Abstract

Interoperability is more than "digital plumbing"-making sure that computers talk so that bits of data flow properly. Fundamentally, interoperability is people talking and sharing information. Sharing information reduces the "paperwork burden" on the citizen, streamlines work processes, and enriches the formulation, implementation, and evaluation of policy. Building on prior theory and research, this research has developed empirically derived, practical findings and recommendations to support the development of appropriate interoperable systems. Concerned parents noticed an abnormally high incidence of leukemia among River Valley High School graduates, in Marion, Ohio. After investigating their hunches, they learned that leukemia rates had increased 122 percent between 1966 and 1995. The parents then learned that the high school was built near a World War II bomb-making plant that had leaked cancer-causing chemicals. Relating disease clusters to environmental factors occurred through personal recollections of friends and family of the cancer victims. It took almost 30 years for the community to accumulate the evidence to tie their hunch to a suspected cause (adapted from Interoperability among epidemiological, Department of Defense, and Census Bureau information systems could have facilitated making these correlations faster and easier, perhaps saving lives. While technology offers great promise for improving government, it is also true that every important new technology invariably poses a challenge to the status quo and David Landsbergen Jr. is an associate professor in the School of Public Policy and Management at Ohio State University. He teaches, conducts research, and consults on the legal and policy issues surrounding the introduction of new information technology. He holds a J.D. and Ph.D. from the Maxwell School at Syracuse University. Email: landsbergen.1@osu.edu. George Wolken Jr. is a research fellow at the Ohio Supercomputer Center and an attorney specializing in intellectual property law with a professional focus on the creation, development, management, commercialization, and litigation involving advanced computing and optical and wireless communications. He holds a Ph.D. in chemical physics from Harvard University. His physics research has involved the use of advanced computer and software technologies to model chemical reactions. He has written more than 40 journal publications related to software and computer technologies. Email: gwolken@skjerven.com. puts a strain on existing social contracts on "who gets what." In the River Valley case, collecting information about cancer-causing agents would have made it much easier to anticipate potential problems. But is this something that is central to the mission of the Department of Defense? How does the Environmental Protection Agency (EPA) make the case to the Department of Defense that it should provide this information to the EPA? This article argues that interoperability is more than getting bits and bytes to flow properly-fundamentally, the goal of interoperability is the much more difficult problem of getting people and organizations to share information in an information-technology environment. To really understand the possibilities, it is important to see the big picture of the opportunities for improving government through the appropriate use of technology. In the Government Information Systems and the Fourth Generation of Information Technology 207 span of some 50 years, our computing technology has already gone through three significantly different generations of computers The third generation of computers-personal computers-increased the ability to collect and store information on each person's desktop. Yet with this exponential growth in the power of personal computers, the fourth generation of computers is even more impressive because it leverages the power of each of these powerful desktop computers. The leverage is gained through the networking of computers. Theoretically, networking makes it possible for the information and processing power on all of these desktop, mainframe, and super computers to become available on each individual's desktop. In practice, the problem of "interoperability" is how to make this promise a reality. This article begins with two illustrations of attempts to improve interoperability. The literature review then builds a comprehensive list of the major threats and opportunities to interoperability, including prior work which has discussed the problem or a proposed solution. 2 The literature review examines the many dimensions of the interoperability question, including previous work on interoperability as well as relevant work on information-technology management, intergovernmental relations, and innovation, information, and information technology policy. After the literature review, the methodology section discusses several phases in the research design. A key element in obtaining rigorous results was the use of a "tight" learning loop-every new insight or practical recommendation learned was immediately included in the set of working findings and recommendations so that it was tested in the next cycle. For example, where interviews were used, each person was asked, do you agree with our working set of findings and solutions? Is there anything else we should know? The use of this tight learning loop was necessary to ensure the validity and reliability of the findings and recommendations. The results section accumulates and integrates what was learned from the case studies, interviews, and policy-infrastructure analysis into four basic themes: • Interoperability is best understood as the sharing of information. • There are critical success factors in the development of interoperable systems. • There is no strong federal/state architecture in place. • In a technology-driven society, technical standards are another kind of public law. The conclusion then simplifies and organizes the findings of this research (see Benefits and Costs of Interoperability Two cases illustrate the range and magnitude of benefits that could be achieved through closer attention to interoperability. These cases were part of the general study of interoperability. For more case-study details, please see appendix A, and for a full description of the case studies, see the full report. Federal-State Cooperative Environmental Information Systems In keeping with the effort of reinventing government, the EPA has declared its policy to reduce industry burden and streamline regulatory programs by using data-transfer technologies to reinvent the process of industry compliance. The overall goal is to cut as much red tape as possible. One of the more interesting projects is the implementation of One Stop Facility Identification and Reporting. 