« i l DÄMM! PUBLIC ADMINISTRATION AND INFORMATION TECHNOLOGY
Compendium to the Erasmus Course
COMPLEX 6/95 INSTITUTT FOR RETTSINFORMATIKK TANO Hovedkontor: Rosenholmvn. 25. 1410 Kolbotn, tlf. (66) 99 80 00 Bergen: Øvre Kråkenes 17. tlf (05) 12 OK 00 Stavanger: Professor Olav Hansscns vei 13. tlf. (04 ) 87 70 00 Trondheim: Vestre Rosten 81. tlf. (07) 88 00 00 ZNgnkfrmwngfor JUSogEDB
Postboks 6728, St. Olavs plass, 0130 OSLO Postgiro 0813 5139654 Bankgiro 8200.42.49727
Foreningen står bl.a. tor salget av CompLex-heftene
Enkel og hurtig tilgang til offentlige informasjonsdata
- en forutsetning for effektiv saksbehandling og riktige beslutninger
Statens Datasentral er Norges ledende databasevert for offentlig informasjon. Vi formidler opplysninger fra Løsøreregisteret, Det sentrale personregister, Motorvognreaisteret, Tinglysingsdata og Lovdata, for å nevne de viktigste. I tillegg til alle tjenester for online bruk av basene, bistår vi selvsagt også med satsvise datauttak. Ta kontakt for nærmere opplysninger!
inform
Statens Datasentral a.s Ulvenveien 89 B, 0581 Oslo telefon 22 95 63 00
Advokatfirmaet Selmer § Co. DA har 40 medarbeidere, hvorav 25 advokater og advokatfullmektigen Firmaet yter advokattjenester hovedsakelig innen de forretningsjuridiske disipliner omfattende bl.a. kontraktsrett, selskapsrett, skatterett, aksje- og børsrett, arbeidsrett, petroleumsrett. miljørett, gjelds forhandling og konkurs, jus og EDB. Våre oppdragsgivere er i del vesentlige innenfor bank, finans, revisjon, industri, forsikring, shipping, energi, handel, entreprise og fast eiendom, forskning og utviklingsarbeid. Firmaet yter dessuten advokattjenester ved generasjonsskifte i familiebedrifter og bistand generelt innen arverett, familierett og skifterett. Firmaet har en stor andel utenlandske oppdragsgivere og er tilknyttet Unilaw, en internasjonal sammenslutning av advokatfirmaer.
Asbjørn Berg Odd-Einar Christophersen Anne-Blanea Dahl Bjørn Kleiven Jeppe Norman Cristian Selmer Jon Skjørshammer Roy M. Slettvold Synnøve Smedal Svein Sulland Eilif Torma Sverre S. Tysland Lars Weyer-Larsen Trygve Øydne
Fullmektiger:
Amt Angell Inger Ekker Bartnes Fredrik A. Borch Gunnhild Buestad Morten Bøsterud Susan Halvorsen Ingebjørg Harto Jan Herud Margrethe Husebø Geir Kruge Anne Helene Osberg
Firmaet har etablert traineeordn ing for studenter
ADVOKATFIRMAET SELMER & CO. SELSKAP MED DELT ANSVAR Kirkegt. I5.tif. 22 42 64 90 Postboks 476 Sentrum. 0105 Oslo f ADVOKATFIRMAET BLOMSETH, KRISTIANSEN, SOLBERG & VALE DA M.N.A.
❖ Telekommunikasjonsrett ❖ Bygg- og entrepriserett kabel-TV, kringkasting m.m. ❖ Eiendomsrett ❖ Opphavsrett ❖ Boligrett ❖ Transportrett - Luftrett ❖ Selskapsrett ❖ Skjønns- og ekspropriasjonsrett ❖ Foreningsrett ❖ Anglo-amerikansk kontraktsrett ❖ Arbeidsrett
Tlf. 2 2 60 85 85 - Fax 22 60 85 20 Pilestredet 57, 0 3 5 0 Oslo Jf
CompLex nr 6/95
Institutt for rettsinformatikk Postboks 6702 St Olavs plass 0130 Oslo
Jon Bing and Dag Wiese Schartum
Public Administration and Information Technology
Compendium to the Erasmus Course
TANO © TanoA.S. 1995 ISBN 82-518-3416-3
Institutt for rcttsinformatikks utgivelser i skriftserien CompLex er støttet av
Den norske Advokatforening Lovdata Kopi nor
Fotografisk opptrykk ved Engers Boktrykkeri A/S, Otta 1995 Preface
This compendium is specifically composed to serve as a syllabus for Erasmus students attending the course on "Public Admini stration and Information Technology" at the Norwegian Re search Centre for Computers and Law, University of Oslo. The book contains a collection of articles discussing various legal problems connected with the application of information technol ogy in public administration. Many of the contributions take, as an explicit point of departure, examples from the Norwegian public sector. However, this should not mislead the reader to believe that these problems are specific for the Norwegian and Scandinavian societies. Even if the articles primarily discuss problems related to the public sector, the reader should, moreo ver, have in mind that similar questions are found in the private sector. Studying these articles should challenge the Erasmus students to find parallel examples and problems in their respec tive home countries and thereby convey new aspects and experi ences to lectures and discussions. We sincerely call on the students to give us feedback on the articles in this book, and hope that reading this book will contribute to their interest and involvement in the on-going and future debates about the relationship between public administra tion and individuals in a computerised society.
Oslo 3 August 1995
Jon Bing and Dag Wiese Scartum
Contents
1. The Norwegian Research Center for Computers and Law: Research Issues and Organisational Outline 2. Data Protection in Norway 3. Privacy and Surveillance Systems 4. Commercialisation of Government Information and Data Protection 5. Automated Systems and Claim Realization in Social Security Organizations 6. Introduction of information technology in social insurance institutions: Political objectives and organizational consequences, history and future 7. Representation of legal rules in computer programs - some implications on legal protection and service to citizens 8. Delegation and Decentralization - Computer Systems as Tools for Instruction and Improved Service to Clients 9. Dirt in the Machinery of Government? 10.Organization of Systems Development - How Do We Secure Legally Correct Computer Programs? 11.Regulatory Co-Operation through computer assisted solutions
Public Administration and Information Technology
The Norwegian Research Center for Computers and Law: Research Issues and Organisational Outline
Jon Bing
1. The initiative
In 1970, computers had become an integrated part of the Norwegian society. For instance, the introduction of a general social benefit scheme comprising the whole society, and introduced in 1967, was based on the maintenance of population-wide data bases for the monthly payment of the different categories of benefits. Professor Knut S Selmer, at this time both Dean of the Faculty of Law at the University of Oslo, and Chairman of the Department of Private Law, had taken sufficient interest to take a course in FORTRAN programming. He was dissatisfied with the results, however, as he felt that the ability to write trivial program statements did not bring him much closer to the two essential questions: How did a computer really work? And what impact might this technology have on society in general and law in particular? Compendium to the Erasmus Course
Therefore, it was somewhat natural that he turned to me when I was hired as a research assistant after my law exam in January 1970. My reputation as a science fiction writer may have made him believe that I might be inclined to take an interest in the area loosely described as "computers and law". He turned me towards the American journal Law and Computer Technology, and 1 started taking classes in COBOL, another high level programming language. March 16, 1970, the Department of Private Law organised the first of what was to become a series of seminars in "computers and law". I spoke in general of the possible issues involved, reviewing developments abroad, and emphasising computerised legal information systems.1 Due to the interest generated by this and other seminars, professor Selmer received a request from a government agency summer 1970 to undertake a research contract analysing the issue known then as "data banks and privacy". It is not usual to accept research contracts for university institutions. But by doing so, professor Selmer made the foundation of what should become the Norwegian Research Center for Computers and Law. And these early developments have imprinted the NRCCL with characteristics still visible: An interest in both legal technology and the law applied to information technology, and a reliance on extramural funds for pursuing its research interests. The activity became sufficiently visible that it became desirable to establish an organisational framework - it was maintained that creating a section of the Department of Private Law mainly was a question of printing a letterhead. From January 1971, the NRCCL a Section of the Department of Private Law.
1 The introduction is published as "Elektronisk databehandling i rettsvitenskapen". Lov og rett 8/1970:369-382. Public Administration and Information Technology
The activities slowly grew and became consolidated. In 1981, the NRCCL moved to new building that also housed the Data Inspectorate, which had been established under the Data Protection Act in 1980, and the Lawdata Foundation, offering computerised legal information services for Norwegian lawyers and public institutions. Together, these three organisations - one private, one public agency and one research center - became a unique symbiotic relationship in computers and law. It may be indicated that professor Selmer remained Chairman of the NRCCL (a position he held from the start until 1989, when he was succeeded by me), was the Chairman of the Board of the Lawdata foundation (a position he held until he retired in 1990), and became the first Chairman of the Board of the Data Inspectorate (a position he still holds). In 1985, the NRCCL was divorced from the Department of Private Law. The reason was related to the fact that research was conducted outside the area of Private Law, the geographical distance made administrative co-operation hardly feasible, and the economic risks implied by the projects funded by extramural sources. It became a "Department for Legal Informatics", but retained the English name. In 1988, the Parliament established a new chair in computers and law, which was given myself, and the year after, I succeeded professor Selmer as a chairman. Before given the organisational status of the NRCCL. we will briefly look upon the areas of research prominent throughout the more than 20 years of activity.2
2 Two books summarise the activities of the NRCCL: Jon Bing and Knut S Selmer (eds) A Decade of Computers and Law, Norwegian University Press, Oslo 1980; and Dag Wiese Schartum (ed) IR/s Spectrum: 20 artikler i anledning Institutt for rettsinformatikks 20 års jubileum, Tano, Oslo 1990.
3 Compendium to the Erasmus Course
2. Research
2.1 Legal Information Retrieval Systems and Services - Regulatory Management
Early the NRCCL took a strong interest in computerised legal information systems. This was stimulated by one of our early associates, Mr Trygve Harvold, who came to the Faculty in Oslo with a MBA from Berkeley, and a background as a system programmer for Lockheed. Exploiting his considerable knowledge, a research team was established which over the next decade explored several issues related to legal information systems. Mr Harvold is today the executive director of the Lawdata Foundation, still pursuing his interests with success in practice. The research can be characterised by some indications: User Research. It was the belief that a legal information system did not have any value apart from its contribution to improve legal decision making. Empirical knowledge about the interaction between conventional information systems and legal decision making was not available. Consequently, a number of projects were devised to study empirically this relationship. The studies took an analytical approach, and contributed to the development of a theory of legal decision making and communication processes. Analysis of text retrieval. No information system was available to the NRCCL at this time, but retrieval was simulated using specially batch programs. The interest was turned towards a better understanding of the performance of text retrieval systems, and what factors would influence performance. This led to a series of different research. A methodology of experiments in text retrieval was developed, and a number of empirical tests were made. The analyses of these led to the development of a recommended strategy for text retrieval known Public Administration and Infornuition Technology
as the conceptor based method performing better than more conventional Boolean strategies. A large number of other strategies were also devised. On the basis of the British text retrieval system STATUS, a Norwegian version was developed - NOVA*STATUS. This was for experimental purposes interfaced with a vector based retrieval machine, and program modules for supporting experiments in text retrieval, calculating performance figures. Interest was taken in the linguistic presumptions behind text retrieval, and a method for automatic splitting of compounded words and the identification of basic morphemes was devised, improving performance further. On the basis of this work, a community of text retrieval experts had been established. A system specification for a state of the art retrieval system was made early 1980, and today the result - SIFT - is a reasonably successful competitor on the European market. An active associate in this work on text retrieval methods was Ms Tove Fjeldvig (who now is an executive related to the SIFT program). The research is summarised in several books.3 Later, the work has had two characteristics: Conceptual retrieval. A strong interest has been taken in the intermarriage of knowledge based technology and text retrieval - so called "conceptual retrieval". Prototype implementation has been made, but what is more important, attempts have been
3 Jon Bing and Trygve Harvold Legal Decisions and Information Systems (Scandinavia University Press, Oslo 1977) won the Royal Academic Gold Medal, Jon Bing Rettslige kommunikasjonsprosesser (Scandinavian University Press, Oslo 1982) is his doctorate thesis, while Jon Bing, Trygve Harvold, Tove Fjeldvig and Robert Svoboda International Handbook of Legal Information Systems (North-Holland, Amsterdam 1984) adds to the theory of legal information retrieval the description of such systems in international organisations and 25 different countries.
5 Compendium to the Erasmus Course made to analyse and discuss the developments and the possibilities. Regulatory management. A strong interest has been taken in the interrelationship between the legal system as such and the legal information service. Detailed studies have been made of several countries, many of them on contract with international organisations like the UN and OECD. This has emphasised regulatory management - looking to those responsible for the management of the body of regulations rather to those lawyers applying these regulations on legal problems. The interest between the institutions and legislative procedures and legal information services characterises the work at the NRCCL in this area.
2.2 Knowledge Based Technology
Basic to the interest in legal information services, was an interest in the decision process. In 1972 a computerised system for deciding a special category of social benefits was introduced, which was almost wholly automatic. Several studies were made of this and other decision systems, and a budding understanding of the relation between legal rules (as extracted traditionally from written sources) and the rules of computerised system started to grow. This was stimulated by a general theory of legal rules, especially legal expert judgements, developed by a gifted colleague, Nils Kristian Sundby. Simplifying his theory, this became the foundation for an attempt of developing a system for analysing legal decisions - by describing the decisions according to the structure proposed in his thesis, it was believed we could then make a computerised analysis disclosing the underlying rule of an expert judgement - which is related to "open textured concepts". Public Administration and Information Technology
The system was successfully implemented by Johannes Hansen,4 and become known as SARA, an acronym denoting a system for analysing legal decisions. It should be noted that this was seen as a tool for legal research, for analysing large volumes of decisions (several hundred) where conventional tools of analysis failed, and where the system would present a model of the underlying legal expert judgement, but also identify inconsistency between the described decisions, precedence relationships etc. In the 1980's, a major effort was made to integrate the different components of legal technology in what has become known as the F*KUS concept: An integration within a work station of a document administrative system, an expert system, and conceptual retrieval. Individual components of this concept were implemented, but the envisioned "lawyer's work station" was never realised. One may note that the NRCCL always have had the strategy not to implement anything but prototypes. The work necessary to polish a production system, and its maintenance, has been seen as outside the scope of a research institution. But there have been several instances of other organisations picking up the concepts and implementing them in their own solutions. This was also the case with the F*KUS concept. The work in this vein was stimulated by professor Andrew Jones applying his knowledge of modal logic to the domain of law, and through the co-operation with research institutions as Imperial College, London. Today, one may characterise the work carried out at the NRCCL by two aspects: The interest in modelling (and understanding) uncertainty in law (including legal expert judgements), and the interest in answering the question: Is modal logic necessary for representing legal
4 Using the object oriented language SIMULA.
7 Compendium to the Erasmus Course knowledge, identifying at least one major example where this holds true.
2.3 Data Protection
The first research contract given the NRCCL was in data protection, and a strong activity has been present throughout. In the beginning, there were many empirical studies designed to disclose how systems for the processing of personal data really worked, how they were interlinked, etc. At this time, there was no dedicated data protection legislation, but there were a large number of policy issues to be analysed. The research paralleled that of the development of new legislation, and it is fair to say that the notion of data protection in Norwegian law is determined by this research: Data protection being seen as an interest of the data subject to control data about himself or herself, the control being exercised concerning decisions relating to the data subject. The analytical approach underlying this rather vague characteristic has become known as the interest model. In 1975, the NRCCL launched research into what became known as transnational data flow and the international, comparative and interlegal aspects of data protection. These two characteristics still describe the research carried out at the NRCCL, a basis of empirical studies of how data processing actually is being carried out, and an interest in comparative and international aspects. Obviously, there are no longer any lack of legislation, both in Norway, in other countries, and international legal instruments are available as objects of dogmatic analysis. Public Administration and Information Technology
2.4 System Development and Quality in Public and Private Organisations
As mentioned above, some of the work related to knowledge based technology was initiated after the analysis of decision support systems operational in public administration. This did not only lead to an interest in how to design systems better to represent legal knowledge. It also leads to an interest in discussing the legal aspects of the computer assisted decision making, especially in public administration. And as mentioned above, research in data protection also was based on an analysis of how personal data actually was processed and communicated, especially within public administration. Combined with the insight from the "knowledge based systems", this made it obvious that there were other legal policy issues than that of data protection. A trivial example would be the representing of an incorrect rule in the program of a decision system: The result would be an incorrect decision. But from the perspective of data protection, nothing much could be done about it as long as the personal data being process was correct and not misleading. This rather naive example must suffice to indicate the rather large area of research currently being called "the informatisation of public administration", where the law of public administrative procedures and the rule of law is used to analyse and assess the development of computerised systems within public administration, especially those related to decision processes. The major work so far is by our associate Dag Wiese Schartum, who will defend his thesis for the doctor juris title later this year. Another rather substantial result is the introduction of a new interfacultary Master's degree, cf below.
9 Compendium to the Erasmus Course
2.5 Electronic Funds Transfers and Electronic Data Interchange (EDI)
As a research student at the NRCCL in 1983, Olav Torvund made a study of the legal issues of funds transfers using cards in automatic tellers, point-of-sales terminals, etc. Now a deputy director of the NRCCL, he will in June 1993 defend his thesis on the law of funds transfers for the doctor juris title. The legal aspects of funds transfers had attracted research projects before EDI became a buzzword, and was probably one of the reasons for the NRCCL to become involved in the first drafts of what today is known as the UNCID rules. Today, several studies of EDI have been made, and the NRCCL participate in two large scale international studies funded by the European Communities. Of interest is also the small study made on contract with the International Chamber of Commerce (Paris) on obstacles in national law to the introduction of paperless solutions that the UNCITRAL found of sufficient interest to feature at their 25th anniversary (New York) in 1992. The interrelationship of the research areas may also be exemplifies regarding electronic funds transfers. Such transactions will cause trivial personal data to be recorded. Seen isolated, they are of little or no interest - seen as a whole, they will describe a personal profile, disclose likes and dislikes, habits, travel routes, etc. But the problem of electronic trails is not addressed in first generation data protection legislation, and is still a challenge to legal policy. Such transactions will also frequently imply cross frontier data Hows. For instance, if a transaction takes place in US dollars - even in Norway among two Norwegians - the transfer will be cleared in New York. The data protection issue of transnational data flows has found one of its most important examples in this type of transactions. Public Administration and Information Technology
2.6 Computer Contracts - Information Services
A traditional aspect of "computer law" has been the contractual law relating to the purchase of computer hardware, programs and related services. This was an early subject of the research at the NRCCL, a compilation of the contracts used by the industry in the early 1970's disclosed, for instance, a heavy influence of US contracting practices, imported through the organisational structure to Norwegian daughter companies. This caused some nice, tidy problems as the law surrounding the contracts was different in may important aspects, for instance are merger clauses not recognised by the law, making the interpretation of contracts rather different. In Norway, a standard contract developed by a government agency has a dominant role in the marked, being used as some sort of measure for any contact.5 This was originally developed by lawyers with a background from the NRCCL, and the relation has been maintained - the latest revision is based on drafts developed on contract with the NRCCL. This strong link to the actual computer business is important, and is emphasised also in may other respects. Also other types of contracts have been examined at the NRCCL, important have been contracts for data base services, both those based on the editorial material of newspapers, legal data bases, and data bases based on public material. The interaction between the private and public sector in information has been a matter of strong interest, both in Norway and the European Communities. There are a number of issues involved, some of them related to the freedom of information that make documents in the public sector available to the private
s Statens standard for edb-avtaler (SSA).
11 Compendium to the Erasmus Course sector.6 Other aspects are related to the liability of misleading information in public sector data bases, on the intellectual property rights in such material, etc. The whole area of contractual regulation of purchases of computers, programs and information services is crammed with legal issues. Many of them are related to the basic problem of classifying computer programs and services concerning the binary distinction between goods and services. Does the provisions consumer sales of goods act apply the computer programs? Does product liability extend to errors in an expert system? One may find several instances where the application of regulation to information technology has made it necessary to go back to basic and re-think the whole justification and interpretation of the clauses in question.
2.7 Intellectual property law
The first monograph on copyright protection of computer programs in Norway was published in 1973 as a result of a project at the NRCCL. The issues of the intellectual property protection of computer programs, data bases and integrated circuits have been with us since then. When the US adapted its copyright legislation to computer programs in 1980, there was already in Norway a Royal Commission working on the revision of the copyright law. There is a tradition between the Nordic countries for legislative co operation, and the commissions of Denmark, Finland, Norway, and Sweden joined in contracting the NRCCL for a report7 that then mas made basic for further work within the national
6 The current regulation was drafted by the NRCCL on contract with the Ministry of Justice. 7 Jon Bing Opphavsrett og edb, CompLex 2/85, Scandinavian University Press, Oslo 1985. Public Administration and Information Technology commissions. In Norway, the draft to the government bill was later developed by the NRCCL: A similarly strong co-operation with the government was later repeated with respect to the legislation on the protection of integrated circuits, where also Jon Bing made up the Norwegian delegation to Diplomatic Conference adopting the Washington Treaty of 1989.8 The issues of intellectual property law are numerous, and offers continuously new challenges for research, many of them originating for the implementation of the directive of the European Communities for the protection of computer programs - for instance the regulation on decompilation, which it may be maintained already also has had an impact on US law. One of the new, interesting areas is the conflict between the principle of free competition and the exclusive intellectual property rights in the area of telecommunication standards, which recently has been solved by the members of the European Telecommunication Standards Institute to adopt an undertaking.
2.8 Telecommunication law
One of the problems of information law is that it is - like all law - closely interconnected, but that the intertwined elements are not always well analysed and available to be applied to the new problems. In 1978, telecommunication still was a monopoly in most European countries. At that time, the NRCCL initiated the first research projects in telecommunication law, mainly because the role of telecommunication in distributed data processing, and the relation to transnational data flows. For the better part of this
8 An ill-fated treaty, as neither the Untied States nor Japan accepted the result, it is doubtful if it ever will enter into force.
13 Compendium to the Erasmus Course time, assistant professor Anne Kirsti Brække has made this her field of expertise.9 The research has been paralleled by the deregulation efforts in Norway and most other European countries, the conversion of monopoly into competition. This has been a rich field for research. A close co-operation has been established between Norwegian Telecom, the Telecom Research Establishment, the Telecommunication Regulatory Authority, and the Ministry of Communication. Within this framework for co-operation are there carried out a number of research projects. In addition, international experts are brought to Norway contributing to a better understanding of the issues involved. Obviously telecommunication today is much more than the physical infrastructure of cables, radiolinks and space segments, and the basic signal transmission. But in entering the legal problems related to value added services or audiotex,10 one again returns to problems related to information services, electronic data interchange, transnational data flows etc. The activity has sufficient scope to warrant the establishment in 1993 of a Section on Media- and Telecommunication Law within the NRCCL under the administration of Beate Jacobsen, whose research fellowship is funded by the Telecommunication Research Establishment.
