Copyright ©Henry H. Perritt, Jr.  
Summer, 1992

ELECTRONIC RECORDS MANAGEMENT AND ARCHIVES
53 U. PITT. L. REV. 963.

Henry H. Perritt, Jr. [FNa1]

  This article is adapted from a report prepared for the Administrative Conference of the United States in 1990, [FN1] which resulted in the adoption of several recommendations.

A. Viewpoint

  More and more records of federal agencies are created, used and stored in electronic rather than paper form. For example, President Bush reportedly used a personal computer to sketch American policy goals in the early hours of the Soviet coup. [FN2] As this occurs, legal requirements and management efforts designed around paper records are progressively less satisfactory to ensure an adequate legal and historical record of government decisionmaking.

  Administrative Conference Recommendation 88-10 [FN3] and the accompanying report address public access and dissemination of electronic information. The 88-10 effort considered statutes and policies that contemplate agencies releasing information they already have to the public. [FN4] The recommendation and report also considered agency acquisition *965 of information in electronic form. The 88-10 effort did not consider internal agency use of information in electronic form, nor did it consider the circumstances under which agencies should create electronic information systems or the characteristics of such systems. The 1990 report and this article complement the 88-10 effort. This article focuses on internal agency electronic records management.

  The basic legal framework contemplates that agencies must retain certain records of public decisions so that there will be a historical record. As electronic technologies become the predominant media through which decisions are recorded, faithful application of the policies embodied in the records statutes requires assessment of whether electronic records should be treated the same or must be treated differently from paper records. An idea shared widely among information systems professionals is that the best way to ensure retention of electronic records having archival value is to design information systems with inherent archival features. Accordingly, this article also considers electronic information system design.

*966 B. Relevance to Private Sector

  It is erroneous to suppose that electronic records and archives management present issues solely of concern to creators and managers of government information. The private sector should be concerned with electronic records management issues for three reasons. First, the government imposes certain record keeping requirements on the private sector that may or may not be met by electronic record keeping strategies. Second, private entities want to maintain records for their own archiving and litigation support functions. Third, some of the same issues regarding reliability of official records kept in electronic form also are raised in connection with private electronic contracting methods. [FN5]

  The federal government imposes diverse records keeping requirements on private sector entities. These requirements include aspects of economic regulation by the tariff regulatory agencies, aspects of environmental, health and safety, and food and drug regulation, and regulations imposed on defense contractors. [FN6] The Occupational Safety and Health Administration (OSHA) requires that hazardous substance exposure records for employees be maintained for 50 or more years. [FN7] The same technologies that permit electronic archiving of government information also permit electronic maintenance of private sector information subject to governmental requirements. Most of the same policy issues concern both private and public information, and the same procedures that make sense for government records managers also may make sense for private sector records managers operating under governmental requirements.

  *967 Purely private incentives also raise electronic archiving issues for the private sector. Private sector entities have an incentive to maintain records in connection with potential litigation, such as medical records, OSHA- related hazardous substances records and banking records.

  A growing number of private and public entities are using electronic messages to make and administer contracts and to engage in other legally significant transactions. One important requirement for these electronic contracting systems is that reliable records of the transactions be maintained in the event of a dispute over whether the transaction occurred, whether a contract was formed, or over the terms of the contract. [FN8]

C. Goals and History

  Records management and archives have the following purposes:

    -- to maintain an institutional policy memory;

    -- to enhance the body of knowledge in a factual and scientific sense;

    -- to maintain official records for possible use as evidence in legal proceedings; [FN9]

    -- to maintain the nation's history;

    -- to promote efficiency and effectiveness of agency operations; and

    -- to increase the return on investment from information in government records which has long-term value.

  Realization of the last goal would be enhanced by expanding access to archived information at reasonable costs. Other conceivable purposes would engender more controversy, such as a goal of providing expanded electronic public access to public information on demand.

  A 1988 National Archives and Records Administration (NARA) Report   [FN10] proposed that agencies systematically maintain documents containing data having informational value [FN11] or that:

    *968 -- facilitate action by agency officials and their successors;  [FN12]

    -- make possible scrutiny by Congress and other institutions "and other persons properly and directly concerned" about the manner in which public business has been discharged; [FN13]

    -- protect financial, legal and other rights of the government and of persons affected by governmental actions; [FN14]

    -- contain essential information on formulation and execution of basic policies and decisions or on major actions; [FN15]

    -- document significant decisions reached orally, face-to-face, by telephone, or in conference; [FN16] and

    -- document important board, committee or staff meetings, or matters considered at or resulting from such meetings. [FN17]

  The evolution of the National Archives shows the emergence of these goals from a starting point that concentrated on permitting agencies to throw away records no longer needed for their operational missions, while giving someone with an historical perspective a chance to preserve some of the records before they were destroyed.

  The Disposal of Records Act of 1939 [FN18] authorized agencies to submit schedules of records [FN19] not currently needed "that appear . . . to have no permanent value or historical interest" [FN20] to the archivist, authorized the archivist to report to a joint Congressional committee lists of records to be destroyed, and authorized agency heads to destroy records, when the joint committee had no objection. [FN21] Legislation enacted in 1940  [FN22] authorized the destruction of records that had been microfilmed in accordance with technical standards developed by the National Bureau of Standards, and made the microfilm records admissible into evidence. [FN23] Legislation originally enacted in 1943 [FN24] authorized *969 the archivist to promulgate regulations establishing procedures for preserving  [FN25] and disposing of records, [FN26] and required agencies to submit records disposition schedules to the Archives, in accordance with Archives regulations. [FN27] Executive Order 9784 [FN28] required agency heads to establish programs for effective management and disposition of agency records.

  The Federal Property and Administrative Services Act of 1949 [FN29] transferred the National Archives Establishment to the General Services Administration [FN30] and authorized the Administrator of General Services to obtain reports from federal agencies on records management and disposal practices, to "promote, in cooperation with the executive agencies, improved records management practices and controls . . . including the central storage or disposition of records not needed by . . . agencies for their current use," and to report to the Congress and to the Director of the Bureau of the Budget on the subject. [FN31]

  The Federal Records Act [FN32] authorized the Administrator of General Services to promote standards for improved records management[FN33] and to "establish standards for the selective retention of records of continuing value." [FN34] It required agency heads to develop records management programs, [FN35] to preserve records "containing adequate and proper documentation of the organization, functions, policies, decisions, procedures, and essential transactions of the agency," [FN36] and to establish safeguards against removal or loss of records under regulations promulgated by the Administrator of General Services. [FN37] The Act authorized *970 the Administrator to accept for deposit with the National Archives agency records determined by the Archivist "to have sufficient historical or other value to warrant their continued preservation . . . . "  [FN38]

  The Federal Records Management Amendments of 1976 [FN39] reorganized and codified the records management statutory provisions [FN40] to clarify the objectives of records management to include:

    -- accurate and complete documentation of government policies and transactions;

    -- control of quantity and quality of government records;

    -- simplification of records management activities, emphasizing the prevention of unnecessary paperwork; and

    -- judicious preservation and disposal of records. [FN41]

Legislation enacted in 1978 made application of general records schedules to agencies mandatory. [FN42]

  The National Archives and Records Administration Act of 1984 [FN43] removed the National Archives from the General Services Administration and set it up as an independent establishment in the executive branch called "The National Archives and Records Administration." [FN44] The Act authorized the archivist to issue regulations, and required agency heads to "issue[] such orders and directives as such agency head deems necessary to carry out such regulations," [FN45] thus preserving the divided responsibility between the National Archives and individual agencies.

