erspective is peculiarlyabsent from archival literature.]

In this paper, I extend my earlier analysis of how information about the content, structure,and context of record s is required in ord er to ensure preservation of evidence.' I explore whatarchivists must understand about the nature of record-keeping sy stems if they are to design an dimplemen t systems that capture, maintain, and prov ide access to evidence. Specifically, I ex-plore how understanding the evidential purpose of record-keeping systems provides criticaltools for articulation of wo rkable strategies for the managem ent of e lectronic records. I alsoargue that the design of appro priate docum entation method s for archives depend s upon appre-ciation of the centrality of record-keeping systems to archival theory and practice and on theconcept of records as evidence.' Ou r society recognizes som e docum ents as records, becausethey carry out or documen t transactions. Because records are accepted w ithin this social andlegal framework a s evidence of an act, they are retained in record-k eeping sy stem s designed toserve the needs of the people and org anizations that created or received them .

To understand record-keep ing systems w e must recognize them first as systems, and then, asinformation system s. Systems consist of interdependent components organ ized to achieve anend; information systems are organized collections of hardware, software, supplies, people,policies and procedures, and all the maintenance and training that are required to keep thesecomp onents w orking together. Record-k eeping systems are organized to accomp lish the spe-cific functions of creating, storing, and accessing records for evidential purposes. Wh ile theymay also be able to retrieve records for informational purposes, they are designed for opera-tional staff, not for archivists or researchers, and thus are optimized to support the businessprocesses and transactions of the crea ting organization rather than generic information retrieval.Although record-keeping systems are not created for archivists, archivists must appraise record-keeping systems and m ake decisions to destroy or preserve the records that they contain. Tra-ditionally, archivists have m ade these dec isions based on the examination of reco rds after therecords have fulfilled their role of supporting the operational needs of the organization thatcreated them. Th e advent of electronic records, which are not susceptible to ready examinationof the physical docum ents, has led archivists to seek alternative approaches to appraisal. It wassoon realized that if archivists could m ake su ch decisions on the basis of analysis of the busi-ness functions and the need for evidence of these functions, they cou ld avoid having to reviewthe records themselves. In addition, they could concentrate their efforts on records system s ofcontinuing value, which are relatively few in number, rather than squan dering resource s on theappraisal of insignificant records systems.


3/21

As a m atter of principle, when archivists do d ecide to retain records, they take special carenot to disturb the relations defined by the record-keeping system. These relations-which inmanual systems are limited to "original order," but which in automated environments mayinvolve many types of relationships-are evidence of how individual records were or couldhave been used within the record system and thus of what they meant in the context of thebusiness process that they docum ent. In manual systems, accessioning records need not disturbthis original order, but in electronic records systems, removing records from the applicationthat supported the relations am ong records, and between a record and the actions that it docu-ments, runs serious risks of destroying the structure and context information that preserves theevidential significance of the record.

The relationships amo ng records, business transactions, and record-keeping systems are il-lustrated in Figure l .Figure 1

are evidence of - re organized in- RECORDS 1Lcreates

dictates functions ofTRANSACTION supports conduc t of

_tmanages

Record-keeping systems are established to serve institutional o r personal purposes and there-fore reflect the functions and activities of the creating organization or individual. For mo rethan fifty years, archivists and records managers have assumed the role of experts who canprovide assistance to the organization in setting up record-keeping systems to serve businesspurposes efficiently and at the same time to satisfy archival functional r eq ~ ire m en ts .~ ecog-nizing that not all records systems serve organizational purposes eq ually effectively, archivistsand records manag ers focused on organizing paper records into series, each of which directlysupported the ex ecution of specific business transactions. Guidelines for effective file m an-agement issued by the U.S. National Archives in 1968 go s o far as to suggest that files thatrequire indexing to provide alternative access points are probably not designed to support aspecific function effectively, since a single fun ction, conducted in a specified way, will need toaccess records in only one--or at most a few ai f f er en t schemes of a~ ~ a n g e m e n t . ~

This may have been true as long as records systems were designed to support isolated busi-ness functions, but the spread of database management systems (DBMS) has been driven bythe informa tion managem ent belief that organizational efficiency can be enhanced by reducingdata redundancy through organization-wide data integration. In an integrated DBMS, each


4/21

RECORD-KEEPING SYSTEMS 19

are a of function al responsibility w ithin the organization is provided with views of the databasethat are limited to the data it requires. Th e software supports the transactions that this func-tional area conducts but records of thes e transactions may not be created o r maintained if thesystem w as only designed to serve as an organizational data resource (i.e., be an informationsystem) rather than to preserve evidence of business transactions (i.e., be a records system).Th e possibility that records could be used by bringing inform ation from various sources to-gether in a logical view at the time of m aking a decision, while not physically creating a record,is new to electronic methods of manipulating data, and presents the first of several seriouschallenges to corporate m emory and operational viability b rought about by electronic record-keeping.Archivists recognize that organizational functions (or "competencies" as the Europeans callthem) a re the roots of business processes, which in turn dictate the way in which transactionsare con du ct ed .V he way that the process is conducted is reflected in the organization of recordsto support a function. In paper systems, the physical records (each document or file) corre-

spond to logical business records (a transaction or case); so the physical organization of therecords in the system, within series, relates records to each other and t o the way in w hich workis done in the organization.' In automated systems, logical records (representing businesstransactions) do not necessarily co nform to physical records (which are s tructured to maximizedatabase efficiency); business reco rds may not on ly involve combining data from more thanone logical or physical record (as they typically d o in relational database manage ment sys-tems), but may also involve processing this data in ways that are only docum ented externally tothe data itself. Information systems might suppo rt the ongoing business of an organization onone level, even though they do not create records esse ntial for accountability.A second reflection of the nature of activity or transactions is what I once called the "form ofmaterial" and which has more recently c ome to be known a s the "documentary form."8 Docu-mentary form s structure the information internal to the individual record, d ictating what data

will be present for specific types of transactions, and facilitate its recognition and use by sig-nalling to readers, by means of typography, da ta structures, and electronic links, where particu-lar information will be located. In the paper w orld, organizations used p articular documen taryform s for specific business transactions, but in automated environments the aim is to free thedata from the form in which it was created, for use in other ways. At the sam e time, automatedenvironments have spawned new, virtual, documentary forms such as dynamic documents,multimedia docum ents, and individuated docum ents with properties that the organization andthe broader culture are only beginning to ~n d er st an d. ~he novelty of electronic documentaryform s means that we cannot mak e assumptions-common in our dealings with paper records,whose forms we understand-about the relationship between form and content, between formand how the record-keep ing system functioned, or between forms and the processes that cre-ated them, just a s other periods of radical chang e in documentary form s and method s of busi-ness communication have disrupted the relationship between the expression of structure indocu men ts and their interpretation by recipients.

Th e relationships am ong records, business transactions, functions, documen tary forms, andrecord series are depicted in Figure 2."'