2.9 Media Law
One thing leads to the other. When the research in telecommunication law was initiated, we became interested in
9 Ms Brskke is currently on leave working for the Telecommunication Regulatory Authority. 10 Andreas Galtung, a senior researcher at the NRCCL, has explored several of the legal issues related to value added services, including EDI and audiotex. Public Administration and Information Technology
the regulation of related fields. For instance, radio broadcasts were an instance of telecommunication, but there was enacted special legislation. It was difficult to understand the interaction of the two sets of regulations without having an insight into them both. Also, a historical accident had made Oslo one of the cities with the largest cable television network in Northern Europe: Oslo is sufficiently close to the Swedish border for viewers nearly to receive the two Swedish channels. In 1970, only one Norwegian channel was available, and the Swedish offer was of interest. Several individuals would co-operate to establish large receivers, and private cable television networks being the result. And when Sky Channel opened its satellite broadcasts in 1981, Oslo was there with a ready-made infrastructure. The regulation of cable and satellite television became an interesting arena of legal policy. Professor Knut Selmer again became a central figure as the chairman of the council organising the cable television through a licensing system. And the NRCCL was given funding from the Norwegian Research Council for a small program on media law - the last of these projects is the work of a research student, Mr Nils Risvand, on the international application of Norwegian broadcasting law, especially related to directed advertisements. Advertisements are commercial information, and as such the NRCCL has targeted this area with a hope to develop a reader for the law of advertisement. So far this has not met with success, but several smaller studies have been conducted.
2.10 Criminal law - information security
At the end of the 1970’s work on computer related crime was initiated when the person at that time being the head of the section of prosecution within the Criminal Investigation
15 Compendium to the Erasmus Course
Department at the Police Headquarters in Oslo contacted the NRCCL.11 Computer crime is often tall stories and spectacular incidents. At the NRCCL, the work has mainly been oriented towards regulatory reform: Are the current provisions appropriate when applied to information technology and the use of computer as the means for a certain offending action? Several interesting solutions have been found, my favourite is the extension of the time honoured prohibition against breaking a seal on a letter to the breaking of any kind of security device, which made this provision applicable to hacking. But there is also a strong relation between criminal law and data or information security, which again is related to data protection. Some of the recent work of the NRCCL has again been on contract with the government (Ministry of Administration) in developing a common legislative structure for data security related to freedom of information, data protection and national security interests. Obviously, information security is also related to many of the other research areas mentioned above - for instance the issue of authentification in electronic data interchanges. Current research (by Beate Jacobsen) is oriented towards the issue of communication integrity.
2.11 Wide Screen Law
Such an enumeration of research areas and issues is not able to convey the interest in the problems encountered, the enthusiasm in new understanding. Neither is this a complete sketch, for instance the considerable work relating to value added tax and general tax law has not been mentioned.
11 In Norway, police officer above a certain rank has to be qualified lawyers. Public Administration and Information Technology
Apart from the two areas for legal technology, the list of research issues may be seen as both broad and confusing, a sort of wide screen of flickering, legal images. And one should perhaps as a conclusion offer a view on information law as an area of research. We think this area of research is defined by two different strategies. Information technology is introduced for new applications. This makes it necessary to understand how the existing provisions regulating the applications are sufficient. Often the major problem is what in the Nordic countries has become known as the problem of description.'2 This implies that there often are severe difficulties involved in describing the application in terms that mach the concepts of law. If one succeeds in this, the legal issues can be unravelled. Often an institution like the NRCCL takes care of the problem of description, relying on its considerable experience in tackling exactly such problems. But when this is completed, the issue reverts to its original status as being something outside the area of information law. But there are some issues that do not have a traditional place in the legal system, and which seem more genuinely like information law - they include data protection, telecommunication law, intellectual property protection for computer programs, data bases and integrated circuits, etc. These are the issues on which a long term research policy should be based.
12 Introduced by professor Mads Bryde Andersen, University of Copenhagen, in his thesis EDB og ansvar, Jurist- og Økonomforbundets forlag, Copenhagen 1988.
17 Compendium to the Erasmus Course
3. Organisation and policy 1993
3.1 Research policy
Much of the research at the NRCCL is funded by extramural sources. This has made it necessary to make up a research policy to defend the integrity of the institution. The two main points in this policy is first, that no contracts are accepted for issues in which the contractor is involved in a conflict, and second, that the resulting report should be published - no confidential research is carried out. In addition, a project memorandum is drawn up by the NRCCL defining the task - the contractor is rarely allowed to describe the project, but has - of course - to agree to the specification made by the NRCCL. In this way one has some degree of control of understanding the implications of the contract. Contracts are generally accepted for a fixed amount of which half is paid on the initiation of the project, and the rest on its conclusion. To manage the research, the NRCCL early adopted a policy of project management. Three research programs have been created - the oldest is NORIS, organising studies in legal information and decision support systems, including user research, projects relating to knowledge based technology etc. In short, NORIS organises the activities in legal technology. The youngest is MERETE, which organises all projects in media and telecommunication law. And the rest of the substantive legal issues are organised within TERESA, including data protection, intellectual property law, computer related crime, electronic data interchange, etc. For each project, a project manager is named, a budget is specified and a termination date indicated. The project name is generally used for reference in accounting and filing, and project management is not very formalised. Public Administration and Information Technology
3.2 Teaching
The law study in Oslo is of 5 years duration (in principle), there being one exam after each year. Currently, there are approximatly 6,500 law students at the Faculty, straining the resources beyond what is comfortable. For third and fifth year exams, the NRCCL offers to optional subjects, one in "information law" (mainly contracting law and intellectual property protection), and one in "legal informatics" (data protection, legal information systems and communication processes). Annually, approximatly 30 students take exams. In addition, "legal informatics" are offered to foreign students - especially within the framework of the ERASMUS program of the European communities. For all law students, the NRCCL organises classes in the use of legal information services in general, and the national service of the Lawdata foundation in particular. I 1992, approximatly 450 students were given this course. For the fifth year exam, the students may replace the optional subject with a written report. The NRCCL has encouraged students to combine this possibility with a research project, and has therefore a fair share of research students. The use and reliance on research students, often for quite ambitious projects, is perhaps characteristic of the NRCCL, reflecting the importance placed on integrating students in the research community. Recently, the Council of the University of Oslo has accepted to establish an interfacultary Master's program. Students are admitted if they have certain exams in social sciences, law and informatics. The program is of two year's duration, and will also be of an interdisciplinary nature. A Chair has been created, and a professor will be appointed within the next 6-9 months. This Master's program will be under the administration of the NRCCL, and will make research in "legal technology" more prominent at the NRCCL. One has, for instance, great
19 Compendium to the Erasmus Course expectations regarding the experimental laboratory that will be established.
3.3 International contacts
The NRCCL has emphasised international contacts. The geographical location of Norway makes it necessary to plan for such contacts, they will not occur by chance. Partly this is implemented by participation in conferences, invitation of guest lecturers, etc. But three other strategies may be mentioned. First, different networks of institutions for computers and law. The NRCCL participate in several such networks. The Nordic countries organise an annual conference, this year's conference will be in Lillehammer, Norway, mid September. It strengthen the co-operation between those working with computers and law to take advantage of the fact that those speaking Danish, Norwegian or Swedish are able to understand each other without great difficulties, the languages are sufficiently similar. A European network - FIRILITE - of institutions is incorporated under Belgian law, and includes centres in Florence (Italy), Hannover (Germany), Montpellier (France), Namur (Belgium) along with Oslo. And a network established under the ERASMUS program of the European communities for the exchange of students and staff has been set up between Belfast (United Kingdom), Milan (Italy), Rotterdam (Holland), Tübingen (Germany), Zaragoza (Spain) and Oslo. There are also bi-lateral agreements with Southampton (United Kingdom) and Bologna (Italy). Second, the NRCCL is co-operating in a number of international research projects (three funded by the ESPRIT program of the European Communities), and accepts contracts from international organisations. Third, members of the staff participate in the work of international organisations as consultants or as Norwegian Public Administration and Information Technology
official delegates, participate in editorial panels for journals (more than a dozen on three different continents), etc.
3.4 Organisation and site
The NRCCL is a department of the Faculty of Law, headed by a Chairman (Jon Bing) and a deputy (Olav Torvund). Recently to sections have been created, Section for System Development and Organisation (under administration by Dag Wiese Schartum) and Section for Media and Telecommunication law (under administration by Beate Jacobsen). The staff annually is approximatly 20 researchers (including contracted research students and part time researchers). The NRCCL is physically located some 20 minutes by foot away from the Faculty in the Center of Oslo. It occupies three floors in an elderly villa. The ground floor includes student facilities like reading and terminal rooms. A special facility for students is maintained by the NRCCL at the central site of the Faculty (JURITEKET), and annually approximatly 20 students are hired as supervisors and instructors for this facility. A library of computers and law is maintained. It is not complete, but is often visited by foreign researchers as it probably is one of the more comprehensive collections available. The NRCCL publishes the report series CompLex (Computare and Lex), currently in co-operation with the publishing house TANO. The first report in this series (succeeding an even older series) was published in 1981, till date more than 100 reports have been issued, about one fifth in English. With the Danish Association for Computers and Law, the NRCCL also edits the quarterly Lov&data, published by the Lawdata foundation. This reports on regulatory reform, cases etc. The NRCCL co-operates closely with the Norwegian Association for Computers and Law, for instance for the annual
21 Compendium to the Erasmus Course
Fora on Computers and Law and other national events of social and professional nature. Public Administration and Information Technology
Data Protection in Norway
Jon Bing
1. Data Protection Regulatory System
/./. Privacy1 in Norwegian law prior to the data protection legislation
Norway is a rather young democracy. In 1814 the long union with Denmark was dissolved and a new constitution adopted. One of the provisions of this constitution was to develop a criminal and civil code. The civil code was never really realised, but a general criminal code was being developed throughout the century and finally adopted in 1902. A central figure in this reform process was the prominent lawyer Bernhard Getz, and at June 29, 1889, the reform process took another step forward when one of the older codes (of 1842) was amended.2 This amendment included the introduction of a provision prohibiting
1 In this paper, the term "privacy" is used to denote the traditional set of rules protecting the publication of intimate personal data, while the term "data protection" is used for the protection based on the data protection act and related regulation. This distinction roughly corresponds to the Norwegian distinction made between "personlighetsvern" and "personvern". 2 Cf Otto Mejlamder Den norske Straffelov, Mailing, Kristiania 1889:59-61
23 Compendium to the Erasmus Course the publication of information3 relating to the "personal or domestic affairs". One may like to reflect on the timing. It was nine years after the milestone paper by Warren and Brandeis,4 which cause probably was the publication in the newspaper Saturday Evening Gazette of details from the marriage of Warren's daughter. The Norwegian reform seems to be of a related nature. The reason for its introduction by Getz is not known, but he probably was inspired by Denmark, where a similar reform took place at the same time. This may be the reaction of legal systems to the dramatic developments in information technology taking place in the latter part of last century. The invention of the rotary press in 1863 made cheap newspapers possible, and society gossip was popular then as now. This created a market for the trade in personal information, causing instances of the invasion of privacy. The legal systems reacted - the US example is famous, but also in Norway a protection of privacy in criminal law was established. This provision - the criminal act sect 390 - remained the major regulatory measure until the enactment of the data protection legislation in 1978, though one should, of course, appreciate that the numerous rules of confidentiality also are part of the traditional legal regime. However, the Norwegian privacy law also has a non-statutory component. Norwegian law will allow the court to apply civil law based on unwritten principles, without authority in statutes. This is rarely done, and a major example is of interest in our context.
3 In this paper, the terms "data" and "information" are used loosely as synonyms, and not according to definitions like those common in computer science. 4 "The right to privacy". Harvard Law Review 1880:193. Public Administration and Information Technology
In 1926, two men upset Norway by brutally slaying a local chief of police and other policemen by axe. The two men - one older and one younger - were caught and imprisoned, the older taking his own life while imprisoned. The incident stimulated the well-known author Gunnar Larsen to write a documentary novel (To mistenkelige personer, or: Two suspicious persons) which was published in 1926, and was very well received. After the war, in 1949, Tancered Ibsen, a pioneer in Norwegian movie industry and the grandson of the dramatist Henrik Ibsen, directed a film based on this novel. Before its opening, the youngest of the two "suspicious persons", protested. He had in the meantime been released from prison and established himself under a new name, married and had children. He feared that the release of the film would once more make his old crime topical, that he would be identified, and his new life shattered. In its decision,5 the Supreme Court found that non-statutory principles of privacy had been infringed, prohibiting the public showing of the film, and emphasising the special nature of film as a medium:6
"The film communicates to a larger number of persons than for instance a printed publication, and is very vivid in its portrayal. It will be seen, discussed and made subject to reviews in the newspapers. By this one will have to assume that details from the real event will be repeated in papers and illustrated journals, and it cannot be avoided that identity of [the younger criminal] - in spite of his changed name - will be unveiled for persons to which it formerly was unknown. Also among those who had prior knowledge of [his] relation with the murders of the police
5 Rt 1952:1217. 6 The spokesman for the Court, Judge Qvigstad, at Rt 1952:1220-1221. The translation is the author’s own.
25 Compendium to the Erasmus Course
officers, but who has suppressed this knowledge from their consciousness, will recall their memories of that conviction, and its characteristic of [him] will once more be brought alive and result in reflections unfavourable to him."7
Again one may reflect on the curious parallel to the earlier Red Kimono Case in the US8 in which Gabrielle Darley, a prostitute, had been prosecuted in a sensational murder trial, but had been acquitted. She went on changing her name, marrying and settling in a town where she was not known. Seven years later, the film The Red Kimono based on the trial, unveiled her identity and destroyed her new existence.9 Again the two cases may be taken as indications of the law reacting to the potential of information technology - this time represented as the popular movies - to invade the privacy of individuals. Norwegian law had at the end of the 1960's therefore two may components in its law on privacy, the prohibition in the criminal code against publishing private or domestic information, and the hovering existence of an unwritten privacy protection the extent of which was only glimpsed through a few court decisions, the most important being the first which has been mentioned above. However, one should include a few more background facts on Norwegian law to appreciate the legal context into which the data protection legislation was introduced.
7 In the citation, the name of the criminal has been replaced by neutral indicators in square brackets. 8 Melvin v Reid, 112 Cal App 285, 197 Pac 91 ( 1931) 9 This decision was based on an obscure provision of the Californian constitution stating that all individuals had a right to "pursuing and obtaining happiness" - it has later been repealed. Public Administration and Information Technology
There are, as indicated, a large number of provisions on confidentiality, they being designed as a prohibition to give information of a specific nature to a third party, exceptions are, however, made with respect to certain third parties, which again are made subject to confidentiality obligations. In this way, a large system of "secure boxes" are created, but there are holes punched in them to channel relevant data into other secure boxes. For instance, the primary medical doctor is subject to rules on confidentiality, but certain patient data are to be communicated to the health authorities for statistical, research or planning purposes. These authorities are subject to other, though in detail somewhat different, rules on confidentiality. The resulting system is complex, though not unique for Norway it is slightly dissatisfying. There is a well-developed law of administrative procedure10 that includes confidentiality clauses, and also an access right for parties to a case processed by the administration, limited to the documents of the case. There is further a freedom of information act11 that give access to anyone to documents in a certain case, the case has to be identified either through the index of the authority in question or other means. This implies that the Norwegian freedom of information legislation is quite more limited than, for instance, the Swedish. It may be argued that these two acts are somewhat related in nature to the data protection legislation, indicating that in Norway data protection is more a matter of administrative procedure and due process than human rights. Finally, it may be appropriate to mention that the Norwegian public administration at the end of the 1960's was introducing computerised system at a large scale. A social benefit system had been created in 1967 that pivoted on a number of large
10 Act of February 10. 1967 - "Forvaltningsloven". 11 Act of June 19, 1970 - "Offentlighetsloven".
27 Compendium lo the Erasmus Course computerised files that included information on all individuals in Norway. Already a Central Personal Register was in existence, and a unique personal identity number (fødselsnummer) has been introduced in 1964 without any real political debate, the reform actually only appeared before the parliament as an explanatory note in the budget of the Central Bureau of Statistics. The main justification was not computerisation, but the complaint from employers who had to keep track of different reference numbers for the reports they were required to make to the tax authorities, the social benefit administration, the employment administration, etc. It should be appreciated that Norway is a small country (4 million people), without any real subdivision into regional or local jurisdictions (though this level issues some regulations, and there are considerable regional differences in this inconveniently long and thin strip of land at the edge of the European continent). At the end of the 1960's it was quite modern, and the rather recent major reforms in the law of administrative procedure and freedom of information emphasised legal issues related to them.
1.2. The development of the data protection legislation
The Council for the use of computers in government administration12 approached summer 1970 the Department of Civil Law at the Faculty of Law, University of Oslo with a request to undertake a study of "privacy and governmental data banks", which an approximate translation of that somewhat dated wording would read. At this Department, professor Knut S Selmer had taken the initiative to an activity exploring issues of
12 Rådet for databehandling i staten. Public Administration and Information Technology computers and law.13 A research contract was negotiated, and it became the occasion of formally establishing the Norwegian Research Center for Computers and Law from January 1, 1971.14 Consequently, the early history of data protection in Norway is also to a large extent the history of the budding research activity of the NRCCL. This research contract resulted in a monograph,15 exploring the existing literature - especially the works of Westin and Miller - and examining the use of computerised data processing within the Norwegian public administration. Before this monograph has been published, two committees had been named by the Cabinet to report to the government on possible legislative action. The first was charged to explore problems related to credit reporting, a problem that was not acute in Norway at that time, but which was much discussed due to the US initiative to legislate - and it was felt (quite rightly) that the issue would become more current also in Norway as credit cards and similar payment instruments became more popular. The chairman of this committee was Tore Sandvik, a professor of law at the University of Bergen. After its establishment, the mandate was broadened to include all data protection issues in the private sector. The committee reported in 1974.16
13 At this time, the author of this paper was his research assistant, and was doing much of the practical work - professor Selmer at this time being both dean of the Faculty and Director of the Department. 14 This was funded as a sub-department of the Department of Civil Law, but was in 1981 made into a full department of the Faculty. 15 Erik Samuelsen Statlige databanker og personlighetsvern. Norwegian University Press 1972. 16 NOU 1974:22 Persondata og personvern. One may note that this is the first time the Norwegian equivalent of "data protection" - "personvern" - is used in a written publication. The term itself was coined by professor Selmer in a talk on the subject.
29 Compendium to the Erasmus Course
In 1972, a second committee was created to explore the issues related to data protection in the public sector. The reason for the two committees should be understood on the background of the research being carried out within the NRCCL, this concerned the public sector, and the committee was not named before it could take advantage of the material produced within the project. The chairman of this committee was Helge Seip, at that time secretary to the presidium of the Nordic Council, an intergovernmental agency of the Nordic countries set up to facilitate co-operation. He had, however, a broad experience that included experience as a minister in one conservative coalition government and president of the liberal party (Venstre). His committee reported in 1975.17 One may note that at the time of the reports, the first legislation in Europe had been enacted. Obviously, the main impact in Norway was of the Swedish pioneering act of 1973, which was enacted before any of the committees reported. There was considerable pressure to adopt an act, but the government was hesitating. The main reason, if would seem, was the choice between a licensing system (as indeed became the result) and a system mainly based on direct regulation in substantial law. Also, though the two committees had co-ordinated their work prior to reporting, there were actually two proposed acts, which had to be co-ordinated. The government bill was published in 1978.18 Under the Norwegian system, the Government bill is communicated to the Parliament, which splits in two chambers for the purpose of adopting legislation. The bill was adopted without major amendments and with full support of the different parties, and became the data protection act of June 9th, 1978.19 Two years
17 NOU 1975:10 Offentlige persondatasystem og personvern, 18 Ot prp nr 2 (1977-78) Om lov om personregistre mm 19 Lov om personregistre mm, 1978:48 Public Administration and Information Technology went by, however, before the act entered into force, mainly due to the efforts necessary to set up the administrative body stipulated by the law, the Data Inspectorate. The act became effective of January 1, 1980. Regulations were issued under the authority of the act by the Ministry of Justice 1979, and enhanced March 10, 1981. A rather modest amendment of the act was made in 1987.20 A new review was made at the 10th anniversary of the Data Inspectorate, this time by the Ministry of Justice (though they were also assisted by a former deputy director, Eirik Dj0nne).21 This was discussed in parliament spring 1992, and several amendments were suggested to the statute - among them the introduction of strict liability and general damages for errors in public registers. It is not believed, however, that a general review will take place until the directive of the European Communities on data protection also is available in its final form.22 One may also mention that in the statutory revision of 1987, a Sect 8b was introduced, giving authority to issue regulations on data security. The Data Inspectorate had originally some limited power to require data security measures in the licences or regulations issued under Sect 8 (see below), but no general authority to require security measures for personal register. The work has been co-ordinated with the revision of the internal government instructions regarding classified documents, both such classifications required by civil considerations (like trade secrets) and those required by military considerations (like national security issues. This has not been a trivial task, as there
20 Act of June 12, 1987:55. 21 St meld nr 43 ( 1990-91 ) Om personvern - erfaringer og utfordringer og om Datatilsynets årsmelding for 1990, Ministry of Justice, Oslo 1991. 22 At the time this paper is written, Norway has just accepted the European Economic Space Agreement, and through this the directive will be binding and require amendments of the national legislation.
31 Compendium to the Erasmus Course are rather different interests to take into account, and co ordination between several ministries. Autumn 1992 a draft bill for both a common regulation and a new statute on information security is to be adopted. It was felt that existing statutory authority was not sufficient for the new and rather comprehensive regulation, therefore the data protection act Sect 8b may have a rather brief life, and actually be repealed before it really has come into effect.
1.3. The data protection concept: A decision oriented interest model
It will be apparent from the brief sketch of the legislative history that research was paralleling the development of the government bill. This research did partly have as its objective to arrive at a better understanding of the concept of data protection by analysing systems for the processing of personal information and the interests of the data subject at stake.23 Today, through contributions of several persons, data protection is understood as a bundle of interests. The presentation in this paper represents the author's own preferred view, but it will only differ in details from what would seem to be some sort of consensus in Norwegian legal theory.
23 Important is a study of the Directorate of seamen which co-ordinated all administrative activities relating to persons employed in the Norwegian merchant navy. Cf Ragnar Dag Blekeli Personvern og offentlig forvaltning. Papers on Computers and Law 11/1975, Norwegian Research Center for Computers and Law, Oslo, 1975. The Directorate has no been phased out. Also the early articles of Knut S Selmer were influential, for instance "Elektronisk databehandling som verktøy i offentlig og privat administrasjon", Samtiden 1976:85, 77-83. The anthology of Ragnar Dag Blekeli and Knut S Selmer (eds) Data og personvern, Norwegian University Press, Oslo 1977 summed up the understanding of data protection at the time the government bill was drafted. Public Administration and Information Technology
The basic premise of the concept is that data protection is related to decisions, and through them, to power. Someone has power with respect to another if that someone may, through his or her decisions, influence the welfare, health or prosperity of another person. In this, data protection becomes an interest related to safeguards to the rule of law or due process, the law of the procedure of public administration being seen as protecting some of the same interests.24 This bundle of interest can be characterised as the interest to control the flow of personal information relating to the data subject. These include control of the primary collection of data (whether by observation, inspection or by submission of the data subject), the use the primary recipients make of the data, its storage by the primary recipient and the communication onwards to a third party. This third party, being the secondary recipient of the data, will then be subject to an interest in exercising an identical control. And so on through the chain of recipients. The control is mainly related to the use made of the data in decisions - though "decisions" is used as a rather inclusive concept, and also for informal acts like "making up one's mind" or otherwise reacting on the basis of the data. The point is that the mere existence of the data is of little cause for concern: Data collected not to be used for individual decisions of any kind (as in statistical surveys) are not very important to the data protection issue, they only represent a potential for mis-use through unauthorised access, etc. The general characterisation can be broken down into three rather more specific interests. First, the individual has an interest in confidentiality. This is the interest in personal data not being collected through observation or inspection without his or her consent, and that
24 One may not exclude that this is a bias inherited from the emphasis on data protection in public administration in the first studies.