  The Presidential Records Act of 1978 [FN46] is a product of Watergate and the Nixon resignation. It established governmental ownership in presidential records, and obligated the president to implement records management controls to "assure that the activities, deliberations, decisions, and policies that reflect the performance of his constitutional, statutory or other official or ceremonial duties [were] adequately documented." [FN47] The Act permits the president to dispose of presidential *971 records no longer having "administrative, historical, information, or evidentiary value," but only if the Archivist of the United States has no objection, or over the objection of the Archivist if the President submits a schedule of records to be destroyed 60 days before the destruction. [FN48] Upon the conclusion of a presidency, the Act makes the Archivist the custodian of the presidential records of the departing president, and obligates the Archivist to make the records available to the public, subject to restrictions on access imposed by the president under specific categories set forth in the Act. [FN49] The categories of authorized restrictions closely track the Freedom of Information Act (FOIA) exemptions. [FN50]

  The Act's provisions were written to follow the guidelines established for congressional regulation of presidential records by the Supreme Court in Nixon v. Administrator of General Services. [FN51] These provisions included affording interested persons, including former Presidents, notice and a meaningful opportunity to assert legal constitutional privileges, including executive privilege, privacy, and association with political party; guarding against disclosures barred by defenses or privileges available to former presidents; and ensuring that private materials were returned. Under such circumstances it was not a violation of separation of powers for the Congress to authorize screening of presidential papers by the Archivist.

D. Vision of the Long-Term Future

  In the long run, wider use of electronic information technology will change administrative agency structures and application of traditional administrative law concepts. [FN52] Agencies will use electronic information *972 technologies for adjudication, rulemaking, internal management and to deliver important services. In electronic adjudication, hearing and pretrial procedures all would be electronic. Record keeping and filing obligations imposed in aid of enforcement [FN53] also would be electronic, as in Electronic Data Gathering, Analysis and Retrieval (EDGAR) and the IRS Electronic Filing Project.

  EDGAR is a system for automating the Securities and Exchange Commission  (SEC). Under EDGAR, all corporations covered by the federal securities laws are required to file documents such as prospectuses and 10K and 10Q forms electronically with the SEC. EDGAR will be fully operational by 1994. The IRS Electronic Filing Project permits taxpayers to file their individual income tax returns electronically through third-party tax preparers.

  The rare exception to electronic methods for adjudication would be a contested matter in which the credibility of witnesses is at issue. Even there, videotaped depositions could be taken in advance, digitized and presented to the administrative law judge through hypermedia electronic filing and data base techniques. There would be a much more complete and much more accessible record of all adjudications than with paper processes. Because of the basic similarity of the adjudication process regardless of agency, certain format and software standards would emerge that would make it easier for adjudication records to be transferred among agencies and the National Archives. Appellate review of agency adjudications would be facilitated by development of judicial capacity to handle the standardized electronic formats.

  Using electronic information technology for rulemaking has an even greater potential to change the way in which government operates. Rulemaking is supposed to involve the general public rather than only parties to a particular case. Electronic information technology greatly facilitates timely public involvement and reduces burdens on agencies affording that involvement. The notice and comment process could be made largely electronic, with notices being posted on electronic bulletin boards and comments being submitted via dialup telephone links. The electronic record of a particular rulemaking decision would be all of the entries in the appropriate table of the rulemaking data base. In April, *973 1992, the Office of Management and Budget used electronic methods like this to provide notice of and opportunity to comment on a proposed revision in its guidelines on information collection and dissemination.

  The record on petition for judicial review would be a set of electronic pointers to appropriate places in the table. [FN54] Electronic pointers are references in one document which specify the place in another document where material can be found. By selecting an electronic pointer, a user can automatically retrieve the referenced material. A reviewing court simply would access the appropriate table though the pointers. The rulemaking process might become less discrete and more in the nature of a dialogue. In the long run, there would be a tendency for the interactive capability to blur the distinction between incomplete and complete rulemaking decisions. The technology permits a dialogue between regulator and regulatee, reducing the need for communications between the two to occur by means of formal, written documents in the form of petitions, comments and final rules.

  Internal management practices would build on concepts developed for the United States Forest Service. Electronic management of agency resources  [FN55] would work primarily by relaxing the need for official paper documents to record and communicate instructions and decisions to lower levels in the organization. Rather than signing a new delegation of authority, a memorandum making a change in organizational structure, or a directive reallocating enforcement resources, an agency head would post a notice directly from her desktop workstation electronically changing a data base record. The state of the organization, of delegations, of authority and of resource allocations would be defined officially by the state of the data base.

  Electronic information technology also can be used to deliver some governmental services: dissemination of public information and electronic transfers of money in connection with public welfare and subsidy programs. Electronic food stamps, direct deposit of social security checks and electronic management of Medicare and Medicaid benefits all are examples. Such electronic information technology applications *974 use well-proven electronic funds transfer techniques, combining them with EDI standards in the case of Medicare reimbursement. (EDI standards are formats agreed to in advance which permit commercial information, like the elements of an invoice, to be represented by fields in an electronic transaction record, which can be processed by a receiving computer without human intervention.) The advantages are faster availability of funds, and improved ability to audit. The problem to be solved is that of electronic contracting and electronic funds transfer potentially increasing risk of forgery. [FN56]

E. Some Basic Concepts

1. Classes of Agency Information

  Four different types of agency information present somewhat different records management and archives problems: policy documents; scientific and technical information (e.g., satellite-collected data); agency administrative records; and social data (e.g., census information).

  Electronic policy documents only recently have presented records management problems. Because such policy documents must get distributed on paper, and because of uncertainty about electronic signatures, such policy documents virtually always are printed on paper before they are signed by the highest level decision maker. The paper medium is the archival record. But this practice will change, as decision makers increasingly have access to documents in electronic form, and as electronic transactions become a regular feature of life. Soon some agency officials may make choices by electronically checking a box on an electronic option paper sent by electronic mail.

  The Forest Service System, [FN57] where directives from the agency head to the field offices do not necessarily exist in paper form at any time, is another good example of changing practices regarding decisional documents. As electronic mail becomes more pervasive, and as electronic publishing of government information becomes more common, the paper version of many official documents may disappear. As electronic mail is used more widely by a greater variety of agency decision *975 makers below the top level, the part of public decisionmaking covered by paper documents shrinks.

  There also is a need for better records management plans to capture  "prefinal" communications to preserve the content of the decision process. Such predecisional electronic transactions likely are protected by FOIA exemption 5, [FN58] but they are an important part of the historical record. FOIA exemption 5 states that internal agency memoranda subject to a deliberative privilege need not be made accessible under the Act.

  Scientific and technical data present very different problems. One problem is the large amount of data collected, particularly the quantities of data collected by satellite. New technology such as weather satellites enormously increase the amount of information that is collected and used by agencies like the National Oceanic and Atmospheric Administration (NOAA), which is responsible for weather forecasting. The other problem relates to the interpretation and analysis of such information. An unskilled person cannot use satellite weather information in its raw forms even if presented on paper media. Even a meteorologist cannot use the raw information without software and careful documentation compatible with the data.

