, Purdue University
, Brown University
Pages: pp. 32-38
Abstract—Despite occasional setbacks, digital government projects now appear firmly on the road to fulfilling their promise of making civil and political processes more accessible than ever.
Digital government or—following current technolinguistic conventions, e-government—can be defined as the civil and political conduct of government, including service provision, using information and communication technologies (ICT). Although the term represents and reflects only part of the long technological transformation most sectors in society have undergone, the government domain is unique with respect to ICT.
Government transcends all sectors in a society. It provides not only the legal, political, and economic infrastructure to support other sectors, but also exerts significant influence on the social factors that contribute to their development. Digital government, as a result, has the potential to profoundly transform citizens' conceptions of civil and political interactions with their governments. Unlike commercial service offerings, digital government services must—in most societies—be made accessible to all.
Further, the implementation of a public service often requires the integration of intragovernment ICT. Consequently, human factors, geography, organization, ontologies, security, data quality, and other issues must all be dealt with in ways possibly more complex than other application domains require.
Holistic visions of digital government have emerged over the past two decades. Simon Nora and Alain Minc, perhaps the most influential early progenitors of a comprehensive vision of digital government, advanced the notion of télématique, the synthesis of telecommunications and computing. Their 1978 report, 1The Computerization of Society, to then-president of France, Valéry Giscard d'Estaing, defined a vision for an industrial policy that would restructure civil and political society via télématique. 2 The report also identified roadblocks to implementing such a policy.
Enormously influential, the Minc and Nora report spurred deployment of the Télétel/Minitel videotext system by the French government in 1979. By 1995, Minitel provided more than 26,000 government services. 3
The current vision of digital government, however, is—like most other IT areas—dominated by the Web. Commercial Web services have clearly raised citizens' expectations of the service level that government agencies' Web-based offerings must provide. 4
Digital government systems can generally be characterized along two dimensions:
Architectures include intranets to support intragovernmental processes, public network access to facilitate government-citizen interactions, and extranets for supporting interactions between the government and nongovernmental organizations.
Generally, current digital government system designs provide one of four service levels: 5
The progress made thus far in providing online access to government has received mixed reviews. As shown in the " Selected Links" sidebar, in the US, unprecedented levels of citizen interaction with government agencies now take place over the Web. Citizens can easily retrieve congressional records, information about agency activities, forms, statutes, and regulations from government Web sites. Digital government also facilitates two-way communication and more complicated processes. Several US government agencies now let citizens file comments online about proposed regulations. The US Department of Agriculture, for example, received more than 100,000 electronically filed comments in 1998 for its proposed organic foods regulations. Citizens in Scotland can now create and file online petitions with their parliament.
Broadly speaking, however, digital government initiatives still need better solutions. Many problems—such as data integration or security interoperability—are ultimately technical in nature, but remain most apparent at developmental and functional levels. Todd Ramsey, IBM's head of worldwide government services, noted recently 5 that "[a]bout 85 percent of all public-sector IT projects are deemed to be failures."
Brown University's Taubman Center for Public Policy and American Institutions recently released the results of a detailed study of more than 1,800 state and federal Web sites in the US. 6 These results point to some major functional shortcomings of Web-based digital government. The study found that, overall, the "e-government revolution has fallen short of its potential." According to this report, areas that need improvement include disability access, the use of security and privacy policies, foreign-language translation, and consistency and standardization of Web site designs across government organizations.
Richard Heeks's sidebar, " Understanding Digital Government Project Failures," underscores the complexity of the developmental factors affecting success and failure in digital government projects.
Development of a distinct digital-government research area has been accelerating during the past decade. Several major research centers have been established, including the Center for Technology in Government (CTG) at the State University of New York, Albany; the Digital Government Research Center (DGRC), jointly operated by Columbia University and the University of Southern California's Information Sciences Institute; and the International Teledemocracy Centre at Napier University in Scotland.
Major governmental initiatives have also been undertaken. The Italian government established the Authority for IT in the Public Administration (AIPA), an independent body charged with planning, promoting, and monitoring IT in the public sector. In 1991, Singapore's government commissioned its IT2000 master plan for research and development of a comprehensive e-government. In the " Electronic Petitions and the Scottish Parliament" sidebar, Ann Macintosh writes about the e-petitioner system the new Scottish Parliament is using.
Several critical conferences and funding sources have emerged as well. In 1997, the Ford Foundation funded an invitational workshop on digital government hosted by the CTG and attended by researchers from Europe and North America. The US National Science Foundation started its digital government program in 1997, held interdisciplinary workshops on digital government in May 1997 and October 1998, and issued a major report by the participants of the latter workshop in March 1999.