33 Compendium to the Erasmus Course also any submission of personal information should be voluntarily. It extends to an interest in control of the use made of the information by the primary recipient, and the data security relating to storage or erasure. It also obviously relates communication from this primary recipient to any third parties. One may note that this "flow control" usually is achieved through the many regulations governing confidentiality of different professional groups, public servants, etc. This traditional regulation really is some sort of "stop-go"-regulation: The confidentiality of the primary recipient is determined by a certain legal regime, but this regime also contains rules penetrating the walls of confidentiality with obligation for communication: For instance, a medical doctor may be obliged to communicate information on certain diseases to specified authorities. The secondary recipient will in the same way be enclosed by walls of confidentiality, but again there may be communication apertures, and these may be different from those opened in the walls of cell of the primary recipient. In this way a rather complex system has been created by conventional regulation, and at least in Norway, none has a clear picture of the system as a whole. Data protection legislation plays a complementary role to this conventional body of regulation. The second interest is the interest in having adequate data for making a decision. This interest may be further broken down into the principle of relevance and the principle of adequacy. The principle of relevance is the obvious interest a data subject has in avoiding information to be taken on the basis of irrelevant information. The information may be irrelevant because it is erroneous or dated - and actually conventional regulations of defame to some extent already protect this interest. The information may also be irrelevant because there is no justified relation between the information and the objective of Public Administration and Information Technology the decision: This is a very viable interest, defended in Norway for instance by making sex or political opinions as relevant criterion for job selection. The principle of adequacy is something further: Though a certain fact may be relevant, it may be misleading without being interpreted in relation to another fact. Omissions may, of course, result in misleading information. And the data subject has an interest in a decision relating to himself or herself taking all types of relevant data into consideration.25 In this aspect, the data protection concept reveals its relation to the law governing the procedure before the courts or in public administration. The courts will often have a responsibility to make sure that all relevant evidence has been presented, and the public administration likewise will be responsible to make its decision after satisfying itself that the facts and circumstances of the case are known. Data protection legislation will again supplement the more traditional regulation. Finally, there is the interest in access of the data subject to the information relating to himself or herself. This has for many jurisdictions been perceived as the major objective for data protection legislation. But again this has to be considered in the context of the national law. In Norway, there already exist two major pieces of access legislation briefly mentioned above. The first is the act of procedure for public administration,26 which in its sect 18 gives any party to a case within public administration access to the "documents" of that case. The second is the freedom of
25 One will appreciate that with respect to a certain data subject, that data subject may will profit from certain facts being omitted that in a typical situation would be considered relevant. Therefore the emphasis on types of data in the text: We are considering, one might say, an abstract or ideal decision. 26 Forvaltningsloven, February 10, 1967.
35 Compendium to the Erasmus Course information act27 that gives anybody access to the "documents" of a certain case. This implies that the person requesting access must be able to identify the case. This may be done by outside sources (like a newspaper report), but may also be done through the journal (the file index) of the authority in question, and this journal is public without restrictions. One will note that both these acts rely on the notion of "document", which is defined in the freedom of information act sect 3 and pursuant regulations of December 19, 1986:2202. The regulations extend the conventional document concept to an analogous concept for computerised systems. It may not formally be a completely satisfactory solution, but it makes these access laws also applicable to computerised systems. One may also note that the right to access one's own file is contained in non-statutory basic principles of law. This was the decision of the Supreme Court in a case from the late 1970's28 where a patient requested access to his own patient journal at a hospital in order to decide whether he should sue the hospital for malpractice. The hospital denied the patient access, but the Supreme Court decided that though there was no statutory access right, this was implied by non-statutory principles. Since then the relevant legislation has been amended and contains today an explicit access right.29 These interests are all related to the individual. During the 1980's, the legal theory has developed three additional interests that also are seen as part of or related to data protection, but which concern not the individual as such, but the individual as related to a group. The first is the interest in controlling the surveillance level in society. It is especially the use of data base surveillance
27 Offentlighetsloven, June 19,1970:63. 28 Rt 1977:1035. 29 Act on medical doctors (legeloven) of 13.6.1980:42 sect 46. Public Administration and Information Technology
techniques30 that has triggered off this interest. The example may be that a convict population with respect to a certain crime is analysed, and certain characteristics having a statistical significant co-occurrence with their crime are identified. Then the same characteristics are used to combing through existing data bases, isolating a "suspect population", which then is made subject to increased surveillance. In this population, the majority probably are not actually guilty of the crime under investigation, but only victims to circumstances. The crime may be tax evasion, social benefit fraud, drug trafficking, terrorism etc, or there may be similar methods used for identifying probable AIDS victims, child molesters, etc. The second in the interest in a robust society. This is the reverse side of the issue of a "vulnerable society" that has been a political issue in the Nordic countries due to the dependency on technology, and the social consequences failure may have. One may mention one special historical episode that has formed the Norwegian attitude to this issue. Not only is there a concern for technology to fail and causing social disturbance, there is also a concern that the technology will aid unlawful elements if it cannot be destroyed in wartime or in a similar national crisis. The background is an episode in the Second World War, when Norway was occupied by Germany. In May 1944, the Germans wanted to mobilise Norwegian youth for "labour service", and three age groups were to be mobilised. There was reason to fear that they would be sent to the Eastern front. There were two alphabetical tabulators in the country that could be used to draft the youth, both by Watson Norsk A/S (better known today as IBM). The data was with the Nazi authorities, but they needed the machines to select the relevant persons. The resistance movement broke into the offices of both Watson and the insurance company operating the alternative
10 What in German is called "Rasterfhandung".
37 Compendium to the Erasmus Course machine (Norske Folk) and set off explosives.31 This episode has made a lasting impression in Norwegian politics, making both surveillance and vulnerability a sensitive issue. The third of these collective interests is the interest in a friendly administration. This is seen as related to the computerisation of public administration, which in its early stages was related a tendency to cryptic codes in form letters, use of punched cards for requesting or applying for certain services, etc. There still is a concern for the introduction of computers to increase the possibilities of the public to communication with the administration (both public and private) rather than to reduce the availability. The Data Inspectorate has often indicated that this has been a concern in addition to data protection understood more strictly. These six interests characterise the data protection concept generally accepted in Norway today. They represent an attempt to make operative the data protection concept, not an exhaustive definition. Especially one has in the last years emphasised that one should not forget that data protection also has a core of privacy (as the two terms are used in this paper): An individual's perception of himself or herself as an autonomous person in a democratic society presumes that the individual may control who get access to personal information, especially of an intimate or sensitive nature. If this control is reduced by pragmatic reasons, this may over time influence this perception, with adverse effects for society. Finally, one may question whether this theoretical discussion is reflected in the legislation itself. The Norwegian data protection act does not - define the concept. But there are several indications of the act operating with a decision oriented interest concept based on the three individual interests. Sect 10
31 Aktuelt 271985:19. Public Administration and Information Technology
prescribes, for instance, a balancing of interests - data protection against the objective served by a registering - in the licensing process. And Sect 7 limits access to registers that do not support individual decisions. These two examples will have to suffice in stating that the legislation reflects the decision oriented interest concepts, and should be interpreted on this background.
1.4. The Data Inspectorate
The Data Inspectorate is established according to Sect 2. It is an independent public authority according to the Norwegian tradition. This implies that is part of the public administration, its budgets are passed up through the Ministry of Justice. The Ministry of Justice also is empowered under the act to issue regulation. The Ministry cannot, however, instruct the Inspectorate with respect to individual cases. The Inspectorate is headed by a director. The first director to be named was Helge Seip, chairman of one of the committees preceding the act. He was succeeded in 1989 by Georg Apenes, a lawyer and former member of parliament for the conservative party (Høyre). The administration in addition consists of lawyers and a few computer scientists, altogether (1992) approximately 15 persons. Its address is Data Inspectorate, PO Box 8177 Dep, N-0034 Oslo. The Inspectorate is governed by a board named by the King in Council. This board has seven members. They are appointed not as representatives of political parties or interest groups, but in their personal capacity - with the exception of there being one member appointed in consultation with the organisation of employers, and one member appointed in consultation with the organisation of employees. As chairman of the board was named professor Knut S Selmer, who at that time was the chairman also of the Norwegian Research Center for Computers and Law, and who had played a major role in the research paralleling the
39 Compendium to the Erasmus Course development of the legislation. When he left his position at the NRCCL in 1990, he retained his position at the board, and consequently has functioned continuously from the creation of the Inspectorate. With only two directors and one chairman of the board, there has been stability in the development of a policy within the Data Inspectorate. The style of the Inspectorate is also to seek contact with applicants and others in order to find consensus for a certain policy. However, the Inspectorate actively attempt to identify issues that are assessed to need a general policy decision, which through the appeal process (cf below) is passed on to a general political level. It is seen as proper that such controversies should be settled by the general policy makers as other major political conflicts of interests. In adopting this view, the Inspectorate avoids taking an adversary position in the appeals - the Inspectorate seeks clarification rather than approval. The Data Inspectorate has, according to the legislation, several major tasks, cf Sect 3 and title 4. It is charged with decision making authority in individual cases, especially on licensing of registers, but also in disputes with respect to access and correction of data in registers. Such individual decisions may be appealed to the Ministry of Justice. In cases whether the Ministry itself is applicant, the King in Council is the appeal authority. It has been questioned whether it is appropriate that a Ministry that itself is responsible for major sensitive registers, for instance related to the police and crime, should be the appeal authority, and this remains an issue of some concern. It is also charged with authority to control and inspect registers in order to ascertain that they are operated according to the provisions in the act, the regulations or the licence. And the Inspectorate is charged to comment on data protection aspects of pending legislation, reviewing the Public Administration and Information Technology technological development, inform the public and take initiatives to promote a data protection policy. These aspects have been emphasised by the Inspectorate in the last few years. In order to strengthen the independence of the Inspectorate, and stimulate the general data protection discussion, the annual report of the Inspectorate takes the form of a statement to the Parliament (Stortingsmelding), submitted through the Ministry of Justice. The annual report for 199032 also contains an assessment of the first ten years of practice, and policy suggestions for further development. The discussion of this statement in the parliament was of major importance.33
2. Some basic concepts and general structure
The data protection legislation is based on two basic concepts, the concept of "personal data" and the concept of a "register". Similar concepts are found in other national acts and in international legal instruments, but they differ in detail. A discussion of these concepts is, therefore, justified.
2.1. Personal data
"Personal data" is defined in Sect 1 of the act as "data or assessments which directly or indirectly can be related to individuals, corporations or foundations that can be identified". Initially one may observe that the definition is conventionally based on the distinction between anonymous and nominal data. We know that this is not a strict distinction. One may see personal data as consisting of an identifier and some related data.
32 St meld nr 43 (1990-91) for 1990. 33 A review of the discussion can be found in Lov&dala 31/1992:4-6.
41 Compendium to the Erasmus Course
Fig 1 - Personal data
identifyer relateddataj Personal data The identifier - which commonly is a name, a personal number or some system-related PIN - is in itself personal data. This is rather clearly demonstrated by the unique PIN assigned each Norwegian, the "fødselsnummer" (see above). This is a 11 digit number. The six first digits represent the date of birth (this explains the name of the number, literally "birth number") as day-month-year. The next group of three digits indicates the sex (even numbers for females, odd numbers for males), the century in which the person is born (to distinguish new born children from those more than a hundred years old), and a serial number within each day. The last group of two digits are controls - the first digit is calculated according to an algorithm using the succeeding 9 numbers, while the 11th digit is the result of a calculation also including the first control digit. The anatomy may be indicated as below, using an authentic PIN:
Fig 2 - Norwegian PIN
sex CQntrol date of birth century .digits sequential number Public Administration and Information Technology
One will appreciate that the "related data" itself does not have otherwise to be related to a person, or have any "private" nature. For instance will all information of an identified property - its size, the vegetation, what houses are built on the land etc - be "personal data" due to its relation to the owner through the land register (which is open to the public). One should bear this in mind when one consider the next aspect of the definition, which certainly is the more controversial: The inclusion of information on "corporations or foundations". This phrase is in the legislative prose intended to include all and any legal person. This implies that according to the statute, data on a legal person is also "personal data". The rate of shares in IBM is consequently an example of personal data, or the area of square meters offered in the company's new premises just outside Oslo. The examples are mentioned mainly to emphasise that the notion of "personal data" is a much wider concept according to the law than its natural language implications, and that this obviously will have some consequences for legal policy issues related to the legislation. The inclusion of data on legal persons is mainly practical. In Norway, a large number of businesses are privately owned and part of the owner's property. It is therefore in practice difficult to determine whether data actually is related to a legal or physical person. This was the major justification for not making the distinction at all. At the same time, it was considered to restrict the concept of "personal data" to that which is private in nature, but again it was maintained that it would in practice be difficult to distinguish between what is private and what is not.34 It is stated, however, that data on legal or physical persons should not be treated equally, as the data protection interests are of different strengths.
34 Cf Ot prp nr I ( 1977-78^ Om lov om personregistre m m , Oslo 1978:25-26.
43 Compendium to the Erasmus Course
This aspect of the Norwegian law has been severely criticised initially when the legislation was adopted, not least by commentators in the United States. In practice, however, there have been few problems.35 Several other countries include some protection of legal persons in their data protection legislation, but few do this by the way of a statutory definition of "personal data”, and it can be queried whether this is the most appropriate solution. The definition mentions two alternatives for data - "data or assessments" ("opplysninger og vurderinger"). This is probably related to a distinction made in the act of procedure in public administration when somebody requests access to documents in his or her own case, a distinction which through a citation of this act in the data protection act Sect 7 also is relevant for the access rights according to these provisions. This somewhat subtle point will be discussed with respect to the access rights (cf below). One should finally mention that the law operates with two general categories of personal data. According to the act of procedure in public administration Sect 13, a general obligation of confidentiality is imposed on all public servants with respect to personal aspects ("personlige forhold"). According to Sect 13(2), some trivial data are excluded from this protection - place and date of birth, PIN, citizenship, martial status, occupation, domicile and place of employment if such information does not imply some further private information (like the fact that the address of the domicile is that of a prison or a hospital). These trivial personal data are, of course, personal data in the meaning of the data protection legislation, but are not subject to confidentiality measures in public administration. However, in a licence the Data Inspectorate may impose restrictive measures for such trivial
35 Cf Eirik Djønne, Tove Grønn and Tor Hafli Personregisterloven med kommentarer, Tano, Oslo 1987:28-29. Public Administration and Information Technology
data contained in a personal register, and will typically restrict communication of also this trivial data from the register to a third party. According to the data protection legislation, a sub-set of personal data is qualified as "sensitive data" (Sect 6(2) - the catalogue is repeated in Sect 9, 16, and 26). Five categories are mentioned:
(1) Data on race, political or religious beliefs (2) Data related to criminal justice (3) Data related to health or the abuse of alcohol or drugs (4) Data on sexual aspects (5) Data on family relation, excepting kinship, economic arrangements between spouses, and family responsibilities.
These five categories are used to trigger certain of the provisions in the legislation, and to ensure special protection of data that is perceived as generally of a sensitive nature.
2.2. PersonaI register
A major issue of legal policy in the 1970's was how to qualify in which situations data protection legislation should apply. It was rather evident that it could not apply to any use of personal data. The computer industry argued that though the computer might be the occasion for the legislative concern, it was not the cause of this concern, which were similarly caused by manual systems. Many national statutes constructed a notion of a "system for
45 Compendium to the Erasmus Course personal data processing", a "file" or some similar concept.36 The Norwegian solution is the notion of a "personal register". This is defined in Sect 1(2) as a "register, list etc where personal data are stored in such a way that data on any individual can be retrieved". The criterion used is that of retrievability, and in this way the Norwegian concept is closely related to the notion of a "system of records" as defined in the US Privacy Act of 1973 (Sect 3(a)(5)). In explaining the concept, the legislative history takes as an example a system for filing letters. If the letters are filed according to a subject index, then the resulting file is not a "register", as one would have to search sequentially through the file to retrieve data on a certain person. If, however, the letters are filed alphabetically according to the name of the addresses, then it would be a "personal register".37 This works fine for manual systems, but with respect to a computerised system, it is difficult to find examples where the test of retrievability does not apply. Taking the example of the file of letters resulting from word processing, one may easily see examples where the document name used by the system contains a subject index, or contains the name of the addressee. In the first case, letters containing data on a certain addressee cannot be retrieved by searching the index of document names. But it would be trivial, of course, to use a universal search function that would search access documents. Such a function is incorporated in any moderately powerful word processing system, but is also available under any operating system: Under MS-DOS this is the command "find".
36 Cf Jon Bing "'Personal Data System': A Comparative Perspective on a Basic Concept in Privacy Legislation", in Jon Bing and Knut S Selmer (eds) A Decade of Computers and Law, Norwegian University Press, Oslo 1980:72-91. 37 Cf Ot prp nr 2 (1977-78) Om lov om personregistre mm, Ministry of Justice, Oslo 1978:69. Public Administration and Information Technology
It has therefore been held that any computerised system containing personal data qualify as a "personal register" in the terms of the law. This also holds true if the system only potentially may be used to retrieve data on a specific person, but in practice is never used in this way.38 Initially, it has been suggested that the notion of a register should be limited to "one physical unit".39 It has, however, in practice often been found necessary to apply a logical register concept. Also, in a computerised system, there usually will be more than one physical file that constitutes what from the user's point of view is one system. There may also be such an intimate functional relation between a computerised and a manual file that they have to be considered as one register. The notion of a "register" implies that more than one data subject are included. But there is not any strict lower threshold. The retrievability test generally also requires the data to be organised in some way that aid the retrieval, but this is not always necessary. If the number of data subjects is low (but still sufficiently high to constitute a register), then the data on any individual may easily be located though the data is not organised to facilitate retrieval.
This was first discussed with respect to the infamous "beer registry case", purchases of beer were registered, and where the Data Inspectorate indicated that if the number of names was low, it would be a personal register even if the names were entered into the register according to the time the purchase was made.40
38 This is unlike how the retrievability test is applied in the US law. 39 Cf Eirik Djønne, Tove Grønn and Tor Hafli Personregisterloven med kommentarer, Tano, Oslo 1987:29-30. 40 Cf Eirik Djønne, Tove Grønn and Tor Hafli (eds) Personregisterloven med kommentarer, Tano, Oslo 1987:30-31.
47 Compendium to the Erasmus Course
It has also been argued that the retrievability test applied to a video recording - for instance made by a surveillance camera in a bank - is a personal register. Normally there would not be an index associated with the recording,41 but it was argued that viewing the recording constituted a search method sufficiently effective to meet the retrievability test. In a Supreme Court decision of 199142 this was one of the issues. The court states that it has difficulties to find sustenance in the statute or its legislative history for the argument that the video recordings in question should constitute registers. Though this is an obiter dictum, it may be taken as rather conclusive for the qualification of video recordings where there has not been generated or established an auxiliary index. The register concept is, in principle, technology independent. But as we have illustrated, any computerised file containing personal data, will be a register. Combining this with the point made above on the inclusive nature of the concept of "personal data", we see that there hardly will be any computerised lile not being a "personal register". In principle, the inclusion of copyright notices and name of programmers in the code of a program is sufficient for this program formally to be a "personal register". In the Royal Decree of December 21, 1979, bringing the data protection act into operation, one category of registers was generally excluded by part 11(1) pursuant to Sect 1(4), these were "books, journals etc which according to the data protection act Sect 1 are or contain personal registers". In this way, one
41 Though in transaction oriented environments there may be generated a computer record of the transactions which include a time-stamp, and which then become a time encoded index to the persons recorded by the camera 42 Rt 1991:616, cf also Lov&data2S/\99\:8-9.
48 Public Administration and Information Technology
avoided that the act actually would apply for the control of the printed word. The register concept is, as its relatives in other national first generation data protection legislation, a child of the mainframe area. But the Norwegian law entered into force approximately at the same time as the personal computer was created. And this explosive development has distributed personal register to desktops, laptops and palmtops, challenging the basic legal policy of the data protection legislation. The register concept was designed to qualify situations where data protection issues were at stake, today it is much too inclusive.
2.3. The system: An review
Using the two basic concepts, the system of the data protection legislation can be roughly sketched into place. First, the legislation applies to both the private and the public sector. In the private sector, it is limited to business activities and activities of associations and foundations. In other words, the personal or family sphere is not governed by the law. Second, the law has three different systems. A set of rather general substantive rules applies to any personal register (cf title 3). Two sub-sets of personal registers are qualified to special attention: Those containing sensitive data or which are computerised. These registers must either be authorised in the regulations issued pursuant to the data protection act Sect 9(2), or must obtain a licence from the Data Inspectorate prior to their establishment. Exception is made for those registers established under statutory authority (Sect 41). This is rather logical: If the parliament has authorised the establishment of a register by statute, this should not be a subject to be reconsidered by the Data Inspectorate. The problem is, however, that many registers
49 Compendium to the Erasmus Course established prior the data protection legislation is related to some activity with which the agency in question has been charged in statute, but it is not clear whether the register established to support this activity also is authorised in the statute. Even when a register is authorised in statute, the Data Inspectorate may regulate the operation of the register as they would in a licence. Also, four types of businesses have been identified:
(1 )Credit reporting agencies (title 5) (2)Computer service bureaux (title 6) (3)Address brokers and direct marketing companies (title 7) (3)Opinion poll and market research companies (title 8)
The data protection act introduces a licensing scheme for these types of businesses - they have to obtain a licence from the Data Inspectorate before starting their operations. If they also need personal registers in their operations - and this will typically be the case - such registers will be subject to the more general statutory provisions, and typically need prior licensing. The Norwegian statute has, therefore, a double licensing scheme. Finally, a separate licensing scheme is introduced in title 9 for export of personal data - this will be dealt with in more detail below. Otherwise, this paper will mainly deal with the rules of establishing personal register, and the right of the data subjects with respect to such registers.
3. Establishing a personal register
3.1. The substantive basis
The establishment of any personal register has to comply with the substantive principle of Sect 6. This implements the principle of relevance mentioned above: The inclusion of Public Administration and Information Technology
personal data has to be justified in the administrative or business activity of the register operator.43 One will note that this is a modest principle. There is no authority to see a register as a violation of the law just because one perceives the business as objectionable. One may think private detective agencies, match-makers or other such operations that may occasionally be somewhat shady as less attractive, but they are, however, justified in creating personal registers supporting their lawful operation. The decision of whether a certain business activity should be permitted in Norway is not one that has been delegated to the data protection authorities. It is a modest principle, but it has nevertheless a substantive content. It applies to all registers, also to those which are operated according to regulations or licences issued under Sect 9, the relevance principle of Sect 6 in such cases constitutes a limitation of the authority delegated to the Data Inspectorate: The Inspectorate is not permitted to grant licence for the inclusion of data not justified by the objectives of the activity in question. It is a partial implementation of the interest in adequacy. It would not be advisable to implement the principle of adequacy in full: Decision will not typically be made on the basis of the registered data alone, but will be supplemented by data from case files, etc. One may, however, see the right of corrections on behalf of the data subject as an expression of the same interest. The personal registers to which this substantive basic principle applies are (1 ) manual registers containing (2) no
43 There is only a formal "controller of the file" where such has been appointed in a license. Otherwise the "operator" will be the person responsible according to the law of the type of organisation applicable, whether limited company, charitable foundation, or public authority.