  To some extent, scientific and technical data share a general problem with electronic records: a need for compatible hardware and software to retrieve the information. In another respect, however, the interpretation and analysis problem with scientific and technical data is different. Scientific and technical data is specialized and its consumers will always be specialists in particular fields such as meteorology and climatology. The specialized nature of this information justifies handling archives of such information separately from the general archives. Indeed, separate handling of NOAA information implicitly is the result of a NARA inability to accept large quantities of satellite data from NOAA. [FN59]

  Administrative information includes purely internal information like information pertaining to agency personnel and contracting. It also includes mission-related information such as claims files for the Social Security Administration (SSA) and the Veterans Administration (VA), and tax records of the Internal Revenue Service (IRS). [FN60] Agencies *976 have their own incentives to maintain administrative information for extended periods of time, independent of obligations imposed by the records statutes. These mission- related incentives relate to a potential for litigation over contested claims, or contested personnel actions. Agencies are aggressively automating the processing of such information simply to meet the increasing burden of claims. Agencies also have incentives to apply new archival technologies to increase storage density for large volumes of information. The SSA, VA and IRS have been aggressive in exploring optical storage technologies.

  Social information has received the greatest attention in terms of electronic records management. Census information was the first type of archival information made available to the public. Furthermore, the private sector invests much effort to ensure the utility of electronic records formats for academic researchers and others.

  A basic crosscutting distinction must be made, under present technology, creation and use patterns, between data base information and word processing documents. Word processing documents are fixed in form, though they may exist in different versions. Data bases are inherently dynamic. A document is more likely to have decisional significance than a snapshot of a data base, "taken" at a time when no one happened to be retrieving information from the data base. Moreover, the data base itself is not what people consume or ultimately use. A data base is analogous to a dictionary out of which users extract pieces of information to assemble into documents with significant information value.

2. Value in Information

  An important part of what makes information useful is value that has been added to form the information. The dimensions of new information technologies can be explored more carefully by thinking about ten discrete characteristics or attributes of information products, or "types of value." These types of value can be unbundled from each other and supplied and assembled by various agents through high-speed digital networks, or CDROMs, for eventual presentation to the end user. Vertically integrated publishers, adding all ten types of value, may be less appropriate as consumer preferences shift toward electronic formats and away from paper ones.

  The value-added analytical framework says that information products are bundles of ten types of value added through creation, organizing, retrieval- and-assembly, and marketing processes:

Process       Type of Value          Print Examples                            

Creating      1. authorship          Content generated by original author       

Organizing    2. chunking-and-tagg-  Organization boundaries: sections,        

                ing                    paragraphs, pagination, chapter         

                                       boundaries; headings and titles, running

                                       headers and footers, page numbers       

              3. internal-pointers   Tables of contents, indexes               

              4. external-pointers   Bibliographies                            

Retrieval &   5. presentation        Print on paper; video displays            

  assembly                                                                     

              6. duplication         All copies after the first                

              7. distribution        Getting the information from the generator

                                       to the consumer                         

Marketing     8. promotion           Advertising; inclusion in lists, product  

                                       reviews                                 

              9. billing             Identifying users; assessing prices,      

                                       collecting money                        

              10.                    Guaranteeing the accuracy, expertise of   

                integrity-assurance    suppliers of each type of value;        

                                       guaranteeing against forgery or         

                                       tampering                               

  *977 A modern print publisher is a broker or assembler of all of the types of value. The publisher arranges with an author to supply authorship value, and has designers and copy editors on staff or under contract to supply chunking and tagging value. Authors also supply substantial chunking and tagging value, typically determining the boundaries of paragraph, section and chapter chunks. Publishers supervise extractors who prepare internal pointers value in the form of tables of contents and indexes; arrange with the Library of Congress for the cataloging-in-publication information and with Books in Print and reviewers for external pointers value; contract with printers and binders to supply duplication and presentation value; work with advertising agencies for promotion value; and handle distribution and billing value through warehousing and order fulfillment. Publishers traditionally also play an important quality control function, not only in the sense of editing and checking for clarity and accuracy, but also in the sense of selecting*978 material, deciding which authors communicate well, and which authors or concepts of information packaging are sufficiently authoritative to be useful to user communities.

  Chunking-and-tagging value, internal pointers value and external pointers value, increase utility to consumers because they reduce the cost of human browsing, searching and retrieving. With print technologies, chunking-and- tagging value involves all basic typographic design features. Scanning a newspaper is easy because the material has considerable chunking-and-tagging value, primarily headlines. It also is easy because newspapers with state-of- the art design features have indexes (internal pointers value), which point to particular pages and story headlines (chunking-and-tagging value). Researchers make use of external pointers value when they consult the Index to the New York Times or the Readers' Guide to Periodical Literature. These also have pointers to human-processable tags and chunks, the call numbers, volume and page numbers, and titles of the articles referred to.

  Information products are bundles of these different value types. For example, a typical book is a bundle of authorship value reflected in the raw text, chunking-and-tagging value reflected in its structure of articles, sections, pages, and paragraphs, internal pointers value in terms of tables of contents and indexes, and presentation value in terms of printing all of this on paper bound between two covers. Unbundling separates these aspects of added value, possibly supplying them separately to consumers.

  This ten-type, value-added concept framework is as useful for evaluating the electronic archives issues. [FN61] Agencies are likely to strip value from electronic records submitted to the archives for different reasons than they may strip value from information released to the public. [FN62] In both cases, however, it is appropriate to understand the nature of the value lost.

  The National Archives can provide for all ten types of added value, or none. For example, the information transferred to the National Archives could include authorship, chunking-and-tagging, internal and external pointers value types, [FN63] and the National Archives itself *979 could add presentation, duplication, distribution, promotion, billing, and integrity- assurance value types as desired by National Archives users. [FN64]

  Alternatively, agencies could provide only the raw content (authorship value) of the information to the National Archives. Then, the National Archives system could provide for adding presentation value, but nothing else. Image presentation is necessary for a person to use information and therefore presentation value is an essential part of any information system including a National Archives system.

  Standards are a necessary prerequisite to transferring authorship, chunking and tagging, and internal and external pointers value to the Archives.  [FN65] These standards relate to the interface between an agency and the Archives. Such standards would permit the structure of a text document, and its formatting, to be tagged in a way that receiving computers could understand. Standard Generalized Markup Language (SGML) is an example. Standards also are necessary if an archival system is to add duplication and distribution value.  [FN66] Here, standards relate to the interface between the archives and the user. Standards relating to presentation on a printer or a video screen, such as Postscript page description languages, are examples.

II. LEGAL IMPACT OF TECHNICAL ISSUES

A. Technical Issues with Legal Implications Raised by Electronic Media

  A 1990 report of the National Historical Publications and Records Commission   [FN67] identified the following problems and issues for electronic records:

    -- System dependencies;

    -- Storage media;

    *980 -- Easy changes and easy loss of information;

    -- Difficulty in defining a record; and

    -- Difficulty in distinguishing originals from copies. [FN68]

  The Commission concluded that the following solution alternatives are worthy of attention:

    -- System design to provide for electronic records management and archival needs from the beginning;

    -- Development of appropriate standards to reduce fragmented system dependency; and

    -- Interdisciplinary projects. [FN69]

  The following sections in this subpart elaborate on the issues raised by the new technologies. They generally agree with the Commission as to the nature of the problems posed. First, however, the background of records construction is addressed.

1. Background

WHAT CONSTITUTES A RECORD?

