Federal open records laws, executive orders on national security classification, and similar policy instruments in the states require that most records at government agencies be available to the public upon request -- subject to exemptions which may entail review. Traditionally this review has been accomplished by redaction (purging) of hardcopy. Resource limitations and increases in the number of requests demand a re-thinking of the process. Offices are experimenting with optical scanning, onscreen purging, and online dissemination. Statutory and common law trends may soon require disclosure of database reports in electronic form, when so requested, rather than printouts. The dissemination of manipulable data of commercial value requires a rationale broader than the "citizenship rights" that justified the Freedom of Information Act. Online publication of agency controlled records also demands careful planning of information systems and public networks. The issues discussed in this article are central to the establishment of an adequate policy for electronic freedom of information.
 

For decades, Freedom of Information Act (FOIA) processing and national security record declassifications have been carried out by Dickensian methods involving grease pencils and scalpels. FOIA releases usually involve no more technology than the photocopier and the postal frank. Given severe staff shortages and low budget priorities, the existence of year-to-year backlogs of requests at some agencies is hardly surprising.⁽²⁾ The basic issue addressed in this article is whether the political, administrative, and legal forces surrounding freedom of information programs will be receptive to automated systems for public access.

Over a hundred Federal agencies are subject to the disclosure requirements of public records acts, foremost among them the FOIA.⁽³⁾ This Act of 1966, as amended significantly in 1974 and 1986,⁽⁴⁾ establishes in section (a) that government records are public records releasable upon request to "any person," but subject in section (b) to nine exemptions which permit withholding of portions of documents.⁽⁵⁾ Although some 30 agencies have a negligible burden and another 30 a minor load of FOIA requests, a dozen or so have tens of thousands of requests and programs costing millions of dollars. Examples at the cabinet level are the departments of Defense, State, and Health and Human Services; at the subcabinet level, the Federal Bureau of Investigation (FBI), Food and Drug Administration (FDA), and Environmental Protection Agency (EPA). Controversy surrounding the Act has been more heated than one would expect from the modest costs involved, largely because the Act has required a marked shift in the conception of bureaucratic responsibility; officials and public servants accountable just to Congress and the Presidency, have become increasingly accountable to the public and its interest groups.

FOIA operations are squeezed in various agencies by hiring freezes, early retirement plans, and budget strictures that respond to plans for deficit reduction and "reinventing government." Vice President Gore's National Performance Review claimed a reduction of 70,000 Federal civil service jobs during 1993, 100,000 by the end of 1994, and a target of 250,000 for the full term. Opening the Internet (forerunner of the information superhighway) to commercial entities, jointly with more user-friendly access software (such as Mosaic), is bound to accelerate demands for online publication of official data.⁽⁶⁾ Under the pressures, FOIA directors are likely to find that automation of FOIA requests is preferable to an increase in staff time necessary to handle the volume of requests.

In order to establish whether the political climate is suited to the development of online FOIA processing and online publication, we need to explore recent policy developments in Congress and the administration.

The concerns of current policy about the control of commercially valuable official data, user fees for release of data, and automation of FOIA responses are different from those of past policy cycles that produced access legislation in 1958, 1966, 1974 and a restrictive amendment in 1986.⁽⁷⁾ From 1948 until the 1980s, the FOIA had been debated quite separately from questions of the automation of federal offices. The first signs of interest in merging the FOIA with online systems appeared in the mid 1980s, when the Office of Management and Budget (OMB) began preparing general guidance for handling records in the electronic information age. OMB's Office of Information and Regulatory Affairs (OIRA) was granted extraordinary control over executive regulations through the 1980 Paperwork Reduction Act and a Reagan administration mandate to review all proposed regulations. In 1985, OMB's Circular A-130, "The Management of Federal Information Resources," raised issues concerning adaptation to the digital information age.⁽⁹⁾ Valuable more for raising interesting questions than for settling answers, the circular established terms of a debate about the role of the public and private sectors in digital publication and database control.⁽¹⁰⁾ It argued for avoiding monopolies of information, for giving value-added resellers a role in the private sector, for the contracting out of databases, and for continuance of inexpensive public access to official records. Viewing the bureaucracy as perhaps too ungainly to drive the information highway alongside the more agile private companies, in maintenance of information resources it urged the Federal government (in flexible ways consistent with the FOIA) to make better use of price discrimination and market forces. The 1985 circular perhaps fit the reformist "reinventing government" paradigm adopted later by the Clinton-Gore administration.⁽¹¹⁾

Electronic FOIA Investigations in Congress.

Grappling with the same issues but from an FOI rather than a management perspective, the House Subcommittee on Government Information (of the Government Operations committee) held hearings, in 1985, on Electronic Collection and Dissemination of Information by Federal Agencies.⁽¹²⁾ The subcommittee concerned itself only with databases normally open to the public under access laws.⁽¹³⁾ Fearing that electronic data systems would increase agencies' control of government information, the 1986 House report based on those hearings sought to maintain a role for the private sector:

While agencies must fully comply with statutory disclosure requirements, agencies should attempt to preserve a role for private sector companies in offering information products and services by engaging in fair competition with the private sector.⁽¹⁴⁾

Agencies use the new information technology to broaden and improve public use of government information; more administrative guidance on the development and use of electronic information systems be provided; agencies consult regularly with those affected by electronic information systems; competitive procurements be used for the acquisition of automated information products and services; and laws that have been interpreted to allow agencies to maintain exclusive control over electronic databases be modified.⁽¹⁵⁾

With the aid of the General Accounting Office, contractors and interested experts outside of government, the Office of Technology Assessment (OTA) reported on a project to examine Federal information policies in the electronic age:

OTA has concluded that congressional action is urgently needed to resolve Federal information dissemination issues and to set the direction of federal activities for years to come. The government is at the crucial point where opportunities presented by the information technologies, such as productivity and cost-effectiveness improvements, are substantial. However, the stakes, including preservation and/or enhancement of public access to government information plus maintenance of the fiscal and administrative responsibilities of the agencies, are high and need to be carefully balanced by Congress.⁽¹⁷⁾

Electronic FOIA Bills in Congress.

Democrats in both chambers produced electronic FOIA bills in the late 1980s. Rep. Bob Wise (D-WV) proposed an electronic FOIA during his two-year chairmanship of the Subcommittee on Government Information. Lacking, however, were pressures from a concerted FOIA lobby or bipartisan coalition reacting to executive secrecy. Public Citizen's counsel resisted spelling out in statute the conditions of electronic freedom of information, arguing that the courts were moving toward giving requesters a choice in electronic formats and that conservatives might interpret a new law differently.⁽¹⁸⁾ The Democrats restored their Senate majority in 1987, but their hold on policy was tenuous. Although Sen. Patrick Leahy (D-VT) chaired the subcommittee on technology and the law, Sen. Orrin Hatch (R-UT) retained conservative influence over information policy through the rival constitution subcommittee. In October 1986, the sudden conservative reform of the FOIA via a rider to the war-on-drugs bill gave ample cause for the FOIA lobby to be wary of a legislative strategy rather than a judicial or administrative strategy for policy change.

Sen. Leahy sponsored a companion bill, S.1940 of the 102nd Congress,⁽¹⁹⁾ in which an early draft proposed that agencies complying with time limits should retain half the FOIA fees they collected. That draft proposed penalties for undue agency delays: courts could order payment of out-of-pocket expenses up to $75 per day; agencies objections, however, led to this being dropped in later versions. The bill made it explicit that agency records included electronic records and defined a search as "a manual or automated examination to locate records."⁽²⁰⁾ It proposed that computer capabilities be rated and that agencies able to deliver data in computer formats do so when requested.

Unlike the Senate, the Bush White House was unfavorable towards an electronic FOIA. During the Bush administration electronic FOIA bills languished. The president's veto threat was conveyed to the 1992 Senate Hearings by Steven R. Schlesinger, Director of the Office of Policy Development, Department of Justice.⁽²¹⁾

In contrast, the Clinton administration has reasserted (at least) the principle of open government being tied to new access technology. On February 22, 1993, in announcing the administration's technology initiative to develop the Internet -- the "information superhighway," the Vice President mentioned creation of the Government Information Locator System (GILS), an online index to materials publicly available at agencies.⁽²²⁾

Executive Reaction to the Senate's Electronic FOIA Proposals.

Agency responses to OMB concerning the Leahy bills were consistent with the long-term pattern of opposition to greater administrative burdens.⁽²³⁾ OMB's draft guidelines of November 1994 set forth 66 principles, which, in sum, were less forceful than the Senate bill in requiring agencies to release digital records. The major step lay in abandoning the Dismukes standard (discussed below) by which agencies could decide upon a release format. According to the leading trade newsletter, many agencies have been voluntarily releasing data in a variety of formats, stepping away from the Dismukes defense.⁽²⁴⁾ Unlike the Senate, which generously opted for requiring "reasonable" efforts to convert data into a previously unavailable but requested format, OMB suggested limiting format conversion to a meager two hours of labor time and to an overall agency resource limit. While this could be adequate for a simple request, one can easily imagine a request involving a large database where it would be reasonable to expect days of programmer time rather than hours. While OMB's approach would recover some expense, it could easily be abused as a means of cutting down on FOIA processing.

Clinton Administration Policy on FOIA Issues.

On June 25, 1993, Sally Katzen, of OMB's OIRA, in a revised Circular A-130, announced a policy favoring agency release of electronic information over having private sector contractors do the job (as the Reagan administration had preferred). Agencies were requested to integrate CD-ROM and online indexing and publication with hardcopy processes, and to charge only for dissemination, not acquisition cost.⁽²⁵⁾

In an October 4, 1993 memo to agency heads, the President said:

The [FOIA] is a vital part of the participatory system of government. I am committed to enhancing its effectiveness in my Administration. ... Federal departments and agencies should handle requests for information in a customer-friendly manner. ... This is an appropriate time for all agencies to take a fresh look at their administration of the Act, to reduce backlogs of Freedom of Information Act requests, and to conform agency practice to the new litigation guidance issued by the Attorney General ... Each agency has a responsibility to distribute information on its own initiative, and to enhance public access through the use of electronic information systems.