5/21

are evidence of

ARCHIVARIA 36

Figure 2are organized inRECORDS

I creates -I BUS= defines info. content

structures

_tarranges

TRANSACTION signals info. relationsdefinesr

FORMI

People (as individuals and in their positions as employees), create documents of varioustypes as a consequen ce of their positions, offices, or roles in life. People also crea te non-recorddocumen ts. Information created by people only becomes a record wh en, and if, it participatesin a transaction. Purely private information, not shown to others, is not a record. In modernorganizations, if records a re created, business practice req uires them t o be "filed," so that inprinciple they are available to others. Archivists and records managers instruct filing clerks tocreate job-, project-, case-, or subject-files around functions of the organizational u nit and tofile individual records into these structures. In bureaucratic organizations, specific forms ofrecords (often literally numbered and pre-printed form s or "form-letters") are linked to particu-lar business transactions conducted by o rganizational units. Procedures may dictate that agiven type of file will always contain certain of these categories of records. Only specificinformation is present in each form of record, although the case as a whole contains all theinformation required for any aspect of the mission of the organization. The same principlesapply to records created by individuals in the m odem world; different form s such as diaries,correspondence, a nd subject files of personal "business" will make up the series of records inthe home of a private person. When we speak properly of the records of a family, we mean bythis that the record system was used by more than on e individual, often siblings or mu ltiplegenerations, of the same family. Otherwise the "family" papers are really an artificial collec-tion, as we call groupings of manuscripts or records made by the collectors rather than thecreators.

I I

While the relationship between record-keeping systems and functions is, therefore, alwaysstraightforward, the relationship between a record system an d organizational units is not. Thishas been the cau se in the past of many of the problems of locating provenance in organizations.Even paper record-keeping systems are not necessarily owned, built, or maintained by theorganizational unit that creates the records they contain, although they will be used by thatorganization. Traditional central registry offices, and the contemporary data processing de-partment which has succeeded them, were assigned the function of maintaining records. The

- FUNCTION1ROLE arrangement supportsdictates access pts. SERIES


6/21


records they m aintain are created and used by num erous, different organizational units. Thusrecords mu st be linked not only to the organizations that created them, but also to those thatmaintained them, used them, and owned them. Each of these organizations may know thesame records system by a variety of different names. The views of information held by theorganization that are ava ilable to a given o ffice of origin are limited by both the record systemand their access to it, usually to those views for which they have a business need. Furthermore,it must be remem bered that, especially for electronic records systems, record-keeping systemsmay encompass records physically located in more than one place. Indeed, such distributedlogical records systems will become increasingly commo n during the 1990s, with the accept-ance of client-server architectures. Logical records systems are even m ore radically the normin object-oriented environments in w hich the record alone will carry the methods by which it issearched, disseminated, and disposed, and the procedures governing the record-keeping sys-tem are distributed to the level of the individual records and d o not exist in a higher aggrega-tion.Figure 3 represents the elements discussed so far and their relations.

Figure 3are evidence of are organized inRECORDS

tstructures----+I I

BUSINESS defines info. content DOCUM ENTARYTRANSACTION signals info. relations FORM

is assigned1

carries defines

authorizes arranges-responsible for manages 1 t___)

FUNCTION/ROLE /ictates access pts.arrangement supports

I I

Although these relations among elements are the same in manual as in electronic recordsenvironments, the character of record-keeping systems is being radically transformed by auto-mation, as is the character of series, forms, and records themselves. Changes th at'are signifi-cant to archivists include the software dependen cy of record-keeping systems, the existence of

SERIES

OFFICE OFORIGINICREATORuses

limits views & uses of informationRECORDSYSTEMS


7/21

record-keep ing system s that serve many different and physically remote offices--each officehaving its own views of the system and also its own functions-and business processes that donot create records although they use information from dynam ic information systems.Before examining the implications of these changes both for archival automation and formanagement by archivists of electronic records of organizations, it is useful to establish therelationship between the concept of a record system an d the fundamental archival principles.11.Archival Documentation and Record-Keeping SystemsProvenance, unarguably the m ost important concept in archival science, dictates that recordsare to be understood with reference to their origins in activity." As a shorthand, archivistsoften equate the provenance of records with the organization in which records w ere created orreceived, i.e., the "ofice of origin." However, as the preceding data model makes clear, theprovenance of archives is better understood by reference to the function of which they areevidence and the record system in which they were created, stored, preserved, and accessed bythe organization. Elsewhere, I have written on why archivists must recognize function, and notorganizational setting, as the locus of provenancial meaning.I2 Suffice it to say here that whatsystems analysts would call the business function being conducted, not the "office of origin,"determines the form an d content of records and dictates the procedures for their creation anddissemination. As a consequence, when functions are transferred from one office to another,the records that document the function typically are stable and record-keeping systems areusually transferred lock, stock, and barrel with the transfer of responsibility. On the other hand,if a new function is assigned to an office, it will usually require new procedures accompaniedby new documen tary forms , new series of files, and often entirely new, separate record-kee pingsystems.

Archives appraise and accession record-keeping systems, not individual records, becauserecord-keeping systems do not just passively reflect how the creating organization used infor-mation; they actively determine it. As such, record-keeping systems are an organic whole.Som e record-keeping systems, such as central registries or decentralized filing systems operat-ing with a shared classification structure (thereby resulting in "virtual" central registries), maybe managed at the corporate level during their active life. Other record-keeping system s, suchas subject files, chronological transaction files, or incoming and outgoing correspondence,may be managed at a work unit, or even a work-group level, with or without reference to alarger corporate records system. In North American organizations it is even com mon for som erecords to be managed by individuals, either because in the prevailing corporate culture larger-scale systems do not exist or because the individuals want to retain control over the informationthat the records contain.13 If information or documents pass across the boundaries betweenindividuals, w ork groups, formal organizational units, o r independent organizations, record-keeping systems should crea te records. However, the definition of a record-creating boundaryis not absolute or fixed and depen ds on the nature of the transaction, aspects of the organiza-tional culture, and boundary perceptions in process definitions.I4

Ability to access and use record-keeping systems, rather than employment within the officeof origin, determines the role that records play in specific business processes during their ac-tive life. Relationsh ips and structures established in record-kee ping system s determin e theconnec tions that can be m ade between records they contain both during and after their activelife. Although archivists know that record-keeping systems provide evidence of the role thatrecords played in the organization, they have not developed tools or techniques for document-ing how record-keeping systems relate to organized activities through established procedures.In traditional paper-based systems, neither archivists nor the operating entity can typicallydocument who accessed record-keeping systems or how the records that they retrieved wereused; although, when a record is thus used, it is participating in a new business transaction and


8/21


should, in principle, become a record of that transaction.15 In electronic information systems,tools for representing such relationships as permissions, views, and actual uses of records ex-ist, and data administrators and configuration m anagers can doc ume nt the participation of recordsin concrete transactions over time.

Archivists have not made use of these tools in automated archival control systems. Indeed,the history of archival automation has not been a story of great successes. Th e relatively earlyadoption of a data content and interchange standard led not to the development of m ethods tobring archival documentation from active office settings directly into archival finding aids, butto a species of rigid text editors designed to create databases of M AR C AMC records.I6 In sofar as arch ival automation exists, it builds databases that replicate the data that was previouslyfound in paper finding aids and indexes, although it may provide m ore access points. Data getsinto these systems by m eans of archivists preparing finding aids, and it generally is used byarchivists acting as refer ence intermediaries.17 Automation-as imple men ted in archives to-day-is not integrated recor d-ke eping syste ms docu men tation, contr ibutes little if anyth ing toarchival productivity, and d oes not insinuate the archival function into the operating environ-ment of the parent organizations.