  Electronic records management concepts are based on the concept of a record copy of a document. It is natural to assume that information comes in packages called documents. But this has not always been so. Until the thirteenth or fourteenth century, information was primarily oral, although much valuable information, especially official records of church and state, was inscribed on a variety of media, ranging from clay tablets and papyrus to hand-written parchment. Information creation, transfer and preservation all were oral. Written information became important in the legal reforms initiated shortly after the Norman invasion of England, especially by Henry II. The printing press, once it was widely accepted, transformed social and cultural concepts of information again, leading to new packages such as newspapers and books. Gradually, new typographic forms emerged to facilitate the use of information by a broad range of consumers. Current cultural and social institutions have grown up around printing press technology, and consequently it is natural to think of information issues only in terms of that technology.

  The electronic information revolution is, however, as profound as the printing press revolution in its potential impact on cultural and social *981 patterns for creating and using information. While new information paradigms have not yet emerged with sufficient clarity to be widely accepted, policy makers and designers of long-term information policies should be wary of accepting printed information paradigms uncritically.

  The appearance of on-demand publishing technologies [FN70] presents major challenges for any archive policy. If books and newspapers can be produced in many different versions, with press runs on the order of a dozen per version, should one archive all versions, and if not, which version should be archived? A compromise possibility is to archive the most extensive version, or possibly a master copy of all fragments that were included in any version, along with a kind of control document that would correlate document sections with particular versions, showing which section was included in which version. [FN71] The problem, of course, is the same problem associated with archiving any data base: Which state of the data base is the archival copy?

  Electronic records can be thought of as analogs of paper memoranda and reports, or they can be thought of as analogs of telephone conversations. The latter analogy obviates concerns with preserving electronic records because telephone conversations are not preserved as a part of records management and archival activities [FN72] even though the technology exists for recording telephone conversations and saving the recordings. [FN73] It may be that some electronic information volatility characteristics impose such challenges to records retention efforts that the telephone conversation analogy is the best place to end. On the other hand, the paper memoranda and report analogy is clearly appropriate for electronic information possessing reasonable permanence in content, even though the final content may have evolved through many versions.

  *982 A more profound example of the inadequacy of customary information concepts is the impracticability of applying them to compound documents created by electronic technologies. In the future, the prevalent means of accessing information may be through a kind of master document which really is little more than a set of pointers to textual, graphic and sound information stored in diverse places. Presently, a single query from a relational data base is a precursor to that kind of compound document, although all elements selected by a data base query usually represent similar forms of information. [FN74] Perhaps the most important long-term objective is to develop conceptual frameworks and strategies for dealing with compound documents.

  The Patent Office system [FN75] is an actual example of a compound document system. In this system, text and images are stored separately and put together into a compound document only when an examiner requests. United States Geological Survey (USGS) maps may be another useful example, because of the desirability of representing maps as modules of small geographic areas. A large map would be a series of pointers to smaller area modules. The USGS publishes maps that cover small areas in great detail, and larger areas in less detail. Through a compound document approach, maps of larger areas would not have the detail, but instead would have pointers or references to maps of smaller areas, which would contain the detail. A map user could "zoom" in on a particular area, causing the more detailed map to be retrieved and presented to the user.

  Hypertext is another clear example. Hypertext is a technique for linking different documents or materials with pointers that can be activated by a user. The linked materials may include graphical images and sound, as well as text. Compound documents are collections of related materials, the content of any one of which can change independent of the other materials in the collection. Both hypertext and compound documents are analogous to a master contract, all of the terms of which are incorporated by reference, as they may be amended from time to time. [FN76]

  Several questions arise regarding the future meaning of "record" *983 for historical purposes. For example, what is the relevant information for historical purposes to show: the state of the data base against which pointers were executed? A collection of pointers is one unambiguous representation at a point in time. But pointers alone lack meaning unless one can restore their referents. [FN77] Should the strategy be to save transactions rather than data bases? [FN78] These issues are not unique to federal electronic records management; they will be pervasive as compound documents come into wider use.

  One possibility, of course, is that new information technologies require people to abandon their preoccupation with the state of information at a particular point in time as being usually represented by a paper document. But if this concept is abandoned, it is not clear how historical developments and changes can be recorded. On the other hand, the new technologies permit more information to be captured about the processes of using information than is possible with paper documentation, which shows only what the information was, not how it was used or by whom.

2. Records Disposal

  A major problem with paper records is storing them. If everything is saved, storage requirements become too great. Moreover, effective access for legal, policy development, or historical reasons becomes much more difficult with larger inventories of paper records, because finding the desired record requires one to review many more records. Thus, an important objective of a sound records management policy governing paper records is to ensure that a substantial portion of records without historical value are destroyed.

  Electronic technologies mitigate this need to some extent. The space required to store information in electronic form is several orders of magnitude less than the space required to store paper information. Moreover, electronic retrieval techniques make it less burdensome to search for a particular record through many undesired records.

  Nevertheless, there is a practical need for considering records disposal as a part of system design. Electronic storage technologies have finite capacities. Frequently insufficient attention is paid to getting information *984 out of a system. As occurred with the Forest Service System when information is not removed from electronic system, storage requirements soon exceed design assumptions.

3. Records Retention

  Records retention requires deciding what records should be retained and then actually keeping those records when other records are destroyed. These two acts, deciding and keeping, can be done at the same time or they can be separated. In an extremely simple hypothetical agency, a decisionmaker could review records, making judgments on the spot as to what records are of historical value and setting those records physically aside for retention while throwing other records in the waste basket. In the real world, however, the decision about what to retain is made in the abstract, through the articulation of criteria for retention. Then the act of keeping records meeting the criteria is made by someone else (perhaps a computer) at a later point in time.

  Electronic technologies raise two generic problems with records retention. First, they make it much easier to destroy records. Second, they also make it easier to apply predeveloped criteria for records retention.

  There is no question that text and document management systems are appropriate adjuncts to a sound electronic records management plan. Otherwise, dispersal of electronic documents on decentralized computer systems, and cryptic file names, make capture of appropriate official records even more difficult in an electronic regime than in a paper regime. Moreover, electronic document management approaches may make it easier to discover and capture official records using electronic techniques than is possible with paper techniques.

  Agencies usually assign a relatively low priority to records management activities. It is not likely to be feasible, therefore, to impose additional burdens on agency employees to code working documents for eventual archival purposes if the coding requirements impose any significant cost, in terms of time or inconvenience. Certainly a post hoc requirement to reformat word processing documents with archival value would impose significant costs. Therefore, the best approach for textual documents is either to design archival coding into the document management system at the outset, or to have an overinclusive schedule that causes most textual documents to be transferred to the Archives on increasingly inexpensive storage media. This strategy assumes that free text searching techniques can only improve and, even with the present *985 state of the art, these techniques would permit appropriate retrieval precision when someone wants to access the documents.

  The ultimate goal should be making records management transparent to agency employees. For example, official correspondence could necessitate an accompanying code that must be sent before the system would recognize it as official correspondence. The code would cause the document to be saved and processed for eventual archiving.

  Designing agency information systems to provide adequate records management capability, transparent or otherwise, implicates a tension between centralized and decentralized automation strategies that is pervasive in office automation. [FN79] Centralizing information system administration makes it easier to ensure that records management policies are followed. But centralization deprives individual users of the autonomy they have come to expect as a result of the PC revolution. Autonomy in using computing resources is not only an expectation; it is a technological capability that enhances individual productivity. [FN80]

  There are two points to be made about this tension. First, the tension should be recognized. It is unavoidable. It also is not new; the conflicting desires for centralized control and decentralized entrepreneurial energy is a central issue of organization design. [FN81] Second, a really good electronic records management program permits substantial amounts of user autonomy with respect to software selection and use patterns.