The Department will no longer defend an agency's withholding of information merely because there is a 'substantial legal basis' for doing so. Rather, in determining whether or not to defend a nondisclosure decision, we will apply a presumption of disclosure. ... Where an item of information might technically or arguably fall within an exemption, it ought not to be withheld from a FOIA requester unless it need be ... Accordingly I strongly encourage your officers to make 'discretionary disclosures' whenever possible under the Act. ... [The serious problem of backlogs] principally appears to be a problem of too few resources in the face of too heavy a workload. ... The American public's understanding of the workings of its government is a cornerstone of our democracy.

In August 1994, during the 103rd Congress, Leahy sponsored S.1782, an "Electronic FOI Improvement Act" that offered agencies incentives to curb delays in processing and obligated agencies to give the requester a choice of electronic data formats.⁽²⁶⁾ A companion bill, introduced as HR.4917 by Rep. Maria Cantwell (D-WA) and co-sponsored by Gary Condit (D-CA) chair of the Government Operations subcommittee on Information, Justice, Transportation and Agriculture, faced no prospect of House passage in the end-of-term rush. Despite passage in the Senate, House leadership seems to have accepted OMB's advice not to press for the bill. After the November election brought an historic turnover of leadership in both chambers to the Republicans, the House subcommittee was reorganized, dispersing its experienced staff of Democratic appointees.⁽²⁷⁾

The fates of a new pair of bills in the Republican-led 104th Congress are in doubt. In general, while the FOIA has long enjoyed a bipartisan coalition in the House, there has been a discord between liberals (e.g. Leahy, D-VT) and conservatives (e.g. Hatch, R-UT) in the Senate since 1981. With conservative Republican leadership in the House it might be feared that the FOIA consensus would disappear; but there is instead some prospect of developing further FOIA reform. First, Speaker Gingrich (R-GA) is not a traditional law-and-order-and-defense conservative but a market-oriented representative of Marietta, a high-tech suburb at the heart of the 6th district near Atlanta. Secondly, although he has called for cutbacks in domestic welfare programs, he clearly retains an interest both in the information superhighway and specifically in providing additional government information service to the public. In January 1995 he endorsed the "Infobahn," as well as expansion of the congressional Gopher and World Wide Web services.⁽²⁸⁾

Thirdly, Speaker Gingrich also has a sure instinct for press coverage, which for 40 years has driven FOIA oversight in Congress. While not actively lobbying these days for extension of the FOIA, the press has a substantial investment in the use of information systems, certainly more so than the public interest groups. A combination of interests in the press might, therefore, focus attention on the current FOI policy cycle: that of extending to the electronic age a law conceived in the era of paper documents.

A Side Issue: Extension of the FOIA to Congressional Records.

The Republicans' pre-election Contract with America included the Grassley (R-IA) Congressional Accountability bill (passed by Congress in January 1995) to apply 11 major employee protection laws to the Congress -- a logic of equivalency between executive and legislative bureaucracies that presumably could be extended in future to the FOIA.⁽²⁹⁾ Indeed, an influential Heritage Foundation paperback lent support for this extension:

Apply Freedom of Information. If there is a single law that most needs to be applied to Congress it is the Freedom of Information Act. Congress gets away with many abuses simply because no one can find out about them: the cozy relationships that comprise the Barnumocracy [sic], the petty acts of retribution that keep bureaucrats in thrall of Capitol Hill. If Congressmen and their staffs were required to keep adequate records and to make them available to the public, many congressional abuses would go away overnight, and questionable behavior wouild be subject to the informed judgment of voters. Again, the light of accountability is the key to reform.⁽³⁰⁾

Electronic records add to a list of questions that await definitive ruling.⁽³¹⁾ What is the purpose and scope of the FOIA? What do "agency records" embrace? Do they include internal databases and software (as distinguished from document files)? And particularly, what types of database search create a "new record" (and are not required since the FOIA is limited to existing records)?⁽³²⁾

As for the purpose of the FOIA, long understood to be that public business is the public's business and that official records are the public's records, an unexpected nuance emerged from the Supreme Court's decision in Reporter's Committee. The Court opined that FOIA was intended to help the public learn what government is up to. Employee home address mailing lists requested for collective bargaining did not contribute to this goal. While this, indeed, was featured in congressional debate as a rationale for passage of the FOIA, the point was intended to establish that official records are public records, not to limit the scope of the Act.⁽³³⁾

Although the FOIA amendment of 1974 substituted the term "records" (implying reproducible materials) for "documents" to ensure that searches would include electronic and photographic representations as well as paper documents, the courts have not ruled on what constitutes: (1) a reasonable database search or (2) reasonable online access for the public.

In Long v IRS (1979) the court found -- and the Supreme Court let stand -- that "computer-stored records, whether stored in the central processing unit, on magnetic tape or in some other form, are still 'records' for purposes of the FOIA."⁽³⁴⁾ In the leading case on agency records,⁽³⁵⁾ Yeager v DEA (1982), the court held the FOIA "makes no distinction between records maintained in manual and computer storage systems."⁽³⁶⁾ While clearly making electronic records subject to the Act, these findings in no way hint that electronic records are often more valuable than hardcopy records. Certainly, they do not establish any duty to create a virtually new record from a database. The Yeager precedent, while establishing a basis for treatment of electronic records, may also have set a ceiling. A goal of the Leahy electronic FOIA was to establish that electronic records are superior in value to printouts and that there is a positive duty to offer digital formats where available.

The leading decision against releasing records in the format preferred by plaintiffs is Dismukes v Dept. of the Interior (1984).⁽³⁷⁾ Here the court upheld the agency's decision to offer microfiche in place of the requested magnetic tape, observing that it would not "unreasonably hamper" the requester's access to obtain records in the format "most useful to the general public." It is this ruling that Senate bills regarding electronic FOI have sought to overturn.

E-mail records present an especially difficult case. Used in place of voice calls, they are generally treated as ephemera, not as agency records. Yet, e-mail records are typed, routinely stored, and centrally archived much like traditional memoranda. Consequently (as Oliver North of the NSC staff testified before Congress during the Iran-contra affair hearings), far from being ephemeral they can be restored even after deliberate deletion at a local terminal. The National Security Archive, among others, had sought and obtained the e-mail records in searchable, digital form.⁽³⁸⁾ In the National Security Council e-mail case, Armstrong v Bush (1989), the district court distinguished between telephone conversations and electronic mail records. On remand (sub nom. Armstrong v EOP, 1991), it found that the e-mail records contain information not found in any hardcopy equivalent.⁽³⁹⁾ This finding is in the tradition of US v Nixon, the Watergate tapes case, where the court (in a criminal investigation) mandated the release of the original tapes -- not expurgated transcripts, as the president proposed. If e-mail records contain more than hardcopy records, then a searchable database must constitute more than a printout of its content.

One of the gray areas of the electronic records law is whether agency databases with no direct impact on the public are subject to the FOIA or may be protected under the (b)(5) exemption for internal memoranda concerning deliberative process. An example arose following numerous, anguished consumer requests for information concerning the discontinued dietary supplement L-tryptophan suspected of causing EMS syndrome. Both the Centers for Disease Control (CDC) and FDA released voluminous paper documents on the subject -- 25,000 pages from the CDC. Then, in Gottlieb v HHS (1993), counsel for SDK, the manufacturer, sought a copy of the database concerning studies on the supplement. Judge Charles Richey found (citing Yeager v DEA, 1982)⁽⁴⁰⁾ that the agency was not obligated to reconstruct a database from the numerous case reports. He also found that the CDC internal database was an agency record furthering the agency's work, but that its deliberative nature -- consisting of notes written by a scientist revealing "their creator's mental processes" during an investigation -- permitted protection under exemption (b)(5) for deliberative privilege.⁽⁴¹⁾ The decision, thus, seems double-edged: recognizing the qualitatively different nature of electronic records, while still applying to them the FOIA provisions of disclosure -- much as the House subcommittee suggested in its 1986 report.

Another question facing Richey: Was the software developed to manipulate the database also an agency record, or was it, as the CDC maintained, "a list of instructions for a computer to manipulate a database" rather than "a medium to record and preserve information?" The court found, following the Supreme Court's decision in Forsham v Harris (1980), that records include "machine readable materials ... regardless of physical form or characteristics," and that because of the unique tailoring of the software to the database it preserved information and perpetuated knowledge responsive to plaintiff's FOIA request.⁽⁴²⁾ In other words, where a database could only be searched and used with a piece of software, the program must be released with the data. Such a rule could have far-reaching implications, for the databases of mainframe computers are less likely to be as standardized in format as those of small personal computers. Thus, mainframe programs owned by, rather than licensed to, the Federal government might have to be released with FOIA data.

The key question for FOIA processing is whether databases must be searched in such a way as to create a new record -- a step not required of paper documents -- simply in virtue of the ease of doing so by computer. Recently, Judge Louis Oberdorfer of the DC Federal district ruled, in Thompson v HCFA, that an agency must conduct a computer search in response to a request for a tailored list, even though the Health Care Finance Administration argued that no such record existed. This extends an emergent, but still uneven, pattern of requiring reasonable efforts to compile a record via a database search.⁽⁴³⁾

Virtually all states and many localities have information access laws that roughly parallel those at the Federal level. Yet, a recent survey found only nine states with specific provisions dealing with access to electronic records. Nonetheless, a market in the compilation and reselling of databases from such governments is already flourishing. This often involves customized maps, mailing lists, or reports for commercial development. State motor vehicle departments regularly sell names and addresses to direct marketers, and approximately 60 percent of counties sell some data.⁽⁴⁴⁾ In one state a private company proposed to become the central data repository and to disseminate data for 11 agencies and functions via a private time-sharing system.⁽⁴⁵⁾

One of the questions surrounding such sales by government is whether the needed return from investment should override the principle of releasing at marginal cost of dissemination. The states can provide some evidence, albeit with mixed results. Creation of mapping software (or geographic information systems, GIS) frequently requires such a large investment that governments are tempted by public-private consortia to spread the costs and generate a revenue stream. To institute commercial changes, Iowa and Oregon have exempted state and local agencies' GIS data from access law obligations, an unfortunate trend unless carefully considered in the broad context of information policy. Given the tax revolts felt in many states in the 1980s, a major factor driving data sales in the 1990s is undoubtedly the desire to raise new sources of state revenue. Occasionally an effort meets with great success. For instance, a state department of motor vehicles can realize in excess of $10 million for its lists.⁽⁴⁶⁾ Although the Texas Open Records Act prohibits the charging of excessive fees for data, the Texas department of public safety realized $52 million in revenues from data sales in FY 1992. It had obtained, in 1987, an exempting law saving it from releasing "class listings" under the act. Confronted with a 1993 request for the entire database of driver license data at the marginal cost of transferring the data to tape, it initially demanded $60 million for the 12 million records, at the standard access rate of $5 per individual record.⁽⁴⁷⁾ High revenue levels are not common, however, and one cannot determine from isolated successes that the Federal government would derive sufficient revenue from data sales to justify overriding the FOIA principle of releasing at minimal cost.