Som e archivists have been work ing to analyze archival system s in a way that would generaterequirements for archival documentation standards, which would move automated archivalinformation systems beyond their role as fast paper? In a recent effort to define the informa-tion architecture of archives in order to provide a framework f or more integrated archival auto-mation softw are, it was consistently found that the need of data archivists to describe the con-text and structure of records originates in documentation of organizational missions and record-keeping systems.19 It was als o foun d that current data mode ls and flow diag rams for archivalinformation systems overlook the nexus of records creation and record -keeping in the recordsystem, and that the archival function was being implemented as if it could be logically segre-gated from the record-keeping system s of the business.This segrega tion is impossible except at the expense of total redundancy, becau se archivalinformation systems have always been information systems about record-keeping system s, orwhat data administrators call "metadata systems." Although data administrators developedautomated systems called Data Dictionaries and Information Resource Directories to docu-ment and m anage electronic record-keeping sy stems, archivists have not adopted these auto-mated system s, but have instead tried to employ traditional methods fo r describing electronicarchival holdings.?" Unfortunately, the prose narrative and the simple data structu res that archi-vists use in traditional finding a ids cannot rigorously describe the my riad links of records w itheach other or transactions that are supported in automated system s. In addition, because theyare constructed after the fact from evidenc e still visible after the records co me into the archives,they also do not documen t the evolution of relations which takes p lace over the life of a system.Wh en automated, these sim plistic representations of information systems fail to help research-ers recon struct archival eviden ce or p ermit archivists to achieve operational efficiencies.By failing to employ techniques of documentation available from the domain of systemsdesign and management, archivists have overlooked a pre-existing source of documentationwhich would, if properly regulated, mitigate the need for archivists to engage in the post-hocdocumentation of accessioned systems." More importantly, archivists have missed the oppor-tunity to m aintain systems that serve as the repository of org anizational memory of functions,structures, and events, even though such databases are much needed by contemporary organi-zations and the data is necessarily present in an ade quate archival information system.?'I know of no archival institutions that serve as repositories of the life-cycle software configu-ration managem ent docum entation essential for establishing evidential context and structure inan adequate archival inform ation system. This is especially unfortunate, as it appears that thedata management requirements, and hence the metadata documentation requirements of ar-


9/21

chives are identical to those of vital records management, privacy administration, freedom ofinformation, and administrative security. If archivists did their jobs docum enting record-keep-ing systems, they could exploit the often greater political and financial clout of constituenciesfor these other interests. Such an integrative function speaks directly to strategic opportunitiesfor contemporary archives.Archivists must find ways to make the data that they manage or create regarding organiza-tional functions and struc tures sufficiently important to the organization that othe rs will keep itup to da te and use it as a n official referent. If archivists do not becom e the authoritative sourcesof information about which record-keeping sys tems exist and how they are implemented, theycannot identify the reco rds that should be preserved archivally. Ultimately, archivists will needto design ways to acquire descriptions of individual records, files, and record-keeping systemsdirectly from the self-documenting features of electronic records systems, because they willotherwise never have the resources to obtain this level of detailed documentation.Documentation of record-keeping systems in metadata systems that contribute to fundamen-tal organizational data management will dictate a very different agenda for standards for archi-val description. Hints of this agenda were present in the report of the SA A Ad Hoc Com mitteeon Description Practices, which defined archival description as:

the process of c apturing, collating, analyzing, and organizing any information that servesto identify, manage, locate, and interpret the holdings of archival institutions and explainthe contexts and record systems from which those holdings were s el e~ te d .~ 'Carefu l readers saw a radical shift from "making" description, to capturing it, and from de-scribing records to documenting contexts and record system s. An extension of this shift offocus led to two critiques of the proposed General International Standard Archival Description(ISADG).24 The critique is equally applicable to the recently-developed Canadian RAD frame-work (which is built around the concept of fonds), the American reliance on record groups, andthe Australian primacy of series.25 I believe that what each framework really needs is theconcept of a record-keeping system.26Reade rs will note that in the model in Table 3, fonds andrecord groups are unnecessary theoretical constructs that do not consistently correspond to anycombination of other concepts. Series only provide context when they are not part of a multi-ple-series record system.

Archivists, like Ptolemaic astronomers, are struggling with "very subtle" notions to makereality fit theory. When applied, the theory of fonds leads to inherent contradictions. This isbecause fonds are defined simultaneously as having what Te ny E astwood calls an external andinternal dimension27or what Terry Cook describes as the product of a "defined creator" and a"linked record-keeping system."2R The effort to define fon ds as being a theoretical constructthat is simultaneoisly organizationa l in context and the relations amo ng records needs to beabandoned in favour of what Angelika Menne-Haritz calls functional provenancez9. Record-keeping systems have the virtue of being the locus of functional provenance and a t the sametime being real things with concrete boundaries in time and space that do not require philoso-phy to locate. Their characteristics are precisely the variables that are involved in definingdocumentary evidence: content, structure, and context.Record-keeping systems defy the traditional approaches we have taken to documentation,and even resist the mo re innovative efforts to forge descriptive systems around the concept ofseries linked to organizational units, which was pioneered by Peter Scott in Australia and ex-plored in North America by Max Evans.'O Record -keeping systems have complex structuresthat give meaning to records. Although some m anual records systems may consist of a singleseries, most involve multiple series, with links between them that facilitate the ongoing workof the organization. Chang es in either documentary form or arrangement that signal a chang ein record series are physically revealed in manual record-keeping system s, but are not self-evidentin electronic systems, where both format and order are logical constructs. In electronic record-


10/21

RETORD-KEEPINGSY ST E M S 25

keeping systems, the documentation that describes what we have come to regard as series(either a "view" or a separate physical file with defined links) may b e part of the logic of thesoftware, the content of tables w hich the softw are reads, a function of the architecture of thesystem, or external to the electronic form record system . Of course, in both man ual and elec-tronic systems, the documentation itself is a record series that is part of the record system;however, electronic records system s retained without approp riate docum entation will hold noevidence. In o rder to retain eviden ce, archivists need to ensure that series of records within arecord system are retained as they were employed together by the creators and users of therecord system . In the process, the separate description of each series, which sufficed for manualsystems, becomes inadequate. Record-keeping systems must be documented using data ad-ministration techniques fo r metadata representation, because relations between series are com -plex data structures with links into elements of the business environm ent in which they oper-ate. This will become more ob vious as MI S offices try to implement "enterprise com puting,"process control, corporate decision support systems, and object-oriented systems.