  A concrete example may be helpful in illustrating this point. Suppose an agency adopts ODA/ODIF as the standard for its textual documents. Office Document Architecture/Office Document Interchange Format (ODA/ODIF) is an international standard for representing the structure and format of office documents. It might consider three alternative strategies for ensuring that agency personnel comply with the standard. One strategy would be to develop new word processing, document management and desktop publishing systems that comply with the standard. A second alternative would be to procure commercially available products that comply with the standard and to permit personnel to use no other products except a single selected product, say *986 Microsoft Word. A third alternative would be to identify a preferred product, say Microsoft Word, and allow users to use any other product as long as files can be transferred from the non-preferred products to the records management system.  [FN82] Of these alternatives, the first is the worst, and the third is the best. The public interest is not served by sacrificing agency effectiveness and productivity in the interest of long-term archives enhancement. Electronic records management practices must serve both goals.

  Considering management of electronic records forces planners to think about a choice that regularly confronts designers of document management systems for office automation systems. Should one invest effort at the beginning in formats and indexing for an eventual retrieval, or should one simply save the information in whatever format is best for its original use, and burden the individual archive user with search and retrieval efforts? [FN83]

  Historically the National Archives took the latter approach. Information was, for the most part, simply forwarded to the National Archives in whatever formats existed, with only limited effort expended toward indexing. Computer data base techniques, however, generally focus on front-end investments in tagging aimed at reducing the cost of eventual retrieval. To the extent that front-end formatting and tagging imposes avoidable human costs, the protocol for records retention is less likely actually to be followed by document generators and receivers. On the other hand, some features aimed at records management can be built into data base systems at a relatively low cost. Data base approaches frequently are preferred by agencies, not for any reasons having to do with record management, but simply to facilitate accomplishment of the agency mission.

  Document management systems that associate profiles with specific textual documents apply some of the data base design approaches to free text documents. The reworked Forest Service system provides virtually immediate dissemination of agency head documents to some 300 field offices by automatically designating appropriate documents as official records and referring them electronically to the archives.

  The philosophy expressed by the NARA guidelines envisions individual microcomputer users following records management plans while creating files within word processing, data base and spreadsheet applications, *987 and ensuring that an appropriate official copy exists before erasing files. Ensuring compliance with such guidelines is extremely challenging. Decentralization of computer file management associated with the microcomputer revolution obviously increases the challenge.

  On the other hand, ensuring compliance with this electronic records management philosophy is no more challenging than ensuring compliance with paper records management philosophies, given the reality that important subsets of agency records exist in filing systems (which may not be all that systematic) under individual control, and not only in centralized agency file rooms.

  Indeed, it may be easier to capture appropriate records when the records exist in electronic form than when they exist in paper form. A paper document existing only in the personal files of the author and the addressee may never be known to an agency's official records management system. An electronic equivalent of the same document is likely to exist on a computer network and can be discoverable by an appropriately designed electronic protocol for electronic records management. Electronic technology permits draft management, because a well-designed office automation system can track who looked at, and commented on, a document and when.

  Automating capture of drafts and of decisionmaking records is much easier on networked computer systems, as compared with free-standing microcomputer systems. On networked systems, the records management functions can be designed into the host or the file server, which keeps most of the files for an entire system of users. On free-standing microcomputer systems, in contrast, the only way to implement automated records tracking functions is to rely on the user of the particular microcomputer periodically to transfer documents via floppy disk or other transferable medium, or to rely on a management function that periodically copies documents from the hard disk of each microcomputer.

  Indeed, Local Area Networks (LANS) force a degree of electronic records management on users and LAN administrators that is not absolutely necessary in standalone computing environments. If users are to be able to find files on a LAN, some attention in designing and running the LAN must be given to electronic records management. [FN84]

  Despite these technological capabilities, the goal of capturing draft *988 documents in order to record the process of policy-making is, to a considerable extent, unrealistic. [FN85] Draft documents do not necessarily reflect the policy deliberations, and forcing the retention of drafts may be impracticable.

  The author has had considerable experience in negotiating labor agreements and participating in policy formulation at the cabinet and sub- cabinet level of the United States government. Both of these processes produce many draft documents. Many of the most important ingredients in the policy development process, however, are oral, and are never reflected in discrete changes between two identifiable draft documents. Moreover, many drafts are never seriously considered, because they were written at lower levels not informed by the most current discussions among decisionmakers, or were drafted by interested parties and submitted to the decisionmakers, or reflected efforts to make more concrete the options at the margin of what policy-makers were willing to consider. Not only do such draft documents not capture the main parts of policy deliberations, preserving them may affirmatively mislead historians as to what was seriously considered.

  Requiring that personal computer users keep all drafts of their documents on a central network server is somewhat like requiring office workers to keep no papers in their desks, but only in centralized filing cabinets. The likelihood of compliance is low. Of course, one could design personal computers so that they do not have disk drives, just as one could design desks so they do not have drawers. Such an approach, however, sacrifices worker convenience and productivity in the interests of records management, and this may not be a sensible organizational strategy.

  The policy formulation process is as much oral as it is written. The new technologies do not change that reality, unless electronic mail induces decisionmakers to deliberate less face-to-face and over the telephone and more via electronic mail. Significant investment or distortion of agency practices is not warranted in order to capture draft documents.

  The NARA guidelines may be too limited in requiring structured and field- based approaches to automatic document retention. It is a mistake to limit thinking about automation retention selection to structured *989 headers. Effort should be invested to develop free text algorithms (probably on an agency-by-agency basis) for selecting documents as candidates for retention.  [FN86] Free text searching technologies available now would permit development of reliable algorithms for individual agencies, based on key words likely to be contained in archival records. Thus, the full text of working documents could be searched at appropriate intervals, with the algorithm deciding what records should be archived. NARA's Center for Electronic Records is working on free text retrieval approaches.

4. Records Integrity and Accessibility

  Records retention in archives is ineffective unless the archived records can be accessed. Three technical issues with legal implications relate to accessibility: ease of alteration, physical deterioration, and retrieveability. All have to do with integrity in some sense. Electronic records are easier to alter than paper records if they are stored on magnetic media. Meaningful access to electronic records requires that formats be processable and readable. Paper records that physically survive remain readable as long as one understands the language in which they are written. Electronic records present larger hardware, software and format compatibility problems. Some electronic records deteriorate more rapidly than paper records. [FN87] Large quantities of records impede practical retrieval unless records are indexed.

a. Easy Alteration

  Magnetic media is easier to alter and there is less likelihood of detecting the alteration than in paper media. If one changes something on a piece of paper, the eraser or the white-out correction is usually evident--at least on the original. Because the physical changes involved in changing the content of magnetic storage are not visible, the possibility of undetected alteration of magnetic records is appropriately of concern. It should be noted, however, that information stored on optical media cannot be altered as easily-- at least on CDROM media. Information recorded on CDROM is permanent because it is represented by *990 bits burned into the surface of a plastic disk by a laser beam. Especially for information that is used for evidentiary purposes in legal proceedings, procedural or technology techniques must be used to reduce the risk of alteration.

b. Deterioration of Media

  Paper. Environmental control [FN88] is important to protect paper and bindings. [FN89] All post-1840 paper is subject to acid deterioration which cannot be halted, except for freezing or helium atmosphere storage. [FN90] In 1986, the Law Library Journal included an updated report on the lifespan of paper. [FN91] It concluded that twentieth century paper possesses a lifespan of thirty to eighty years, depending on the production quality of the book and of the individual paper (in the case of photocopies or unbound works).  [FN92] Another report indicates that the maximum longevity of twentieth century paper is fifty years. [FN93]