4 Traditional media-based regulations identify polluting facilities by any means relevant for that particular media. The location of a discharge pipe, emitting stack, or the street address are all typical means for identifying a facility. This leads to confusion over coordinating policy and management. For example, it has been reported that a facility was commended for waste reduction by one program and simultaneously fined by another program, unknown to the respective programs that were dealing with the same facility. Government Information Locator Service People who surf the Web are often amazed at the wealth of information available. But for those who are looking for specific, relevant, and useful information, the experience is often disappointing. They often find something close to what they want, but not exactly. The abundance of information gets in the way because of the amount of time required to sift through all of the irrelevant material. As the Internet makes it easier to access information, the important tools and skills will shift from gaining access to information to knowing how to use this information. Meta-data is an important building block to achieving interoperability because it will make information easier to use by providing context to data so that it can be more easily shared across diverse information communities, and by providing pointers to that information so that it is easier to find. A distinction is often made between data and information. For example, "there are 1 million people unemployed" is a datum. For this datum to be informing (information), we need to know the context; for example, how much is that figure as a percentage of employment, or is unemployment rising or falling. Meta-data play a critical role in providing, or pointing to, important contextual information so that information can truly be shared and understood among users. Government Information Locator Services (GILS) is a technology that not only provides pointers to contextual information but also makes it easy to locate relevant public information. 5 Like the library card catalog we learned to use in elementary school, GILS collects important meta-data 6 about the information. Some meta-data questions include: • How was the information collected or computed? • Who is the author of this document? • Whom can I contact to gain more information? • How much will it cost to obtain? These instrumental questions go towards the more important questions: • Is there relevant information out there? • Is it quality information? • Will it be useful, and can I gain that information and its context? GILS was chosen as a case study so that we can begin to understand how meta-data systems support the construction of interoperable systems. What Are the Benefits of Interoperability? As fourth-generation networked technologies become more dominant, the benefits of interoperability cluster around three fundamental benefits. Effectiveness. While many federal agencies were created to respond to the particular crises of the day-antitrust, drugs, labor, education-new policy approaches are based on the understanding that most social and regulatory problems go beyond the jurisdiction of one agency or bureau and require integrated policy approaches. More effective government will require leveraging information (Kraemer and King 1986), knowledge, and technology across these agencies and their divisions to provide an integrated response to our more complex social problems. Efficiency. Once information is collected electronically, it becomes very easy to duplicate and manipulate. Thus, interoperable systems have the potential to reduce the paperwork burden on the public and private sectors. Interoperable systems will allow fourth-generation computer networking technologies to facilitate access to information by all stakeholders. These new systems reduce transaction costs, thus lowering costs and increasing participation Responsiveness. Better access to information will allow government to act faster and more effectively to identify problems and respond to them What Are the Barriers to Interoperability? 7 There are significant barriers to achieving interoperability on a broader scale. In some cases, these barriers are based on quite legitimate concerns about how we should realize our constitutional values. In other cases, the barriers are the result of a mismatch between an older generation's way of doing things in tension with the promise of fourth-generation networking of computers. We draw upon the work of Political Privacy. 8 The Orwellian threat of Big Brother government is always associated with a sophisticated governmentinformation system. Less well understood, but perhaps more threatening, are the lack of controls on the private sector (Federal Trade Commission 1998). Before an interoperable government can be achieved, privacy rights and their protection must be clarified. Beyond the difficult problem of actually putting appropriate systems in place, there needs to be much more-a genuine feeling of trust and control by citizens-before it becomes politically and operationally possible to have interoperable systems Ambiguity about Statutory Authority. Legally, agencies cannot act outside the power delegated to them under their authorizing statute. Derivatively, agencies cannot collect information outside the subject delegated to them Government Information Systems and the Fourth Generation of Information Technology 209 about federal agencies' statutory authority to collect and share information extend to federalstate interoperabilities and information sharing. While there is certainly ambiguity in any statute, the large amount of ambiguity, and even a lack of vision of "this information sharing thing" invites abuses by overreaching administrators or nonaction