Metadata documenting a record system needs to link organizational structure and function,business and archival processes, software procedures, and documentary form s. As such , itneeds to be represented in a relational data model supporting processing along connectionsbetween the files. In defining what data is needed to describe the record sy stem entity in sucha mod el, it is clear that this data is different from data describing an o rganization, a recordscreator, or an accession-although record system docum entation is linked to docum entation ofthese entities in the metadata system ."Figure 4 below illustrates the data about records sy stems that we m ight need in a m etadatasystem. As laid out in the illustration, it appears to be a flat record of the sort we m ight "write"in an archival finding aid; however, readers should no te that the field nam es indicate that nu-merous record types are present an d linked, and that other attributes of the entities referencedby the first word in the field label would be present in a fully coherent meta-documentationsystem. The data values in the working metadata system would not contain the sorts of wordsused in this example for the purpose of helping archivists to imagine the meaning of thesefields, but rather wo uld consist of pointers to other records and d ata represented in a fashionthat enables it to be processed consistently. In this illustration, the d ata resembles ou r currentarchival finding aids more than that found in Information Resource Directory Systems, but isincluded to introd uce archivists to the range of co ntent that is necessary to describe a recordsystem, rather than to suggest an actual data structure for an archival metadatabase on reco rdsystem entities. Even so, it is noteworthy that these fields of data about record-k eeping sy stemsare absent from RAD, MAD, and APPM.

Figure 4

Metadata files partially describing record-k eeping system s with descriptive text of thesort found in archival finding aidsSystem Name: Environmental Disaster Record SystemSystem O wner Name: Health & Public S afety DivisionSystem Own er Business Function: Service D eliverySystem Authorized Record Creator Names: State Dept. of Environmental AffairsHealth & Public Safety Division, City Police Department, State Department of High-ways, S tate Department of Ed ucation, City Welfare Services DivisionSystem Implem entation Date: April, 1972System A bandonment D ate: activeSystem ...


11/21

26 ARCHIVARIA 36

User Name:User Views: Accident Report; S ervice Cost Analysis; Application fo r Assistance; Ap-plication A pproval Hearing evidence; Grant Award; Disbursement Authorization; C aseFile Summation; G eographic Locations ReportUser Permission View Files: Accident report file, claim file, hearing file, client file,incident file, agency fileUser Permission Update Functions: relief recipient dataUser...Hardware Configuration CPU:Hardware Configuration Storage Devices:Hardware C onfiguration ...Data Configuration..Data Elements ...Data Output Products:Report 534; Report 9876; Repo rt46; G Is forms 2,9- 14,63,66-87;Stat Report form s 1-23 1Data Input Products: Screens 1-56Data..Software Configuration...Documentation Products: Disaster Relief Coordination System Procedures Manual101;Disaster Response System Softwa re Documentation; System Perm ission Configu-ration Audit TrailDocumentation Data Test Set:File 1344Documentation Data Audit Set: File 87654Documentation Data Configuration History:File 76

As can be seen from the above list of files and fields. which retxesents a small portion ofwhat w ould be required to document a record-keeping system, it is not possible to implement ametadata system in a "flat" format. Such a descriptive approach would not link the views thata given department had with the content of the data in those views, and the state of the softwareconfiguration a t any time. It would be unable, for example , to determine how the input from theState Department of Environmental Affairs would be acted upon by the system, and thus whetherthe input files (the case record as retained in the database) or the output in response to particu-lar user queries made as part of certain service delivery processes, would be the evidencerequired to document the function.

While archivists will not need to create or maintain all of this metadata about electronicinformation systems by themselves, they will not be able to define what metadata would berequired to documen t record-keeping systems, nor how it w ould need to be represented, with-out understanding the functional requirements for archival record-keeping systems. Thesefunctional requirements dictate what documentation w e actually require in order to preservethe evidential value of records.


12/21

RECORD-KEEPING SYSTEMS

111. Functional Requirements of Record-Keeping SystemsIn contemporary organizations, electronic records systems create, store, disseminate, and re-trieve records. Softwa re applications developed specifically for organizations and generic com -mercial applications are operating on a wide variety of hardware to support these systems.Archivists would like to ensure that electronic record-keeping systems developed or acquiredto support other functions of the organization are impleme nted and man aged in such a way a sto ensu re that records are captured and preserved. However, electronic records systems differfrom their man ual counterparts in several ways that are of considerable significance to archi-vists, including that they are typically des igned and o perated by p eople other than either archi-vists or records creators. In addition, they are typically dependent for functioning on the hard-ware and software in which they were implem ented. The professionals who manage electronicinformation systems demand that archivists articulate their functional requirements so thatdecisions ca n be m ade whether, to what extent, and how they should be satisfied.

Th e failure of archivists to understand records systems in their practice with paper recordshas left them without analytical tools with which to approach electronic records. Instead ofdefining the functional requirem ents for archivally sound records sys tems, archivists have beentrying to preserve "machine-readable records" or output products from systems. Instead ofdefining how s ystems would self-document the content, structure, and context of records, ar-chivists have tried to docum ent their provenance, their depend encies, their relationships, etc. indescriptive activity. Without unde rstanding the record system in relation to processes and ac-tivities of the organization, however, it is not possible to identify what data in the systemconstitutes evidence of an activity, and which activities and com petencies spawned o r used therecord. From outpu t products it is not possible to reconstruct the record a s evidence; in addi-tion, looking at output products has obscured the need fo r archivists to develop m ethods thatwill permit long-term retention of and access to system s. Moreove r, the least effective way todocument systems is after they have been retired; ongoing documentation, maintained fromdesign spec ifications onwards, is a much more reliable and effective means of systems control.Because records systems are a logical construct rather than a physical one, they may spanmany "volumes" in computer disks and many offices in location; however, a single documen -tation o r description will define the selection of records to the system, their arrangement withinit, and the methods of access to it. Such documentation enables systems staff to operate thesystem, to integrate it with other system s, and to modify its functionality and ultimately "mi-grate" the data that it contains to a new h ardware and software environment. Unfortunately,archivists are not conversant with such docum entation or with the formal properties of record-keeping systems. Docum entation of record-keeping systems is not easily isolated from docu-mentation of the software application as a whole, because most software applications havehistorically stored data in their own record system."Regardless of the implementation e nviron men t, the archival manage men t of electronic recordsis an inseparable component of ongoing data management in electronic record-keeping sys-tems. It should be app roached first with a clear definition of what w e want "archivally respon-sible" s ystems to do. Onc e we enumerate these functional requirements, we should ask when(in the life of the system ) and how (by w hat means) w e could intervene to satisfy the require-ments. The n we sh ould test these intervention strategies in installed record-keeping systems inthe real world in order to refine heuristics that can be used by others.In a study based on these premises curren tly underway at the University of Pittsburgh," wehypothesize that the functional requirements apply to any reco rd-keeping system. They are notunique to electronic record-keeping systems , although the means for satisfying a requirementwill be dependent on the way that the system is implemented. Th e methods available to satisfyfunctio nal require men ts include policy, procedu res, system design , and standard^.'^ In elec-tronic systems these are often referred to collectively as "data management practices." Weexpect that suc cess in using data managem ent practices to satisfy archival functional require-


13/21

28 ARCHIVARIA 36ments will be a factor in the interaction of the choice of strategies with the features of thebusiness application, the software application, and the corporate culture. Different businessapplications will have differing levels of risk associated with non-satisfaction of each require-ment. Different software applications will have different barriers to use of design, implementa-tion, and standards-based approaches to requirements, and w ill be correlated with im plementa-tion at different levels in the architecture. Different organizational culture s will be correlatedwith different approaches to satisfying each requirement. A representation of this researchproject, showing the variables, their anticipated interactions, and the hypotheses of the re-searchers, is shown in Figure 5 below.15