  Microfilm. Microfilm exceeds all other recording media in terms of longevity. Estimates range upwards to three hundred years for silver microfilm and to one hundred years for diazo microfilm. Silver film, however, is more easily destroyed by fungus and scratches than its diazo counterpart. [FN94] Microfilm requires particular storage conditions, and appropriate processing and packaging in order to avoid quick deterioration. [FN95]

  Magnetic Tape. Conservative estimates concerning the archival lifespan of magnetic tape range from five to ten years. [FN96] There are other estimates asserting that the tape may last up to twenty years. [FN97] *991 Reliability of estimates are questionable, however, because of the wide use of the Arrahenius test, [FN98] which incorrectly presumes that accelerated testing produces the same or similar results as actual long-term exposure. [FN99]

  Like any recording medium, the better the quality of the product, the longer it will last. [FN100] Additionally, appropriate storage conditions extend the lifespan of magnetic tape. The key to successful storage is the retention of the tape at least every two years. [FN101] This presents a burden for constant retentioning on a records custodian like the National Archives, which may possess millions of tapes.

  Magnetic Cartridges. Magnetic cartridges are better for storage than magnetic tape. An accelerated test performed by IBM indicated that the 3480 class tape cartridge (3480) has a life expectancy of over thirty years. [FN102]

  From an archives institutional perspective, cartridges are better than magnetic tapes because no attention need be given to retentioning of cartridges. [FN103] Also, the binder (i.e., casing) of the cartridge prevents many of the dangers associated with handling of tapes. [FN104] For example, there is less danger of dust or fingerprints damaging the tape because it is sealed in the binder, unlike the magnetic tape on exposed reels. While cartridges require ambient storage conditions similar to those of magnetic tape, they can withstand more severe conditions than magnetic tape. [FN105]

  In order to ensure reliability, vendors suggest that cartridges be exercised annually or semiannually. [FN106] Archivists must consider which type of tape to use in cartridges. Chromium oxide tape is less stable *992 than iron oxide tape, particularly in the long run. [FN107] "If archivists use a chromium dioxide medium, they must monitor it rigorously for deterioration and insure proper storage and handling." [FN108]

  Diskettes. Neither the American National Standards Institute (ANSI) nor the National Institute of Standards and Technology (NIST) has compiled data on the archival life expectancy of magnetic diskettes, though ANSI currently is conducting tests on diskettes to determine their longevity. Regardless of the type of diskette, [FN109] the archival life expectancy of a magnetic disk allegedly is forever. [FN110] Diskettes possess an estimated 1,600 hours of usable life; that equates to over 32 million revolutions. [FN111] Under the proper storage conditions, a diskette may never lose its stored information due to non-use. These conditions should be approximately 75 degrees Fahrenheit and 40% relative humidity. [FN112] The main problem with diskettes is their susceptibility to damage caused by improper storage or handling. [FN113]

  Compact Disk Read Only Memory (CDROM). There are no confirmed data on the lifespan of CDROMs. [FN114] The CDROM technology is so recent that there has not been ample opportunity to conduct tests on its reliability and storage life expectancy. [FN115] NIST agrees with this proposition but notes that there is an industry consensus that information stored on a CDROM is secure for at least three to five years, with some estimates reaching as high as twenty years. [FN116] The relatively short estimates for CDROM at least superficially are inconsistent with the virtually unlimited estimates for magnetic disks and WORM.

  Write Once Read Many (WORM). WORM technology is an optical disk technology, in which user drives permit information to be recorded permanently on a plastic disk. Once the information is recorded *993 it cannot be erased, but it can be read an indefinite number of times. Like CDROMs, there are few available data on WORMs because testing is currently being performed. Kodak, however, offers a thirty-year warranty on Kodak WORM products, though it claims that information stored on the WORM can last forever. [FN117] Like all other computer readable media, the longevity of a WORM depends on the product's quality, on careful storage and handling, and on the amount of use (more use leads to shorter life expectancy).

B. Retrieveability from Diverse Storage Formats

  The major issue confronting an increasingly electronic National Archives is how to maintain hardware and software necessary to access electronic records, which over time will embody a diverse universe of storage technologies. Even now, in the infancy of electronic technologies, it would be difficult in an ordinary office environment to find a computer and the necessary software to read a document stored on an eight inch floppy disk used by a mid 1970s Wang word processor, or a Radio Shack Model II microcomputer. Similarly, it would be difficult to access a document saved in 1984 by the Perfect Writer word processing software running on a CPM microcomputer and saved on a low density 5.25 inch floppy diskette.

  Generally, there is concern that any storage medium may become outdated by the quick advances of technology in computer hardware. Already, the federal government is having problems finding computers to read tapes from the 1960s.  [FN118] There is a report, perhaps apocryphal, that there are only two computers left in the world that can read 1960 census data, one in Japan and one in the Smithsonian Institution. The potential for adaptability should be taken into account when selecting a computer readable medium for long-term storage of information. The Internal Revenue Service requires that corporate taxpayers who submit electronically must maintain not only the electronic records supporting the filing, but also a complete copy of the system that can retrieve the information and the operating system.

  Mass electronic storage is not the problem; the problem is ensuring access to electronic information over periods exceeding thirty to forty years. Several years ago, a NARA advisory committee said the only *994 way to assure permanent accessibility was to record information in human readable form. It expressed confidence that conversion between human readable form and electronic forms would be nearly transparent. NARA rejected that conclusion, believing that technologies would develop to the point that accessibility of information would not depend on the availability of the same hardware and software that created it. The current solution, and an acceptable one for the future, to impermanence of electronic information in various media is to recopy or "refresh" electronic information at periodic intervals. One simply recopies magnetic tape or files on magnetic disks. Recopying is a solution not only to impermanence but also to technological obsolescence. When a format standard is becoming obsolete, information can be recopied to a new format standard. Of course this recopying requires resources and effective implementation of a program that ensures the copying and appropriate intervals.

  NARA Bulletin Number 88-8 [FN119] rejects optical disks as media for NARA excision of permanent records "because of the hardware and software dependence of current optical disk systems and the absence of standards that ensure affordability of data to one system to another." Paragraph six of the Bulletin reports that industry standards for optical disk systems are now under development. The Bulletin also explicitly permits agencies to use optical disk media for storage and retrieval of permanent records while the records remain in an agency's legal custody, although no permanent records may be destroyed after copying onto an optical disk without NARA's approval.

  The current staff thinking at NARA is that appropriate standards are the best way to assure infinite accessibility notwithstanding changes and hardware and software platforms. What is needed is a new richer standard to serve the function now served by flat file ASCII.

C. Impact of Emerging Freedom of Information Act Policies for Electronic Records

  The Freedom of Information Act (FOIA) [FN120] influences electronic records management in two ways. It indirectly can require the preservation of files that otherwise might be destroyed, and it permits nearterm public access to electronic files.

  The FOIA does not directly require preservation of agency *995 records. The FOIA requires that the public be given access to records but does not require them to be maintained, preserved, or created in the first place. However, the FOIA arguably can be used indirectly to require the preservation of electronic files or paper. If an FOIA requester requires access to a large number of electronic files, a reasonable inference is to require the agency possessing the records to preserve them until its FOIA access obligations are determined. Those files that are exempt from public access under the FOIA could be destroyed.