by timid bureaucrats. This is clearly an area where a model code of law needs to be developed to support lawmakers and administrators. Openness to Public Scrutiny. The development of interoperable systems will allow for heightened scrutiny. For many agencies, easy access to information raises the spectre of exacerbating the "goldfish bowl phenomenon" (Bozeman and Bretschneider 1986) through more public requests for information, increased micromanagement of the agency's decisions, and opportunity to attack an agency for motives other than bona fide concern for the performance of the agency's mission. Organizational Trust. Before agencies can share information, they must develop mutual trust Lack of Experience. Our field research indicates that interoperable systems are very much the exception. Consequently, an experience base and institutional memory that reflects the benefits and costs of interoperability has not yet developed. When faced with the unknown, agencies prefer to attend to problems where there are relatively wellknown ways of solving them and a relatively clear understanding of the trade-offs involved (Weiss 1987). Lack of Awareness of Opportunities to Share. Good information-resource management requires an inventory of those resources Economic Lack of Resources. Given the ever-increasing accountability of agency action combined with less resources, most agencies are hard-pressed to pursue interoperability where the benefits are ill-defined and the costs are unclear and uncertain. Why should an agency expend its own scarce resources to make information available for the benefit of another agency when there are much more pressing and concrete information system needs? (Strategic Computing and Communications Program 1998). "Low-Bid" Procurement Methods. "Life-cycle" or "best-value" procurement approaches are still the exception in purchasing information technology in the public sector (NASIRE and NASPO 1996). Even more of an exception is the consideration of interoperability from a governmentwide instead of an agency perspective. Often, interoperability costs more in the short run, posing difficulties for agencies oriented toward "low-bid" approaches (Rosen 2000). Technical Hardware or Software Incompatibility. Technical incompatibility between hardware and software will always be an important consideration to interoperability. Most basically understood, it is getting a Microsoft Windows program, for example, to read a Macintosh data set. On some level, most of these problems can always be solved, but the real question is how much cost will be involved Data-Sharing Standards. One of the most difficult and widespread barriers to interoperability are differing data definitions Summary While this listing of benefits and barriers provided a useful starting point, several questions still remained: 1) are all of these problems equally prevalent; 2) is there a simple way of organizing and understanding these problems; and most importantly 3) what are the leaders in interoperability now doing to manage their way through these concerns and opportunities? Method: Integrating Policy Documents, Interviews, Case Studies, and Prior Research 9 Research Phases The research employed several phases (see The third phase of the study developed an interoperability policy model. Although listed as the third phase, the development of both the de jure and de facto descriptive model of the federal law and policy governing interoperability was commenced soon after the literature review revealed which law would be important to research. In developing the findings and recommendations, both direct and indirect barriers 10 to interoperability were sought out by analyzing the relevant laws on interoperability; that is, the statutes, administrative rules, Office of Management and Budget circulars, and executive orders. A descriptive model of federal policy was then developed to comprehensively identify potential legal and policy barriers to interoperability. The fourth phase of the project involved the use of case studies. Our strategic choice was to eschew a broad-and necessarily shallow-survey, in favor of an in-depth probe of selected case studies. Our focus was less on what the average agency thought of interoperability and more on what the innovators and experts could explain and recommend. The case studies allowed the research to focus on particular state approaches to observe the microdynamics and interplay of state attitudes towards technology, politics, and policy. Two case studies 11 were chosen to study the most critical aspects of interoperability: 1) cross-agency regulatory enforcement where significant penalties could incur (environmental reporting) 12 and, hence, significant conflict over access to information could be expected; and 2) drawing upon the studies of scientific-and technical-information dissemination, innovation, and technology transfer, "For the purpose of this section, the term 'hazardous substance' includes radioactive waste and mixed radioactive and hazardous waste." "The term 'hazardous substance' means (A) any substance designated pursuant to section 31 1 (b)(2)(A) of the Federal Water Pollution Control Act (33 USCS Section 1321 (b)(2)(A), (B) any element, compound, mixture, solution, or substance designated pursuant to section 102 of this Act (42 USCS Section 9601), (C) any hazardous waste having the characteristics identified under or listed pursuant to section 3001 of the Solid Waste Disposal Act (42 USCS Section 6921) (but not including any waste the regulation of which under the Solid Waste Disposal Act (42 USCS Sections 6901 et seq.) has been suspended by Act of Congress), (D) any toxic pollutant listed under section 307(a) of the Federal Water Pollution Control Act (33 USCS Section 1317(a)), (E) any hazardous air pollutant listed under section 112 of the Clean Air Act (42 USCS Section 7412), and (F) any imminently hazardous chemical substance or mixture with respect to which the Administrator has taken action pursuant to section 7 of the Toxic Substance Control Act (1 5 USCS Section 2606). The term does not include petroleum, including crude oil or any fraction thereof which is not otherwise specifically listed or designated as a hazardous substance under subparagraphs (A) through (F) of this paragraph, and the term does not include natural gas, natural gas liquids, liquefied natural gas, or synthetic gas usable for fuel (or mixtures of natural gas and such synthetic gas)." USCS 7274d USCS 9601 Government Information Systems and the Fourth Generation of Information Technology 211 the researchers identified the use of meta-data (data about the data) as an interesting case study (GILS) on the problems and opportunities of building an information infrastructure for government. GILS was also interesting because it required interoperability to implement, but it also facilitated interoperability. A summary of the case studies can be found in appendix B, while the full case studies are discussed in the original report. 