Figure 5University of P i t tsburgh Electronic Records Study

Archival Yfmc t i ona lRequirements1-n

AccountableReswnsibleImplementedRel iab le

CaptureComprehensiveCompleteIden t i f i ab l eAuthentic

MaintalnsoundAuditableExportableRemvable

Accessnvai abieUsableUnderstandableRedactable

TBCtICS II BUSlneSs/ runct ,ens/[given i n App l ica t ionsproposal

[to be definedPo l l cy u l t h L l te re ture

and analysis1DeSlgn

exanpies maylnplementation include:Standards Financ~al

ManagementServiceDellvery

Research 8Development

PersonnelManagement

SoftwareApplications[ t o be definedu i th survey/ana lys is lexanples mayinclude:TransBCtlonOriented

Data Oriented

Object Oriented

Procedural

BUSlnessSector[g lven i np r w s a l lCoverrrnentCwmercialNon -P ro f l t lUn ivers i ty

IOrganizational ICUI ure I[ t o be definedw l t h l i t e r a t u r eand Gper ts l

Full Bureaucracy/Uarket Bureaucracy/Workflow BureaucracyPersonnel-

BureaucracyCentral/Decentral ized

l a ) The functional requirements f o r archnval management o f electronic records a r e the same as fo r t rad i t tona l recordslb) Many functlanal requirements u i l l not be satasfied by tra dit ran al records system2s) I t u i l l be possib le to s a t is fy each o f these functional requlrements fol lo wing any of the four t act ics2b) Many requirements w l l l be more fu l l y sa t is f led far e lect ron ic record than they cou ld be for paper records3a ) Different business applications # i l l s ha re different sets of func tlo nal requirements, and3b) Di ff er in g degrees of r > s k a r e associated u i th non-sat8sfact lon of requ i rements i n d i f f e rent business appl ica t ions4a) Diff ere nt software applications u l l l no t d lc ta te d~f fe re nt unctrona l requ irements, but46) Dif fer ent packages with," app llc atl on categories u l l l sa t ls fy the functnonal requ lrements to d~f fe re nt egrees4 c ) softuare dependent data objects are not records and as evldencc u l l l general ly be saved ~n an independent format5a ) Fmct lonal requirements u i l l be the same f o r each business sector, and5b) Di f fe rent sec tor s u i l l not determine cholce of tactics as much as dif fer ent corporate cultures6a) The best way to sa tl r f y func t~o nal equirements u i l l depend heav, ly on corpora te cu l tu re6b) The technological cap abl l l t les of the a r c h ~ v e s nd v ts agen ts w i l l be l ess c r i t ~ c a l " sa t l s fy l ng archivalrequlrementr that u l l l be the acceptance of archivat res pon s-b l l l ty by manegers throughout the organlzatron


14/21


The se functional requirements w ere initially identified through a review of the literature onelectronic records managem ent, archives, and organizational information sy stems ma nagement.A draft statem ent of the functional requirements for archiving36was then s ubmitted to critiqueby a gro up of experts in the field. After two day s of deliberations, a revised statement of func-tional requirements for record-keeping was prepared, as illustrated in Figure 6 below.

Figure 6Functional Requirements for Record-Keeping"

COMPL IANT ORGANIZATIONS

ACCOUNTABLE SYSTEMSresponsible implemented reliable

FUNCT IONAL RE CORDSCAPT URE MAINTAIN ACCESScomprehensive sound availablecomplete auditable usableidentifiable exportable understandableauthentic removable redactable

To understand how these functional requirements relate to the concept of record-keepingsystems, it is important to free ourselves from a physical m odel of record-keeping sy stems tiedto a specific implementation. We need to adopt a conceptual framework in which a sys tem isunderstood to be the totality of people, policies, hardw are, software, and practices surroundingthe creation (or acquisition) and the use of information within any organization. The businessapplication for which these particular functional requirements are being specified is archiving.All other business applications of the organization, such as correspondence management ororder fulfilment, are presumed to have their ow n functional requirements in add itionto archiv-ing requirements.Th e requirements are purposefully stated as outcomes rather than as methods. As mentioned

earlier, each requiremen t could b e satisfied through either policy, systems design, system s im-plementation, or standards-or through a combination of these functions. Indeed, it is as-sumed that no organization would seek to satisfy all of these requirements using a single strat-egy. In this the functional requirements d epart significantly from the approaches that have beenused by archives to achieve these ends in manual record-keeping systems, which have oftenassumed that all the (unarticulated) functional requirements could be satisfied at once, in thesam e way, and in the s am e place in the overall system design.This has significant implications for the architecture that we envision to satisfy the require-ments. Insofar as system s design, implementation, and standards (rather than policy) are em-ployed to satisfy these requirements, the functionality required for archiving may be locatedwithin the Application Software, in a service located in the Application Programm e Interface,in any of the services of the Application Platform (such a s the operating system, user interface,

network services, etc.), in the E xternal Environment Interface, or in the E xternal Environmentitself (for example, in the communications systems or the telecommunications e n v i r ~ n m e n t ) . ~ ~


15/21

Each individual functional requirement m ay be satisfied by solutions implem ented w ithin oneor more software layers, and no two functional requirements need be satisfied in the same way.By taking the view that each transaction gene rates a record-rather than the perspec tive of thedocument, which views documents as participating in many transactions-we save ourselve sthe very complex m odelling requirements posited by Richard Barry's work with state transi-tion diagrams.39

Except that it is only possible to satisfy functional requirements relating to storage, preserva-tion, and access of evidence insofar as those relating to its creation h ave been satisfied, there isno presumption that any system w ould, could, or would want to satisfy all these requirementsfully. It is known that these functional requirements are not completely satisfied within exist-ing paper-based information systems, on which we have long relied. For example, few paper-based systems m aintain evidence of w ho used the records in the course of w hat decision-making(although som e registry functions retain this data with files). Virtually no paper-based systemcan document whether the individuals or offices named in a distribution list for a documentactually received it (or who even sent it). In electronic record-keeping systems it may be e asierin some cases, and more difficult in others, to satisfy these functional requirements. Always,the decision regarding the degree to which any functional requirement will be satisfied is abusiness decision grounded in risk assessment. Whether risk management methodology isformally applied or not, costs and benefits, specific liabilities, and organizational needs andpriorities w ill always be taken into consideration. Decisions not to satisfy functional require-ments are just that; they do not invalidate the requirement.

It is the intention of the University of Pittsburgh research project, for which this articulationof functional requirements was undertaken, to examin e business functions, software app lica-tions, and organizational culture variables relating to the satisfaction of these functional re-quirements, in order to develop heuristics that can guide practice. In Figure 7 below, wepresent these requirements as currently articulated. They suggest some of the power of theconcept of record-keeping systems, as the locus of provenance, to define effective strategiesfor electronic records managem ent.