While Federal agencies are often prohibited from copyrighting their data, states are not necessarily so restricted. Some states have, in recent years, made great efforts to "reinvent" themselves by marketing intellectual property. Colorado copyrights all its databases; Washington and many other states copyright their revised codes. Florida -- cited repeatedly in Reinventing Government for establishing state profit centers⁽⁴⁸⁾ -- is reported to have had a significant trade in the sale of data and software. Hawaii Inc., a public-access network, charges for access to five databases while making others available free of charge. Some states permit differential charges for requesters planning to resell the data, although enforcement of the distinction about whether or not the data are copyrighted is difficult.⁽⁴⁹⁾ The courts appear to be generally counting data as intellectual property, but a number of counter cases permit third parties to redistribute government data.⁽⁵⁰⁾ There is some precedent for the Federal government to begin copyrighting its data, but it is not a short-term prospect. In terms of the data superhighway, the question is whether the government revenue from toll booths (less than the cost of collection) is worth more than the long-term economic stimulus from supplying both infrastructure and data gratis. Taxpayers may question the injustice of paying twice: once for creating the data with public staff and then again while purchasing the data as consumers.

Some governments have formed independent sales agencies, or a "joint powers authority."⁽⁵¹⁾ The Vermont State Center for Geographic Information, Inc., a joint venture of the state and its university, has as its mission the coordination of the strategic use of GIS and the selling of hardcopy and online services that cannot be provided by the private sector.⁽⁵²⁾ The San Diego Association of Governments and the Atlanta Regional Commission formed nonprofit organizations for market research, accumulation of data, and data sales.

The contract solution is a more traditional articulation of public and private. In Nevada, a private firm sells dial-up access to diverse government databases, among them driver and vehicle registration, public works procurement, court dockets and building permits. Los Angeles Urban Research Division and Santa Cruz County are also implementing such data publication contracts.

It is unclear whether Federal access law dominates when both Federal and state sources fund a database. The Federal agency involved has no ability to copyright, while the state agency probably does. As for court-held data, common types of data request are those for landlord and tenant records, renter credit histories, traffic scofflaw lists, and "rap sheets" (criminal histories). The Information Industry Association (IIA) argues that agencies should cease dictating the data format, since the format greatly influences the market value of the data.⁽⁵³⁾ This interest was absent from the 1950s and 1960s FOIA debate, when commercial value was merely a by-product of the desire for public records access.

A 1989 study identified two dozen Federal agencies operating data acquisition and release programs.⁽⁵⁴⁾ On the other hand, recent studies in Nevada and Maine each identified over 200 pieces of legislation including FOI and Privacy acts that exempt many types of data from disclosure.⁽⁵⁵⁾ Privatization of data dissemination potentially involves loss of control over personal privacy, unless privacy rules in a contract can be enforced much as national security classification rules are enforced in defense contracts. The Privacy Act of 1974 -- which is limited to government records, unlike the European Data Protection Acts, which apply to company records -- prohibits redirection of personal data beyond the purpose for which it was originally collected, except for "routine uses" among Federal agencies. Obtaining consent from all donors of personal data may be impractical, and this prohibition may be largely honored in the breach.⁽⁵⁶⁾ Some such sales of data are benign: for instance, the Postal Service's sales of address corrections to direct marketers saves the Service millions of dollars by avoiding misaddressed mail. On the other hand, such sales may render citizens more reluctant to divulge information to government, and may result in the nuisance of lawsuits.

The courts, presumably wishing to encourage the dissemination of information, have generally held that there is "no implied warranty" for information in books. Thus, authors, publishers, and booksellers are not information merchants and not liable for inaccuracy in print.⁽⁵⁷⁾ Since courts treat defective machinery in stricter terms when determining liability, it is possible that, in future, they could treat data on a stricter basis. An ominous precedent is the liability of aerial map publishers for an inaccuracy determined to be the cause of an airplane crash. If the courts were to extend the doctrine of liability to official data, the implications -- for example, for maintenance of safety data on car crash tests at the National Highway Traffic Safety Administration, pesticides at the Environmental Protection Agency (EPA) and food and drugs at the Food and Drug Administration (FDA) -- could stifle the willingness of these agencies to process FOIA requests.

Federal concerns over record formats, software as data, and contractors as data custodians are reflected, to some degree, in the states. In Ohio, agencies are not required to respond to a request if it necessitates generating new records, formats, or analyses.⁽⁵⁸⁾ A Maine Justice Department study found that the FOIA could be interpreted as requiring the production of custom software to serve commercial demands to access a public database.⁽⁵⁹⁾ In the states, as in the Federal government, where electronic records are privatized, there is a legitimate concern that revenues may prove insufficient to cover long-term database maintenance.⁽⁶⁰⁾ Termination procedures should perhaps be specified in information system contracts.

New Information System Design

States in the 1980s acted in other policy areas as laboratories of innovation, sometimes running ahead of the Federal government. In the area of official information this is less common, though Connecticut, for example, has proposed that all state agencies obtaining new information systems demonstrate that they have considered the access rights of the public in analyzing new system requirements and designing or acquiring systems.⁽⁶¹⁾

Agencies developing in-house systems for disclosing electronic records typically are subject to lags in technology, largely because of onerous procedures for planning, bidding, and approval. As Sen. William Cohen (R-ME) of the relevant oversight subcommittee put it, "It takes the federal government almost four years to purchase computers, compared with 13 months in the private sector. Computer technology is developing rapidly, while the government' acquisition process moves at a glacial pace." His panel found the government spending at a current rate of $25 billion a year on computers and related services, totalling some $200 billion over the decade. Yet, some agencies suffered from truly outmoded technology or hardware incompatibility problems.⁽⁶²⁾ A case in point is the FDA's acquisition of a functioning minicomputer system for its overburdened FOIA staff only after 13 contracts and seven years.⁽⁶³⁾ With generations of computer equipment succeeding each other at three year intervals (20% compound growth of technology), obsolescence is endemic.⁽⁶⁴⁾

Just as agencies are experimenting with electronic publishing to supplement other forms of publication, so FOI shops need to explore procedural changes affecting the electronic submission, storage, requesting, and releasing of public documents. First, computer tracking and indexing databases are more conducive to efficient FOI searches than are traditional filing systems. Especially at technological entities such as the Defense Department, computer database practices are already commonplace. A limitation, however, resides in the prevalence at most agencies of multiple records systems. It is uncommon to have an index that flows across different data processing systems, let alone different hardware components, field offices, or agencies, and consequently the domain of an adequate search has to be locally determined by rule of thumb.

Second, since efficient request processing relies heavily on front-line staff rather than on technicians, user-friendly database software is the key to improvement in FOIA processing. The generation of "new" documents, a task performed at agency discretion where a consolidated list is more convenient to release than entire sets of documents, became in the heyday of mainframe computing a matter of writing database query and report code. Now, in the age of high-resolution workstation displays and graphical user interfaces, it is becoming a matter of menu and button choices.⁽⁶⁵⁾ Agency processing needs to evolve, for the courts or Congress may soon require reports created if they involve "only a reasonable amount of effort." What is possible with a reasonable amount of effort continues to expand rapidly.

Third, electronic data submitted to government by contractors and by regulated industry can either be arranged into formal database fields, or else marked up to preserve structure less formally. In the interests of flexibility and utility, modern practice favors informal structure and embedded codes and tags that widen the basis for helpful searches and extractions.

Fourth, after redaction, if needed, data can be placed on a public server reachable via communication services. A menu will enable the requester to browse indexes and abstracts and to download material without making a formal FOIA request. Records will be fetched from state-of-the-art storage devices (perhaps optical drive arrays) and displayed, copied, or printed as desired.

Where data change weekly or daily, schedules must be more flexible. Some agencies already have standing orders for continually released information of such urgency that periodic copying and shipping of mainframe computer tapes is not acceptable. Such agencies consider attaching storage mechanisms to servers on either the Internet or a bulletin board system with telephone lines. To ensure data security, the servers must be dedicated to public access. Reviewed data would be uploaded to them on a daily or weekly basis. Given commercially valuable data, the higher the updating frequency, the higher the demand: a possible disincentive to increased efforts by FOIA staffs. The answer to this disincentive, it would seem, is to have automation reduce the staff time. Standard menu-driven server software (such as the hierarchical Gopher, the hypertext World Wide Web, and the full-text search Wide Area Information Server, or WAIS), together with continuing advances in hardware,⁽⁶⁶⁾ are making server dedication increasingly cost effective.

Fifth, the most popular classes of unpurged documents -- for example, the FDA's 510(k)s, requested by the tens of thousands -- would be fed in redacted form to a database open to all requesters, obviating the need for subsequent searches and redactions. The value of such a system for frequently requested data is readily apparent from the success of service companies (e.g., FOI Services Inc. of Rockville, MD) that store released documents, index them on computer, and, for a substantial fee, fill requests faster than they would be filled by the source agency.

Redaction, as currently practiced with scalpels, grease pencils, and repeated copying absorbs some 25 percent of FOIA staff time. Online processing could better this and cut down on trips to the office photocopier and warehouses. Accuracy of redaction, a sore point at present, could easily be increased by having legal citations inserted from a palette of buttons onscreen. Many subsequent requests could be filled without human interaction and fees collected from commercial requesters should increase as records take on more convenient formats.