13 We decided to conduct both the GILS and the environmental reporting case studies in the same states to control for the unique characteristics, laws, and history of each state. We selected states based on: 1) whether GILS and environmental reporting had been implemented to a fairly significant degree so that we could learn from the lessons of implementation; 2) the state had secured a grant to participate in the EPA One-Stop Reporting program; and 3) we also wanted to maximize differences in size and have some regional representation. We decided on five states: Kansas, Massachusetts, Ohio, Texas, and Washington. In Phase Five, we took the lessons learned from Phases Two through Four to see if a more general statement about problems and solutions to the interoperability question could be made. In these interviews (outside of the GILS and environmental-reporting interviews) we asked whether our findings about interoperability corroborated what managers and policy makers saw. After each interview, the working model describing barriers and solutions was updated and tested anew. To begin the interview process, the research team developed a generic list of government information-technology positions to be interviewed, including chief information officers, persons in charge of enterprise architecture planning, individuals responsible for managing meta-data, and state librarians. The goal was to find those information-technology professionals and managers who had experience with both the technology and policy spheres and who could comment on federal/ federal and federal/state interoperability. Federal officials were fairly easy to locate by searching federal Web pages and through a series of phone calls. To help in identifying the state officials, the National Association of State Information Resource Executives 14 furnished us with a list of names for the officials identified above. This list was expanded by asking, at the end of each interview, who else could comment intelligently on the kinds of questions that were just discussed. A full list of interviewees can be found in appendix B. Before each interview, a statement on the general thrust of our research was faxed or emailed. It included a short description of the Intergovernmental Enterprise Panel and the Technology Policy Group at the Ohio Supercomputer Center. This was accompanied by a working set of findings and recommendations. To secure the interview and assure that it would not take too long, it was made clear that the interview would not cover every point; the goal was to obtain their most salient reactions-what was most horribly wrong and what was the most energizing insight. It was explained that these were working conclusions, and the purpose of the interview was to receive comment on the conclusions most salient and relevant to their work. Despite the proffer to discuss only the most salient points, interviews usually touched upon most of the general findings. The interviews lasted an average of 1 hour, though some were as short as 45 minutes and one lasted 2.5 hours. A research judgement was made to conduct the interviews pursuant to an understanding that specific comments and/or recommendations would not be attributed to specific officials, states, or agencies. The decision was to err on the side of obtaining a richer conversation about the difficulties and opportunities in achieving interoperability. Findings and Recommendations This section presents the final results of our research and develops a model that builds on the work by The findings and recommendations reveal four distinct themes: 1) a vision that interoperability really means the sharing of information; 2) identifying and working with critical success factors in the development of interoperable systems; 3) developing a strong federal/state architecture to support interoperability; and 4) recognizing that in a technological society, technical standards are another kind of public law, and better systems need to be in place to support government standards setting where that is important to sharing information. All of these components are necessary to building an information infrastructure that can support real information sharing. Interoperability Means Sharing Information Interoperability among federal, state, and local information systems is more than "plumbing"-that is, making sure the information pipes fit together through compatible hardware and software. The interviews revealed that despite the current popular focus on hardware and software compatibility, fundamentally, interoperability is really "the sharing of information." According to these experts, the technical plumbing issues, while not trivial, are receding in importance relative to the information-management issues. In other words, because the technology is allowing some degree of interoperability to occur, a host of information-management issues are now appearing. 