Figure 7Functional Requirements for Record-Keeping Systems

Record-keeping is a critical function that is performed through the collective action of indi-viduals and systems throughout all organizations. Record-keeping is not the province of archi-vists, records m anagers, or systems adm inistrators alone, but an essential role of all employeesand of individuals in their private lives.Record-keeping systems a re information systems that are distinguished by the fact that theinformation they contain is linked to transactions that they document. Records may be con-

sulted for documentation of those transactions or because they contain information that is use-ful for som e completely separate purpose, but record-keeping systems d o not just contain datato be reused; they maintain evidence over time.Record-keeping systems support the corporate memory of organizations by supporting thebusiness functions of the organization. All business functions require records of business trans-actions in order to continue their day-to-day operations, satisfy administrative and legal re-quirements, and maintain accountability. The following functional requirements for record-keeping systems d efine a corporate requirement for any record-keeping system, not the appli-cation requirements of archives and records management systems. Archives and records man-agement are only o ne business application within the organization, just as are manufacturing,sales, service delivery, or personnel managem ent. In designing and implementing inform ationand record-keeping system s, the functional requirements for any particular business applica-


16/21

RECORD-KEEPING SYSTE MS 3 1

tions must be considered together with various corporate functional requirements. Archivesand records man agement systems have functional requirements specific to their business appli-cation-such as storage managem ent, records retention and scheduling, reference manage-ment, and access control-which are not discussed in this documen t. The functional require-ments presented below, on the other hand, are universal for any record-keeping system. Theymay be of special interest to archivists, records managers, security officers, freedom of infor-mation and privacy adm inistrators, auditors, lawyers, and others with special obligations to-wards records, but they should be of value and relevance to programme managers at all levels,from corporate manag ement to line supervisors.

These functional requirements were specifically developed in order to provide guidance forthe managem ent of electronic record-keeping systems, although they are equally app licable tomanual systems. Information systems professionals should note that business functions, busi-ness processes, business transactions, and business records-rather than system functions, sys-tem processes, system transactions, or system records-are the consistent focus of record-keeping.

Articulating functional requirements is the first step in effecting adequate control of record-keeping systems. The next step is to determine an organizational strategy for satisfying thefunctional requirements insofar as is ap propriate. Strategies m ight include adopting policiesand procedures, designing new systems, implementing systems in a way that supports satisfy-ing the requirements, or developing standards. Each of these four strategies may be appliedseparately or in comb ination to each separate functional requirement. The choice of strategywill depend on the degree of risk involved in failure to satisfy a requirement w ithin the busi-ness function that the record-keeping systems is to support, the existing systems environme nt(including hardware, software, and a rchitecture), and the corporate culture in w hich the strat-egy must succeed.Record-keeping systems capture, maintain, and access evidence of transactions over time, asrequired by the jurisdiction in which they are implemented and in accordance with com monbusiness p ractices.

Fun ctional Requ irements for Record-Keeping SystemsI. CompliantRecord-keeping systems comply with the legal and administrative requirements for record-keeping within the jurisdictions in which they operate, including specific requirements notreferred to below.11.AccountableResponsible: The organization m ust have policies, assigned responsibilities, and formal meth-odologies for managem ent of its record-keeping systems.Implemented: Records must have been created and maintained in the normal course of busi-ness, and documented procedures that were followed shou ld conform to common p ractices inthe industry.Credible: The system mu st control quality characteristics of information being input and proc-ess information in a fashion that is consistent and accurate.111.FunctionalRecord-keeping systems must capture, maintain, and access evidence ov er time. If they do so,records will be:


17/21

Complete: Records accurately capture all information recorded or generated by their creators.Records incorporate or link to, a representation of the software functionality that created them,other versions or views, a data model of relations between elements of information within arecord, eye-readable conventions such as placem ent or fon t, and other structural informationthat adds to their meaning. Records incorporate, or are linked to, information about the contextof their creation.Identijiable: A distinctive and bounded record exists for every business transaction.Authentic: The system m ust validate records creators andlor authorizers.Communicated: The system must capture a record of all communication in the conduct ofbusiness between tw o people, between a person and a store of information available to others,or between a source of information and a person.Sound: The integrity of records is protected from accidental or purposeful dam age or destruc-tion, and from any modification after they have been received by anyone other than the creator.Auditable: Record documentation traces the processes in which records participated, includ-ing indexing, classification, filing, viewing, copying, distribution, disposition, use, and de-struction, throughout the life of the record. Managem ent controls preserve auditability of inter-actions external to the system (such as during media m igration or transfer).Understandable: Records documentation should permit stored business records to be logicallyreconstructed. Information content, plus any structure and context, must be preserved in mean-ingful and documented relations. For records with functionality, business application proce-dures must be do cum ented so that they can be correctly associated with th e status of the systemat the time of record creation and later.Removable: It must be possible, w ith appropriate authority, to remove records from the sys-tem, leaving only audit trails to docum ent their prior existence.Exportable: Record content, structural representation, and representation of context must beexportable in standard protocols, if such protocols exist.Available: The system must docum ent all logical archival records that it contains, indicate theterms under which they are available for research, and retrieve them for authorized users.Renderable: The system m ust render records by display or otherwise as they appeared to crea-tors with views in effect at the time any record w as used, or retain structural data necessary todetermine such views.Redactable: The system must support delivery of redacted, summarized, or censored copies,and keep records of the version released.

ZV Some Strategic Implications of Focusing on Records SystemsThe concept of record-keeping systems as the locus of provenance provides tools for under-standing archiving requirements, which a re missing if we retain traditional definitions of prov-enanc e equating it with records creators or fonds. The recognition that records systems haveconcrete properties directly related to their ability to capture, maintain, and access records isthe first step in directing archival intervention so that evidence can be saved. Wh en archivistsunderstand the concept of record-keeping systems, they are freed from imagining that suchintervention only takes the form of a unified policy, an isolated "archival" application, or auniversal archival standard. By taking a systems approach, it becomes evident that the satis-faction of each separate archival requirement can be approached separately. Thinking in sys-


18/21

RECORD-KEEPING SYSTEMS 33tems terms permits us to imagine architectures for satisfying these requirements, in whichsatisfaction of the overa ll requirement is achieved by s atisfying particular requirements a t vari-ous different places, and in different ways, within the sy stem of people, procedures, har dware,software, and data.