At present, exempt removals are indicated in hardcopy releases, and marginal notes remark the legal justification. On-screen redaction would bring an end to the possibility of redacted text showing through blackout markings, but lacking safeguards, text might be excised electronically without leaving any trace of the extent of the deletion. The scale of redaction from an electronic record must be indicated, since a "segregable portion" legally may be anything from a single character to multiple pages.

Against any savings over conventional processing are the investment and operating costs of servers, networks, and workstations. Maintenance and training costs of an automated system may be at least 20 percent of the capital cost, possibly much higher. The life span of typical system components will probably not exceed five years because of technological obsolescence. In addition, improved FOIA service might lead to an increased demand for service. The whole system must as usual be backward compatible with traditional mailings and forward compatible with improving technology standards. FOIA officers will continue to rely on trained staff for redacting documents: given the many types of situations involved in requests, expert-system software probably cannot provide effective onscreen coaching.

Where would online systems find application? First, where entities hold their data primarily in electronic form (e.g., the Department of Defense). Second, where an entity faces political pressures to improve its FOIA responses (e.g., the State Department in 1992).⁽⁶⁷⁾ Third, where the requests are predominantly from corporations and lawyers with computer resources (e.g., the Securities and Exchange Commission, although its efforts focus on online publication rather than requests under the FOIA.) Fourth, where a substantial proportion of requested material is already purged and released ("fill-froms" in FOIA parlance). Fifth, where an interagency database presents an opportunity to reduce duplication of effort via cooperation among agencies.⁽⁶⁸⁾ (Note that public-interest groups claim, and some national security agencies fear, that multiple FOIA requests to different agencies can sometimes result in a fuller "jigsaw puzzle" or "mosaic" of an extensively redacted document.)

To illustrate an online processing context, consider the FDA. The majority of the agency's 47,000 yearly FOIA requests come from sophisticated commercial sources at least as computerized as FDA itself. One type of document accounts for most of the backlog, and a quarter of requests are for previously requested material. A profitable, computerized "sidecar" business (FOI Services Inc.) accounts for 27 percent of the requests to FDA and from that business' store of previously released FDA documents serves customers faster than the FDA can itself. Since the early 1990s, the FDA has implemented the Image system for scanning and onscreen redaction of its most backlogged document category (the 510k form in its Center for Medical Devices). Although the FDA claims that Image has increased efficiency, a doubling of the FOIA request workload at the Center for Medical Devices, and the implementation of a procedure for pre-disclosure notifications to companies that submit data have obscured any quantifiable improvements.⁽⁶⁹⁾ The FDA would appear to be a prime candidate for further automation.

The FOIA has, since 1974, required release of all segregable declassified portions.⁽⁷⁰⁾ Nonetheless, because the National Archives and Records Administration has considerable holdings of documents generated elsewhere, but only a small staff, declassification and release are conducted by whole document. Hence the agency -- a strong candidate for automated tracking, scanning, and dissemination -- has only a weak case for online redaction.

Putting aside the perennial problems of budgetary famine, hardware obsolescence, and lengthy procurement, there are still major obstacles in the way of electronic redaction and disclosure. The large existing inventory of hardcopy documents is a serious problem, although the success of scanners in achieving throughputs of 120 pages per second is easing the bottleneck. Storage, security, transmittal within the executive, and public disclosure also pose challenges, many of which can, or soon will be, surmounted at reasonable cost by a combination of commercial off-the-shelf ("COTS") hardware and software. Compression of files, linkages to searchable if imperfectly recognized text, secure fax lines between agencies having joint custodies over a document, and a "firewall" system protection between internal and external file servers will address most of the technical problems.⁽⁷¹⁾ Human review remains the fundamental bottleneck.

Automation is least appropriate where submitters and requesters are individuals rather than resourceful corporations, where the data are in haphazard hardcopy form, and where particular material is seldom requested.

The following subsections focus on current development of three processes, each derived from a clause of the FOIA. The FOIA's section (a) applies to all agency regulations and decisions having the force of law, in addition to all records (subject to nine listed exemptions composing section (b)). It requires that they be published, made available for public inspection, or -- the most time-consuming and controversial -- be disclosed upon request. Potentially, all these requirements can be assisted by digital means. Although controlled primarily by executive order, national security records may be requested under (a)(3) and are given broad exemption from disclosure by the (b)(1) clause of the FOIA. Classified material, while not necessarily commercially valuable except to spies, also invites online solutions by virtue of huge scale, digital origin, and sporadic need for rapid review.

Confronted by an ever-growing stock of classified records, the Carter administration called for an end to overclassification, and subjected most national security documents to age-determined downclassification and complete declassification after 20 years. In 1982, the Reagan administration however, abandoned automatic reclassification and encouraged overclassification.⁽⁷²⁾ The Clinton administration moved against this trend in a March 17, 1994 draft executive order on classified national security information, which took effect in 1995.

In the spirit of the Clinton/Gore initiatives combining open government with new technology, a number of agencies holding classified material jointly contracted with Mitre for the design of a declassification review system. The scale of the problem was put this way:

In Fiscal Year (FY) 1993, 6.6 million pages of previously classified documents were declassified and made public (30 percent fewer than in FY 1992). In both FY 1992 and FY 1993, however, classification actions (including original and derivative decisions) exceeded six million. Because most of these actions involved multiple pages, and each is subject to many reproductions, the number of new classified pages produced in these years greatly exceeded the number of pages declassified.⁽⁷³⁾

The initial study sketched several trade-offs: text versus graphics handling, scanning versus keying in data, and printing versus digital dissemination.⁽⁷⁴⁾ Generally, the system proposed is akin to the State Department's system. It is noteworthy that full-text search programs (e.g., Unix WAIS) already exceed the needs of typical FOIA declassification teams, as do available communication services. Given Mitre's assumptions for a declassification system with capacity of one million pages per year, system plus operating cost is estimated at between $2.2 and $3.9 million for the first five years. (Six such systems could match the current declassification throughput of the Defense Department. Declassification costs are not well estimated, but for comparison Defense reported FOIA costs of $30 million in calendar 1993.)⁽⁷⁵⁾ Generally, at other entities this is likely to be substantially less than the salary cost of FOIA staff. Text recognition at only 97 percent accuracy could still enjoy some improvement, but human review for declassification at a typical 100 pages per day and a maximum of about 200 pages per day -- as with FOIA processing -- will continue to limit processing throughput.⁽⁷⁶⁾

The FOIA's section (a)(1) requires that the general structure, functions, procedures and rules of each agency be published in the Federal Register. No unpublished regulation which adversely affects the public can be enforced in the courts. The Register, thus, provides an extensive but unwieldy source of regulations across the Federal government. But by the 1990s its hardcopy format lagged badly behind electronic versions offered by the private sector. In 1993, Congress became frustrated with the mail services and storage required by the daily Congressional Record and Federal Register. The Government Printing Office Improvement Act (PL 103-40) required them to be produced online in searchable format with an urgent deadline. This transcended the GPO's sales of diskettes and magnetic tapes, as well as its computer bulletin board service for thousands of official publications.

On June 8, 1994, Federal Register and Congressional Record isssues dating from January 2, 1994 were added to a Wide Area Information Server, accessible via Internet or modem. GPO's Electronic Information Dissemination System met the deadline by exploiting a limited system already in use on Capitol Hill.⁽⁷⁷⁾ The Register, already produced in digital form, was stored using Standardized General Markup Language (SGML).⁽⁷⁸⁾ The resultant database can be searched by supplying arguments such as subject, date, agency name, table of contents, Code of Federal Regulations number, or embedded character string. The Record database has a different structure, necessitating less convenient searches.⁽⁷⁹⁾

The Register's technical graphics were compressed and linked to the text of the Record by tags keyed to a glossary at the top of the document. While the graphics are still not accessible via modem dial-in, they can be downloaded via proprietary software linked to the Internet.⁽⁸⁰⁾ As the Register staff ruefully note, the 1993 Act gave them an unfunded job with a short deadline. Alas, despite planned enhancements, their database will be limited to the two most recent years of publication.⁽⁸¹⁾

The principle of coexistence with the private sector was upheld. Despite the GPO system and the rapidly developing congressional online system, Legi-Slate Inc. continues to serve congressional offices, and others, with a more sophisticated database of Congressional Record issues, bills in progress and Federal regulations. Commercial-rate access permits detailed searches to be guided from a menu of over forty entries. During 1994, also, Legi-Slate uploaded a limited dataset to the Minneapolis Gopher (an hierarchical, menu-driven Internet server) with access priced modestly for public libraries, thereby competing more directly with the GPO system. In addition, the venerable legislative reference service, Congressional Quarterly, added an Internet service in 1994.

The FOIA's section (a)(2) requires any materials not published or released upon request, to be made available for public inspection and copying. Normally an agency places the materials in public reading rooms. A computer bulletin board system is an obvious analog of a reading room, but one that can serve distant customers and merge public inspection with publication. The system may yield savings in both citizen time and agency staff time.

Among the first digital public inspection systems was MEDLARS, the National Library of Medicine's (NLM) Medical Literature Analysis and Retrieval System. Dating from the early 1960s, it fulfills the NLM's legal mandate to "publish and make available ... catalogs, indexes, and bibliographies" that "aid the dissemination of scientific and other information important to the progress of medicine and to the public health."⁽⁸²⁾ Of more recent systems, the EPA has its public online Toxic Resource Inventory, the SEC is phasing in its EDGAR system for record submission and dissemination, and the Federal Maritime Commission has an ocean tariff database open to uploading of import data. At the Patent and Trademark Office, a feasibility study has sketched the design of an online database; owing to the rapid pace of technology transfer between developed countries, the design is being coordinated with European Union patent offices. Other agencies such as the Bureau of the Census and FDA are experimenting with publication via compact disk or electronic bulletin board.⁽⁸³⁾

These new online databases raise difficult questions about the boundary between the public and private sectors, and between rights to access and commercial property rights. Expertise in up-to-date technology installation often resides with contractors in the private sector, yet contractors' profitability may depend on their ability to restrict access to the database and to charge fees above the marginal cost of dissemination. In 1986, House report 99-560 argued against permitting agencies to charge fair market value for data and against awarding contractors monopoly rights to official data:

First, the agency must be certain that it has fully met its obligations to provide public dissemination of its data ...Once adequate public disclosure has been assured, the private sector should be allowed every opportunity to redisseminate and to add value to the agency's data.