15 This focus on information sharing implies a much richer set of political, organizational, and economic issues because now the focus must be on the ultimate goal of sharing information. Sharing information can be any of the multiple uses 16 of information that is generated from another organization to improve policy or management. This reframing implies a whole infrastructure that must be in place to support effective information sharing. Agency information policy makers and managers must make this first conceptual leap before any real progress in improving interoperability can take place. There are several practical implications that directly result from reframing the definition of interoperability. Once policy makers and managers focus on interoperability as information sharing, it becomes easier to generate appropriate operational subgoals and success criteria. The operational subgoals of interoperability should include the extent to which the tools that enable information sharing are in place. These tools include: 1) meta-data about the information that will allow a user to identify the existence, nature, and quality of information so that the prospective user can assess the utility of the information; 2) clearer law and policy for when and under what conditions organizations should make their information available; 3) economic and budgetary mechanisms to support the identification of costs and benefits associated with sharing information; 4) success criteria to show the extent to which information is shared; and 5) managerial tools to provide incentives and controls over the sharing of information. Another simple but important practical recommendation is that once interoperability is seen as a problem in sharing information, interoperability projects should not automatically be delegated to strictly technical teams. Stated positively, teams working on interoperability should include persons who understand the strategic information needs of the organization and how technology can be used to get them there. Identifying Information Identifies What the Organization "Knows." Information-sharing needs must also be driven by a clear understanding of what the organization "knows," which is essential to the process of continually improving policy and management The Prevalence of Ad Hoc, Informal Contracts to Negotiate Agency Needs. The research revealed the existence of many informal agency "sharing agreements," memoranda of understanding, and informal contracts to specify how information is to be shared. Questions remain as to whether these informal contracts are legally binding, but it is clear they do provide expectations about how to share information. For example, environmental protection offers many instances of intraagency cooperation agreements as agencies transition from media-based data collection and storage systems (air, water, etc.) to a facility, one-stop form of information management. 17 lntraagency pooling of information and information technology typically have a common supervisor (maybe as far up as the director of the agency). Additionally, responsibilities for environmental protection are typically spread over several state agencies (typically agriculture, health, natural resources, fish and wildlife, occupational safety, and the like). Further complicating matters, states often have local or regional authorities with responsibility for a particular coastal area or river region that also must interface with environmental information systems and needs. Coordinated environmental protection agencies in adjacent states also have common environmental needs (for instance, Ohio River matters are relevant to six bordering states). Ad hoc sharing agreements were Government Information Systems and the Fourth Generation of Information Technology 213 one solution to coordinating the information-sharing efforts of these many agencies and divisions. Unfortunately, the interviews also revealed that, in many cases, agencies go through the process of developing interoperable systems alone. This results in unnecessary duplication of time and resources as each sharing agreement is individually negotiated anew. Moreover, the complexity can also be a severe disincentive for smaller agencies. Ad hoc approaches invite the creation of inconsistent policies and practices across agencies and levels of government. One interesting possibility to address the prevalence of ad hoc agreements is to accumulate existing sharing agreements and develop a "uniform contract" or "formbook" from which others can draw expectations about future behavior. One of a lawyer's best kept secrets is the formbook. If a lawyer has a problem to solve, chances are there will need to be some kind of document to solve that problem. If the lawyer has to go to court, s/he consults a formbook for the many kinds of documents that must be filed with the court. It may be as simple as selecting the form and filling in the names of the parties. If a contract has to be negotiated, the attorney can consult a contract formbook, which provides all kinds of contract clauses to solve all kinds of contract problems. The responsibilities laid out in a contract formbook have been time tested as practical by the courts and are backed as being the law. In the parlance of programmers, the formbook provides a set of "object code" modules that can be plugged into a contract. The formbook approach provides practitioners with a commonly understood set of practices. Over time, interoperability agreements could become more routine because the full list of concerns that must be considered is made explicit, the respective responsibilities are made clear, and useful tools, accumulated over time, are located in one place. In the case of interoperability contracts, the formbook approach might be indexed by the many possible issues in sharing information identified earlier. It would suggest clauses to anticipate those potential problems so they could be avoided, or if they occur, how they could be mitigated. Interoperability must move from an ad hoc development to a systematic strategy with clear performance expectations. General Attitudes toward the Sharing of Public Information. A significant "cultural" barrier to sharing information is the belief that providing undigested data too freely to the public will result in distortion and misuse, perhaps by persons advancing a political agenda rather than governance. In many cases, however, organizations found the fear of distortion and misuse unfounded when they actually began sharing more information. For politically salient problems, where distortion and misuse would occur regardless of the degree of information provided, agencies should provide more contextual information and meta-data. Public access to even more information corrects distortion. Critical Success Factors in the Development of Interoperable Systems There are factors that tend to support the identification, planning, and implementation of data-sharing arrangements. This is the knowledge about how to share information. If there is a natural learning curve, agencies can use this knowledge to learn from the mistakes of others by climbing the learning curve faster and more intelligently. Our interviews identified several factors in managing this learning curve: 1. Sharing is more likely when agencies have a history of working together. 