Record-keeping systems-based strategies may have fundamental implications for archivalp ro gr am m e s t r ~ c t u r e s . ~ "ocusing on functional requirements allows us to emphasize out-comes of archival actions rather than outputs, and sugg ests a framework for regulation in whichthe archival function of the organization can require other units to address these functionalrequirements for those record-keeping systems identified as linked to mission-important func-tions, but not dictate specific solutions or records that must be sav ed. The programm e units ofthe organization must then cons ider the risks and the opportun ities, and develop plans for datamanagement that address each functional requirement to the degree required by the businessfunction, and in a way sug gested by existing technology ap plications and the corporate culture.Archival data management would complement data management requirements of other cor-porate control functions and of op erational managers, and lead to construction of archival in-formation systems that are operationally useful sources of information about record-keepingsystems or metadata system s. Because the information that these systems contain about record-keeping systems is inherently part of the documentation of these systems themselves, archi-vists would less have to "describe" records systems than to "gather descriptions" of them.Archivists will find natural allies in their documen tation efforts because the s ame docum enta-tion of record-keeping systems required to support archival needs a lso supports FOI, security,vital record, and privacy requirement^.^'With control coming early in the l ife of the system, responsibili ty being ~ c e p t e d y linesupervisors and senior management, and documentation collated in metadata systems, archi-vists would have less reason to accession records from record-keeping systems. The existingrecord-keeping systems would enable archivists to exploit search mechanisms already con -

structed by program me offices to retrieve records. Patrons could thereby be assured of eviden-tially reliable records throu gh mec hanisms that themselves are evidential, and archivists wouldeliminate the need to create external search systems that introduce artifacts into the searchprocess, and could retrieve informa tion that is not a record. Th e methods employed withinrecord-keeping sy stems can be augmented by informa tion obtained through full-text analysis,statistical analysis, or artificial intelligence, from records maintained by the record system.Such me thods could also be employe d for retrieval in situations where the user of the archivesis interested in information that may be contained in record-keeping systems, rather than inrecords themselves. Archivists would then be seen as professionals who assist in mining therecords of the organization for evidence and information, rather than custodians wh o overseethe destruction or storage of old docum ents.Record-ke eping systems -oriented thinking not only gives archivists a tool that supports docu-

mentation, appraisal, preservation, and retrieval. It also defines for them a unique role am onginformation profess ionals as defenders of records, rather than processors of information. Itdefines special skills that archivists can learn in their educational programmes and apply intheir professional lives, and w hich are not the province of the other inform ation professions. Italso levers the most important traditional archival concep ts into tools for the information age ,making it clear that the record-keeping system is the locus of provenance.Notes

* A draft of this paper was presented at the Ontario Association of Archivists Conference on Archivesand Automation , Toronto, 13 May 1993.1 The concept of record systems, and especially of filing systems, was present in the United States'archival literature through the 1950s but has disappeared since. The only direct treatment of record-

keeping systems that I have discovered was published in American Archivist 13 (1950), pp. 259-67).


19/21

The author, Helen Chatfield of the United States Bureau of the Budget, discussed "The Development ofRecord Systems" with attention to the history of chronological, subject, and alphabetic classificationschem es in government offices, and the role of self-indexing or separate indexes in each configuration.2 David Bearm an, "Information Technology Standards and Archives," Janus (1992), pp. 161-66.3 This will likely involve considerable rethinking of archives in the United States and Canada. I found theindex entry "systems" utterly ab sent from indexes to the Ame rican Archivist since its inception.Archivariapresents the same picture. It is noteworthy that the Australian literature up to and including the latestedition of Keeping Archives is replete with references to record-keeping systems, but (following PeterSco tt) nonetheless focuse s on the series as the fundamen tal unit of archival control. Description islinked to records about organizations and their functions rather than to documentation of record-keep ingsystems. I imagine that recognizing the role of record-keeping systems will be easier for archivists inAustralia, who already acknowledge the record-keeping system as an identifiable element in controlbut have not brought it into a rigorous model of appraisal or documentation.4 An example of the kind of analysis of record-keeping systems that was once basic to archival practiceis found in Howard Crocker and Kenneth L. Brock, "Building a Records Filing System for New YorkState Schools," American Archivist 19 (1956), pp. 249-60.5 U.S. National Archives and Records Service, Guid elines For Effective Files Managem ent (Washington,

DC, 1968).6 Luciana Duranti, "Diplomatics: New Uses for an Old Science," (in six parts), Archivaria 28-33(1988-1992).7 T.R.Schellenberg, Modern Archives: Principles and Techniques (Chicago, 1956 ), p. 77, quotes Aus-tralian archival managem ent guidelines for registry systems, which stated that they should "be planne din relation to the functions and activities of the department" and "as far as possible reflect the organiza-tion of the department."8 I used the term in my draft of the NISTF Data Dictionary, but my definition of it appears under theheading "form" alone in Nancy Sahli, ed., MA RC for Archives and Manuscripts: The AMC Format(Chicago, 1985). For a discussion, see David Bearman and Peter Sigm ond, "Explorations of Form ofMaterial Authority Files by Dutch Archivists," American Archivist 50 (1987 ), pp. 249-53 and DavidBearman, "'Who about what' or 'From whence, Why and How': Intellectual Access Approaches toArchives and their Implications for National Information Systems," in Peter Baskerville and ChadGaffield, eds., Archives, Auromarion & Access, Proceedings of a Conference held at the U niversi& ofVictoria, British Columb ia, March 1-2 , 1985 (Victoria, 1986).9 Ronald Weissman, "Virtual Documents on an Electronic Desktop: Hypermedia, Emerging ComputerEnvironments and the Future of Information Management," in Cynthia Durance, ed., Management ofRecorded Information: Converging Disciplines (New York, 1990), pp. 37-59; also David Bearman,"Multisenso ry Data and Its Managem ent," pp. 11 1-19.10 See for example, Barbara L. Craig, "The Introduction of Copying Devices into the British Civil Serv-ice, 1877-1889,'' in Barbara L. Craig, ed., The Archival Imagination: Essays in Honour of Hugh A.Taylor (Ottawa, 1992). pp. 105-33; also Frank Burke, "Chaos through Communications: Archivists,Records Managers and the Comm unication Phenomenon," Ibid., pp. 154-77.

1 1 "Provenance" is defined in Louis J. Bellardo and Lynn Lady Bellardo, comp., A Glossary or Archi-vists, Manuscript Curators and Records Managers (Chicago, 1992). as "the organization or individualthat created, accumulated andor maintained and used records" - ithout any reference to record-keeping systems. Eastwood argues for both the external (organizational) and internal (systematic) ref-erent for provenance, but asserts their equality as elements of the fonds: Terry Eastwood, "GeneralIntroduction," in The Archival Fonds: From Theo ry to Practice (Ottawa, 19 92), pp. 1-14.12 David Bearm an and Richard Lytle, "The Pow er of the Principle of Provenan ce," Archivaria 21 (Winter1985-86), pp. 14-27.13 David Bearman, "D iplomatics, Weberian Bureaucracy and the M anagement of Electronic Records inEurope and Am erica," American Archivist 55, no. 1, pp. 168- 180.14 Jon Harrington, Organizational Structure and Information Technology (New York, 199 1 , discusses theconcept of "perceptual" boundaries in organizations and how these can be changed (or not) by imple-menting electronic information systems. When the perceptual boundary is not changed but the patternof work is, the system will often fail. One notes that the concept of organizational and perceptualboundaries employed by Harrington conforms to what we are introducing here, e.g., business rulesestablish how a record-keeping system functions.15 David Bearman in UN ACCIS, Electronic Records Management G uideline s: A Manual for Policy De-velopment and Implementation (New York, 1990 ), pp. 17-70, 89-107.