  The requester would argue that the Freedom of Information Act would be defeated by permitting an agency to destroy records that eventually would be subject to access under the Act. Under this argument, an injunction against destruction or erasure would be necessary to permit FOIA rights of access eventually to be realized. This is essentially the rationale adopted by the D.C. Circuit Court of Appeals in American Friends Service Committee v. Webster. [FN121]

  The absence of explicit linkage between the FOIA and records statutes creates problems for sound records management. Agencies are concerned that declaring something to be a record for records management purposes will increase the likelihood of access under the FOIA. They tend, therefore, to be underinclusive in their records management practices. Ironically, the FOIA definition of "agency record" has been interpreted more broadly than NARA interprets the "records" definition under the records statutes.

  The Privacy Act [FN122] has little effect on electronic records management practices, except that it has marginally improved agency practices by creating an additional incentive to designate certain electronic information systems as containing records, thus making NARA aware of them.

  Administrative Conference Recommendation 88-10 [FN123] was developed in part to consider the treatment of electronic records under the FOIA. Recommendation 88-10 impacts electronic records management in four major ways.  [FN124] First, Recommendation 88-10 strongly embraces *996 the proposition that electronic information should be treated as agency records for FOIA purposes. The basis for this conclusion supports treatment of electronic information as records for records management purposes.

  Second, it encourages agencies to design FOIA retrieval capabilities into their information systems. Designing FOIA capability into systems is one aspect of designing records management capabilities. The major difference between FOIA and archival needs is that identification of information desired in connection with an FOIA request is made at the time of the request by the consumer of the information. Designation of information for records management purposes is done in advance, as a part of classifying types of information to be contained in a system, or as a part of the authoring step.

  Third, Recommendation 88-10 encourages agencies to experiment with electronic techniques for improving public participation in agency rulemaking and adjudicatory proceedings. As this occurs, more of the public record public input to agency decisions will exist in electronic form. Moreover, the data base structures necessary for effective management of public comments also will facilitate management of this information for archival purposes.

  Finally, Recommendation 88-10 encourages agencies to use accepted standards for formats and implicitly encourages agencies to participate in improving standards. This is consistent with the need for better standards for agency electronic records.

D. Copyrights

  One difficulty in moving toward electronic archives is the possibility that intellectual property rights in public data, data structures, or retrieval processes will inhibit appropriate levels of public access. Two underlying intellectual property issues are involved with this concern: Whether public information can become private intellectual property when it becomes part of a compilation authored by a private entity, and whether intellectual property interests in retrieval processes (software, formats and indexes) may be used to retrieve public information with *997 which they are associated without infringing private sector intellectual property interests.

  Opposing these concerns about intellectual property impeding public access are concerns to stimulate private sector investment in making electronic public information more readily accessible. In some cases federal agencies create incentives to stimulate investment in electronic publishing of public data. In such cases, as Administrative Conference of the United States (ACUS) Recommendation 88-10, paragraph F recognizes, a sponsoring agency may consider protecting markets for private sector providers prepared to make a commitment to appropriate product features. Such arrangements may permit a private entity to reserve a copyright [FN125] or other proprietary interests in specific value added features, [FN126] but not in the raw agency information or in valueadded features developed by the agency.

  These intellectual property interests should not be protected in a way that eliminates public benefits from government-funded development or restricts the public availability of public information. For example, members of the public should be permitted to use value-added features to facilitate disclosure-level information release, much as libraries disclose copyrighted works to consumers while not interfering with copyright owner rights.

  The potential conflict between protection of intellectual property and ensuring public access to archived electronic information is especially acute when new formats are involved. It is important, on the other hand, for agencies to embrace new formats so that the public can have the benefits of the new format. The new formats can be embraced only by respecting intellectual property. Intellectual property protection can impede public availability to new formats in three ways. First, the existence of the intellectual property may prohibit agency release of the information to the public. This is unlikely because copyright interests alone should not exempt information from accessibility under the FOIA. [FN127]

  *998 Second, the agency or the user may not be able to get the proprietary software needed to use the information, either because it is no longer in print, or because it is too expensive. Finally, the proprietary software needed to use the information may be incompatible with the user or the agency's current hardware or operating systems.

E. Use of Electronic Records as Evidence

  Public records serve an important evidentiary role in resolving disputes among citizens, and between citizens and the government. It is important, therefore, that public records be accepted by dispute resolution tribunals. Changing to electronic formats should not impede the acceptability. There is no reason to be concerned that well-designed electronic records systems present any new problems with admissibility of electronic records under the rules of evidence.

  A major difficulty in appropriate records management strategies is a lack of familiarity by the government's lawyers with the technology, and conclusions reached about legal treatment of the technology. For example, agency lawyers repeatedly "reinvent the wheel" regarding admissibility of electronic information and the internal procedure necessary to authenticate electronic information offered as evidence as well as electronic signatures.

  There is a natural tendency when a new technology raises legal concerns to forget basic principles and to embark on unnecessarily difficult and complex de novo searches for legal principles to govern the technologies. This natural tendency should be resisted. A de novo inquiry is not necessary; the areas of concern unique to electronic methods can be addressed best by narrowly focusing on the differences between electronic techniques and conventional techniques.

  Reliability of computer records is suspect due to the inherently greater volatility of computer and telecommunications data, as compared with written data. But relative reliability of different means of proving legal facts is not a new problem for the law. Historically contract law promoted reliability of contract proof by insisting on written documentation of certain contracts under the Statute of Frauds [FN128] or of other signature requirements. The same underlying issues are addressed *999 by the parol evidence rule [FN129] and by rules of evidence pertaining to the inadmissibility of hearsay, subject to exceptions for business records. [FN130] The standard for the business records exception to the hearsay rule is not necessarily the same as the standards for the Statute of Frauds or the parol evidence rule. Nevertheless, all three legal concepts share a common concern with reliability. All three recognize that some forms of evidence have greater reliability than others.

  Document authentication [FN131] or proof of the inclusion of particular terms in documents requires that the communications making up the transaction be recorded and retrievable in some reasonably permanent way. Statutes of frauds require a "writing" as the permanent means of recordation. Statutes of frauds also require signatures, which also are intended to serve an evidentiary purpose.

  Writings and signatures serve somewhat different purposes, although their purposes also overlap to a considerable degree. A signature protects against outright disavowal of a document by a party who denies authorship of a writing. A writing memorializes the terms of a document and thus protects against disputes over the terms of a document, admitting that a document exists. Reliability and authentication issues are related to the signature requirement under the Uniform Commercial Code, although the relationship is not altogether clear. [FN132]

  Hypertext and compound documents are emerging concepts that represent a high level of incorporation by reference, and major challenges in terms of creating a reliable transaction record.

  The Justice Department has not provided recent official guidance on the admissibility of electronically filed federal records as evidence. The most recent effort by the department seems to be a November 6, 1986, draft memorandum. [FN133] The points made in the draft are relatively non- controversial. The provision of the draft NARA regulations *1000 concerning judicial acceptance of electronic records is a step in the right direction; to give clearer and more official guidance on evidentiary treatment of electronic records.