This special issue contains an international spectrum of reports from the evolving area of digital government research.
Systems integration has been a major hurdle in implementing broad and comprehensive digital government infrastructures. In "Enabling Italian E-Government through a Cooperative Architecture," Massimo Mecella and Carlo Batini discuss a government-wide effort in Italy to address this issue. Related hurdles include combining systems and ontological integration with workflow support. The DGRC's contribution, "Simplifying Data Access: The Energy Data Collection Project," presents solutions to government-wide ontological integration. In "Managing Government Databases," Athman Bouguettaya and colleagues address the development of digital services that aid citizens in receiving services that require interactions with multiple agencies.
Security and privacy are critical and constant issues that arise in digital government. In "Digital Government Security Infrastructure Design Challenges," Arif Ghafoor and colleagues present a survey of issues and models being used to address security in a governmental context.
Two sidebars further illuminate the challenges and successes digital government projects have met with worldwide. In " Disseminating Information but Protecting Confidentiality," Alan Karr and colleagues write about the development of statistical security techniques to protect citizens' privacy in digital government data systems. The sidebar by Joaquim Pinto and colleagues, " Portuguese Parliamentary Records Digital Library," describes a new digital library system being developed to provide access to the Portuguese Parliament's records.
These sources provide more information on the promise and challenges of digital government. For a complete set of links, see http://www.cs.purdue.edu/icds.
Recent survey analysis 1 suggests that a fifth to a quarter of digital government projects are either never implemented or abandoned immediately after implementation. Beyond these total failures, a further third fail partially, falling short of major goals, causing significant undesirable outcomes, or both. A few projects incur sustainability failures, succeeding for only a year or so. Some experience replication failures that make them unrepeatable beyond their pilot location.
To avoid such outcomes, we need to understand, predict, and assess digital government project failures. With these aims in mind, concerned researchers commissioned and analyzed an international set of digital government case studies. 1
The framework for understanding these failures centers on the difference between design ideas and organizational realities. The case studies showed that the larger this design-reality gap, the greater the project failure risk. Conversely, the smaller the gap, the greater the chance of success. The gap consists of up to seven "ITPOSMO" dimensions: information, technology, processes, objectives, skills, management systems, and other resources (time and money).
The following specific cases illustrate a failure and a success in navigating the design- reality gap and the ITPOSMO dimensions.
A Swedish city government launched a digital government project using a Web site and online discussion forum to increase citizen participation in governance of three city districts. 1 The project team encountered design-reality gaps in the following dimensions:
Overall, the design-reality gap resulted in a partial failure of this digital government initiative, with objectives significantly unmet. Despite covering tens of thousands of inhabitants, district Web sites registered no more than 40 hits per day. Similarly, debate contributions numbered just a few dozen over an entire year, with only one discussion in one district having input from local politicians.
An intranet-based digital government project initiated by the Johannesburg Metropolitan Council in South Africa 1 provides a contrasting example. This project attempted to break apartheid-legacy information flows and give all staff access to information sources. Here, design and reality matched well, significantly narrowing the gap between the two along the following dimensions:
The project's success made council processes more inclusive and transparent. The project is now being extended to encompass local community leaders and other citizen representatives.
Converting each of the ITPOSMO dimensions into a set of rating scales helps predict the potential failure of digital government projects. Key project stakeholders then discuss and score these scales in a facilitated workshop that uses an iterative approach.
Once participants identify major design-reality gaps, the workshop formulates advice for closing those gaps in one of two ways:
Take the fairly straightforward example of a financial gap. A design change could reduce this gap: scaling down the project remit and thereby reducing cost. Or a reality change could reduce the cost: using public-public or public-private collaboration to increase available financing.
More generic gap-closure techniques include
The work of the International Teledemocracy Centre at Napier University in Edinburgh focuses on how democracy can leverage new developments in technology. (A. Malina and A. Macintosh, "Teledemocracy: Energising the [New] Public Sphere(s), Civil Society, and Citizen Activity," http://www.teledemocracy.org/ourwork/our-work-pubs.htm).The ITC, in partnership with British Telecom Scotland, has developed an e-democracy toolkit that contains three tools: e-consultant, e-voter, and e-petitioner. The Scottish government uses e-consultant to consult young people. The Highland local authority uses e-voter to elect young people to a youth council. The Scottish Parliament first piloted e-petitioner as a tool for citizens to petition Parliament electronically.