2. Sharing is more likely when there is unearmarked, centralized money than when agencies are asked to fund sharing out of their own budgets. 3. Federal/state sharing will only be accomplished when it is perceived to be in those agencies' own self-interest. 4. Multilevel or multiagency information-sharing arrangements benefit by making all the interested organizations equal parties to the agreement. 5. Interoperability projects are more easily implemented when the focus is on sharing transaction information (as opposed to operational or strategic information) since these are the processes where tangible benefits in reduced costs and improved service can most easily be documented and justified for budget reasons. 6. Interoperability projects are more easily implemented when the projects occur within an agency, when budget savings can be more easily shared, and when they result in short, or even same-year benefits as opposed to long-run benefits, and when all participants have common executive leadership. 7. Crisis (real or created) is an essential catalyst and opportunity, for intergovernmental collaboration points toward raising awareness and providing the political capital necessary to sustain progress. There are the beginnings of a federal policy architecture to support information sharing, 18 but law is not always sufficient to ensure implementation There Is No Strong Federal/State Architecture in Place to Support Interoperability At the federal level, legislation mandates a strong, agency-centered strategic-planning effort, which lays the foundation for smart strategic information planning. But this strong emphasis on accountability (indeed, "hyperaccountability") that requires a clear, explicit connection between resources and outcomes, measures for success and a full set of reporting mechanisms, actually has the perverse effect of excluding states because it focuses the federal agencies' efforts on meeting their own accountability requirements. Unfortunately, if states are not included early in the planning process, there is little chance for them to have a real voice. State officials also expressed the hope that federal requirements for providing data should be rationalized. Far too often, states view these requirements as unfunded mandates because there are significant programmatic and fiscal responsibilities that come along with these information requests. In some cases, the states are not sure the information is even being used. State and local governments will need to see a benefit in providing this information to the federal government. Finally, states are concerned about the sometimes inconsistent data requirements-agencies explicitly requiring different kinds of information from states. It is essential that states be able to provide input early in the federal information strategic-planning process. In a Technological Society, Technical Standards Are Another Kind of Public Law While there are many formal procedures surrounding the creation of public law, the procedures surrounding the creation of technical standards are less clear. Technical standards are public law because they dictate, for the indefinite future, the kinds of activities in which society can engage. Almost all of the standards organizations are private sector, and this is appropriately so, but there is still a need for the government to make sure that in some settings, the government's and the broader public interest should be represented. While the federal government has made significant progress in redefining how government participates in the standards process (National Technology Transfer and Advancement Act Currently, there is sporadic representation by public agencies in the standards-setting process. Where there is representation on the federal level, the National Institute of Standards and Technology (NIST) 20 does most of the representation to non-security-related standards discussions, but NIST is limited by resources from doing more. Most state agencies communicate their preferences for technical standards directly to their vendors, but they feel that other avenues to ensure good standards should be cultivated. Alternatively, public servants communicate their needs for technical standards through their professional communities, but they agree that this results in a splintered representation of public interests. States do not necessarily want the federal government to become involved in the standard-setting process on their behalf. They also desire that standards should be established only when pilot projects have proven the workability of those standards. Often, standards practice is divorced from best practices and the specific needs of organizations. The interviews revealed that far better coordination could be achieved by learning about best practices and then bringing that knowledge to bear on the standards process. A clearinghouse that investigates best practices could also provide advice to the standards-setting process. The process supporting adoption of a standard, the gathering of data on best practices, and the actual implementation of a standard that will uncover difficulties is a very delicate one and requires close participation among the user and the standards communities. Conclusion Information technology is one of the few exciting answers to the question of how to improve governance. If government's great experiment with information technology does not result in significant improvements, it will owe largely to the fact that interoperability was too difficult to implement. At the same time, if interoperability is realized, it will result in a fundamentally different way of doing government. Interoperability is thus a key issue in understanding how information technology will truly affect the public sector. Dawes (1996) discusses how individual agency-toagency decisions are made about whether and how to share information (see