51 Compendium to the Erasmus Course sensitive data. These are only governed by the substantive provisions of the statute.
3.2. Regulated registers
All other registers are in principle governed by regulations or license, these are (1) registers that are computerised, or (2) registers containing sensitive data, manual as well as computerised. Sect 9 requires that such registers are only established according to a prior licence, but according to sect 9(2), there may be issued regulation excepting certain types of registers from the requirement to obtain prior license. The regulations define such registers according to their purpose, and the following types of registers are governed by the regulations:
• Membership registers of associations • Registers of customers, subscribers or suppliers • A banks register of customers • The registers of life insurance companies relating to insurance or loans • Register of tenants • Register for distribution of publications • The registers of a library relating to indexes to their books and other materials, their lenders and what on loan • The manual files of lawyers • The manual files of medical doctors, dentists, psychologists and other authorised health personnel • Registers established pursuant to the limited company act Sect 3-8. • Registers of current and prior employees or representatives • The personal files or indexes operated by publishing houses Public Administration and Information Technology
• The personal files or indexes operated by the daily press • Registers of elected or appointed representatives according to the statues governing municipalities, counties, The Church of Norway or containing data on Members of Parliament
The scheme of the regulation is to indicate (1) the technology permitted, (2) the types of personal data permitted to be included in the register, and (3) the communication permitted from the register to a third party. In general, personal data from the regulated registers may not be communicated to a third party unless there is consent from the data subject or the communication is required by law. The registers are only permitted used for the purpose indicated by their category name. There are, however, many variations in details among the different regulations. For instance, a register containing information on employees indicates 22 different types of data that may be recorded (Regulations, Sect 2-12). But this list is supplemented by data that are permitted in agreements concluded between the organisations for employers and employees. There are major general agreements governing conditions for work etc, and the regulation actually delegates to the parties in further to detail regulate the lawful content of employee registers.
Actually, an agreed framework for data protection was negotiated and entered into force spring 1975 - five years before the legislation. The basis was a research project conducted by professor Kristen Nygaard at The Norwegian Computing Center, an independent research institute, and the Union of Iron and Metal Workers. In the act on working environment of February 4, 1977:4 Sect 12(3), the introduction of planning or management systems is made subject to a special participation procedure. The
53 Compendium to the Erasmus Course
delegation to the parties in the working place is therefore in accordance to a long tradition, and can only be appreciated in this context.44
Another example is the regulation for libraries (Regulations, Sect 2-8), which do not permit inclusion of historical lending data in the registers.45 This is motivated by a concern for the combination of lender and the book information over time creating profiles of the likes and dislikes of the lender. It is an early example of the concern for the information created in "electronic footprints" that has been enhanced by consumer oriented electronic funds transfer, automatic traffic control, pay-by-view television and other transaction oriented systems that generate trivial data well suited for the establishment of profiles valuable for surveillance, direct marketing, etc. Some of the regulated registers may contain sensitive information. This is the reason for only manual registers operated by lawyers or health personnel are permitted by regulation - computerised versions will have to be licensed. Registers of the press and publishing industry may, however, contain any information as long as they comply with the principle of relevance - and the justification is obviously the freedom of the press, which should not unduly be reduced by data protection legislation. One should note, however, that in those cases sensitive data is included, a qualified relevance principle applies, see below.
44 Cf Niels Schweigaard "Bedriften, de ansatte og datamaskinen". Ragnar Dag Blekeli and Knut S Selmer (eds) Data og personvern, Norwegian University Press 1977:241-257. 4> Cf Eirik Djønne, Tove Grønn and Tor Hafli (eds) Personregisterloven med kommentarer, Tano, Oslo 1987:80. Public Administration and Information Technology
3.3. Licensed systems
If not positively excepted, a computerised personal register or a manual register containing sensitive data will have to obtain a licence prior to its establishment. One should note that in licensing a system, the authority of the Data Inspectorate is limited by the relevance principle of Sect 6( 1). If the data is sensitive, the authority of the Inspectorate is limited further by a qualified relevance principle, in these cases it is not sufficient that the activities of the operator justify the inclusion of the sensitive data, it must be "necessary" (Sect 6(2)).46 In deciding whether to award a licence, the Data Inspectorate is to identify data protection issues. Such issues should then be addressed in the licence, and Sect 11 mentions a number of typical provisions that may be included. If such measures are not sufficient to relieve the data protection concern, the Data Inspectorate has to make a further assessment: The advantages obtained by establishing the register should be balanced against the disadvantages in terms of data protection concern. One should note that a licence may be granted even if there are data protection concerns which cannot be solved. Such discretionary balancing of different interests is rather typical for the decisions by Norwegian public agencies, and nothing very special within the data protection legislation. If a licence is awarded, the Data Inspectorate includes in the licence a number or provisions for the operation of the register. These must as a minimum which type of data is permitted
46 Curiously, this provision follows the substantive provision in Sect 6(1) containing the relevance principle. But as all registers containing sensitive data falls within the scope of Sect 9 on licensing and regulations, it is not really a substantive provision addressed to operators of registers, but a limitation of the authority of the Data Inspectorate to either issue regulations or license registers.
55 Compendium to the Erasmus Course included, and to what purpose the register may serve. In Sect 11(2) it is suggested that additional provisions are considered for a number of aspects:
• collection and control (verification) of data • processing and storage of data • matching with other registers • use of PIN • disclosure of data to third party and communication to another register • the volume and nature of the information to be communicated to the data subject • access by the data subject • correction of data and routines for maintaining data quality • updating of the register • deletion or non-use of data after the passage of time, and communication from the register to the governmental archives
One may see the licensing system as an exploratory tool, which keeps the Inspectorate in touch with current developments. On the other hand, the scheme generates a lot of work that may reduce the Data Inspectorate to a bureaucratic organisation spending most time on routine matters. The regulations may be seen as a possibility of excepting typical routine registers from the licensing procedure, leaving them to be wholly governed by substantial law. But there are also two other strategies used by the Inspectorate. One is the standard licence. For typical activities, like the records of primary schools, there have been developed standard licences that are applied if there not are strong reasons to deviate. For other extensive operations, as projects financed by the National Research Council, there has been issued a blanket Public Administration and Information Technology
licence that imposes a certain internal procedure on the establishment of registers by these projects, but which do not require a licence to be obtained in the individual case. One may consider these developments as a gradual trend towards sectorised legislation. In fact, a standard licence is very similar to a regulation. It has certain advantages, for instance a greater flexibility for adapting its provision to the case at hand and to change its content. It also has drawbacks, a major disadvantage is that it is not published and made available to the public at large as regulations - and it may therefore be somewhat more difficult to check whether the provisions are satisfactory or that the operator is complying whit the provisions. In the future development of the Norwegian legislation in order to harmonise or comply with the forthcoming directive of the European Communities47 one may expect a development of the sectorial approach either in the statute itself, or in a somewhat more extensive set of regulations.
3.4. Public sector information services
The regime sketched above may appear rather strict, but one should bear in mind that registers authorised in special legislation do not fall fully within the scope of the data protection legislation. Norway is a small and open society, and there is a higher degree of trust - perhaps not wholly justified - between the citizen and the public administration, which gives that administration a great deal of leeway. Mention has already been made of the Central Population Register. Its services are made available to many public agencies, but also private sector organisations may have limited
47 At the time this paper is written, it has not been decided whether Norway will accept the European Economic Space Agreement or apply for membership in the Communities at a later date.
57 Compendium to the Erasmus Course______access to this source of information. There are several other major systems that are made available to the private sector. For instance are the key figures resulting from the annual taxation review available to the public. Traditionally, the lists have been displayed in the local tax office, but there has been published lists of taxable income and property since the 1920’s. The tax authorities have created a system whereby this information may be purchased at a certain price, also in machine readable form.48 This is, of course, example of a commercial utilisation of a special type of personal register. There is also a traditional system for real property modelled on the German "Grundbuch" system, which makes available information on ownership, leases, mortgages etc (the act is of June 7, 1935:2). During the past years, the local systems have been converted into a national computerised system, the conversion was finalised in October 1992. To achieve this, a private limited company was set up, Tinglysningsdata, which now is owned wholly by the state. In regulations issued pursuant to the real property registry act. this company has been granted monopoly for the sale of the information from this register (regulation issued by the Ministry of Justice June 29, 1989:527), and regulations have also been issued by the Ministry of Justice pursuant to the legislation governing the payment for the services of the justice administrative system governing the fees that can be claimed by Tinglysningsdata for their on-line services and print-outs. Again a special scheme for the exploitation of a personal register has been created. A register has also been created for personal chattels used as security for loans, etc with related information (Losoreregisteret). This will, for instance, contain information on court orders giving chattel as security to creditors when
4K Cf regulations issued by the Ministry of Finance pursuant to the freedom of information act Sect 8(3). Public Administration and Information Technology debtor has failed to meet his or her payments and related information. These data bases are daily copied to another state owned company in Oslo, The government computer center (Statens datasentral) pursuant to a regulation issued by the Ministry of Justice (1989:527), and information is communicated from this register to, for instance, credit reporting agencies for payment decided by the company. In this case, the access of the credit reporting agencies are governed by licence provisions issued by the Data Inspectorate. A similar scheme has been introduced for a register on those persons in quarantine due to insolence proceedings in which they have been involved, a decision that is made by the court. In this case, however, information from the register is sold by the central registration unit situated in Brønnøysund. These are some of the major services made available from the public to the private sector on the basis of registers containing personal data. There are several other examples,49 but perhaps not as clearly within the scope of the data protection legislation. One should note that this is a political issue that still is not settled, and that one may expect further developments. One will note, for instance, that the draft directive on data protection from the European Communities will have provisions that make it questionable whether the established solutions can be retained.
4. Data Subjects
4.1. Access right
First we will again mention that the access right in the data protection legislation can only be understood as part of the other
49 Cf Jon Bing Juridiske aspekter ved etablering og distribusjon av elektroniske tjenester. Prosjekt 5: Standardiserte formidlingskanaler. Nasjonal infrastruktur for edb, Statskonsult, Oslo 1992.
59 Compendium to the Erasmus Course access rights, and in the introduction, the two major examples - access according to the freedom of information and the procedure in public administration legislation was briefly mentioned. The access right is governed by the data protection act Sect 7. It is one of the provisions in the act that most directly concerns the data subjects, and it is therefore regrettable that it is difficult to understand and interpret. The person requesting access is all data subjects, and the data subject is to be given access to all data relating to himself or herself in the register. Only the person himself or herself may exercise this right, at death the access right expires. For instance has a widower not been granted access to the data of his dead wife.50 The data subject may, of course, appoint a representative to exercise his or her access right, the operator of the register will, however, have to have sufficient evidence for the authority of the representative. A register may exist in different versions due to back-up procedures, etc. In the legislative history there is discussion of whether a data subject will be permitted access to historical versions that are printed out, which is suggested not to be contained in the access right, though this is not directly reflected in the statutory text.51 A schematic representation of the access right is indicated by the figure below:
5(1 Data Inspectorate's decision, Jnr 80/530. 51 Cf Ot prp nr 2 (1977-78) Om lov om personregister mm, Ministry of Justice, Oslo 1978:75.
60 Public Administration and Information Technology
Fig 3 - Access according to the Norwegian data protection act
—F S Jl— - "I . Personell rcgi
! Factual data ) Documents stored in internal registers o ms 4o—> C/3
Statistics Planning Research
■*—>o oQJ C/3
• »-HCJ sp CU ( Factual data ) Documents stored in internal registers
Computerised registers Manual register
Two broad distinctions are established, one between registers in the private and in the public sector, second between computerised and manual registers. In this way one has four general categories of registers. All registers in the public sector are in principle subject to access right, while in the private sector, this applies only to the computerised registers.
61 ■ Compendium to the Erasmus Course
The data protection act makes itself a general exception for one type of registers. If the register only serves the purpose of generating statistics, general planning purposes or is only used in research, a limited access right is applied. This gives the data subject a right to learn which categories of personal data are contained in the register, but not to learn what those categories contain with respect to himself or herself. The data subject may learn the structure of the register, so to say, but not the contents of the structure. The justification for this rather broad limitation of the access right is twofold. First, the decision oriented data protection concept basic to the legislation would indicate that these are registers on which no individual decision is made, consequently the data protection issues are not prominent. It should be noted that the characterisation of the registers is not formal, but substantial - if a register established for research in fact is used to make one individual decision, it is not excluded from full access by data subjects. Second, there was suggested a practical problem by the Central Bureau of Statistics (Statistisk sentralbyrå): Their files were organised in such a way that the production of statistics was optimised. They could easily produce a breakdown of the correlation between geographical location and the age of teachers, but would have grave practical problems retrieving all data relating to a specific identified person. The Bureau suggested that if their files were made subject to access, they would have to reorganise all their files, with substantial cost as a result, and also a file structure that increased the risk of unauthorised disclosure.52 This view was, perhaps, dictated by
52 C f Ot prp nr 2 ( 1977-78) Om lov om personregister mm, Ministry of Justice, Oslo 1978:55.
62 Public Administration and Information Technology
the data base technology available at the end of the 1970's, but in fact this limitation has not really been challenged. In the Royal Decree ofDecember 21, 1979, bringing the data protection legislation into operation, part III made a further general exclusion of a type of registers - those necessary for the security of the nation or military preparedness are excluded from access. The reason for this exclusion is rather self evident. The Decree is interpreted not to exclude the register as such from access, but only to the extent this actually would endanger security. There is a dispute resolving mechanism for cases where the operator of the register and the Data Inspectorate do not agree to the qualification of the register, such disputes are settled in accordance with the general scheme by the Ministry of Justice.53 Though formulated as an exception of a type of registers, it is interpreted as an exception of certain (or all) data contained in the registers. Likewise the act itself limits access with respect to certain data that it is deemed "unadvised" to communicate to the data subject. There may be to reasons for this decision, one is the health of the data subject, the second is protection of closely related persons. Restricted access due to the health of the data subject is considered to be an exception that only rarely may be relevant. One should also be aware of the fact that access to medical files often will be governed by the health legislation (cf the note above on access to the journals of medical doctors). An example is the licence provisions for the National Cancer Register, which does not limit access, but which channels the information through the more recent medical doctor treating the data subject in order for him or her to exercise the assessment of whether
53 This exception is not reflected in the simplified figure above.
63 Compendium to the Erasmus Course access may have an adverse effect on the health of the data subject.54 Restricted access to closely related persons is an exception with a different justification. As we mentioned above, personal data as name is not protected by the administrative confidentiality. The data subject may therefore in accessing his or her file gains information on other persons of a non-confidential nature. If this was the case, one might in some instances have difficulties in investigating certain matters. For instance, a sister or a spouse may be willing to assist in the investigation of a child abuse case, but might be more reluctant to do this if the data subject (the person under suspicion of child abuse) through accessing his or her file could gain information on this assistance. The adverse effect on the relations between such persons will be clear if we propose that the suspicion is cleared, but that ill feeling is generated through the access. The rather vague expression "closely related persons" are interpreted as denoting those persons actually close to the data subject, not only family by kin or marriage. The exception is rarely applied, but the Data Inspectorate has in licences for Social Welfare Offices authorised the operator of the register to exclude access to information on the source of certain personal data in exceptional circumstances.55 For systems in the public sector, a rather more subtle exception is made. The data protection act Sect 7(2) makes the principle of the administrative procedure act Sect 18 applicable also to access based on the data protection act. The Sect 18 exception relies on a distinction between "internal" and other "documents". An internal document is an interim note or other materials that are produced while a case is under preparation.
54 Data Inspectorate license, jnr 80/1115. 55 Data Inspectorate license, jnr 80/490.
64 Public Administration and Information Technology
The party to the case has access to the documents of the case, but not the "internal" documents. An analogous application is not quite straightforward, as the access based on the administrative procedure act refers to "documents", while the data protection act refers to "personal data in a register". The documents may, however, be computerised and consequently part of a register.56 In that case, access is limited for internal documents. However, the administrative procedures act Sect 19 makes a further exception to the exception in Sect 18, and though the data protection act Sect 7 does not contain any reference to this, it is interpreted also to be applied with the administrative procedures act Sect 18. This grants access to "facts" contained in the internal documents, but not to "assessments". This distinction is generally taken to exclude "legal subsumption" from access. It is suggested that whether a person is "untidy" or "debauched" are examples of facts, while whether the person is a "thief or guilty of "social benefit fraud" are examples of assessments. On the basis of this interpretation, little is left of the exception for reduced access to internal documents, as all "facts" are available, only the rare cases of "legal subsumption" are excluded. But this explains the qualification of personal data as "data and assessments", where "data" refers to the facts, while "assessments" refers to the legal subsumption.
56 The definition of a "document" is found in the freedom of information act Sect 3, and this definition also applies to the administrative procedures act. Pursuant to the freedom of information act Sect 3, regulation ofDecember 12, 1986:2202 was extended to computerised material (see above). Any document contained in the file of a word processor, electronic mail server etc will consequently also be part of a register in the meaning of the data protection act.
65 Compendium to the Erasmus Course
One might respectfully submit that it has not been necessary to draft this provision like a legal puzzle with a somewhat uncertain solution. The other side of the coin is an extended access right in manual, private files. An employee in the public sector has access to his or her data in the manual personnel register, while the act does not grant employees in the private sector the same right. When the regulations were revised March 10, 1981, Sect 1 -5 was amended to give employees the same access to the registers of private employers. The provision is modelled after the administrative procedures act, and incorporate therefore an exception for "internal documents", which are somewhat more detailed specified in the regulations Sect 1-5(2), and the corresponding extension of the access right to facts in Sect 1-5(4), though the provision only recommends access to facts rather than requires such access.
This interpretation of the statute has recently been challenged with respect to a conflict with the Jehovah's Witnesses. Their manual file contains sensitive data, and is subject to licensing. According to Sect 11(2)(7) the licence may specify further the access right of the data subjects. But this provision cites Sect 7, and according to Sect 7 there is no right to access a private, manual register - though that register cannot be operated unless a licence has been obtained. It is therefore argued that the Data Inspectorate lacks the statutory authority to impose on the operator of such a register the duty to accept requests for access. Sect 7(5) and the Royal Decree of December 21, 1979. part I do, however, give authority to the Ministry of Justice to issue regulations on the right to access register, as is the case with respect to the personnel register
AA Public Administration and Information Technology
mentioned above. The Data Inspectorate has suggested that the Ministry of Justice should issue such regulations.57
Also in licences, access rights may be extended beyond that of the statutory provisions. With respect to schools, there is standard licence that extends access rights. As mentioned above, personal data also includes data on legal persons. This has caused IBM to apply for exception to the access rights for two registers. One of them is a register of companies and their installed computer facilities, both from IBM and competitors. The sources are newspapers, reports from the marketing division etc, and the register is used to support marketing. The other register contains some information on their own customers used in marketing support. The Data Inspectorate emphasised that the registers only contained information on legal persons, and that they were used in a competitive situation, granting an exception to the access rights for these two registers. The access rights in Norway are, taken as a whole, quite comprehensive. But the data protection legislation has made its provisions obscure and unnecessarily complex. A revision will probably result in a simplification.
4.2. Right o f correction
The data protection act Sect 8 contains a rather obvious provision that protects the data subject from retaining false information in a register. Four categories of data are indicated.
• Data which are incorrect • Data which are incomplete • Data which are unlawful
57 C f Lov&data 32/1992:7.
67 Compendium to the Erasmus Course
• Data which have lost their relevance
Obviously, any of these qualifications may be a matter of dispute. Whether data is incorrect, may be disputed (though some data may represent notorious facts where the correctness may be objectively ascertained). Also, whether data are incomplete, may be a matter of dispute - this is, however, an important issue related to the interest of adequacy: The data in the register alone may be quite misleading.58 The data may be unlawful on the basis of the relevance principle of Sect 6 - but it may be disputed if for instance a business objective justifies the inclusion of certain data. And finally, it may obviously be disputed whether data have lost their relevance because they have become dated.
An example of the latter consideration is a provision in the licence for the aliens register operated by the police, where data is to be deleted if the alien is granted Norwegian citizenship: One is not have different classes of citizens.59
The correction of the data may also take three forms:
• Correction • Deletion • Supplement
58 In Norway a child may be permitted to kindergarten partly on social indications. As there is a lack of capacity, the parent may emphasise the problems which constitute these social indications. The same information mas - lawfully - passed on to the social welfare authorities. In this case the Data Inspectorate recommended that the data subject was allowed to supplement the data. 59 Cf Data Inspectorate license, jnr 81/1896. Public Administration and Information Technology
An example may illustrate the choice between alternatives. There exist also in Norway computerised information retrieval services based on the editorial material of newspapers, journals, etc. The first was based on Aftenposten, the largest daily newspaper60 for its ATEKST service. The problem of rectification realised itself - obviously a newspaper item might contain data on a person that the person contested, or which actually was amended by subsequent items. This was solved by the data subject being granted a right to append to the item a comment, and the fact that a comment is appended, is noted in the title of the document containing the item. Likewise, the newspapers will communicate sensitive information. It was feared that the information system might in practice function like an informal criminal justice record. This motivated the inclusion of a provision in the licence that excluded from the public version of the system61 information on crimes when seven years has passed from the time of the crime. This exclusion applies not to crimes related to execution of jobs or public office, or when the case is of "major importance to the public".62 It follows further that if the data has caused to the communication or use of erroneous or incomplete data, the operator of the register should seek to limit damage with respect to the data subject. The provisions of Sect 8 may, as mentioned, easily be contested, and there may not only be a difference of opinion, but also a conflict of interests between the data subject and the operator of the register. If the operator will not make the
60 Licensed December 30, 1985, jnr 85/702-10. The subsequent licenses corresponded to this. 61 In the version available for the newspaper's own journalist, the information was still accessible. 62 Jon Bing Journalister, aviser og databaser, CompLex 14/87, Norwegian University Press, Oslo 1987:23-25.
69 Compendium to the Erasmus Course
correction, deletion or supplement required by the data subject, the data subject may appeal to the Data Inspectorate, which is empowered under the act to make a decision to what measures shall be taken. This is, of course, a decision that the operator may appeal - first to the Ministry of Justice, and then challenge the decision before the courts. This is, however, rather theoretical - in practice the operator hardly will pursue the matter beyond the appeal to the Ministry of Justice. When the decision becomes final, the operator is liable to criminal sanctions according to Sect 38 if he or she does not comply with the measures indicated.