Second, the agency must avoid any arrangement that affords itself or a private company with any monopoly power over the data. Where the agency operates the electronic data system, copies of the entire data base should be made available so that there can be more than one source. Where a contractor operates a dissemination system, the contractor must not be able to assert monopoly controls such as high prices or restrictions on redisclosure.⁽⁸⁴⁾

Concerning contractor-maintained databases, it urged agencies to ensure by regulation some form of inexpensive access. For agency-maintained databases, it asked that sales of online access be balanced with inexpensive public access via reading room terminals. It cited with approval the access methods adopted by SEC, FDA, Bureau of the Census and USDA:

All provide for a basic disssemination service at usage charges that do not reflect the value of the database. All provide for private operation of the system. All are structured to permit additional distribution of the database and competition in the retail market for value-added services.⁽⁸⁵⁾

In principle, the subcommittee disapproved of loading the cost of new databases onto the private sector at the price of agreeing to restricted access, a potential infringement of the FOIA. The report criticized the NLM for denying a FOIA request by charging a portion ($50,000) of the database overhead rather than the (nominal) marginal cost of dissemination.⁽⁸⁶⁾ It criticized the Patent and Trademark Office for bartering information with service companies and raising the price of access to its trademark database from a nominal charge to $42,300.⁽⁸⁷⁾

The model that the subcommittee upheld was the open, competitive one adopted for the ocean tariffs and EDGAR databases. Reagan administration thrusts towards greater use of the private sector and towards reducing the net cost of the FOIA while raising fees and reducing disclosures, were largely counteracted by the House's insistence upon more traditional creation of public records and inexpensive dissemination to all comers.

Securities and Exchange Commission: Edgar database -- (a)(2)

The SEC's EDGAR will be, in 1996, the largest online public-access database in the Executive Branch. It functions as a clearinghouse for nearly 12 million pages of documents annually submitted by 15,000 publicly held companies.⁽⁸⁸⁾ Given report and disclosure requirements based on hardcopy standards, the SEC opted for a database with free-form text and capabilities for full-text search. Inputs are tagged with SGML, a mark-up language, to effectively enclose each portion of a document for routing purposes. In one planned enhancement, machine-parsable financial tables will permit an individual firm's data to be compared with aggregate statistics.⁽⁸⁹⁾

Submitting companies upload by modem reports averaging 42 pages. Lengthy submissions optionally can use computer tape.⁽⁹⁰⁾ Filing is distinct from subsequent processing; dissemination is accomplished online and through reading rooms. Mead Data Central, a contractor, maintains the database and provides access service commensurate with annual subscription.⁽⁹¹⁾ As the SEC staff put it, data are sold "by the bucket or the wheelbarrow," and some clients sell value-added analyses. The project has been criticized by one of Ralph Nader's public-interest groups for categorizing consumers,⁽⁹²⁾ and by others for illustrating with delays and cost-overruns the pitfalls of the open-bid procurement process.⁽⁹³⁾ Nonetheless, Congress generally considered it as a positive model.

FOIA requests increased far more than staff availability at most agencies throughout the 1970s and 1980s, pausing temporarily in the 1990s. FOIA section (a)(3) requires disclosure of portions of requested records unless they are already published; and the 1974 amendments imposed strict time limits: ten working days for determination of response. Coupled with a staff shortage and litigious requesters, this limit obliges agencies to consider investing in computers for onscreen redaction and tracking. Even the FDA, short of computer resources, is scanning the most popular hardcopy documents (medical device approval forms) into graphical files.⁽⁹⁴⁾

Declassification of national security material either pursuant to executive order or in the course of FOIA review, perhaps the most labor-intensive of all review tasks, could also be transformed by onscreen digital redaction and disclosure of records. The burden of document review was insisted upon by Congress during a major debate of the 1970s. In EPA v Mink (1973), the Supreme Court declined to review classified documents and thereby denied Rep. Patsy Mink (D-HI) and others some unclassified material in a classified file. Reversing this thrust, the 1974 joint conference committee adopted the Senate's doctrine of release of segregable portions of records. It also clarified that courts -- although according substantial weight to the agency's opinion -- would be expected to review classified material in camera. The 1974 amendment read:

Any reasonably segregable portion of a record shall be provided to any person requesting such record after deletion of the portions which are exempt under this subsection.

The Muskie amendment of 1974 also required that classified material be, in fact, properly classified pursuant to executive order, thereby calling for judicial review de novo:

Sec. 2. (a) [FOIA exemption] (b)(1) ... is amended to read as follows:

"(1)(A) specifically authorized under criteria established by an Executive order to be kept secret in the interest of national defense or foreign policy and (B) are in fact properly classified pursuant to such Executive order;".⁽⁹⁵⁾

State Department: Redac and Inform Systems -- (a)(3) & (a)(2)

While other agencies explore digital redaction and disclosure, the only full-scale onscreen redaction system presently in operation is the 1990 REDAC system for declassified records at the State Department. This system permits redacting (blacking- or whiting-out) files onscreen before printing a purged copy for disclosure. It retains copies of the original, with and without the overlays created by the review staff. Created with graphical software on personal computer workstations, the overlays comprise polygonal covers on redacted material, with annotations for the rationale behind such excisions. Reviewers forward redacted records to the Senior Reviewer for quality control checks. Three types of printout are used: the original, the requester (disclosable) version with opaque covers and selected annotations, and the (internal) "Justice" version which highlights excised areas and includes all annotations.

The system has 60 workstations and is growing rapidly. At reading room stations, FOIA requesters can access disclosable records. One mainframe serves files, two cover the relational databases, one merges the document image with the overlays to produce excisions, and one prints the results. A communications server array provides network gateways and secure fax, an increasingly useful capability where reviewing tasks are shared.⁽⁹⁶⁾

A 1980 system, INFORM, tracks the progress of FOIA requests. Via diskette, a subsystem can transfer tracking data from presidential libraries to State, where it is merged with the main database. Statistics on activity and backlogs in the FOIA and Privacy Act programs are prepared by INFORM.⁽⁹⁷⁾

Social Security's Correspondence System -- (a)(3)

The Social Security Administration's (SSA) need for online FOIA processing is limited, despite the mainframe computers that the agency operates, by the low-tech nature of incoming requests. Many requests arrive on postcards or colored stationery that resist scanning, and typically they ask for data on one individual, such as a former husband's location. SSA improved its internal document processing in 1993-1994 with its OSCAR (Optical Scanning, Correspondence Analysis and Review) system for incoming and outgoing letters. The systemsupplies 17 FOIA workstations and 100 other correspondence staff with an e-mail routing program that sends scanned requests and analysts' responses to senior reviewing staff for editing and quality control, and then to printing. SSA has the usual FOIA control and tracking system using Oracle database software, as well as a method of querying the mainframes when requesters lack a needed social security number. The agency is considering the establishment of a network of shopping-mall kiosks where individuals might extract non-private information directly, but data security and budget considerations delay a final decision.⁽⁹⁸⁾

Justice Department: Juris Database -- (a)(3)

The Justice Department has had unfortunate experiences with its contracted database, JURIS. Created and maintained by a contractor, West Publishing, the database parallels in function the company's main commercial online database, Westlaw. Arguing that Justice could better purchase access to Westlaw, the contractor declined to extend the contract, obligating Justice to delete the database. In 1994, Tax Analysts, an open-government interest group founded by professors Susan and Phillip Long, sued for access to the database as an agency record. The Department of Justice denied the request on the grounds of the FOIA's (b)(4) exemption for confidential commercial information. West Publishing has intervened in the suit to protect its database. The case is pending.⁽⁹⁹⁾ Generally, the courts (following such precedents as Ciba-Geigy Corporation) have held that contractors' data are agency records under the FOIA insofar as they are owned by, controlled by, in daily use by or paid for by the government. Clearly the contractor's right to protect its intellectual property needs to be carefully traded-off for public access rights to official data. Where an agency rents access to a private contractor's database instead of holding its own, it should specify in the contract the conditions for low-cost public access under the FOIA.

Information-based services are gaining strength in the advanced economies. Competitively, the United States has economic advantages to gain in promulgating societal use of official data. Further, the Clinton administration has argued for the construction of an information superhighway to extend the existing Internet. The judicial and political environments, although not active, remain favorable -- if not for an electronic FOIA itself -- at least for the administrative strategy of automating official record disclosure. The principle danger from the current political environment is that enthusiasm both for contracting out and for raising revenue through user fees, will overwhelm the desire to ensure that people of limited means retain inexpensive access to official records. The two desires are mutually compatible if automated systems are properly planned.

A few systems now in operation suggest the nature of the electronic future. Digital handling of document input, transmission, manipulation, storage, and retrieval may substantially improve use of limited FOIA staff time as well as aiding in reducing backlogs.

In designing online publication systems, public servants and contractors would do well to consider carefully the interests of multiple publics. The needs of journalists, academics, interest groups, and citizens may well be satisfied with agency reading room terminals, online public libraries, and dial-up services to databanks at nominal cost. Commercial enterprises will seek fast online access and some the right to market value-added data. Contractors will seek a stronger role in developing and maintaining public databanks. Agencies will demand secure physical or logical "firewalls" between internal and externally accessible databases. The courts will find themselves adjudicating an increasing number of disputes between these interests. Automation systems will continue to expand in capability and decline in cost. Among the likely resultants of these forces will be multi-tiered access, multiple tariffs, and data dissemination in a rich variety of formats. Contractors doubtless will be expected to help fill the gap between the trudging pace of Federal procurement and galloping pace of technological progress. But, guided by the purpose of the FOIA to establish disclosure of agency records to "any person" as an end in itself, contractors like agencies also must be required to respect the rights of individuals to access official data at low cost. Lest policy be driven more by commercial capability than by FOIA principles, the time has come for FOIA staff directors to engage those to whom they report in establishing serious planning initiatives for the development of electronic FOIA programs.
 