20/21

RECORD-KEEPING SYSTEMS 3516 See the Directory of Sofm are for Archives and M useums, 1990-91 edition and 1992-93 edition (Pitts-burgh, 1990, 1992), as evidence for the paucity of archival software. In 1979-80, when he was involved

in drafting a standard for archival data interchange, the author convinced his colleagues that one of theprincipal benefits of such a standard would be to increase the availability of archival description soft-ware; he was wrong.17 North Carolina State Archives MARS system is an exception; its very status as the first, and still theonly widely available online public access catalogue for archives proves the point. See David Bearman,"MARS: The Archives and Manuscript Reference System," Archives & Museum Informatics 4, no. 4(1990), pp. 10-1 1.18 See the reports of the Ad Hoc Committee on Descriptive Standards, American Archivist 52, no. 4(1989) and 53, no. 1 (1990).19 Archival Information Systems Architecture Working Group, working paper, unpublished, Utah StateHistorical Society, 1990-93.20 David Wallace, "Metadata and the Archival Management of Electronic Records: A Review," elsewherein this issue of Archivaria.21 David Bearman, "Documenting Documentation," Archivaria 34 (Summer 1992), pp. 33-49.22 The AT&T Bell Laboratories archives got itself on the main menu of every employee of the companyby establishing a service which reported daily on the important activities in the corporation, includingmajor product announcements, policies, and reorganizations. This function, initiated by archivist MarciaGoldstein and implemented under then-librarian David Penniman, is one of the most successful appli-cations of a suggestion made by Lytle and Bearman (supra, note 12), although theirs was an independ-ent invention.23 Reports of the Ad Hoc Committee on Descriptive Standards, American Archivist 52, no. 4 (1989) and53, no. 1 (1990). (Emphasis added, though present in accompanying explanations.)24 Bearman, "Documenting Documentation," and David Bearman, "ICA Principles Regarding ArchivalDescription," Archives & Museum Informatics 6, no.1 (1992). pp. 20-21.25 Bureau of Canadian Archivists, Planning Committee on Descriptive Standards, Rules forArchiva1 De -scription (Ottawa, 1990.); Steven Hensen, comp., Archives, Personal Papers and Manuscript C ollec-tions, 2nd ed. (Chicago, 1990); Judith Ellis, ed., Keeping Archives, 2nd ed. (Melbourne, 1993).26 Interestingly, Bruce Dearstyne observes of Holmes's five levels of description that "modem archives

are inclined to add a sixth level, usually below the subgroup, of 'information system'. The term [is]primarily associated with electronic records and databases...": Bruce W. Dearstyne, The Archival En-terprise: Modern Archival Principles, Practices, & Management Techniques (Chicago, 1993), p. 132.Unfortunately, when I inquired, Dearstyne was unable to suggest any examples of a sixth level ofdescription in either theoretical papers or actual information systems.27 Eastwood, The Archival Fonds: From Theory to Practi ce, pp. 1-14.28 Terry Cook, "The Concept of the Archival Fonds: Theory, Description and Provenance in thePost-Custodial Era," in Terry Eastwood, ed., The Archival Fonds: from Theory to Pra ctice (Ottawa,1992). pp. 34-85.29 Angelika Menne-Haritz, "Introduction," in Angelika Menne-Haritz, ed., Symposium on the Impact ofInformation Technologies on Information Handling in Ofi ce s andArchive.7 (New York, 1993). pp.9-25.30 In this I believe Max Evans ("Authority Control: An Alternative to the Record Group Concept,"Ameri-can Archivist, 49 [1986], pp. 249-61), may be more at fault than Peter Scott, from whom he borrowed(though he probably had not read the five-part series in Archives & Manuscripts, vols. 7-9, publishedbetween April 1979 and September 1981, in which the full explanation of Scott's position was laidout). Evans essentially proposed a mechanical data representation solution for linking series to organi-zation, elaborating on Bearman and Lytle, while Scott's model had a place for record-keeping systemseven though he did not employ a formal methodology for describing them.3 1 The author has been engaged in an effort with the Metropolitan Toronto Archives to define an archivalcontrol system that incorporates both RA D and the representation of record-keeping systems in order toenable the jurisdiction to utilize metadata created with records during their active life, and integrate itinto the records management and archival life cycle control process.32 In the future, architectures that utilize the "client-server" model of computing will use specializedapplications running on servers; these have the sole purpose of filing and retrieving data for otherapplications (running on "client" machines), which will process, analyze, or disseminate it. When thisconcept is widely implemented, it will become easier to define record system properties, but it will stillbe necessary to understand how applications were executed using the record system. Similarly, when


21/21

object-oriented systems com e into general use, it will be easier to ensure that data objects obey archivalretention and access rules, provided archivists learn to articulate explicit requirements that can be trans-lated into object-oriented methods and classes.33 Richard Co x, "Research Prospectus: Variables in the Satisfaction of Archival Requirem ents for Elec-tronic Records Manag ement," in University of Pittsburgh, Electronic Records M anagement Study (Pitts-burgh, 1993), typescript.34 See David Bearman, "Archival Principles and the Electronic Office," in Angelika Menne-Haritz, ed.,Information Handling in Ofices and Archives (New York, 1993). pp. 177-93.35 David Bearman [University of Pittsburgh, Electronic Records Management Study], "Project Method-ology Overview," (Pittsburgh, 1993). typescript.36 Archivists have generally resisted the use of the term archives in the form of an active verb "archiving."I am consciously adopting this terminology both because our customers use it, and we must communi-cate with them, and because the implication of these requirements is that they will be met throughactive involvement with systems at the level of policy, design, implementation, or standards develop-ment- nd "archiving" is an active verb.37 David Bearman, "Functional Requirements for Record Keeping Systems," Version 1.0 ,23 May 1993.This draft was prepared following the meeting of an experts advisory panel on 20-21 May and incorpo-rates their recommendations. It was widely circulated for professional critique during the summer of1993.38 National Institute of Standards and Technology (NIST), Application Portability Projile (APP):TheUnited States Government's O pen Systems Environment Profile (OSW l Version 1.0, NIST S pecialPublication 500- 187, Washington, DC.39 Richard Barry, "Electronic Document and Records Management Systems: Towards a Methodology forRequirements Definition" [typescript draft of a paper for OIS931, examines the concept of a recordfrom the point of view of the docum ent. This reveals that a docum ent may participate in numeroustransactions du ring a life cycle that is not a linear sequence but a peripatetic path. Representing thesestates of the document requires state-transition diagramm ing. Wh ile ultimately this is equivalent to the'result of viewing records from the point of view of transactions, B arry believes that the latter is signifi-cantly less complex to model and implement.40 The author has recently published a series of articles on the possibility of new programme structuresand organizational models for electronic records programmes. See David Bearman, "New Models forManagement of Electronic Records by Archives," Cade rnos de Biblioteconomia, Aq uivistica, eDocumentacao (1992). no. 2, pp. 61-7 0; "An Ind efensible B astion: Archives as a Repository in theElectronic Age,'' in David Bearman, ed., "Archival Managemen t of Electronic Records," Archives &Museum Informatics Technical Report 13 (Pittsbugh, 1991); and the as yet unpublished introductoryessay on alternative programme m odels co-authored with Margaret Hedstrom, which will appear inArchives & Museum Informatics T echnical Report 18 (forthcoming 1993).4 1 David Bearman, "Organizational Accountability in the Evolving Electronic Comm unications Environ-ment,'' Archives & Manuscripts (forthcoming).

Date post:	07-Apr-2018
Category:	Documents
Upload:	dee-lovett
View:	217 times
Download:	0 times