4.3. Junk Mail Block
In the revision of 1987, a new Sect 8a was introduced in the data protection statute. This simply states that anyone can require his or her name blocked for the use of a register for direct mail or the distribution of similar material. This supplements the licence provisions for operating a direct mail or address brokerage (see above). It has been introduced in the general and substantive part of the statute in order for the data subject to be given this right also when the operator of the register is not such a business, for instance when a register of the customers of a department store or a bank is used for direct mail purposes for the services offered by the store or the bank. The licence of the professional operators includes requirements of citing the source of the address on the label, and not to include the PIN on the outside of an envelope. In practice, the blocking takes the form of establishing one special register containing the names of those persons who have made the request. Any output from the register is matched to this "exclusion register" to delete these names from the mailing. This method has been chosen in order to also block the names if updating of the register would re-introduce a blocked name. Public Administration and Information Technology
4.4. Liability>
In the perspective of the data subject, the criminal sanctions of the data protection legislation (Sect 38-40) are of less interest, and will not be discussed in this paper. Under Norwegian law, liability is generally based on negligence. There may be argued that in general, negligence is not sufficient for finding an operator of a register liable for an economic loss that is suffered by a data subject due to erroneous data in the register. This would represent an instance of pure economic loss, and it is argued that there - in addition to negligence, there should also be justified cause for relying on the data. This discussion has emerged without any special reference to data protection, and is mainly developed in the doctrine without statutory authority, and with few cases on which to rely.63 The data protection legislation has only one special liability clause in Sect 40, which relates to credit reporting services. Strict liability is imposed if a data protection violation causes economic loss for the data subject. T he provision is strengthening the position of the data subject compared to the background law as negligence (and justified reliance) is not required, the liability is strict. But it is limited to economic loss. In the ease of credit reporting, erroneous personal data may more easily cause such loss than in general. But nevertheless, the characteristic of a data protection violation is not the economic loss, but rather the immaterial aspects. However, in Norwegian law, liability for anything but economic loss (including general damages) would require statutory authority.
63 Cf Jon Bing Juridiske aspekter ved etablering ug distribusjon av elektroniske tjenester, Nasjonal infrastruktur for edb - Prosjekt 5: Standardiserte formidlingskanaler. Statskonsult, Oslo 1992:80-98 and Viggo Hagstrøm "Informasjonsansvar - om villedning av annen enn kontraktpart"; Tidsskrift for rettsvitenskap 2/1989:204.
71 Compendium to the Erasmus Course
The general statute on damages64 Sect 3-6 is a statutory authority for awarding damages for non economic loss. This section refers in rather general terms to loss suffered by libel and invasion of privacy, but the legislative history makes it clear that it is presumed that such damages only can be claimed if the provisions of the criminal code with respect to libel and privacy invasion can be applied.65 This leaves the data subject in a somewhat unsatisfactory situation. Though he or she may claim for damages for economic loss under the general non-statutory law on "pure economic loss", this is not a typical effect of the failure of an operator to comply with the data quality requirement of the data protection legislation or the conditions set out in a licence. It is hoped that the right to damages with be strengthened in the future revision of the law.66
5. International data traffic
5.7. Conventions and other international agreements
Norway is one of the original parties to the Council of Europe Convention for the Protection of Individuals with regard to Automatic Processing of Personal Data of June 27, 1980. It has also been active in drafting the different recommendation that the Council of Europe has adopted in pursuant of the convention. It also is a member of the OECD, and has not taken any reservations with respect to its Guidelines Governing The
64 Skadeserstatningsloven, June 13, 1969:26. 6> Cf Jon Bing and Cato Schiøtz "Oppreisning for enkelte personvemkrenkelser"; Jussens venner 1977:241 -270. 66 In the debate following the 10th anniversary report in the parliament, several members suggested strict liability for loss caused by errors in public register, cf Lov&data 31/1992:4-5. Public Administration and Information Technology
Protection of Privacy And Transborder Flows Of Personal Data of September 23, 1980. No other international instruments specific to data protection apply. But, as noted above, Norway will through the European Economic Space Agreement have to comply with the directives of the European Communities, including those on data protection.
5.2. Transnational data flows
The data protection act title 9 governs export of data from Norway. When the regulations were introduced, a licence system was set up pursuant to Sect 36 and 37.67 But at the revision of the regulations in 1981, this was replaced by a simplified system of notification, cf regulations Sect 8-1. According to the regulations, notification is necessary for two situations. The first is the export of a register that either is computerised, or which contains sensitive data, ie those registers that are subject to license according to Sect 9. But there is a duty to notify the Data Inspectorate also for those registers that do not need a licence according to regulations issued pursuant to Sect 9. The second is the export of personal data collected in Norway if the purpose of the export is the inclusion in a Sect 9 register (see above). The provision is interpreted to require some sort of organised activity in Norway (collection). The Data Inspectorate has, for instance, not seen the establishment of a post box address in Norway as sufficient to qualify the en bloc transfer of
67 Cf Jon Bing "Transnational Data Flows and the Scandinavian Data Protection Legislation"; Folke Schmidt (ed) Scandinavian Studies in Law, 1980:67-96.
73 Compendium to the Erasmus Course the main from the box and to Sweden as "collection" in the meaning of the act.68 One will also notice that the purpose of the export has to be the inclusion of a Sect 9 register. This may be a rather slippery criterion: The sender may not know the intention of the foreign institution requesting the data; the recipient may decide at a later stage to create a register; or the register may be manual, the data collected in Norway may be trivial, but data included in the register on the nationals from other countries may be of a sensitive nature, making the register qualifying under Sect 9. In spite of the possible difficulties in interpreting and applying this provision, it has in practice not created many problems - mainly because it is not very actively promoted by the Inspectorate. The regulations Sect 8-1(2) make an exception from the duty to notify the Data Inspectorate if international agreements, or membership in international organisations oblige Norway to make the data or registers available. This, for instance, will apply to organised international co-operation in health care or among police authorities. Notification should be made on a specified form sufficiently time prior to exportation give the Data Inspectorate time to deny export. The Data Inspectorate is actually given the authority to deny export even when this would be in violation to the Council of Europe treaty. This is, however, only a theoretical possibility. The Inspectorate has been rather reserved in exercising its authority - examples include a case where a foreign bank established itself in Norway, but maintained its customer files in Belgium. In this case, the Inspectorate required the content of the register to be within the scope of the regulatory provisions applying to the registers of banks (regulation Sect 2-4), and that
68 Cf Eirik Djønne, l ove Gronn and l or Hafli (eds) Personregisterloven med kommentarer, Tano, Oslo 1987:157. Public Administration and Information Technology the registers were used according to the general principles for such registers (regulation Sect 2-1 ).69
5.3. Territorial application of the data protection act
The provisions on transnational flows are somewhat unexciting. The issue of the territorial application of the data protection act is much more stimulating. The general problem is what sort of relation there should be between a certain state and the operation of a register sufficient for that state to apply its national data protection legislation to the register. The issue of the proper law o f data protection has simply not been solved on the international level, though it was discussed in the work preceding the Council of Europe Conventional and the OECD Guidelines.70 This attractive and juicy problem will not be addressed here in general. It must suffice to indicate that the criteria for qualification - typically access by national users to the register - make positive conflicts of authority rather probable. Norway has some few associated territories, the more interesting is Spitzbergen, an Arctic island with mining and research activities. Spitzbergen is according to a treaty of 192071 under Norwegian sovereignty, though there is a rather large Russian mining colony at Barentsburg. An interpretation of the Spitzbergen act implementing the treaty in national law72 Sect 2
69 Data Inspectorate decision, jnr 84/894 70 Cf Jon Bing "Reflections on a data protection policy for 1992"; proceedings of Access to public sector information, data protection and computer crime: Legal challenges and opportunities created by the prolific growth of electronic information services. Commission of the European Communities and Council of Europe, Luxembourg 27-28 March, 1990. 71 The Spitzbergen treaty of February 2, 1920 between the United States of America, Denmark, France, Italy, Japan, the Netherlands, United Kingdom, Ireland, Sweden, and Norway. 12 Lov om Svalbard, July 17, 1925:11.
75 Compendium to the Erasmus Course implies that the data protection act does not apply to Spitzbergen. This is currently under reconsideration, and the first data protection act directly to apply to Russians may conceivably be the Norwegian. There is a rather extensive offshore activity on the Norwegian continental shelf of the North Sea. This activity is governed by the petroleum act73 Sect 2 of this act make the data protection legislation applicable to activity related to the exploration, production, and transportation of petroleum products. This is extended outside the continental shelf itself as far as this follows from international public law or a bilateral agreement with another state, cf Sect 1 (2). The data protection act is also interpreted to apply to Norwegian representations abroad - ie embassies and consulates.74 Even more tenuous is the relation with Norwegian merchant vessels in foreign waters. One interesting case concerned Norwegian cruise liners in the Caribbean. The shipowner was a Norwegian company, who had contracted with a local catering firm to employ personnel like cabin attendants. The local company maintained a register of passenger ratings of the attendants. The Norwegian Ministry of Justice held that the data protection legislation applied to this register,75 though the ship never entered Norwegian territorial waters, the catering company was foreign, and the employees were foreigners.76
73 Petroleumsloven, March 22, 1985:11. The former act governing offshore activity was interpreted in the same way. 74 Cf circular of the Data Inspectorate, jnr 86/323. 75 Letter of November 6,1981. 76 Cf Jon Bing "Impact of Developing Information Technology on Data Protection Legislation", Organisation for Economic Co-operation and Development, OECD/ICCP(86)5, Paris, 1986. Public Administration and Information Technology
As stated, a general discussion will not be offered. Today, it still is an open issue what link is necessary to establish that Norwegian data protection legislation applied. In the past, there has been suggested for Swedish law that any system that may be accessed by terminals on the territory, falls within the scope of data protection.77 Bergmann78 has suggested several possible additional types of relevant relations: Domicile of the data subjects, business site of the operator of the register, the most favourable data protection law, the place of the data processing and the lex rei sitae of the computerised system. One may also indicate that it may be too simplistic to operate with "data protection law" as subject for the choice of law, this must perhaps be further qualified.79It would seem that at the moment, a territorial relation is sufficient to establish application of the Norwegian legislation, but there may also be an additional policy element. Both with respect to the offshore activities and the case of the Caribbean yacht it may have been a relevant element that data protection is seen as related to the provisions governing an employment situation with respect to security and welfare, and that therefore the relation to the territory has to be less than in cases where it is questioned whether the data protection legislation applies to contractual situations between parties, which lack the "power relation" between data subject and operator of the register thought to be so important with respect to the Norwegian concept of data protection.
77 Cf Michael Bogdan "Dataflykt över gränsema och den svenska datalagstiftningen", Förvaltningsrättslig Tidskrift 1978/1-26. 78 Michael Bergmann Grenzüberschreitender Datenschutz; Nomos, Baden-Baden 1985. 79 Cf Jon Bing "Impact of Developing Information Technology on Data Protection Legislation". Organisation for Economic Co-operation and Development, OECD/ICCP(86)5, Paris, 1986:51-53 and François Rigaux "La loi applicable à la protection des individus à l'régard du traitement automatisé des données à charactère personnel". RCDIP 1980:443-478.
77 Compendium to the Erasmus Course
6. Recent developments in case law
Over the past few years there has been an interesting development with respect to data protection at the Norwegian courts. The development concerns the relation between some elements of information technology used for supervision, either of the public or of employees. In closing, a few remarks on this development may be appropriate. The first case is in itself quite famous, generally known as the Photographic evidence case.80 In several areas of Norway, automatic traffic control systems have been introduced. They are rather conventional in their technology, a car passes across a wire loop underneath the road surface, inducing sufficient electricity to make it possible to time its passing. Using two such loops, its speed may easily be calculated. If the speed exceeds the maximum, a photograph is taken by an automatically operated camera. The owner of the car in question protested when presented a fine for speeding, as the car had been reported stolen in relation to the burglary of his home. This made the criminal police interested in the photograph, which turned out to portrait the driver of the stolen car, and to include a number of objects in the car that had been reported stolen from the home of the owner of the car. The photograph was consequently claimed as evidence by the police. The problem was that it had been presumed in the discussion of the regulations applying to automatic traffic control that such evidence only should be used for road traffic purposes, and this presumption was explicitly expressed by the relevant committee in the parliament.81 The use of excess information was to be limited due to privacy considerations. The Ministry of Justice
80 Cf Norsk Retstidende 1990:1008, Lov&data 25/1991:3-4. 81 Innst S nr 98 (1986-87) page 14. Public Administration and Information Technology
issues instructions limiting the use of the photographs according to this view. The Supreme Court observed that this represented an attempt to restrict the freedom of the courts to admit evidence at their discretion,82 and such a restriction could not be legally imposed without authority in law, and the instructions of the Ministry of Justice lacked such authority. On this rather formal basis, the Supreme Court held that the court was authorised to permit the evidence. The second case has become known as the "Snack bar case".83 This case relates to a possible embezzlement from the cash register of a snack bar. In the snack bar, a video camera has been installed which displayed on a monitor in the back room a view of the shop. In this way the person on duty could relax in the back room and be altered when a customer entered the shop. The owner of the shop noticed that the trade seemed to be less than indicated by the purchased goods. Without altering his staff, he connected the video camera also to a recorder, and in this way he recorded 14 hours. Extracts of this record were copied over to a tape, and was maintained to prove that cash was handed over to the operator of the snack bar without that person entering the sum in the cash register. The Supreme Court discussed at some length whether any statutory provisions applied to the case, and concluded that they did not. But the Court went on to state that such surveillance represented a major violation of privacy of a non-statutory nature. Again we see that information technology provokes a reaction from the legal system - the use of video recordings for surveillance is seen as a violation of the non-statutory privacy
82 There are few formal rules o f evidence in Norwegian law, the court is traditionally allowed rather freely to determine what evidence is to be permitted. 83 Norsk Retstidende 1991:616, Lov&data 28/1991:8-9.
79 Compendium to the Erasmus Course protection. And on this basis, the Supreme Court refused the recordings to be permitted as evidence. This was a criminal case, and the rules of evidence may there be more severe, and more favourable to the accused than in civil cases. Our third case is, however, a civil case, decided by the Appeal Court of Agder.84 In this case, the owner of a pub suspected that beer was sold without payment being registered. The employer compared the consumption of beer with the total income from this sale, and concluded that trick was related to the way in which the employees filled the glass from the tap (in Norwegian this gadget is called "the tap tower", and the case takes its name from this phrase). Again the employer installed a video camera showing in its frame the tap tower and the hands of those employers drawing beer. A case of dismissal was initiated against three of the employees, and the first instance court decided to allow the video records to be used as evidence, drawing a distinction between the Supreme Court criminal case and the current civil case. The Agder appeal court was, however, rather definite in its decision. It stated that the Supreme Court decision held such clandestine recordings as a grave violation of the privacy of the employees. It would be inappropriate to allow such recordings as evidence in civil cases, the court argues, as this might encourage circumvention of the rule laid down by the Supreme Court. The appeal court makes, however, a reservation with respect to possible extraordinary circumstances that would justify such recordings being permitted as evidence. One should note that before the last decision, an amendment of the general criminal code had taken place.85 This amendment prohibits constant or regular video surveillance in a public place (as the criminal code defines that phrase) unless information of
84 Cf Lov&data 33/1992. 85 Statute of March 15, 1991:5, taking effect from July 1, 1991. Public Administration and Information Technology
this is given by signs or other adequate means. Following the snack bar case, a further proposal has been introduced to make a similar rule govern video surveillance by employers. Therefore, it may be argued, this is a special case in which the courts have been in line with the legislators, though slightly ahead of their time. However, there is a fourth case that emphasises this line of argumentation, and which is not related to video recordings, but another example of information technology turned to the purpose of surveillance of employees. This is the E-mail case, decided by the first instance court of Asker and Bærum.86 This case is also a dispute of whether a dismissal was justified. A conflict was brewing between the executive director and an employee of the Norwegian branch of Memorex. The employee felt that the director was being unfair, and was reading his private electronic mail. To prove his point, he wrote an e-mail addressed to the European corporate headquarters in Italy, but did not transmit the mail. The letter he stored in his private area of the e-mail system. The director actually accessed that area, read the letter, and dismissed the employee. The court held that as there was no agreement or warning that the private areas could be accessed by management, such access was a violation of privacy. Again, no statutory authority could be cited, but the court went ahead and applied non-statutory principles. Looking at these four cases, one may see a certain pattern. All of them relate to the use of information technology for surveillance. The difference between the "Photographic evidence case", where the photographs were allowed used as evidence, and the three others, would mainly seem to be the fact that in the automatic traffic control, due notice was given. Automatic traffic control is, of course, mainly a preventive measure, and in order
86 C f Lov&data 32/1992:8-9.
81 Compendium to the Erasmus Course to the control to have a preventive effect, signs are displayed as clearly as possible. This was not done in the other three cases, here the surveillance was clandestine. Consequently, one may conclude that this recent case law indicates that electronic surveillance - by any means - is a violation of privacy if the subject of surveillance has not agreed, or at least been notified, of the surveillance. The recent statutory amendments and bills also support this conclusion. And in closing it may be of some minor interest to note that the interaction between data protection (or privacy) and information technology has not found a balance in the current legislation. Both technological developments, and external influences like the directives expected from the European communities, will cause major changes in the current legislation. We may expect such changes in the relative near future. Consequently, this paper has a limited interest in time as a discussion of Norwegian law. But the development has for those of us interested in data protection primarily a promise: Data protection will remain a current issue for the foreseeable future - in Norway, and in Europe. Public Administration and Information Technology
Privacy and Surveillance Systems1
Jon Bing
1. Data protection and surveillance: Big Brother Is Watching Your Electronic Trails
One of the most popular images of the surveillance society was offered by George Orwell in his important novel 1984.2 Though the computer is absent from the novel - ENIAC - was actually completed only two years before the novel was written3 and hardly heralded its possible use as surveillance technology. In the novel of Orwell, it was the "telescreens" that provided a kind of two-way television system, enabling "Big Brother" and his many representatives to spy into the private homes of the
1 This paper was originally given to V° Congresso Internazionale: Informatica e Attivita' Giurdica, organised by Corte Suprema di Cassazione, Centro Elettronico di Documentazione, Rome May 3-7, 1993. 2 Harcourt, Brace and Company 1949. 3 The novel was written in 1948, and the title is derived from a juxtaposition of the last two digests of the year on which this utopian novel is a comment. ENIAC is generally thought to have been built at the Moore School of Electrical Engineering had completed at the Ballistics Research Laboratory of the University of Pennsylvania in February 1946, though in the patent litigation following this development it was deemed that ENIAC had been in public use since December 1945, cf Stephen Saxby The Age o f Information, Macmillian, London 1990:102.
83 Compendium to the Erasmus Course citizens.4 Another important utopian novel overshadowing our policy discussions on data protection and surveillance is Yevgeny Zamyatin We.5 Written in 1920 as a bitter comment upon the Russian revolution, of which he himself had been a foreground figure, Zamyatin describes a society under strict control - a surveillance society. And though computers also are absent from this vision, this may be the first novel where the name of the individuals is replaced by numbers - his narrator is only known as D-503 who is tempted off the straight and narrow by the dark-haired anarchistic female E-330. The replacement of names by numbers is in the novel a symbol of removing the individuality and humanity from the persons. And this symbol is still today very strong, perhaps emphasised by the registration numbers tattooed onto the very skin of individuals in wartime prison camps. We mention these two references as an introduction to the discussion of data protection and surveillance. The issue is strongly related not only to general policies, but also the social psychology of Western societies, as the two novels illustrate - giving a voice to some of the concerns in discussion also of legal policies. Data protection has been an issue related to surveillance, but only somewhat indirectly. One has had intense discussion on
4 And in this way one might suggest that Orwell implied a society-wide I DSN cable network, though the capacity requirements for the network necessary for implementing this surveillance technology obviously is not discussed in the novel. 5 The novel was written in 1920, and published in an English version in the United States in 1924. Several English editions were published before Zamyatin died exiled in Paris in 1937, and it is probable that both Aldous Huxley's Brave, New World ( 1932) and Orwell's 1984 are influenced by this novel.Orwell himself commented on We saying "[i]t is in effect a study of the Machine ..." C f Brian W Aldiss with David Wingrove Trillion Year Spree: The History o f Science Fiction, Gollancz, London 1986:244. Public Administration and Information Technology
whether the files of civil or military security police should be governed by the data protection legislation - in my own country files necessary for military preparedness or national security are excluded.6 In establishing a digital switched network for voice telephony, information on calling and called terminal point will be retained for billing relative to real connect time, which offers interesting possibilities for analysis and pattern-matching which has made data protection authorities wary of allowing the detainment of such files for more than the brief period necessary for billing purposes. This is, of course, only one example of what the Germans call Rasterfahndung and which the Americans often discuss as data base surveillance. The use of computerised transaction systems will generate a large number of "electronic trails". There are several important sources for these electronic trails. One has been mentioned - the telecommunication networks, including the billing data of voice telephony, but also including data of cable television networks, coded satellite broadcasting (which addresses the individual decoder at the home of the consumer) and the even more detailed data implied by pay-by- view systems.7 Cellular telephone systems also track the subscriber in some detail, and will disclose his or her movements. A second important source is consumer oriented funds transaction systems, including automatic tellers and point-of-
6 C f Royal resolution of December 21, 1979 section III. 7 In many jurisdictions, including Norway, the records of a library linking books and lenders are deleted on the return of the book to exclude the possibility of deriving profiles of likes and dislikes from the choice of literature over time. A similar problem is related to any telecommunication service where the billing information discloses personal choices, not only pay-by-view systems, but also many adiotex- systems.
85 Compendium to the Erasmus Course sales terminals. Such data identify time and place of the transaction, and often the name of the establishment will imply the type of goods or services purchased. A third major source is the automatic traffic control systems, which will record the passage of a vehicle through toll gates, identify speeding vehicles, vehicles violating a red light, etc. The identification may be made by a passive integrated circuit, typically placed inside the front window of a car and polled by the control unit. These are only three examples. In practice, any type of data base can be used for surveillance purposes - the classical example is the use by the German police of the billing records of the Hamburg electrical board to locate the terrorist Rudolph Clemens Wagner.8 The use of data base surveillance has attracted the interest of commentators on data protection, both the uses for locating an individual (as in the case of Rudolph Clemens Wagner), and for analyses with the objective of identifying a suspect population.
This distinction is in my view important to make. When, for instance, the police are looking for a terrorist, any source of information may be relevant. There are traditional procedures to be followed for obtaining this information similar to those governing search and seizure- procedures. The crime is known, the interest in solving the crime may be balanced against, for instance, the interest i privacy. This may be termed individual data base surveillance. More difficult is the collective data base surveillance, situation where no suspect has been identified prior to the
8 Cf for instance Jon Bing "Data Protection and Social Policy"; XIV Colloquy on European Law DE/14(84)3, Council of Europe, Strasbourg 1984. Public Administration and Information Technology
data base surveillance. In this situation, an analysis takes place to identify a suspect population, the members of which are examined in more detail. An individual may be included in the suspect population without having acted in such a way that these acts would have made the individual a suspect - there is just a statistical correlation between certain characteristics of the individual in question and those of a typical suspect. An infamous example of this use of data base surveillance is the Swedish Kungsbacka incident, where a matching of data bases identified 1,000 persons as possibly guilty of social security fraud, but with only 10-20 actual convictions: The discrepancy being caused by matching files containing incompatible data.9
Data base surveillance has become visible in the data protection debate due to the use offiles for this type of surveillance. But the discussion also indicates the limitations of traditional or first generation data protection regulation. A regulation of the collective data base surveillance is obviously not mainly an issue of confidentiality, of the data on the individual being correct etc. the issue is not appropriately discussed in the perspective of the individual data subject, but is part of a broader issue - the level of surveillance that should be accepted in a society, what procedures should safeguard against misuse of such methods, etc. A distinction may be made between data protection and privacy, using the first of these terms (which generally are used as synonyms, though "data protection" often is preferred when discussing the European type legislation), using the first term for
9 The incident is discussed in the memoirs of Jan Freese, who was head of the Swedish Data Inspection at that time, cf Den maktfullkomliga oformogan, Wahlstrdm och Widstrand, Stockholm 1987:97.