Traditionally, the public used paper-based petitions. Several thousand people often sign a petition, and UK parliaments receive more than a thousand petitions each year. However, the UK has only now begun to accept petitions electronically, thanks to the success of the e-petitioner system.
In July 1999, the Scottish Parliament opened officially in Edinburgh, assuming the devolved power for specific government areas from Westminster Parliament in London. The Scottish Parliament aspires to use all forms of information and communication technologies "innovatively and appropriately" to support openness, accessibility, and participation.
The ITC works with Parliament to ensure that the requirements of citizens petitioning Parliament electronically are met and that Parliament has confidence in the electronic petitioning system's integrity. The first electronic petition, which related to the environment, was submitted to the Scottish Parliament on 14 March 2000. As a result, Parliament agreed to take citizen input via e-petitioner for a trial period.
E-petitioner ( http://www.e-petitioner.org.uk) is a Web-based electronic petitioning tool that lets users create, view, sign, add background information to, and submit petitions. The site also supports a discussion forum. The tool hosts valid petitions and submits them to the relevant authority. E-petitioner's design balances wide accessibility with restrictive security checking. Implemented using Microsoft SQL and ASP, e-petitioner does not use frames or contain large graphics files, thereby ensuring easy access by community centers that run slower machines. E-petitioner also supports use by the partially sighted.
Security issues centered on how much trust the public and Parliament should place in this new technology and how much name checking would be necessary. Matching such requirements to those currently available for paper-based petitions raises the issue of what level of checking gets applied to paper-based names and addresses beyond manually reading often illegible handwriting. On the other hand, to say that everything must be checked—as is the case for electronic voting—may be unnecessary for petitions, which are not legally binding in the UK.
Nevertheless, e-petitioner performs name checks prior to submission of a petition and allocates a confidence rating to each name. The tool bases this rating on seven parameters, such as the number of times the same IP address has signed within a certain time limit.
E-petitioner has demonstrated that straightforward computing techniques can enhance public participation in the new Scottish Parliament. In addition, e-petitioner now hosts the first-ever e-petition to the British prime minister.
Federal statistical agencies have longstanding concern about the confidentiality of data such as sample surveys and censuses. Both the identities of data subjects and sensitive attributes in the data must be protected. 1,2 But the agencies also have an obligation to report information to the public.
This tension between confidentiality and dissemination of statistical information 3 arises equally sharply in nongovernmental contexts, from electronic medical records 4 to e-commerce transaction data. Advances in information technology, such as powerful capabilities for record linkage across multiple databases, threaten confidentiality. Other new technologies, however, not only protect confidentiality but also meet user needs in innovative ways. For example, the National Institute of Statistical Sciences ( http://www.niss.org/dg) is developing a system for the National Agricultural Statistics Service that disseminates geographical survey data on agricultural-chemicals usage but protects the identities of farms in the survey.
The NASS database contains 194,410 records from 30,500 farms. It details the use of 322 chemicals—fertilizers, fungicides, herbicides, pesticides—on 67 crops from the years 1996 to 1998. Record attributes are Farm ID, size in acres, crop, chemical, pounds of the chemical applied, state, county, and year. Users can query for application rates in pounds applied per acre of certain chemicals on particular crops, ideally at the county level. Currently, NASS releases application rates only at the state level.
For the application rate in a geographical unit to be disclosable, NASS requires that two widely employed rules 2 be satisfied. The N-rule requires that the unit contain at least N = 3 surveyed farms for the specified chemical, crop, and year. The p-rule prohibits disclosures containing a dominant farm comprising more than p = 60 percent of the total acreage of all farms surveyed in the unit. At the county level, these rules do not work: More than 50 percent of counties are undisclosable. Our system aggregates undisclosable counties with neighboring counties in the same state to form disclosable supercounties, thereby allowing NASS to release data at the highest resolution consistent with the risk criteria.
The system must compute aggregations automatically in response to user queries. It can formulate computing aggregations as an NP-hard combinatorial optimization problem over the edge-set of the adjacency graph of the counties in a state, and use simulated annealing methods to solve them. Long running times make this approach infeasible in practice, however. Instead, we employ heuristic "greedy" algorithms 5 that produce aggregations differing only insignificantly from those that simulated annealing produces.
We have developed two heuristic algorithms that share a common structure. The algorithms examine the undisclosable supercounties in a random order and merge them with a neighboring supercounty until only disclosable supercounties remain. The algorithms differ only in the rule that governs merging. The pure rule favors leaving disclosable counties unmerged, thereby preserving their data's purity, but this can create large supercounties comprised of many undisclosable counties. The small rule, by contrast, favors forming small supercounties by merging an undisclosable region with a neighboring region most likely to achieve disclosability.