 

1. Direct correspondence to Jeremy R. T. Lewis, Department of History and Political Science, Huntingdon College, Montgomery AL 36106. Lewis wishes gratefully to acknowledge Fritz H. Grupe for material on the state FOIA trends for this article.

2. For example, the Department of Justice Annual FOIA Report to Congress for CY 1993 (Washington DC, 1994) indicates 22,660 unprocessed requests on hand at December 31, 1993, of which nearly half came from the Immigration and Naturalization Service (INS) alone. Some other entities have similar levels of backlogs.

3. 5 USC § 552.

4. The 1966 FOIA is found at Pub. L. 89-487, 554; 80 Stat. 383, 80 Stat. 250; 5 USC § 1002 (1964); and 5 USC § 552 (1970). 80 Stat. 250 was recodified within two months to 80 Stat. 383; then in 1967 recodified by 81 Stat. 54 to 5 USC § 552. See Harold C. Relyea, "Faithful Execution of the FOI Act: A Legislative Branch Perspective," Public Administration Review, 39 (July-August 1979):328-332 and his "Access to Government Information in the Information Age," Public Administration Review, 46 (November-December 1986):635-639; and David M. O'Brien, "Freedom of Information, Privacy, and Information Control: A Contemporary Administrative Dilemma," Public Administration Review, 39 (July-August 1979):323-328. The 1974 FOIA amendments are found at Pub. L. 93-502; 88 Stat. 1561. The 1976 FOIA amendments are found at 90 Stat 1241, which begins with the recodification of 1966. The 1986 FOIA Reform Act is found at Pub. L. 99-570 and 100 Stat. 3207.

5. A less-used section (c) dating from the 1986 Act allows some national security agencies to exclude documents from FOIA responses, disguising whether those documents exist. This practice, known as "Glomarization" after the Glomar Explorer case involving a CIA covert operation to salvage a Soviet submarine, has been of limited value since notification of its very use would indicate to a sophisticated requester the existence of an investigation or similar undercover operation.

6. Development of the Internet, especially the World Wide Web is proceeding so rapidly that magazine sources are preferable to books. For a fine topical symposium for the layperson of practical applications of World Wide Web servers for an office, see PC Magazine 14(9):195-244 (May 16, 1995).

7. The first cycle began in 1948 with agitation in the press, then developed with the creation of the FOIA subcommittee in 1955 and a series of public hearings. In 1958, it produced trivial FOIA legislation that removed the 1789 Housekeeping statute as a legal justification for keeping official records from the public. The removal had little effect because a stronger justification was found in section 3 of the 1946 Administrative Procedure Act (APA) (the centerpiece of postwar administrative law). APA section 3 allowed records to be withheld "for good cause found" and "in the public interest," a rationale for almost unlimited official secrecy and a loophole highly unwelcome to journalists and historians in a political culture otherwise characterized as one of "luxuriant publicity."⁽⁸⁾

8. On the 1950s reaction of the press to executive secrecy in the context of luxuriant publicity, see Edward Shils, The Torment of Secrecy (Glencoe, IL: Free Press, 1956)and Francis E. Rourke, Secrecy and Publicity: Dilemmas of Democracy (Baltimore: Johns Hopkins University Press, 1961). -- -- ' -

9. Understanding of A-130 has benefitted from a 1989 interview with its principal author, J. Timothy Sprehe, OMB-OIRA.

10. Not all perceptions of Circular A-130 have been positive. A perceptive if anonymous reviewer of the draft of this article preferred to substitute for "landmark of policy" the term "quagmire."

11. The reinventing government movement is most familiar in the arguments of management "guru" Tom Peters, David Osborne and Ted Gaebler's Reinventing Government: How the Entrepreneurial Spirit is Transforming the Public Sector from Schoolhouse to Statehouse, City Hall to the Pentagon (Reading, MA: Addison-Wesley, 1992), and Vice President Gore's National Performance Review (Washington DC: Vice President's Office, 1993). While the practical benefits are often greatly exaggerated, the more academic heritage of the movement lies in Peter Drucker, The Age of Discontinuity: Guidelines to Our Changing Society (New York: Harper & Row, 1968).

12. U.S Congress, Electronic Collection and Dissemination of Information by Federal Agencies, H.Rept 99-560, 99th Cong., 2nd Sess., 1986. [Hereinafter H.Rept 99-560 (1986).]

13. H.Rept. 99-560 (1986), note 1, p. 2.

14. H. Rept. 99-560 (1986), p. 2.

15. Ibid.

16. U.S. Congress, Electronic Collection and Dissemination of Information by Federal Agencies, 99th Cong., 1st Sess., (1985) [hereinafter H. Hearings (1985)]; and H. Rept. 99-560 (1986), p. 2.

17. Office of Technology Assessment, Informing the Nation: Federal Information Dissemination in an Electronic Age: Summary. Document OTA-CIT-397 (1988) p. 1; Fred B. Wood, "Technology, Public Policy, and the Changing Nature of Federal Information Dissemination: Overview of a New Office of Technology Assessment Study," Government Information Quarterly, 4 (1987):83-96.

18. Interview with Patti Goldman, counsel to Public Citizen, 1988.

19. In 1980-1986 the Senate subcommittee (of Judiciary) on the Constitution had been a hotbed of conservative activism to restrict freedom of information. This subcommittee was chaired by Sen. Orrin Hatch (R-Utah) while that on Technology and the Law was chaired by Sen. Patrick Leahy (D-VT). On the work of these two opposing subcommittees in the 1980s, the author interviewed Randall Rader, counsel to Sen. Hatch, and the following counsels to Sen. Leahy: Ann Harkins, Marc Rotenberg, and (by telephone in 1994) Beryl Howell. A number of other interviewees such as Alan Adler of the ACLU and Patti Goldman of Public Citizen FOI Clearinghouse, and the staff of OIRA at OMB and of the House FOI subcommittee also contributed their interpretations.

20. Documents from the Senate Subcommittee on Technology and the Law. See also Tim Minahan, "Senators Say Electronic FOIA Bill Will Cut Processing Delay," Government Computer News, (February 1994).

21. Given that Senator Leahy had proposed, without administration support, an additional bill that would extend the FOIA both to Presidential records and the Congress, the following came as an unwelcome coda:

Mr. Schlesinger: Well, the President has said that he will veto any future legislation that extends special treatment to the Congress and changes to FOIA that increase the obligations of the executive branch and don't apply FOIA and the new obligations to the Congress ----

Sen. Leahy: Well ----

Schlesinger: [continuing] Would constitute that kind of legislation.

22. The announcement was widely publicized in the press and on the Internet. For the Vice President's reinventing government initiative generally, see World Wide Web at www.sunsite@unc.edu. The sunsite home page holds extensive documents from and about the Vice President's National Performance Review.

23. Responses are summarized in Access Reports: Freedom of Information (May 11, 1994), p 10.

24. Access Reports: Freedom of Information, 20(4) (December 7, 1994):1-2

25. Revised A-130 was published in the Federal Register, 58(126)(July 2, 1993):36068-36086. It can also be downloaded by anonymous ftp from nis.nsf.net as /omb/omb.a130.rev2 or by sending a no-subject e-mail message to nis-info@nis.nsf.net with send omb.a130.rev2 as the first line. For those using the Commerce Department's Fedworld BBS it is available at (703) 321-8020 (with settings N-8-1, and log on as "new"). The original A-130, "Management of Federal Information Resources" Federal Register, 50 (December 24, 1985):52730 ff. OMB's authority over Federal records was assigned in the Paperwork Reduction Act (44 USC 35) of 1980.

26. The author is grateful to Beryl Howell, counsel to the subcommittee on Technology and the Law, for a telephone interview and for providing the text of and documents relating to the electronic FOIA bills, (August 1994). Leahy had long maintained an interest in the issues raised by changing information technology. At a 1994 conference he argued, "If we don't update it in the way we get information and keep information, then the FOIA is useless to us. But we can't just get out of the way of this thundering train, because it can run roughshod over personal privacy." See Dianna Hunt, "Computer Inaccess a Barrier to Open Government: Emerging Technology Stirs Privacy Debate," Houston Chronicle, (Jan 27, 1994) via Lexis/Nexis.

27. This paragraph has benefitted from telephone interviews with participants, coverage of the events by Access Reports, and the comment of an anonymous reviewer on an earlier draft of this paper.

28. At present, limited largely to recent press releases and selected member biographies, this server is expected to hold greatly increased and enriched content by the end of 1995. The gopher is already linked by menu to the White House World Wide Web home page, the Fedworld gateway, and the Library of Congress gateway. It is also commonly listed on gophers at academic sites such as the popular University of Minneapolis node.

29. See for example, Kenneth Pins, "Gingrich Says He's Taking the Information Superhighway," Des Moines Register (December 27, 1994), p.1. Gingrich's e-mail address is georgia06@hr.house.gov, while the Contract with America is listed under the hierarchical menu system at gopher.house.gov. Lewis's exploration of the House resources on the Internet in December 1994, however, found minimal information. Few members had so far participated in the scheme. While the new leadership appreciates the fact that the Prodigy service found more Republicans than Democrats among (relatively educated) Internet users, they also fear "astroturf lobbying" (false, orchestrated grass-roots lobbying) will be greatly facilitated by Internet access. Lewis's perusal of the Senate Gopher server turned up similarly limited resources. One may relay comments to guide@scc.senate.gov. The most useful resource on the Hill is undoubtedly the Library of Congress Gopher server at marvel.loc.gov and its legislative server at locis.loc.gov which can be reached by telnet.

30. Eric Felten, The Ruling Class: Inside the Imperial Congress, (Washington DC: the Heritage Foundation, 1993) p.93.

31. The most definitive rulings in the district courts arise in the District of Columbia (DC), where the FOIA explicitly allows venue. DC judges have developed expertise in FOIA matters. From 1974 to 1984 (when a minor clause was repealed, having served its purpose) FOIA cases had to be expedited above all others. Thus judges became familiar with its disclosure purpose. However, a spate of appointments by the Reagan administration counterbalanced the liberal consensus in the District with conservatives, and it became in the mid-1980s less predictable for public interest groups that the liberal thrust of the Act would be upheld. 1988 interviews (among others) with Kathy Meyer and Patti Goldman, counsel to Public Citizen and frequent litigators in FOIA cases.