87 Compendium to the Erasmus Course the modem control of computerised files, especially relating to decisions having an impact on the health or the welfare of a data subject, and using the latter term for the more general protection against undue invasion of home, the integrity of personal communication, etc. In this paper, some trends in recent Norwegian court practice will be discussed. These trends are related to the use of electronic systems for surveillance, and therefore the collective data base surveillance mentioned above. They also point towards the broader issue of surveillance control and its relation to privacy. Obviously, the cases have modest interest as sources of law for other jurisdictions. But they may have an interest as examples. In discussing the cases, one should be aware of the fact that a Norwegian court in civil law may apply "general principles of law" without citing a statutory authority. This makes the legal system include some of the characteristics of what otherwise is seen as typical for case law jurisdictions. This also allows a court to make policy decisions and actively take part in the forging of emerging law, as indeed is the case concerning the decisions discussed below.
2. Recent eases on electronic surveillance
2.1. The Photographic Evidence and the Mobile Telephone Cases
During the last few years, systems for automatic traffic control have been introduced, following a policy discussion in Parliament in the session 1986-87. The most common of these systems is based on placing two wire loops under the road surface. The speed of a car is measured simply by measuring the Public Administration and Information Technology time lapse between the electrical impulses generated by the car passing the two loops. If the speed exceeds the legal limit by a pre-set amount, a conventional camera is trigged and a photograph is take of the licence plates of the car. In discussing the introduction of this system, the relevant committee in the Parliament emphasised a concern for the privacy of the drivers - for instance, not only the driver, but the passenger, would be identified. It was made explicit that the information gained in this way, only should be used to pursue violations of the road traffic act - excess information should not be used for other objectives. The Ministry of Justice subsequently issued instructions based on the expressed view. Routinely processing traffic offences in the city of Stavanger, the owner of a car received a ticket. However, at the time recorded for the offence, his car had been reported stolen. The owner notified the police of this circumstance, and pointed out that the car had been stolen at the time his house had been burglarised. When examining the photograph, the police were able to identify the driver of the car as well as several of the objects in the car as those being stolen in the burglary. When the prosecutor claimed the photograph as evidence, the defence pointed out that according to the instructions, excess evidence from the photograph was not allowed used for purposes not relating to the road traffic act. The first instance court of Stavanger did, however, permit the evidence, and the defence appealed the decision to the Supreme Court. The Supreme Court made a rather formal argument.10 First, the Court pointed out that the instructions of the Ministry of Justice were not based on statutory authority, but on the general executive power. Secondly, the Court held that the exclusion of the photographs as evidence in the instruction was an attempt of
10 The case is published in Norsk Retstidende 1990:1008 and Lov&data 25/1991:3-4 ("Fotobevisdommen").
89 Compendium to the Erasmus Course limiting the broad discretionary power of the Norwegian courts to permit or reject evidence offered by any party.11 This authority of the courts is based on the statutory rules of procedure in criminal cases, and therefore could not be limited by ministerial instructions according to the general principles of the Norwegian legal system in which statutes have priority over instructions. Consequently, the Supreme Court found that the first instance court had the necessary authority to accept the evidence. In this way, the Supreme Court did not discuss the substance of the legal issue: Whether it was an infringement of privacy to use the excess personal information for another objective. However, it if the Court had felt that this conflict was relevant, it would have been free to consider this issue of substance, saying that (1) the court had the authority to admit or reject the evidence, as it did in its decision, but going on and (2) discuss the reasons for excersising the authority to reject the decision. Therefore this decision justified some concern for the development of privacy policies. On the basis of the photograph, the driver of the stolen car was convicted for receiving stolen property - the photograph did not actually prove that he was the burglar. A more recent case concerns a person charged with giving false evidence and insurance fraud. His mobile telephone was not removed from his person when he was imprisoned pending trial. From his prison cell he made several calls, and the billing information recorded for the mobile telephone system would identify which subscribers had been called. The prosecutor requested the consent of the court for seizure of the records. The telecommunication operator maintained that this would be in violations of their duty of confidentiality.
11 There are few formal rules of evidence in Norwegian law, the court is given wide discretionary authority in this respect. Public Administration and Information Technology
One should note that the reason for this issue to appear before the courts at an earlier opportunity is that the common billing system for voice telephony is based on a mechanical device counting units connect time, but keeping no record of which subscriber has been connected. The Gulating Appeal court decided to admit the evidence, and the telecommunication operator appealed the decision to the Supreme Court, which confirmed the decision.12 The Court did not find privacy arguments removing a printout of the billing data to fall outside the scope of the general provisions governing seizure of evidence. The Court emphasised that it was not an issue of wiretapping, but of subsequent access to recorded data.
2.2.The Snack Bar and Beer Tap Cases
In the Photographic Evidence Case rather conventional surveillance technology was employed. Similarly, video cameras have for a long time been used for surveillance purposes. The Data Inspectorate licensed the first video recording system in 1986, and this concerns recordings made at a home for the mentally disabled. The Inspectorate has held that such a video recording constitutes a "personal register" according to the legal definition in the Norwegian statute sect 1(2). The definition is based on what is known as the "retrievalbility criterion", which require that the data is stored in such a way that it may easily be retrieved. In a board decision of December 13, 1990 the Inspectorate held that video recordings from road toll stations contained personal data as the drivers could be identified
12 The decision was made by the Appeal Chamber of the Supreme Court, and is published in Lov&data 34/1993; 11.
91 Compendium lo the Erasmus Course through the licence plates of the cars with sufficient certainty, and that the data were stored in a way that satisfied the criteria in the statute. An application form for "optical surveillance systems" was developed, and the owners of petrol stations and parking lots started to apply for licences. At the same time, the Ministry of Justice proposed an amendment of the general criminal code Sect 390b, which made video surveillance of a public place illegal unless notices were clearly posted.13 Therefore, at the time the Snack Bar Case came before the Supreme Court, the Data Inspectorate held the view that video recordings qualified as "personal registers" according to the data protection act, and there was no regulation of video surveillance at the place of work. The owner of a snack bar had become suspicious, comparing sales figures with purchases, that he was being embezzled by one of the employees. The bar had a back room in which the employees relaxed. The owner had installed a video camera in the shop and a monitor in the back room for the staff to be able to relax in the back room when there no customers in the shop, and be notified by the monitor of anybody entering the small shop. By tilting the camera slightly, the owner was able also to fit within the frame of the picture the cash register. Without informing the staff, he also linked a video recorder to the camera. The prosecutor wanted to present as evidence in the criminal case an edited version of the recordings made in this way, but this offer was rejected by the first instance court of Ringerike. This decision was appealed to the Supreme Court. In its decision14, the Supreme Court makes several interesting observations. Several statutory provisions are considered, among
13 Statute of March 15, 1991, taking effect from July 1, 1992. 14 The decision is published in Norsk Retstidende 1991:616, Lov&data 28/1991:8-9 ("Gatekjokkendommen"). Public Administration and Information Technology them the data protection legislation and the working environment act, hut none was found applicable. The Supreme Court stated in an obiter dictum that it held it for improbable that such a recording met the retrievability test contained in the legal definition of a personal register, consequently invalidating much of the work of the Data Inspectorate in that respect. It supports, however, the first instance court, a decision based on general principles of law - ie without reference to statutory authority. The final arguments of the spokesman for the unanimous decision, Justice Aasland, may be cited:
"This case concerns ... a serious infringement of personal integrity. To have been subject to a clandestine video recording at one's place of work must be experienced by an employee as a great strain. In my opinion, major privacy considerations favour a policy by which the courts reject evidence collected in this way. By that attitude, the courts will counteract this type of infringements. It would also be offensive if evidence collected in this way was admitted, which could be experienced as a new infringement by the person subject to surveillance. Opposed to these considerations is in particular the objective to make the best possible information available to the court. It has also in this case been emphasised that the director had a justified suspicion of embezzlement, and that such methods were necessary to supply evidence. I cannot find these arguments sufficiently strong in this case. I do not discuss further the reasons for the suspicion and the evidential value of the recordings; neither have these issues particularly well presented for the Supreme Court. If those arguments would have been decisive, it would necessarily lead to an undermining of the protection of privacy, as I have discussed. The use of such evidence would obviously be particularly topical in cases where an
93 Compendium to the Erasmus Course
employer presumes it is possible to prove criminal charges with candelstine recordings. If evidence is permitted in such cases, it will increase the risk for extensive use of candlestine recordings also in other cases, where the recordings subsequently are not deemed appropriate as evidence. 1 cannot find that this case is of a scope or a character which make it justified to deviate from what I held must be the general rule with respect to such offers of evidence as this case concerns."
The defendant was actually acquitted in this case. This was a criminal case. But a rather similar case was heard before the appeal court of Agder in 199215 concerning the issue of unfair dismissal, the Beer Tap Case. The management argued that the reason for the restaurant not making a profit was loss in drafting beer. Controls disclosed that the volume of lager and strong beer being drafted did not tally with the number of unites recorded by the cash register. There might, however, be more than on explanations, and without informing the employees, the management made video recordings of the area close to the beer tap. Obviously the parties argued with respect to the Snack Bar Case, it was emphasised that this was not a criminal case, and that the Supreme Court decision could not be seen as an example of applying a general rule. The first instance court of Skien and Porsgerunn admitted the evidence, but in hearing the appeal, Agder deemed the case similar to the Snack Bar Decision, which was qualified as a precedent binding the court. The court also emphasised that this would discourage use of candlestine video recordings in cases of unfair dismissal, an argument similar to that made by Justice Aasland concerning employer-employee relations.
15 Agder Appeal Court, decision of October 5, 1992, published in Lov&data 34/1993:8 ("Tappet&msaken"). Public Administration and Information Technology
On the basis of these two decisions, it is rather obvious that one may argue that Norwegian law has been found to contain a non-statutory general principle directing the courts to dismiss evidence from candlestine video recordings made by employers in cases relating to their employees. Such an argument would not be very adventurous as the Ministry of Justice is in the process of introduce an amendment to the general criminal code sect 390b: As mentioned above, this imposes a duty by notice to inform that a public place is subject to video surveillance. The principle is, according to the proposal, also to be applied to video surveillance of the working place. At the same time, the Ministry of Justice has launched a proposal to introduce a new Sect 37b in the data protection legislation, authorising the Data Inspectorate to issue regulations on the use of video recordings, including data security measures, the purposes for which the recordings can be used, the communication of recordings to a third party, deletion of recordings and giving the recorded persons a right to access the recording.16 This may imply that legislation is replacing the court-driven innovation of the law applying to "electronic surveillance". But one may see the Snack Bar and the Beer l ap cases as two examples of the application of a principle more general than video surveillance. And there is a last case that may illustrate this point.
2.3. The E-Mail Case
Mcmorex Norway employs an electronic mail system that links all the sites world-wide. Authorisation to access this system is
16 Cf Häkon Knudtsen "Fjernsynsoverväking og personreeisterloven", Lov&data 34/1993:6.
95 Compendium to the Erasmus Course issued by a superior within the organisation, in Norway by the executive director. Authorisation is made by issuing a user identification and a pass word. The pass word may subsequently be amended by the user, but the person responsible for system maintenance, would be able to identify the new word. Within the e-mail systems there are a number of standard files. In the user documentation, the file AO is described as "private", this may not be accessed by anyone who are not grated read-and-write privileges by the user. The files of an employee may contain information not stored anywhere else. The plaintiff was employed by Memorex Norway, and a conflict developed between him and the executive director. The plaintiff suspected that the executive director "snooped" in his private files, and laid a trap for her: He wrote a letter that passed itself off as a copy of a letter to the European management in Italy severely criticising the executive director that was stored in his AO file of the e-mail system. The letter itself was not mailed. The plaintiff was dismissed in November 1990, and the first instance court of Asker and Bærum heard the case in 1992.17 It was not actually disputed that the executive director has read the files in the e-mail system, but it was maintained that she had a right to access the files of the employees, and there were sufficient additional reasons for the dismissal of the plaintiff for this to be justified. The court held that to access private files without informing the employees were an infringement of privacy. Again, there was no statutory authority to cite, but the court referred to the non- statutory privacy protection, stating that an import aspect of this is
17 Asker and Bærum first instance court, published in Lov&data 32/1992:8- 9. Public Administration and Information Technology
"... that the individual knows what knowledge others have with respect to one self. This implies that an employee should know to what information and documents the employer has access. The Court holds therefore that to access the private file of an employee in the computer system, without giving the employee prior notice in the form of an instruction, a generally known practice or by securing prior consent, must qualify as a violation of the protection of private information which an employee has with respect to his employer."
The relation to the principle of the Snack Bar Case is rather obvious, but that case it not cited by the first instance court of Asker and Baerum.
3. A tentative conclusion: Candlestine electronic surveillance
In this brief paper, an outline of some recent developments in Norwegian case law has been set out on the background of the discussion of data base surveillance in general, and some aspects of national data protection legislation relating to video recordings. The policy issue concerned is the use of electronic devices in surveillance. The Photographic Evidence Case and the Mobile Telephone Case both concerned data that were recorded with the knowledge of the data subject: For automatic traffic control, some of the desired effect is derived from road signs altering the drivers to this type of control, which - of course - has as its main objective to make drivers respect the speed limit. Also the billing system of the mobile telephone system was well known to the subscriber.
97 Compendium to the Erasmus Course
In the Snack Bar Case, the Beer Tap Case and the E-Mail Case alike, the surveillance was candlestine. The decisions should be seen as part of a legal policy aimed at discouraging candlestine surveillance by electronic means. That the legislator is codifying the rules on video surveillance only emphasises this conclusion, it may be seen as one more element in the policy working itself out. For Norwegian law, it must be justified to maintain that the non-statutory privacy protection as a general rule directs the courts to reject evidence gained by candlestine, electronic surveillance. In a jurisdiction where the law of evidence contains few principles apart from the mere reference to the discretion of the courts, this development is of some general importance. But this may also be seen as an example of the privacy issues related to information technology that are not addressed by first generation data protection legislation, which consider the world through the conceptual glasses that make anything not organised as a "file of personal data"18 invisible. This is not satisfactory, we need also to include the surveillance aspect in our policy consideration. As the introductory examples illustrated, the concern for exaggerated surveillance has been basic to the utopian novels that perhaps were the first to formulate the fears in our societies. It is, I am afraid, only one of several examples of the first generation of data protection legislation not being wholly appropriate.19 It is important that the policy discussion is kept
'* National legislation will use different concepts, but it would seem that all are based on some definition of a "file", "register", "system of records" or similar. This is also basic for the international instruments, the OECD Guidelines and the Council of Europe Convention of 1980 as well as the pending draft directive from the European Communities 19 Other examples are given in Jon Bing "Reflections of a Data Protection Policy for 1992", Kenneth Russell Yearbook o f Law, Computers & Technology>, Butterworths, London 1991:164-179 and Jon Bing "An Pnnfinnoc novt norro Public Administration and Information Technology alive. Data protection has been created to meet the challenge of the social impacts of information technology - this is the kind of challenge cannot be defined and met once and for all, it is as dynamic and multi-faceted as the information technology itself.
Continued from last page
Assessment of the Belgian Data Protection Bill", J Dumortier Recent developments in data privacy law, Leuven University Press 1992: III- 138.
99
I
Public Administration and Information Technology
Commercialisation of Government Information and Data Protection
Jon Bing
1. The Nature of Government Information Services based on personal data
1.1. "Personal data " - a wide concept
Personal data1 is basic to many forms of government information services. This is partly due to the rather general understanding of person data with respect to data protection legislation. A typical example of this definition is given by the proposal by the European Communities for a directive on the protection of personal data:2
" ’personal data' means any information relating to an identified or identifiable individual ('data subject'); an
1 In this paper, the terms "data" and "information" will be used as near synonums, and not, for instance, is the defined and distinct meanings these terms have within computer science. 2 Cf "Amended proposal for a Council Directive on the protection of individuals with regard to the processing of personal data and on the free movement of such data", COM(92)422 final SYN 28 (15 October 1992), cf art 2(a).
101 Compendium to the Erasmus ( 'ourse
identifiable individual is one who can be identified, directly or indirectly, in particular by reference to an identification number or to one or more facts specific to his physical, physiological, mental, economic, cultural or social identity; Data presented in statistical form, which is of such a type that the persons concerned can no longer be reasonable identified, are not considered as personal data."3
The definition is by purpose as general as possible, it includes "all information" about the data subject, the only requirement is that the data is linked to an individual through an identifier. This identifier is typically a name or an identification number, but there are many alternatives:4
"A person may be identified directly by name or indirectly by a telephone number, a car registration number, a social security number, a passport number or by a combination of significant criteria which allows him to be recognised by narrowing down the group to which he belongs (age, occupation, place of residence, etc.). The definition would also cover data such as appearance, voice, fingerprints or genetic characteristics."
In this way, a large number of possible identifiers are indicated, including the combination of personal data which by themselves are no identifiers. The identifier itself is also an example of
3 This latter paragraph replaces the definition of "depersonalize" in the original proposal for the directive. 4 Cf "Amended proposal for a Council Directive on the protection of individuals with regard to the processing of personal data and on the free movement of such data", COM(92)422 final SYN 28 (15 October 1992). commentary to art 2(a) page 9. Public Administration and Information Technology personal data. Consequently, personal data is any information that can be related to an individual by an identifier, including that identifier. There are in these definitions no restrictions with respect to the type of data that may become "personal" through the link provided by the identifier. Take a simple example, that of a government data base on the environment specifying characteristics of the territory of the sate in terms of farmland, forest, type of trees, recorded damage by acid rain, water quality, etc. This information is by its nature nearly the antithesis of person data - it is not data relating to living persons, but to plants, land, and watercourses. The location described by these data will typically be geographical co-ordinates. Also land registers, specifying who owns a certain lot of land, may be based on co-ordinates for borders between lots, location of buildings, etc. This information will typically be organised for access using a personal identifier. Consequently, using the personal identifier, all the data of a legal nature, general found in a land register, will be "personal data" subject to the data protection regulations. But less obviously, the land register may be linked through the geographical co-ordinates to the environmental information system, making all the data in this system also "personal data" under the broad definition indicated above, bringing the data protection regulation to bear also on this data - which nature one perhaps would have presumed that was not of a personal character. Consequently, as long as the link between the identifier and the data can be traced with certainty, the data retains its personal nature. It may be queried wether this is appropriate, whether data should lose its personal character at some point where the relation to the data subject is rather tenuous.
Actually, definitions as the one above may by courts, perhaps, be interpreted in this way, seeing the phrase
103 Compendium to the Erasmus Course
"relating to an individual" as an addition criterion, arguing that it is not sufficient to be able to link the data through a long chain of pointers to a personal identifier, the data should also be of a personal or private nature. It would seem that there are examples in national legislation for such additional criteria being included in the definition of "personal data".5
A characteristic of the public sector6 is policies of co-ordination and integration. For data bases in the public sector this implies exchanges of data between different data bases, matching of data bases, use of co-ordinated concepts and definitions, etc. The data protection issues relating to the matching of files are frequently discussed, and reflect the possibility of integration. Already, probably the majority of data bases within the public sector contain personal information. In our context it has the implication that the linkage of files may have as its consequence that even a larger number of information systems are governed by data protection regulation, as indicated by the not so hypothetical example of the environmental information service above. The implication of this is that data protection regulation becomes, with intellectual property regulations, some of the more important background law in the trade of information, also information which one at first might not qualify as personal. This is in itself not necessarily neither good nor bad, but only
5 Cf Jon Bing "'Personal data system' - a comparative perspective on a basic concept in privacy legislation"; Information Privacy December 1979, reprinted in Jon Bing and Knut S Selmer (eds) A Decade of Computers and Law, Norwegian University Press, Oslo 1980:72-91 - see especially the discussion of the definition in the Danish act on private registers p 79 6 "Public sector" is defined in the original proposal to the directive on data protection (art 12(g)), but is here used here rather more informally. Public Administration and Information Technology emphasise that the trade issue still is very much present in the data protection policies.7
1.2. Some examples of governmental information services based on personal data
In discussing government information services offered at a price to private sector, one might like to have concrete examples. One should be aware of the fact that there is a slippery slope from general access rights in freedom of information legislation, administrative procedure law, or data protection legislation to commercial services. There may be a fee for accessing or using certain systems. The relative seize of this fee will determine the commercial nature of the scheme: If the fee is set at marginal costs or lower, it is presumed not to be commercial; if it is set at average cost or higher, it may be qualified as commercial. But the relative level of the fee may also be determined by factors of a very different nature, there is an example from Norway on access to material in the State Archives, where it is stated that the fee is necessary to limit the copying of older material of general interest, as this represent considerable wear and tear of the paper documents.- The following examples are based on a survey of the Norwegian situation.8 The examples will not be analysed in detail, as they are offered here as illustrations only.
7 At one time, it was seen as a central issue as it was claimed that some nations tried to regulate their international trade through data protection, cf for instance Rein Turn Transborder Data Flows (vol I-II). Afips, Washington DC, 1981 and Report of the Committee on Data Protection (the Lindop Committee), HMSO, London 1978. These and other reactions are discussed in Jon Bing Data Protection in Practice: International Service Bureaux and Transnational Data Flows, CompLex 1/85, Norwegian University Press, Oslo 1985. 8 Cf Jon Bing Juridiske aspekter ved etablering og distribuasjon av elektroniske tjenester, Statskonsult, Oslo 1992:142-144.