Both algorithms randomize the order in which candidate mergers are considered, and both break ties randomly. Each can produce aggregations in which some supercounties can be decomposed. 5 To alleviate this, our implementation first runs the small algorithm, then runs the pure algorithm within each supercounty produced by the small algorithm. This composite procedure works fast and well.
Figure A shows two screen shots from the prototype NASS system ( http://niss.cnidr.org).
Figure A The NASS data-dissemination system prototype. (1) Input screen, from which users select a state, year, crop, and chemical. (2) Output screen with map that displays the requested application rate, using artificial data.
The user first selects from the left panel of Figure A a state and year of interest. JavaScipt routines then dynamically generate drop-down menus of relevant crops and chemicals. The user next selects either a crop, in which case the system regenerates the chemical menu to contain only chemicals applied to that crop, or a chemical, which causes the crop menu to be regenerated. Finally, the user selects an output format: map (the default), onscreen table, or XML download. The XML DTD mirrors the hierarchical nature of the aggregated data. If not available from a previous query, the system computes the aggregation on the fly. The system stores the result in case it receives the query again, then writes transaction information to a query history database. Map output appears in the right panel of Figure A. Each supercounty's color indicates the application rate of the chosen chemical on the chosen crop; the color bar also shows the statewide average rate. Supercounty and county-within-supercounty boundaries are shown, but differently. Multiple years appear on separate maps with a common color scale.
The federal government has an essential responsibility to provide citizen access to data and information. The system we've described is a step toward using the Web to meet that responsibility. More complex databases, queries, and privacy concerns lead to additional challenges. These include queries whose disclosure risk depends on which queries have been answered previously, risk computation and reduction for queries entailing integration of multiple databases, and problem formulations and tools that let agencies balance the value to society of releasing information against the risk of disclosure.ReferencesFederal Committee on Statistical Methodology,Report on Statistical Disclosure Limitation Methodology,May1994,http://www.bts.gov/NTL/DOCS/wp22.html.L.WillenborgandT.de WaalStatistical Disclosure Control in Practice,Springer-Verlag,New York,1996.G.T.Duncanet al.,"Report of the Panel on Confidentiality and Data Access,"J. Official Statistics,vol. 9,1993,pp. 271-274.E.Lander"Who Knows Your Medical Secrets?"Consumer Reports,Aug.2000,pp. 23-26.A.Karret al.,Web-Based Systems that Disseminate Information from Data but Protect Confidentiality,tech. report, National Inst. of Statistical Sciences,2000,http://www.niss.org/dg/technicalreports.html.
Placing information on the Internet and providing universal access to all citizens has become a major political imperative in Europe. The development of a digital library project and the display of written information concerning parliamentary proceedings have a double objective: to give citizens access to what happens in the Portuguese Parliament and to create a working instrument for historians and researchers.
The system architecture has four basic modules:
The records library organizes the digitization of speeches by members of Parliament into small documents with an average of 50 pages each. The library represents each page individually so that when a user submits a query, the individual page where an expression occurs can be quickly located. This methodology was also applied in the JSTOR Online Archive. 1 The structure is similar to the structure of the original paper documents, an extremely important feature that curtails user rejection of the system.
One system objective is to allow the display of pages in either text mode or original format. This feature requires storing every page simultaneously in two formats. The library stores the image files in TIFF bitone with CCITT Group 4 compression. The library stores the text files in HTML, extending them with extra information so that the system can handle each file again as a whole after individual pages have been accessed. The extra information also helps maintain the relationship between text and image files. To facilitate such relationships, developers use the Dublin Core Metadata Initiative 2 set of qualifiers to add a set of metadata fields to each page.
The system implementation is still in progress. Further system design modifications will likely be implemented after records of the parliamentary sessions are placed online and when the digitization of the gray literature surrounding these records begins.ReferencesS.W.ThomasK.AlexanderandK.Guthrie"Technology Choices for JSTOR Online Archive,"Computer,Feb.1999,pp. 60-65."Dublin Core Qualifiers,"July2000,http://www.purl.org/dc/documents/rec/dcmes-qualifiers-20000711.htm.
This work is supported by the National Science Foundation under grants 9972883-EIA, 9983249-EIA, and 9974255-IIS; ARO/EPRI under grant WO8333-02; a State of Indiana 21st Century Research and Development fund grant; and grants from HP, IBM, Intel, NCR, Telcordia, and Walmart.