  Despite the applicability of these general legal concepts, however, because electronic records are particularly susceptible to purposeful or accidental alterations, or incorrect processing, authenticating them may be more difficult than for paper records using the same standards of authentication. [FN134]

III. INSTITUTIONAL ARRANGEMENTS

A. Institutional Responsibility

1. National Archives and Records Administration

  The National Archives and Records Administration was created by the National Archives and Records Administration Act of 1984 [FN135] as an independent executive agency of the United States Government. NARA succeeds the National Archives Establishment, which was created by act of June 19, 1934. [FN136] The National Archives Establishment was subsequently incorporated into the General Services Administration (GSA) and renamed as the National Archives and Records Service by section 104 of the Federal Property and Administrative Services Act of 1949. [FN137]

  Under existing law, the Archivist of the United States has the authority to accept federal records and private materials that the Archivist has determined to have sufficient historical or other value to warrant their continued preservation by the United States government. The Archivist then causes the transfer of accepted records [FN138] to the National *1001 Archives. As a general matter, NARA handles records according to agency instructions, subject to FOIA obligations imposed on NARA. [FN139] Only permanent federal records defined as such on an SF115 (request for records disposition authority) and approved by the Archivist, are transferred to the National Archives. [FN140] Records already deposited into the National Archives may be designated for transfer to educational institutions or associations. [FN141] The Archivist is authorized to promulgate regulations establishing procedures for preserving records. [FN142]

2. Agencies

  Federal agencies work with the Archivist to determine what records will be deposited into the National Archives. Each agency has a records manager who is responsible for inventorying all records in the agency, proposing disposition instructions for approval by NARA, and creating a comprehensive agency records disposition schedule. This schedule is submitted to NARA for approval. [FN143]

  The current text of the Paperwork Reduction Act makes records management a part of information resources management and gives the Office of Management and Budget (OMB) authority to review compliance with records management laws and regulations. [FN144] The version of H.R. 3695 that passed the House amends the federal records act to authorize the Archivist of the United States to issue regulations establishing standards for interpreting the term "records."  [FN145]

  The OMB has directed that "[a]gencies should incorporate records management and archival considerations in the design, development, and implementation of electronic information collection systems in accordance with the Federal Records Act." [FN146]

  *1002 On May 8, 1990, NARA published a final rule revising its regulations on electronic records. [FN147] The rule established procedures for management of electronic records, for selection and maintenance of electronic storage media and for compliance with legal requirements for disposition of electronic records. Identical regulations were issued simultaneously by the General Services Administration. [FN148]

  The new NARA rules, like longstanding statutory and regulatory records management policies, impose on individual agencies the primary responsibilities for development and implementation of electronic records management plans. Under the regulations, the head of each Federal agency must ensure that his or her agency manages its electronic records so as to satisfy certain requirements:

    -- (a) Assigning responsibility to develop and implement an agencywide program for the management of all records created, received, maintained, used, or stored on electronic media; and notifying [NARA] . . . of the name and title of the person assigned the responsibility.

    -- (b) Integrating the management of electronic records with other records and information resources management programs of the agency.

    -- (c) Incorporating electronic records management objectives, responsibilities, and authorities in pertinent agency directives and disseminating them throughout the agency as appropriate.

    -- (d) Establishing procedures for addressing records management requirements, including recordkeeping requirements and disposition, before approving new electronic records systems or enhancements to existing systems.

    -- (e) Ensuring that adequate training is provided for users of electronic records systems in the operation, care, and handling of the equipment, software, and media used in the system.

    -- (f) Developing and maintaining up-to-date documentation [FN149] about all electronic records systems that is adequate to: Specify all technical characteristics necessary for reading or processing the records; identify all defined inputs and outputs of the system; define the contents of the files and records; determine restrictions on access and use; . . . describe conditions and rules for adding . . . changing . . . or deleting [system] information; and ensure the timely, authorized disposition of the records.

    -- (g) Specifying the location, manner, and media in which electronic records will be maintained to meet operational and archival requirements, and maintaining inventories of electronic records systems to facilitate disposition.

    -- (h) Developing and securing NARA approval of records disposition schedules, and ensuring implementation of their provisions.

    -- (i) Specifying the methods of implementing controls over national security-classified, sensitive, proprietary, and Privacy Act records stored and used electronically.

    *1003 -- (j) Establishing procedures to ensure that the requirements of this part are applied to those electronic records that are created or maintained by contractors [and] . . .

    -- (l) Reviewing electronic records systems periodically for conformance to [legal requirements] . . . [and to] determine if the records have been properly identified and described, and whether the schedule descriptions and retention periods reflect the current informational content and use. . . .  [FN150]

  Revised 36 C.F.R. §  1234.22 specifically addressed creation and use of text documents. It requires that systems maintaining the official file copy of text documents on electronic media to meet the following requirements:

    -- identify electronic versions of documents appropriately; [FN151]

    -- Provide a method for all authorized users of the system to retrieve desired documents, such as an indexing or text search system;

    -- Provide an appropriate level of security to ensure integrity of the documents;

    -- Provide a standard interchange format when necessary to permit the exchange of documents on electronic media between agency computers using different software/operating systems and the conversion or migration of documents on electronic media from one system to another; and

    -- Provide for the disposition of the documents including, when necessary, the requirements for transferring permanent records to NARA. . . .  [FN152]

  The NARA regulations show strong concern about the permanency of record media. Revised §  1234.28(b) (Selection and Maintenance of Electronic Records Storage Media) requires consideration of the following factors in selecting a storage medium or converting from one medium to another:

    -- (1) The authorized life of the records, as determined during the scheduling process;

    -- (2) The maintenance necessary to retain the records;

    -- (3) The cost of storing and retrieving the records;

    -- (4) The records density;

    -- (5) The access time to retrieve stored records;

    -- (6) The portability of the medium (that is selecting a medium that will run on equipment offered by multiple manufacturers) and the ability to transfer *1004 the information from one medium to another (such as from optical disk to magnetic tape); and

    -- (7) Whether the medium meets current applicable Federal Information Processing Standards. [FN153]

  Agencies must "ensure that all authorized users can identify and retrieve information stored on diskettes, removable disks, or tapes by establishing or adopting proceduresfor external labeling." [FN154]

  Agencies must "ensure that information is not lost because of changing technology or deterioration by converting storage media to provide compatibility with the agency's current hardware and software. Before conversion to a different medium, agencies must determine that the authorized disposition of the electronic records can be implemented after conversion."  [FN155] NARA declined to amend §  1234.30(b) (retention of electronic records) to require transfer of a copy of the software and operating system in order for the records to be read at the Archives. Transfer of the software and operating system, whenever necessary, can be specified by NARA when approving a records schedule.

  NARA had received comments from 18 federal agencies, two private organizations, and one Member of Congress on its draft regulations. Most commentators supported the concepts embodied in the regulations, while suggesting less detail. Because the detailed requirements apply only to unscheduled or permanent electronic records, NARA noted in its preamble that agencies could reduce the burdens of compliance by scheduling records as soon as possible, thus reducing the universe of records subject to the detailed requirements. Furthermore, NARA also urged agencies to transfer permanent electronic records to NARA as soon as possible after their creation, further reducing burdens on agencies.

  NARA rejected a comment recommending a new paragraph related to electronic signatures, on the grounds that it was premature to issue regulatory guidance on this subject. NARA noted objections to its proposed ban on floppy disks for permanent storage and modified the language to give agencies more discretion,  [FN156] while expressing continued *1005 concern about careless handling leading to loss of data stored on floppy disks.

B. Schedule Twenty

  NARA has created general records schedules, which govern the disposition of certain records common to many agencies. [FN157] General Records Schedule 20 deals with the disposition of Machine-Readable Records.