105 Compendium to the Erasmus Course
• Land register. Land registers contain information on the ownership of real property, mortgages and other easements in such property. The Norwegian system is similar to the German "Grundbuch"-system, where the information in the register by law is given both positive and negative certainty, ensuring that a third party may in hood faith act on the basis of the information in the register being the full representation of the legal status of the property. This makes the information vital for credit reporting, financing, etc. There is a scheme for entering information into the register and other transactions for maintaining the register, and there is a certain tariff of payments for such actions. This is not of central interest in our perspective. But the land registers have been computerised on a national basis. A limited company9 funded in 1987 to develop this system, and the necessary computer systems for the first instance courts that are those agencies charged with the maintenance of the system. The computerisation was financed by loans against the security of future income from the sale of information from the new system. The limited company was granted a monopoly for the sale of information from the computerised system, a number of contracts were concluded, and regulations issued under statutory authority to secure this outcome. One should note that the information to be sold did not have the legal effects by law given to the "official" documents issued by the courts on the basis of the system. But it was thought that the interest of banks and
9 Tinglysningsdata, today the state is the sole share-holder in this company.
106 Public Administration and Information Technology
other financing institutions to obtain speedy access to information on real property would be sufficient to generate substantial income, which in turn would finance the computerisation of the courts. This prognosis failed, and the state took over the company in 1992. A audiotex system based on the same information has been announced, but yet not implemented. In Finland, there is such a system for access to the car licence system. • A register o f movables. A similar national system for categories of moveables was established by a statutory amendment in 1975 in the northern Norwegian town of Brønnøysund, conceived to utilise computer systems. This information is a great practical interest to credit reporting agencies and similar organisations. Regulations issued under statutory authority allows the state to negotiate a contract with a distributor in the private sector.10 The official data bases are every night copied into the computer facilities of this private company,- from which the data is sold according to tariffs laid down in the regulations. • The liquidation quarantine register. According to the bankruptcy legislation, the court may during liquidation proceedings make the director, or members of the board of the bankrupt company subject to a quarantine, denying them the right to similar offices for a limited period of time. The register is also maintained in Brønnøysund, and may be accessed for a certain fee. • Tax lists. A practice that may illustrate cultural differences, is that of publishing the income tax
10 This is currently the State Computing Centre, which in principle is a private company with the State as the sole shareholder Compendium to the Erasmus Course
decisions relating to individuals. The tax legislation specifies that the list shall contain name, address, date of birth, the net value of personal property, the net annual income, and the current taxation.11 These lists may be accessed at the local tax authorities by anyone. But the authorities are not under obligation to permit mass access (access to the information of more than 5 persons). Such requests shall be effectuated by companies12 with which the Directorate of Taxation has contracted for the purpose, and which has a licence from the Data Inspectorate. From such companies, the tax data may be purchased in the form of paper lists or in machine readable form. There are no limitations in the tax regulations to what purposes the data may be used by those gaining access, or how the data may be further processed, but such limitations may flow from licences or regulations issued under the data protection regulations. • The Patent Office also gives access to registered patents, trade marks and industrial designs. This is a rather old information service based on a statute from 1910,13 and allows the Office to charge those seeking access to this rather special kind of personal information. One may in passing note that a patent holder may be both a physical person and a legal person, but under the current data protection legislation of Norway, data on legal persons are qualified as
11 The statute on the inland revenue administration (lingningsloven) 1980:24 sect 8-8 no 2. 12 In 1990, two companies were contracted for a three year period ending 1992. 3 July 7, 1910:7 chapter III and regulations issued under this statute. 13 July 7, 1910:7 chapter III and regulations issued under this statute. Public Administration and Information Technology
"personal data" according to the data protection legislation. This service is in nature not too dissimilar to the services offered by libraries, where obviously books may be seen as personal information on their authors, and where there also are other types of personal information available in the form of encyclopaedia, memoirs, biographies, etc. This also holds true for archives in general, where access is given to uniquae like letters, documents, certificates, manuscripts, etc - a letter would represent personal information at least related to the sender and the addressee.
The examples are rather diversified. They illustrate that the original cause for the collection and registration of personal data often is related to the possibilities of the data subject to pursue their legitimate interests: Data on real property is necessary for offering security for a loan, the tax authority collects data under the shadow of the severe sanctions directed towards them who withhold relevant information. The data subjects and the public authorities are concerned with transaction in which an information service based on the collected data clearly presents itself as secondary. In some cases, as the land registry, the information service may be seen as an accessory to the primary function of the scheme. In other cases, like the organised access to tax information, the information service would seem to have a different objective than the primary activity.
1.3. The market value of personal information
One may ask what constitutes the market value of the personal data in the commercialised services. There must obviously be some reason to believe that private sector is interested in Compendium to the Erasmus Course purchasing the information. Using the examples above as a basis, three different reasons for purchasing personal data may be discerned. First, the legal function of the public service. The typical example is the land registry. The public sector is charged by law to maintain a certain information system relating to property to assist the workings of the financial sector. In addition to the mere information on who, for instance, in the register is named as the owner of a lot of land, the scheme extends the warrant that this information can be relied upon, backing this warrant by strict liability, negotiability functions, etc. Second, the information may be relevant to credit reporting and rating. Actually, the organising of the computerised land registry in Norway offers an example of one and the same data base supporting first, a legal function, and second, an information service for credit rating. It would seem that this is true for a number of the information services mentioned - it is obvious with respect to the liquidation quarantine information, the tax data on income, etc. Third, the information may be relevant to marketing, especially direct marketing. Again the tax data offers an excellent example, a marketing campaign directed towards individuals with high incomes would obviously be an effective sales strategy for many products and services. There may be other causes for purchasing information from public sector, but the examples above do not include these. We will return to this possibility below, especially the possibility of cost savings in private sector by using data from the public sector in their own operative systems. But as an intermediate conclusion it would seem that the market sectors perceived by public sector are rather traditional, and sectors that is easily recognised in the perspective of data protection: credit reporting and rating, and direct mailing and marketing services. Such Public Administration and Information Technology
services have been seen as two major private sector activities intensively utilising personal data. The credit reporting is important as the data is used as the basis for decisions with a direct impact on the welfare of individuals, the data protection issues were addressed in early examples of legislation like the United States Fair Credit Reporting Act of 1973. Direct mailing is important for another reason: The junk mail problem, the visible and often irritating proof of somebody out there having your address and using this to offer you products or services of a quality that may make the data subject believe there is a misconception of his or her interests, sentiments, inclinations, etc. There are, therefore, rather different causes for data protection policies within the two areas. But the data used may be identical, as the service based on tax lists illustrates. This, therefore, may be an indication that regulations of such services may have to distinguish between uses for different purposes. The legal function illustrated by the land registry - but including the intellectual property registers - is something of a qualitative different nature. This is also based on a data base, and - as emphasised with respect to the land register - the data base may in practice be identical to that basic to an information service serving credit reporting businesses. But in these cases one should not see information alone as what is offered, it is the statutory scheme of warranties (which may include strict liability, negotiability related to the registrations, and other legal functions, for instance validity of transactions before bankruptcy) which is the service offered. This is an important point, perhaps not as much with respect to data protection as to intellectual property. For instance, the
111 Compendium to the Erasmus Course
proposed directive on data bases14 includes a non-voluntarily license for a data base
"... made publicly available by a public body which is either established to assemble or disclose information pursuant to statute or regulation, or is under a general duty to do so."
It would seem that to what extent such a clause would be applied to a service based on a data base, but serving a statutory, legal function would rely on the possibility of distinguish these cases from the case of a proper information service.
2. Data protection regulations governing government information services based on personal data
2.1. The statutory basis of an information service
Obviously, data protection legislation will vary between countries. But in discussing the data protection regulations relating to government information systems in a comparative perspective, one cannot limit oneself to the data protection legislation proper. As illustrated by the examples above, many of the services have a statutory fundament outside the data protection legislation - in the law on taxation, on land registry, on intellectual property, on the registration of automobiles, etc. In such cases, data protection considerations may have been taken in the special legislation basic to the information service in question. Often there may lack specific reference to which extent the special legislation is exhaustive, or whether the data protection legislation applies in addition, wholly or in part.
14 OJ No C 156, 23.6 1992 p 4, cf art 8(2). Public Administration and Information Technology
The brief discussion of the tax information service above indicated that the tax regulations cited the data protection regulations, and made them to some extent applicable. In that case, the tax authority only can contact for the information service with a private party that the Data Inspectorate has licensed for the processing of personal data for third parties.15 The restrictions on such processing imposed by the Data Inspectorate must also be respected. But the Data Inspectorate cannot in any way prohibit or block the information service based on the tax register - this scheme is based on explicit statutory authority that cannot be reconsidered by the Data Inspectorate. In the Norwegian scheme, this flows from the legislative power of the parliament that cannot be limited by executive agencies. This small example illustrates that there may be a number of intralegal conflicts to be solved with respect to government information services to the extent that such services have a basis in some statutory provisions. The relation between the information service offered and the statutes providing the legal basis for an agency and its activities may be rather tenuous. Especially when the statute in question is older that than data protection legislation itself, and consequently the relation between the two statutes has not been addressed explicitly when drafting the special statute, the intra-legal conflicts may be difficult to solve. Principles like lex specialis (the more specific legislation being given priority over the more general legislation) for solving such conflicts may be difficult to apply, as there may be a difficult to decide which of the two statutes is the more specific.
15 The Norwegian legislation has a special licensing scheme for three types of businesses, computer service bureaux processing personal data for third parties being one of these types.
113 Compendium to the Erasmus Course
2.2. Relating data processing to its purpose
One principle embedded in many examples of data protection legislation - and also in the proposed directive of the European Communities - is the principle restricting the data processing to the original purpose. In the proposed directive, one of the basic principles- governing the original collection of personal data is that these are collected for16
"... specified, explicit and legitimate purposes, and used in a way compatible with those purposes".
One of the basic principles relating to the ground for processing data is the consent of the data subject,- and such consent has to be based on knowledge of the "purpose of the processing".17 The draft directive therefore illustrates how the original purpose communicated to the data subject creates a limitation of what other purposes may be pursued on the basis of the data collectcd. There are exceptions to these limitations, exceptions that will not be discussed here. But these exceptions are limited, and generally related to contractual relationships in which the data subject takes part, or as part of the exercise of public authority. There is in the draft directive a general exception:18
"... processing is necessary in pursuit of the general interest or of the legitimate interest of the controller19 or of a third party to whom the data are disclosed, except where such interests are overridden by the interests of the data subject."
16 C f art 6( 1 )(b). 17 C f art 7(a). 18 Cfart 2(g). 19 Cf art 2(d) - that body which processes personal data Public Administration and Information Technology
This draft is rather vague. But the test can be confronted with the two main causes for making commercialising public sector information indicated above. Possibly a service provided for assisting credit reporting agencies might be conceived as in the general or legitimate interest of the agencies not overridden by the interest of the data subject, as this might be necessary for operating such a service, which in general also is an advantage to the data subjects. The corresponding argument hardly would hold with respect to direct marketing schemes. And it is suggested that the thrust of the draft directive may be indicative to the policy behind data protection - the Commission concludes its remarks on the proposed provision by stating:20
"In any event the data will usually have been processed for specific purposes, and should not be used for different purposes except in accordance with the other provisions of this Directive."
It may be argued that the restriction to the original objective easily may be overcome by indicating a rather vague objective. But the principle is generally interpreted as requiring quite specific objectives, for instance the commentary to the draft directive exemplifies that to state the objective as "for commercial purposes" would not satisfy the requirement.21 It
20 Cf art 7(0 21 Cf "Amended proposal for a Council Directive on the protection of individuals with regard to the processing of personal data and on the free movement of such data", COM(92)422 final SYN 28 (15 October 1992), commentary to art 7(7) page 17.Cf "Amended proposal for a Conicl Directive on the protection of individuals with regard to the processing of personal data and on the free movement of such data", COM(92)422 final SYN 28 (15 October 1992), commentary to art 6( 1 )(b) p 15 Compendium to the Erasmus Course may therefore be maintained that some of the examples above of Norwegian government information services would have difficulties meeting the test of the principle of a specific purpose - the data has been collected for one purpose (tax administration), but are offered to the private sector for different purposes, mainly credit rating and direct marketing.
2.3. Disclosure to a third party
A governmental information service based on personal data represents, obviously, a disclosure of to a third party. In traditional data protection, the principle of confidentiality is one of the basic interests. This interest is generally implemented in national legislation by a set of "stop-go" regulations. When collection original personal data, the collector - which may be a public officer, a medical doctor, a priest, etc - is made subject to confidentiality provisions. One may see this as the creation of a cell with secure walls, and a safe in which the data are stored under secure conditions. Once within the secure cells of the original collector, one will perceive a few holes in the walls through which personal data are permitted to be communicated. Typically, a medical doctor will communicate data to hospitals and colleagues (with the consent of the data subject), and to public authorities for national medical surveillance systems, which may be general or rather specific. The communication result in the data ending up in another secure cell, but where the provisions making up the walls of the cell, and those making further holes in the cell may be rather different from the original cell. General limitations on the disclosure of data to third parties from personal data files22
22 This is defined rather differently in the data protection legislation of different countries. For the "Amended proposal for a Counicl Directive on the protection of individuals with regard to the processing of personal data Continues next page Public Administration and Information Technology are occasionally included in data protection legislation, and are integrated in the draft directive of the European Communities. This may be seen as a deviation from the traditional "stop-go" strategy, imposing some general requirements on the communication to third parties. It may, perhaps, be suggested that one might further analyse the relation between this general strategy and the conventional strategy of regulating each cell without really taking into account that that cell is part of a complex communication network spanning the distance between the original collection of data and the cell in question.
3. Potential of services based on personal data
3.1 Slotting public data into systems in the private sector
Above, we have identified two major reasons to pay for personal data in the public sector - credit rating and direct marketing. Another reason would be to offer public to the private sector to reduce the cost for maintaining private sector systems. The public sector is maintaining files containing data which to a considerable extent is required and used by the private sector. Simple examples are addresses, social security numbers, income figures, car licences, etc. To some extent this data is used as a basis for information services, as discussed above. But the potential cost reductions cannot be realised fully until a more integrated scheme is launched. An obvious development would be a syntax for public sector data - some sort of standard for electronic data interchange (EDI) similar to the EDI FACT standard for financial transactions. This
Continued from last page
and on the free movement of such data”, COM(92)422 final SYN 28 (15 October 1992), a definition is included in art 2(c).
117 Compendium to the Erasmus Course would allow a system in the private sector to access a system in the public sector, access the relevant data element and slot this element into the right record of its own system. Such systems arc currently coming into use, as the data flows between private and public sectors are being computerised. Currently this is only used to facilitate the exchange of data in the handling of individual cases: A medical doctor may specify the data on illness in an message for the local social benefit office to justify the appropriate benefit for his patient, or an importer of goods may communicate the necessary data for the customs authorities as an EDI message.23 It would seem quite logical that one would consider such an integration, making the public EDI sector serve the private sector by offering the data maintained within the public sector not only as unspecified information, but as formatted and tagged records. If we consider decision making systems, expert systems or other similar systems in the private sector, we will perceive that a EDI standard would make it possible for such systems to put the public sector data to efficient use. Obviously this possibility also implies policy issues. One major issue is related to the principle of the draft directive relating to automated individual decisions.24
Member States shall grant the right to every person not to be subjected to an administrative or private decision adversely affect him which is based solely on automatic processing defining a personality profile.
23 Both these examples are valid for Norway, the system of the custom authorities is designed for the possibility of releasing the goods on the basis of the automatic screening of the messages by the system. 24 Cf "Amended proposal for a Conicl Directive on the protection of individuals with regard to the processing of personal data and on the free movement of such data", COM(92)422 final SYN 28 (15 October 1992) art 16 and the comments in the note page 26. Public Administration and Information Technology
The principle is not very clear, and in our context the details of this principle will not be pursued. But obviously the principle has as an objective to limit the use of decision or expert systems making final decisions on the basis of variables that determine a standard profile for the individual. Obviously this also will be an issue for other public policies, for instance whether the public sector is under some sort of political obligation to make the data collected by a common effort available to the private sector in order to reduce processing costs. Or - quite the opposite - whether the public sector should not make data available to the private sector without explicit consent, or perhaps statutory authority, as the data has been obtained under their authority, which limits the liberty of the public sector of making it subject to trade - especially when the transfer to private sector also would imply use to different purposes, or loss of control to what purposes the data would be used.
3.2 Constructed documents
In establishing EDI systems for serving the private sector, the public sector would find that data protection regulations are substantial limitations. But public data may be used in such a way that no "personal" data are communicated to the private sector. The point may be illustrated by the way of the Canadian federal freedom of information legislation. According to this legislation a person may access any "record" in the public sector, including computerised records. The legislation also allows access to records that do not exist at the time the request is made:
119 Compendium to the Erasmus Course
"... for the purpose of this Act, any record requested under this Act that does not exist, but can, subject to such limitations as may be prescribed by regulation be produced from a machine-readable record under the control of a government institution using computer hardware and software and technical expertise normally used by the government institution shall be deemed to be a record under the control of the government institutions ..."25
The interesting implication of this principle is that, for instance, a private organisation may request a survey based on data in the public sector. The data basic to the survey may be personal, but the result will be in statistical form, and consequently are not personal data according to the definitions commonly in use.26 Considering the detailed and diversified data collected by national statistical agencies, this possibility would seem to be rather important. The data could be used for market surveys, political polls, and numerous other purposes where co occurrences, often of complex nature, would be of interest. It is easy to imagine that due to the varied objectives of the public sector, and the authority implied by public agencies collecting data, one might be tempted to create systems where the public sector maintained filed of personal data, and where the private sector purchased surveys based on such data. One should, however, consider that the Canadian example was derived from the freedom of information legislation. This indicated that there may be other interests than the commercial in such systems. Perhaps a country has a macro-economic model
2> OECD ICCP (83)15 Legislation on Access to Administrative Documents and the Protection of Privacy; OECD, Paris 1983. 26 For instance the definition in "Amended proposal for a Counicl Directive on the protection of individuals with regard to the processing of personal data and on the free movement of such data", COM(92)422 final SYN 28 (15 October 1992) art 2(a) cited above. Public Administration and Information Technology
based on data collected continuously be the statistical bureau. Perhaps the government is basing its economic policies on the results produced by this model. What if the opposition requested alternative runs, based on different variables in the model? This simple example must suffice in indicating that the exploitation on personal data of an individual nature to produce surveys and other forms of statistical results may be a policy issue of more than an academic interest.
121
Public Administration and Information Technology
Automated Systems and Claim Realization in Social Security Organizations
Jon Bing and Dag Wiese Schartum
Paper presented to the workshop: "Information Technology Within the Service Sector -Work Conditions and the Quality of Services," Berlin, October 25-26, 1984.
1. A Point of Departure: Fixed Rules and Discretionary Rules
Rules may be regarded as tools for decision-making. These tools have different qualities depending on how they are built and structured, and it is possible to separate these qualities into many different categories. This paper will focus on that quality which dictates the extent to which the rule determines the result of decisions, that is, the extent to which the decision-maker is restrained by the rule.
123 Compendium to the Erasmus Course
Recent Norwegian legal research has developed a theory of norms.1 This theory may be useful for further discussion of the degree of liberty in decision-making according to legal rules in social security organizations. In this paper, the theory as such will not be discussed, but rather discussion will be limited to presenting two categories of rules that are formulated in the theory. The first category is what may be referred to as "fixed rules," that is, rules structured to indicate a complete list of legal and relevant factors to be considered. For example, the Norwegian National Insurance Act sets forth the parties protected by the National Insurance Scheme ("Scheme") as follows:
1. All persons resident in the Realm shall be insured in accordance with this Act. The same shall apply to any person not resident in theRealm if he
a).... (or) b ) .... (or) c ) .... (or) d) ....2
© Copyright reserved 1986 by Jon Bing and Dag Wiese Schartum. * We would like to thank the Stiftung Volkswagenwerk, West Germany, which has financed the project "Work Process Information Technology and Service Quality in Social Welfare Institutions," of which this article is a part. 1 See T. ECKHOFF & N. SUNDBY, RETTSSYSTEMER ( 1976); N. SUNDBY, OM NORMER (1974). 2 Lov om folketrygd (National Insurance Act) of June 17, 1966, No. 12, § 1-2. Public Administration and Information Technology
This structure of a principal rule and four alternative exceptions is similar to a check list with a YES or NO answer to each alternative. In interpreting this rule, no additional considerations could legally be made other than those listed in the section. For example, the decisionmaker is not in a position to determine whether the result is reasonable and fair. Even if a person, according to the decision-maker's sense of justice, should be protected by the Scheme because of a high degree of similarity with the cases covered by the Act, the result is fixed because the facts do not coincide with those described. Other rules render some freedom to the decision-maker regarding the considerations that could be taken into account. Such rules function only as a framework for decision-making. In accordance with the norm theory mentioned, this category is comprised of "discretionary rules.” 3 An example of a discretionary rule may be found in the National Insurance Act which provides, "[w]hen it is reasonable to do so, the National Insurance Institution may decide ...."4 This rule is added to a set of fixed rules and gives some freedom to decide within the frame "reasonable." Thus the result is not predetermined but requires a legal evaluation. Discretionary rules lead to decisions where the following two characteristics should be stressed:
1. It is not possible in advance to give a complete list of the factors to be considered. In principle it is always possible that new, relevant factors will occur.
3 See Bing, Legal norms, discretionary rules and computer programs, in B. NIBLETT, COMPUTER SCIENCE AND LAW 119-36 (1980). 4 Lov om folketrygd (National Insurance Act), supra note 2 at § 5-4 (amended June 19, 1970, No. 67, and June 6, 1975, No. 33).
125 Compendium to the Erasmus Course
2. It is not possible to decide in advance what weight the different factors should carry.
In this short description of fixed and discretionary rules certain qualities are emphasized to expose their principal differences. Although in Norwegian statutory rules a spectrum of fixation/liberty can be observed, the distinction between the two categories is useful in a further investigation of rule qualities.
2. An Example Of Rule Combinations In Norwegian Statutory Law
In traditional Norwegian statutory law both fixed and discretionary rules are frequently found. Very often combinations of the two types of rules are used. To a large extent, political arguments can be formulated in fixed rules. Regardless of how accurate and detailed the fixed rules are, however, it is probable that some individuals are not covered by the rule even if they have the same social needs as the groups included. In European political tradition, principles such as equality and fairness hold a strong position; therefore, a result such as the one mentioned will probably conflict with the named political principle. Norwegian statutory law uses discretionary rules in order to prevent unfair results. This is often done by formulating rules such as, "[w]hen it is reasonable to do so, the National Insurance Institution may decide...."5 Thus, in combination with a fixed rule, a discretionary rule functions as a safety net for individuals whom fixed rules are unable to catch. This makes it possible to formulate legal rights in accordance with political intentions.
5 Id. Public Administration and Information Technology
Superior authorities may give directions to their subordinates as to how a discretionary rule should be interpreted. The staff, comprised of civil servants, would be obliged to follow these directions. In this way, the safety net can be used as an active and flexible tool for the top-level administration. This could be done by making the width of mesh smaller or bigger in response to societal changes. It is also possible for feedback from society to influence the contents of rules to make them binding on the civil servants. Based upon Norwegian experience, such feedback can take at least three different forms. First, it is possible for the civil servants to channel their experiences upward to higher levels in the hierarchy. One reason for this occurring could be remonstrances from working with rules that give unfair results. Secondly, results from fundamental complaint cases could influence the contents of rules. These are changing processes activated by citizens. The last point is represented by changes activated by public discussions in mass media. In principle, this is a possibility for those social groups that often act as political groups and interest organisations. If adjustments are made, they will be put into action by circulating a memorandum, or "circular," to all relevant staff members. Changing an act is a complicated task. Formulating a circular is a much easier way of changing the contents of rules. Formally and legally the use of circulars to change discretionary rules must be done within the framework of the act, for example, within the frame "reasonable" and the definition of this word as stated in the preliminary works of the act. In reality, and especially when stricter discretionary rules are made, the interpretation according to the circulars may move beyond a strict and legally correct interpretation. Although one may criticize this result, a vital practical function is served: as society changes, pressure from the citizens on the public administration may activate changes in rules, even if the demands for changes
127 Compendium to the Erasmus Course
go beyond the frames set up in the act. Often this will occur as a prelude to the slow process of changing an act. Changes in the interpretation formulated in circulars result from a hierarchical decision-making process: the top level decides and instructs the subordinate staff. There is, however, the possibility that the civil servants will act independently and channel their experience up to the top level administration. This presupposes that open channels exist and that these are regarded as effective ways to be heard. If not, it is probable that the civil servants will find other strategies to cope with an unsatisfactory set of rules and thereby improve the interaction conditions with clients. Discretionary rules make it possible to find better coping strategies because the interpretation might easily be changed. This could simply be done by bringing in new factors or, more importantly, by changing the weight presumably to be carried by the different factors. It will be very difficult for the superior administration to control and prevent such alterations.