32. US Department of Justice, FOIA Case List 1993 Edition, (Washington DC: US GPO, 1993) [hereinafter cited as Justice Case List]. Allan Adler, ed., Litigation under the Federal Open Government Laws, (Washington DC: ACLU, 1993) [hereinafter cited as ACLU, Litigation.]

33. The Reporter's Committee approach could by extension potentially undercut FOIA processing of business data and business requests. The Court's misinterpretation could dovetail with the Circular A-130 emphasis on marketing valuable data to companies rather than merely giving it via the FOIA to citizen's organizations. Also, much data requested by companies tells more about the work of other companies than about the conduct of government, and hence such requests are considered by the FOIA lobby -- including Thomas Susman, the architect of the 1974 amendments -- as a by-product of more politically significant requests. Interviews with Thomas Susman (Senate counsel) 1979 and subsequently, Robert Gellman (House counsel) 1979 and subsequently, and counsel (Patti Goldman 1988, Kathy Meyer 1979 and subsequently, Eric Glitzenstein 1979 and subsequently, Alan Adler 1988, and others) for the ACLU and Public Citizen.

34. Long v IRS, 596 F.2d 362, 365 (9th Cir. 1979) (Kennedy, J.), cert denied, 446 U.S. 917 (1980).

35. Harold C. Relyea, Evolution of the FOIA Amendments of 1974 (PL 93-502), Congressional Research Service, Library of Congress. Document JC 660 76-172G [hereinafter Relyea, Evolution]; ACLU, Litigation; Justice Case List.

36. Yeager v Drug Enforcement Administration, 678 F.2d 315, 321 (D.C. Circuit, 1982). See also ACLU, Litigation; Justice Case List.

37. Dismukes v Dept. of the Interior, 603 F.Supp. 760 (DC District, 1984.) See also ACLU, Litigation.

38. Interview with Sheryl Walter, counsel, National Security Archive (1993). To indicate the scale of the Archive's operations and significance for the intelligence community, the NSA has published 35,857 documents averaging 5.15 pages, according to Glenn Cooley et al., Interagency Declassification Support System: Estimated Costs and Implementation Considerations, Mitre Report 94W0000101 (McLean, VA: Mitre Corp, October 1994), p. A-37.

39. Armstrong v Bush, 721 F. Supp. 343 (D.D.C. 1989), affirmed in part, reversed in part, 924 F.2d 282 (D.D.C. 1993.) Armstrong v EOP, 810 F. Supp. 335 (D.D.C. 1993.) North's experience doubtless urged President Clinton to claim the NSC a Presidential advisory body, not an agency under the FOIA as the Supreme Court, and seemingly the NSC itself, had assumed. Simultaneously in March 1994, he ordered creation of an alternative disclosure process. North's thousands of often casual messages were obtained from the National Security Council Staff's PROFS system through a combination of FOIA requests to the NSC as an agency (within the Executive Office of the President) and congressional and criminal investigation. Indeed, the messages reinforced his claim to have informed his superiors at inordinate length of his activities in the Iran-contra affair. Recall that in the Iranian arms sales and the support of the contras, the NSC staff took an executive role, not merely an advisory role. In fact, it was used largely because it might evade the prohibition that the Boland amendments laid upon "agencies or entities involved in intelligence activities." In August 1993, the US Court of Appeals for the DC Circuit issued an opinion that the NSC had routinely conceded its status as a FOIA agency, and the Supreme Court appeared to have assumed this status without considering it. See Jim McGee, "Clinton Tries to Limit Access to NSC Data," Washington Post, (March 26, 1994). On March 24, 1994 Clinton directed a new disclosure process in NSC to mirror the FOIA, given that NSC would be considered purely advisory and not an agency. Such a process would lack the statutory safeguards of the FOIA, and Tom Blanton of the National Security Archive derided the proposal as a "trust-me FOIA." Judge Richey of the DC federal district court ruled in February 1995 that the NSC did indeed have agency status under the FOIA. NSC had a statutory staff, structure and budget and it exercised substantial independent authority in rulemaking and adjudication. Richey found the government's characterization of the NSC as a non-agency, was arbitrary and capricious: unsupported by a reasoned explanation. Scott Armstrong v Executive Office of the President Civil Action No. 89-142 (CRR), U.S. District Court for the District of Columbia, (February 14, 1995). For commentary, see "Court Rules NSC Agency Subject to FOIA," Access Reports 21(4) (February 15, 1995):1-3.

40. Yeager v DEA, 678 F. 2d 315. (D.C. Circuit, 1982)

41. Gottlieb, Steen and Hamilton v Dept of Health and Human Services, No. 92-2192. U.S. District Court for the District of Columbia, (Nov 26, 1993). See also case discussion in Privacy Times, (February 23, 1994), p. 4.

42. Gottlieb, Steen and Hamilton v Dept of Health and Human Services, No. 92-2192. U.S. District Court for the District of Columbia, (November 26, 1993). Forsham v Harris, 445 US 169 (1980). See also case discussion in Privacy Times, (February 23, 1994), p. 4.

43. Thompson Publishing Group v Health Care Financing Administration, No. 92-2431, U.S. District Court for the District of Columbia (March 15, 1994).

44. Brian Miller, "Profits in Government," Government Technology, 2(2)(February 1994):1, 42.

45. Data Resources Unlimited, An Innovative Approach to Information Networking of the State of Nevada's Public Information (Las Vegas, January 1993).

46. Nora Sherwood Bryan, "Local Government Data Pricing and Distribution Strategies Vary," GIS World, Special Issue (July 1992):32-38.

47. Dianna Hunt, "Computer Inaccess: a Barrier to open Government," Houston Chronicle, (December 12, 1993) via Lexis/Nexis. Part of a series.

48. For numerous references to Florida's reinvention of government programs, see Osborne and Gaebler, Reinventing Government, pp. 1, 17-19, 30, 38-9, 69, 109, 112, 116, 121, 128, 135, 141, 154-5, 163, 177, 202-3, 210, 211, 213-14, 222, 224, 232, 234, 236, 242, 248, 262, 264, 293-4, 297, 303, 321, 335, 343, 359.

49. Lori Peterson Dando, "A Survey of State Open Records Laws In Relation to Recovery of Data Base Development Costs," Proceedings of the Urban and Regional Information Systems Association IV (1992): 85-110.

50. Patricia Samuelson, "Is Information Property?," Communications of the ACM 34(3) (March 1991):21-26.

51. Office of Information and Technology, Connecticut, Public Act 91-347: Computer Stored Public Records. (April 13, 1992)

52. Bruce Westcott, "The Vermont Center for Geographic Information, Inc.: A Non-Profit to Support GIS Applications and Education or a Wolf in Sheep's Clothing?," Proceedings of the Urban and Regional Information Systems Association IV (1992):133-144.

53. The IIA now numbers 650 members in the official data brokerage business, broadly defined.

54. Henry H. Perritt, "Electronic Acquisition and Release of Federal Agency Information," Administrative Law Review, (Summer 1989):253-314.

55. Deborah C. Friedman, Confidentiality of Public Records, (Augusta, ME: Maine Office of Policy and Legal Analysis, December 1992).

56. George B. Trubow, "Personal Privacy and the Second Use Dilemma," IEEE Software, 9(4) (July 1992):73-73. George B. Trubow, "Protecting Privacy in the Information Age," IEEE Software, 9(3) (May 1992):106-107.

57. Pamela Samuelson, "Liability for Defective Electronic Information," Communications of the ACM 36(1) (January 1993).

58. Lee Fisher, An Ohio Sunshine Laws Update (Columbus, OH: Chief Counsel's Staff, undated).

59. U.S. Congress. Senate. Judiciary Committee. Subcommittee on Technology and the Law. The Electronic Freedom of Information Improvement Act: Hearings on S.1940. S. Hearing 102-1098 1992 (Washington DC: GPO, 1993).

60. National Association of Government Archives and Records Administrators, "New Approaches to Management of Electronic Information Systems," Bulletin of the American Society for Information Science (October/November 1991):17-18.

61. Office of Information and Technology, Public Act 91-347: Computer Stored Public Records.

62. See Robert Pear, "U.S. is Said to Waste Billions on Computers," New York Times, (October 12, 1994). Those with truly outmoded technology included the Internal Revenue Service, Federal Aviation Administration, Social Security Administration, National Weather Service and Immigration and Naturalization Service. Among those with incompatibility problems it cited the Agency for International Development and the Agriculture Department. A special case of incompatibility it found at the National Institutes of Health where researchers favoring microcomputers, shun its new mainframes.

63. Repeated interviews since 1984 with Gerry Deighton, Director, FOI Staff, US FDA.

64. Steven Frantzich, "Technological Innovation in Government: The Impossible Dream?" Paper presented at the annual conference of the Academy of Business, Wembley, London, June 1994.

65. A common example of a mainframe database requiring programming is Oracle; the common equivalent on a micro was dBase III. The modern graphical database which is programmable by the end user is exemplified by Access, Approach, Paradox Windows, and Filemaker Pro. Indeed, some microcomputer databases now allow query by drag-and-drop (using a mouse cursor to connect fields and arithmetical operators), and voice-activated query (manipulating a database using English-like commands without typing) will soon be implemented. Note that the new feature that transforms the situation is the user-friendly software and inexpensive database manipulable by clerical staff and analysts.

66. There are powerful microprocessors such as the Intel i486, Intel Pentium, DEC Alpha and the Motorola/Apple/IBM consortium's PowerPC.

67. The State department was pressed to improve its FOIA processing by Senator Leahy (D-VT) during the confirmation hearings for Secretary James Baker in 1992.

68. Glenn P. Cooley et al. Interagency Declassification Support System: Functional Requirements and Alternative Architectures, MTR 94W0000062 (McLean, VA: The Mitre Corporation, May 1994).

69. Interviews by telephone with Les Weinstein, and comments of Fred Sadler, FDA Center for Devices (Jan 13, 1995).

70. In the case of submission of material to an agency by an outside party, it may be possible for the source to annotate documents with proprietary claims before submission, but strictly limited to the guidelines set by the agency. In the past, submitters often expanded their claims to proprietary data far beyond the realm of the possible, necessitating official review of requested documents. Thus, training submitters may be more cost-effective than conducting in-house line-by-line review of documents.

71. Cooley et al. Interagency Declassification Support System.) Interviews with Rex Klopfenstein of the Mitre staff.

72. President Carter's Executive Order 12065 of December 1, 1978, proposed limited classification periods conforming with the 20 year classification limit of the 1978 Presidential Records Act (92 Stat 915), but subject to exemption for records of permanent value that could be classified for 30 years. President Reagan's Executive Order 12356 of April 2, 1982 extended classification coverage to material that could "reasonably" be covered by extant standards -- thus affording broad discretion to officials. The Reagan administration's efforts to reverse the trend of Carter's restriction on classification also took other forms. National Security Decision Directive 84 of March 1983 proposed that some 120,000 officials handling Sensitive Compartmented Information above Top Secret should sign nondisclosure agreements and be subject to polygraph (lie-detector) testing. Owing to opposition from Secretary of State Schultz, among others, NSDD 84 was not fully implemented. National Security Adviser Poindexter also on October 29, 1986 signed a directive NTISSP No.2 establishing the category of sensitive but unclassified material, thereby extending the system to more innocuous material below Confidential. It was rescinded in March 1987, but material sensitive when viewed as a mosaic of security information remained a category protected by NSDD 145 of September 17, 1984. Despite some backtracking on these and on computer security matters, the Reagan administration overall enjoyed substantial success in tightening national security secrecy. See Peter Hernon, "Protected Government Information" in United States Government Information Policies: Views and Perspectives, edited by Charles McClure, Peter Hernon and Harold C. Relyea, (Norwood, NJ: Ablex, 1989). Also, Richard Ehlke and Harold Relyea, "The Reagan Administration Order on Security Classification: A Critical Assessment," Federal Bar News and Journal 30:91-97 (1983).

73. Cooley et al. Interagency Declassification Support System. p. v.

74. Using solid state technology on a large scale, government agencies are expected to adopt racks of 12" optical WORM (write-once-read-many) drives to serve copies of scanned classified documents from a chain of databases to reviewing staff. Scanning is expected to be used even where data exist in digital form, since there are advantages to redacting graphical images so as to leave no trace of original (classified) pixels.

75. Department of Defense, Directorate for Freedom of Information and Security Review, Office of Assistant to the Secretary of Defense (Public Affairs), Freedom of Information Act Program CY 1993: Report to Congress (March 1994).

76. The NARA differs from other agencies: it makes decisions on the whole document rather than the FOIA's segregable portions; second, it receives but does not originate classified records, and has a lesser interest than do other agencies with "equity" in the documents in protecting them. This appears to render NARA's call for a community-wide declassification system improbable.

77. For practical purposes it was facilitated by the proximity of the Register offices to the GPO.

78. SGML was developed from a markup language written by IBM for the Department of Defense, with modern extensions for hypermedia.

79. Carol Mahoney, Technical Information Specialist, Office of the Federal Register, National Archives and Records Administration, interviewed by telephone (August 19, 1994); William Thompson, Program Analyst, Electronic Information Disclosure Service, USGPO; interviewed by telephone (August 19, 1994); literature transmitted by USGPO, including the following (1994): "The Federal Bulletin Board;" "Why You Should Use GPO's Customized EINet WAIS Client Software;" and "List of File Libraries on the Federal Bulletin Board as of July 28, 1994."

80. Dial in to (202) 512-1661 for WAIS server. EINet (for DOS, Windows and Macintosh) software, required for Internet access, is available from GPO for $15. It was selected above competitors for its ability to retrieve all 150 graphics files linked to a test document. It has since been enhanced to retrieve 200 with a single document, and will soon be enhanced for modem access in addition to the Internet channel. The graphics are in compressed TIFF format.

81. Carol Mahoney, interviewed by telephone. The Office of the Federal Register does expect to develop a long-range plan to have the Register fully coded in SGML, and probably to contract for custom software for marking the text. This would for example permit searching in only the document number field for a number, instead of across the entire text database. The GPO plans to include enrolled bills in addition to those passed for presidential signature, and has received many requests to put the Code of Federal Regulations and Commerce Business Daily online. It also plans a locator service for 12,000 other government publications at other agencies. Depository libraries across the country have been offered gateways and hookups to the system -- free of charge provided they make it available to the public. Progress will not be speedy, for Congress did not rescind the planned further cut of 200 GPO staff in 1995.

82. 42 USC § 275-276 (1982). See also H.Rept 99-560 (1986).

83. Food and Drug Administration, Announcing FDA on Disc: FDA Documents and Manuals. Leaflet. FDA Parklawn Computer Center, Rockville MD 20857. Orders: HRSA-DFS, Collection Officer. Presently priced at $300 per quarter, with an expanding set of documents. FDA's Bulletin Board contract service is supplied via the FDA's Press Office.

84. H. Rept 99-560 (1986), p.61.

85. H. Rept 99-560 (1986), p.64.

86. SDC v. Matthews 542 F.2d 1116 (9th Cir. 1976) in which the Court of Appeals upheld the agency's withholding of tapes; for commentary see H. Rept. 99-560 (1986) p.32-36. "The committee concludes that the decision of the court in SDC v. Matthews is incorrect as a matter of law and as a matter of policy." H. Rept. 99-560 (1986) p. 35. SDC v. Matthews has generally not provided an influential precedent; Dismukes v Interior (supra) has provided a narrower and more acceptable precedent for withholding, finding that computer tapes although agency records could be protected since release of microfiche fulfilled the FOIA's requirements.

87. Office of Management and Budget, User Charges, Circular A-25 (September 23, 1959). H. Rept. 99-560 (1986) p.48.

88. As of August 1994, 3,500 companies were filing online, with cohorts of the remainder being phased in. Companies submit 10-K forms seven or eight times annually and 8-K reports before annual shareholders' meetings.

89. David Copenhafer, Director, Office of Planning, Administration and Security, SEC, interviewed by telephone (August 12 and 17, 1994). The contract is now spread over eight years with an overall cost of $90 million of which SEC funded the $72 million capital cost (i.e. for the establishment of the system.) The balance -- representing the operating cost -- is funded by the private sector, which covers the contractor's dissemination through user fees. This represents a significant increase over the 1985 estimates given to Congress of $50 million over five years and the first arrangement for $70 million over seven years. See testimony of John S.R. Shad, Chair of SEC 1985, H. Hearings (1985) p.20.

90. Uploading requires EDGARlink software developed by a contractor, BDM. A hardcopy filer manual is available for the price of a paperback book. SEC EDGAR Filer Manual, $20 from Disclosure Inc at (800) 638-8241. Long uploadings by tape include those by mutual fund companies filing perhaps eight hundred prospectuses at a time.

91. Complete online access is currently (October 1994) $138,000; one-day delayed tape service is $78,000; and eight topical subsets ("slices") of the database are available for $38,000. The hardware consists of five interlinked Stratus machines running the VOS operating system.

92. Public-interest groups of course themselves have an interest in obtaining data at subsidized cost. But the SEC staff finds it difficult to imagine a giant database of commercially valuable information being handled otherwise: the scale, cost and clientele needs dictate an EDGAR-like system.

93. See H.Rept 99-560 (1986) p.3. See also H. Hearings, (1985), Testimony of SEC Deputy Executive Director Kenneth Fogash, p. 55. The development of the system was a cautionary tale for further systems under official control. The winning bid (by BDM) for establishment of the full-scale system was $20 million lower than SEC's authorization of $72 million from the GSA. (The $72 million authorization itself was increased by GSA to cover a seven year project, up from five years and $50 million in the 1984 estimates found in the 1985 House hearings and 1986 House Report.) An Andersen bid was deemed more realistic in view of the company's experience with the pilot scheme, but after scaling up the BDM bid to match an appropriate size of the project, the SEC was obliged to accept BDM.

94. Interviews annually since 1984 with Gerry Deighton, Director, FOIA Staff, USFDA.

95. 88 Stat. 1563 (FOIA amendments of November 1974.)

96. Internal documents and overlays are stored on a 50-platter juke box with 12" optical disks giving a capacity of about 3.2 gigabytes. This discussion of the REDAC system at the State department has benefitted from a telephone interview (August 19, 1994) with -- and a number of documents provided by -- Jacqueline Lilly, Chief of the Technology Applications Branch, State Department. Also from discussion at the Annual Symposium August/September 1993 of the American Society for Access Professionals, particularly comments by Frank Machak (FOI staff, State Department) and officers of Severn Corporation, the contractor. State Department, Technology Applications Branch, "FOIA Imaging System Hardware Description and Function," (Undated document). Schematic diagrams, explanations of the system, and untitled documents, were provided by the Technology Applications Branch of the State Department. Department of State, REDAC System Overview, 1992.

97. Expanding by 5,500 cases and 150,000 document entries per year, the system accepts regular downloadings from the main foreign policy database, ADS, which itself is being replaced by a new system, OASYS. There are plans to integrate the REDAC system with the older tracking system, thus automating an important element of the process. Since the department has numerous cables in digital form, it hopes to download them directly into the optical-drive jukebox, eliminating printing and scanning hardcopy. Eventually it hopes to link both microfilm readers and offices where searches are commonly made. The current system comprises a CCA model 204 DBMS application running on a classified IBM 3081 mainframe (the RED system) with output on IBM 3262 Tempest (secure) printers. The workstations are WANG 386 personal computers and IBM 3290 remote terminals. The original contractor was TAB Associates, while maintenance is performed by CBMI Inc.

98. The correspondence system comprises personal computers running WordPerfect for Windows with Enterprize which is a combination of database query and Staffware, a British windows workgroup routing application. Telephone interview with Priscilla Rieger, FOIA Staff, Social Security Administration, (January 13, 1995).

99. American Society of Access Professionals, "Waystations Along the Electronic Highway," ASAP Newsletter (May/June 1994).