3. Norm Representation in Automated Systems
In Norway, the majority of legal decisions in social security organizations are produced by computerized systems called deontic systems. There are actually several types of deontic systems, but they have one common characteristic: their programs represent legal norms. To the extent that a computerized system represents legal norms, these norms must be represented in the program by the terms of the programming language. This language may be conceived as an artificial language not unlike some form of stunted English. But it has a property not usually found in languages: it is unambiguous. There is a direct one-to-one relationship between the terms chosen in the language and the actions carried out by the computer. Given the same instructions, Public Administration and Information Technology the same computer will always carry out the same sequence of operations. When programming legal rules in deontic systems, all factors that could be considered must be represented in the program. In addition, all relevant factors must be given a certain weight. This means that discretionary rules cannot be represented in a deontic system. This does not imply that one cannot make decisions with a certain affinity to discretionary decisions. By reformulating the norm, it is possible to make a list of all factors that could be considered when deciding what is "reasonable." The numbers of factors can be made almost unlimited and the factors can be given a mutual relative weight. This could result in a flexible decision-making process within the framework of the program; however, when society changes, new factors cannot be considered without changing the program. If the discretionary norm is not reformulated, the discretionary decision will generally be defined as a manual sub process. The form to be filled in by a civil servant will have a line for the criterion "reasonable?" This is filled in with a "yes" or a "no," and the computerised system takes it from there. The legal norms are represented by the computer programs. But the facts of the case to be decided are not part of the program. The program has, so to speak, neatly labeled boxes in which the facts are to be placed according to a well-defined schcme. In principle, this makes the computer program quite similar to a form that has preprinted legends explaining which data are to be filled into its various boxes. Somebody has to fill in this form. A distinction may be drawn between two strategies for capturing the necessary data, one strategy on the matching and linking of different computer systems and the other on a civil servant's providing the necessary data as if by filling in a form. When done manually by a civil servant, the filling in of forms is an expensive part of the process of decision-making. It
129 Compendium to the Erasmus Course is therefore tempting, as a cost-saving measure, to eliminate the civil servant by computerization of these operations. As mentioned, this could be done by matching and linking different computer systems to other registers. In the following sections, some implications of this strategy are explored.
4. Matching and Linking
The Norwegian system for housing aid ("bostøtte"), which is computerized to quite a high degree, may serve as an example of a complex system of matching and linking.6 The system has been operated by the State-operated Norwegian Mousing Bank since the early 1970's. It is based upon a plenary decision by the Parliament that grants aid to certain households representing a fraction of the difference between "real housing costs" and "reasonable housing costs." To apply for such aid, one need only identify oneself to the system. The system then accesses the population register to check the number of individuals who are living at the applicant's address and their relation to each other, the register of the tax authority to calculate the household's income bracket and the register of the social security administration to check the status of the applicant or other members of the household with respect to the different benefits. Finally, the system accesses the registers and tables of the Housing Bank to retrieve the correct rates for calculating an eventual grant. Twice annually more than 100,000 decisions are made in this way. Once part of the system, the decisions will be made during subsequent periods until the applications are explicitly withdrawn. The decisions are printed out and mailed by the computer system. In the case of a negative
6 Hafli, Bostøtteordningen, 32 PAPERS ON COMPUTERS AND LA W ( 1978); Rynning, En fremstilling og vurdering av reglene om tildeling av bostøtte. 17 PAPERS ON COMPUTERS AND LA W (1976) Public Administration and Information Technology decision, the computer program identifies the exact reason for the decision, and in the case of a positive decision, a postal check is produced. To make such automation possible, all vague criteria must be altered. Vagueness could be eliminated by constructing a sub-norm that decides what the vague expression should mean in that special context. In the Norwegian housing aid system, one vague criterion is that aid depends on the income of the "household." To automate the decision, it was necessary to decide what "household" should mean in this context. This was accomplished by entering the population register's survey of persons residing in a single apartment and matching these to persons indicated in the application form. By linking and matching to another register, the vagueness is eliminated. This example also serves to illustrate the definition aspect of such linkage. The small boxes in the different systems all have labels. The system designer is hunting for another system that already has a box with the same label as the box in the new system; however, two criteria labeled with an identical word in two different acts may not necessarily be identical. The context or purpose may cause the detailed interpretation of identical words to differ in important aspects. The linking of two systems actually states a legal proposition that within the context of both systems this criterion is identical for all possible cases. This in itself is a strong proposition, but one should also be aware that this interpretation embraces more than the meaning of the criterion in a strict sense; one must also include the pragmatics of the system. Two examples illustrate this. First, the data may refer to different time periods. When utilising income figures of the tax administration, one will always utilize dated figures. In Norway these will reflect an income situation that is between nine and twenty-one months old. This has consequences when used in certain systems. For instance, the system for housing aid bases its income criterion on these
131 Compendium to the Erasmus Course
figures. Should a two-income household be reduced to only one income (which is not unusual in the case of childbirth or job layoff) there will be a time lag of between nine and twenty-one months before this is picked up by the routines of the system. Actually, in this respect a manual safety-valve has been introduced to avoid some of the adverse consequences of a rigid application of the routines. Second, the requirements of verification may be different. An anecdote, which is unverified, may illustrate this. In Norway, the Central Bureau of Statistics ("CBS") regularly communicates the identities of deceased individuals to the National Insurance Institution ("Nil"). The Nil checks whether these individuals still draw social benefits, and if so, a request for an explanation is made to the local social security office. In one case, the local office responded with some amazement, as the Nil had implied that all the old age pensioners of its community had died. This is obviously caused by the more lenient verification procedures of the CBS; the result in the form of slightly unreliable statistics is obviously easier to accept than old persons' suddenly being denied their rightful pension, with results endangering their welfare or health. These two examples may suffice to emphasize the importance of linking and matching registers, but this practice also creates a complex pattern that may be difficult to overview because systems relate to each other in complex ways and may create anomalies in the normative structure. This has been discussed especially with respect to the interwoven fabric of tax and social security law, where anomalies known as "poverty traps" have been identified. A major reason is that the rules rely to a large extent upon the same criterion: the gross income of an individual or a family. As the gross income increases social benefits are excluded, and this may cause a negative growth in the available income for abnormal situations. An example has been reported from Denmark, where the available income in 1972 for a single supporter with three children was reduced in Public Administration and Information Technology the interval of gross income between 30,000 and 90,000 Danish crowns.7 An example having the potential for such an effect may also be found in the Housing Aid System. To be eligible for housing aid, one must belong to certain qualified groups of persons. One criterion is that a loan based on social indications for the rehabilitation of an old house or a fiat has been granted by the Housing Bank. These loans are quite inexpensive compared to the usual bank loans; however, if no such loan has been obtained, one may nevertheless have to take on the more expensive loan. The outcome is that if one obtains the inexpensive loan, one also may benefit from housing aid; whereas if the same person has to take a more expensive loan (and as a consequence has a relatively higher increase in housing expenses) no housing aid would be forthcoming. The system's tendency to reinforce itself emerges: those aided once are automatically qualified for further aid. The reason behind this situation will rest fully on whether secondary use of the decision by the Housing Bank based on social indications is appropriate with respect to the Housing Aid Scheme. One of the major concerns with computerised systems within public administration is this growing complexity and interdependency. Some aspects have been mentioned, but there are perhaps more serious issues at stake. Such structures make it difficult for the rulemaker to foresee the consequences of changes in one system. This lack of insight may make the rulemaker reluctant to advance drastic changes, as the effects are difficult to predict. The computerized systems may contribute to a growth in complexity of the legal system that he does not want but finds difficult to avoid. They also introduce a new kind of administrative consequence of amendments. Amendments not
7 Byrgesen, SISIFOS-A Law Preparation Game, IBM Costomer Executive Seminar on Social Security, Brussels (1975).
133 Compendium to the Erasmus Course
provided for in the existing programs will require reprogramming, which is time-consuming and therefore introduces rigidity in the public administration. Such issues may make the growth of computerized systems within public administration a concern with respect to the rule of law and appropriate control of the decision-making systems. To the top administration, the introduction of deontic systems, utilizing matching and linking strategies, means possible reduced wage expenditures while potentially increasing economic transfers to the citizens. The information structures linked together in integrated systems could be used to point out those individuals entitled to a social benefit. This knowledge could be used to inform the relevant groups of people by distributing "tailored" information to groups with an assumed underconsumption problem. In Norway such an arrangement is made for old age pensioners. Every person receives information concerning old age pensioners' rights two months before reaching the pension age of sixty-seven years. This arrangement could be intensified by splitting the pensioners into smaller groups in order to sort out only relevant information. If matching and linking is not possible, a similar cost saving method could be used. If vague criteria are split into a handful of strict criteria, the citizens themselves could more easily fill in the forms and feed the required information into the deontic system. In this way the citizen is prevented from making his own evaluation. Besides, strict criteria can more easily be controlled.
5. Controlling Results by Altering Input Criteria
We have previously emphasized the difference in quality between discreetionary and fixed rules as it affects the degree of freedom afforded the decision-maker. We have also showed how Public Administration and Information Technology deontic systems require a fixed representation of norms. This does not mean, however, that the decision-maker (civil servant) is prevented from influencing the results (the output) in deontic systems. Indications of attempts at "back-seat driving" have been experienced. If the civil servant is aware of the relationship between the values ascribed to criteria and the outcome of the decisions, the civil servant may be tempted to describe the case in such a way that the computerized systems arrive at the result preferred by the civil servant. This may be seen as a struggle between the human decision-maker and the computerized system, where the civil servant uses his position as a reality interface to reclaim the control over the result that has been lost to the computerized system. Use of the metaphor "back-seat driver" implies that civil servants do not have the right to control the results. It is possible, however, to give civil servants a limited duty to ensure "fair" results by controlling the input of criteria. By putting the civil servant in the "front seat," vague criteria can be used as a means of injecting flexibility into the deontic system; however, this presupposes that it is possible for the civil servant to overview the program, thereby enabling him to predict the importance assigned to different classifications. In addition, this presupposes that the classification is made in regard to a specific decision. If criteria are accessed by matching and linking other registers, it will be impossible for the classifying civil servant to overview the effect in regard to the many different decisions to be taken on the basis of that single classified criterion. For the citizens, the deontic systems will most likely mean that the information required will be handed over to the welfare administration by filling in a form. Citizens with sufficient insight will then have approximately the same possibilities for "back-seat driving" as civil servants; however, only a small percentage of the population will definitely have the
135 Compendium to the Erasmus Course knowledge required. This could split people into two groups, the initiated and the uninitiated. As for matching and linking, the citizens will lose any control of the information used in the decisions. Even if they know that information on their income is handed over to the tax authorities and used in taxation, they will not know that a number of other organizations, for example, welfare authorities, will make decisions based on the same information.
6. Reviewing Decisions Made by a Deontic System
Traditionally, the citizen has the right to appeal a decision made by public administration. Such an appeal generally will be made to a superior public agency, which will look at the decision making procedure, the data on which the decision was based and the law which was applied. One of the major problems of computerized systems in public administration is the application of the appeal system to the decisions stamped out. Obviously, one may test in a traditional way whether the data specified for the decision were "true" or adequate. This could be done by reviewing the filling in of the form discussed above, and it should be appreciated that this in itself is important. As mentioned previously, however, there is a tendency to design deontic systems so that the specification of facts is limited to strict criteria or only to the secondary use of data collected from other systems. To the extent that this is achieved, the review of the specified data easily becomes a-trivial task of checking whether the data are correct. In addition, to the extent that data have been communicated from another system, one may have a problem of authority. If a decision of agency A has been appealed, obviously that agency may amend an erroneous decision; however, the Public Administration and Information Technology result may hinge upon a fact that has been lifted out of the system operated by another agency, B,.and agency A does not have the authority to review decisions of agency B, even when they disagree with that decision. Therefore, the linking of systems is also to some extent a delegation or division of authority between agencies that may have an effect on the control regime for the decisions supported by the deontic system in question. One may also review in a traditional way whether the correct procedures are followed, but this will be limited to that part of the system which is not computerised. In the computerized part of the system, the same procedure wilt have been followed for the case under review as for all the other cases that have been Hushed through the system. This will hold true as well for the law applied to the extent it is contained in the program. Consequently, the meaningful review will be directed only towards the human subsystems of the total decision system. The activities of the civil servants acting as the reality interface to the deontic system may easily be trivial. It obviously is of little interest to take the same data and run these for a second time through the computerised system; the results would necessarily be identical to the first. This may lead to a situation where the traditional review of decisions is found to be inadequate. At the same time, there has been a relatively high appeal rate with respect to computerized systems. This is probably related to the form in which decisions have been communicated to the citizen and may be an indication of badly formulated decisions and documents, all of which may have been corrected over time as systems of an improved design have been introduced. Actually, computerised systems may contain tailor-made and lengthy explanations and be much more informative and individual within the context of mass administration than preprinted forms or other traditional means.
137 Compendium to the Erasmus Course
As to the Norwegian Housing Aid system, initially decisions could be appealed to the Housing Bank. The system made approximately 100,000 decisions every six months, and of these some ten percent were appealed. The Housing Bank did not have the necessary staff to handle that number of appeals manually. Thus the review would seem rather futile, as all data were collected for secondary use from other systems, and no human intervention had taken place. The review was restricted to checking that the computerized system had not broken down. Finally, the appeal system was abandoned.8 Today, it is possible to bring: a decision under review only if the data on which it is based are proved to be erroneous, which is rarely possible for the reasons stated. This provides a justification for changing the perspective from decision-makers to rule-makers. The norms embedded in the computerised system would obviously still be open to challenge. The public agency in question would probably be reluctant to admit that the program was an inappropriate representation of the relevant law, but this issue may also; be taken to court. In a situation where the focus is on the program and its representation of the law several issues will become of interest. One issue will be the importance placed by the court on the fact that this program has not only produced one, but several hundred thousand decisions. In most legal systems reference is made to the use of a legal norm. If a legal norm has been applied over a long time and in a great number of cases, it will be recognized as law. But this presumes "use" in the sense that civil servants or lawyers have repeatedly applied the norm. In the case of the deontic system, there may be only one programmer making a few inappropriate statemcnts-and this is then repeated over and
8 A. FRIHAGEN, FORVALTNINSRETT, del 2 (1975). Public Administration and Information Technology over again by the computer. Will or should the use of a computer be recognized as relevant use? Another issue will be the ex officio amendment of prior decisions. If, for instance, a court finds that the program misrepresents the law in some detail, this would in a traditional system have the effect of amending the decision under review. The public agency in question would have no practical means of identifying other decisions to which the same reasoning could be applied and corresponding amendments made. It would probably rely upon persons reacting to the information derived from the court's decision and applying to the agency for a review of their own prior decisions. If the decision of the court were found to be relevant, the public agency would amend its prior decision accordingly. With respect to a computerized system, however, the public agency will be able to identify all the prior decisions in which the critical element may be found. The question arises whether the agency should then be obliged ex officio to make such amendments. Also, if the amendment has a negative economic effect what relevance may this have for the review by a court? Instead of correcting one erroneous decision, the court may find itself correcting several thousand decisions with economic consequences of quite another scale. One alternative to use in reviewing mass administration is to change the perspective from the individual decision to the aggregated mass of decisions. An example is a qualitative review made by the National Insurance Institution in Norway, based on two rather simple questions:9
1. "Are all necessary facts collected?"
9 RIKSTRYGDEVERKET (NATIONAL INSURANCE INSTITUTION NORWAY ), KVALITETSKONTROLL AV SAKSBEIIANDLING- SLUTTRAPPORT FRA EN UNDERSØKELSE I 14 TRYGDEKONTORER (1983).
139 Compendium to the Erasmus Course
2. "Is the result in conformity with the law?"
Without discussing this in detail, it is interesting to note that the fractions of errors with respect to the two questions ranged from 0.5% (question 1 with respect to refunds of doctors' bills) to 23.8% (question 1 with respect to training, reeducation, etc.-and in this case, the statistical significance is somewhat doubtful due to the sampling). This qualitative review was made of decisions in a traditional mass administrative system with a computerized system supporting the manual decisions. The computerised systems do not make mistakes, but this is not very reassuring if the data specified to the civil servants have the quality indicated for some categories in this review.
7. Deontic Systems from the Legislator’s Point of View
When statutory language is translated into programs for deontic systems, it will always have the form of an interpretation. The programmer will read the natural language texts and present his understanding of the norms in the unambiguous terms of a programming language. This could mean a change of content of the statutory norms. In this way, the programmer actually may function as a legislator; however, this function is rarely realized. When unfocused, one may claim that the Parliaments legislative power is weakened as a result of a programming procedure placed outside political control. Without pursuing this example, it serves to emphasize that programming is actually interpretation, and that an authoritative interpretation determining in detail the legal norms is actually rule-making. A decision to develop a deontic system must also be interpreted as a decision to delegate some Public Administration and Information Technology
legislative power on a detailed level to those responsible for the programming. By looking at the situation of the legislator from another angle, it could be claimed that there exist close links between traditional Norwegian political and legal language. Both "occupational" languages have a discretionary nature that makes it rather easy to translate political goals into legal rules. Automation and use of deontic systems could mean that these links will be weakened. Again, an example from the Norwegian housing aid system can illustrate this point. In the Parliamentary document concerning the housing aid system10 the following political goal was formulated:
If the housing aid system shall limit the housing expenditures for those in need of aid, to a reasonable degree, and thereby have an effect as an element in the equalizing policy, it is necessary that the aid, at any time, but within the limits of practicability, be adjusted to the household resources and housing expenditures.
It may be argued that this statement contradicts what actually has been the result of the housing aid system. Because the income figures used come from the tax administration, the information will reflect the situation between nine and twenty-one months prior to the time of application. As a result of this linking and matching with the tax administration's registers, people with a deteriorated economic situation, with an often acute need for help, are not caught up by the housing aid system. This effect could easily be predicted. The organisation of the system, however, makes it impossible to bring aid in accordance with political goals without large expenditures. One may claim that this is an example of an automatized system that has dictated
10 76 STORTINGSMELDING at 80 (1971-72).
141 Compendium to the Erasmus Course legal rules that are out of line with the original political intentions, thereby making difficult a translation of political views into legal rules.11 In our political tradition, the principle of equality is an important maxim to the legislator. It is often said of computerized systems that they will always decide equal cases equally and thus represent a guarantee with respect to the political principle of equality. This is true, with the important reservation that only cases described in an identical manner will receive the same treatment. The computerized systems rely upon the specification of facts by some human somewhere in the system. Only if the descriptions are adequate will the decisions fulfill the principle of equality. The computerized systems will see only those aspects of the case that are made visible through the legal norms; those facts identified as relevant by the norms will be the windows of the system into reality. The rule-maker may choose his facts to make the computerized system as efficient as possible, for example, by choosing facts that could be accessed by linking and matching the tax administration's register. It may be appropriate to conclude this paper by pointing out that the computerized system is a dynamic force that changes not only the decision-making system, but also the rule-making and the legal norms by which these decisions are made. Such changes will of course also influence all levels and all parties in social security organizations.
11 See, e.g., Bing, Edb: Mulighet og problem ved forenkling av regelverk, COMPLEX 13 (1983). Public Administration and Information Technology
Literature
Bing, Legal norms, discretionary roles and computer programs, in B. NIBLETT, COMPUTER SCIENCE AND LAW (1980).
Bing, Edb: Mulighet og problem vedforenkling av regelverk, COMPLEX 13 (1983).
Byrgesen, SISIFOS-A law preparation game, IBM Customer Executive Seminar on Social Security, Brussels (1975).
T. ECKHOFF & N., SUNDBY, RETTSSYSTEMER (1976).
A. FRIHAGENS FORVALTNINGSRETT, del 2 (1975).
Hafli, Bostøtteordningen, 32 PAPERS ON COMPUTERS AND LAW (1978).
B. NIBLETT, COMPUTER SCIENCE AND LAW (1980).
Rynning, En fremstilling og vurdering av reglene for tildeling av bostøtte, 17 PAPERS ON COMPUTERS AND LAW (1976), reprinted in 14 JUSSENS VENNER 47-91.
RIKSTRYGDEVERKET (NATIONAL INSURANCE INSTI TUTION, NORWAY), KVALITETSKONTROLL AV SAKS- BEHANDLING-SLUTTRAPPORT FRA EN UNDERSØKEL SE I 14 TRYGDEKONTORER (1983).
N. SUNDBY, OM NORMER (1974)
143
Public Administration and Information Technology
Introduction of information technology in social insurance institutions: Political objectives and organizational conse quences, history and future
Dag Wiese Schartum
Introduction From 1977 two on-line computer systems have been developed and implemented into the local offices of the Norwegian na tional insurance service. This technological change has not only influenced questions concerning cost saving and efficiency, there is also a connection between the use of these computer systems and the realization of other political objectives such as delegation of authority and tasks, decentralization and questions concerning equality and social justice. First, before the computer systems are described, I will outline the National Insurance Scheme and the administrative structure of the national insurance services. My description of the computer systems will focus upon special relations between the properties of the existing computer routines and political objectives such as the quality of services provided and a socially just national insurance system. Discussing these relations, I will point to requirements that are not satisfied by the current com puter systems, and intimate in which direction the systems should be developed.
145 Compendium to the Erasmus Course
I emphasize that my aim is not to conclude on expert systems in law in particular, but rather - from a practical and le gal point of view - point at some of the actual imperfections of the current systems as well as the uncovered needs that exist.
The National Insurance Scheme The National Insurance Scheme constitutes a cornerstone in the Norwegian welfare state and encompasses compensation for loss of income in various life situation such as old age, disability, sickness, rehabilitation periods and unemployment. It also pro vides child benefits and benefits to single supporters, as well as certain other types of economic support. A major part of the health institution and primary health care expenses are also paid over the National Insurance Scheme. The pension benefits of the Scheme may be split into two categories; the basic benefits and the supplementary benefits. Every insured person receives basic pensions which are estab lished and adjusted by Parliament at least twice a year, usually according to the general development of wages and prices. Peo ple with income exceeding a certain amount, earn so-called pension points, and the sum of these points constitutes the basis for the assessment of supplementary pension payments in case of retirement, disability etc. Short term benefits are, roughly speaking, based on the income figures at the time of sickness or unemployment. People pay contributions to the Scheme accord ing to earned income, i.e. the basic pensions are non- contribu- tionary. In relation to tax deduction, the majority of the benefits of the Scheme are regarded as income, which implies that taxes are deducted from pensions and short term benefits as if it were ordinary wage. This relation between income and national insur ance benefits may illustrate the status of these benefits. One of the main objectives of the development towards the National In surance Scheme has been to make the means tested poor relief Public Administration and Information Technology benefits of former days' redundant, and instead build up a con tractual-like system without social stigmatizing effects on the recipients. The National Insurance Act was put into force in 1967. The estimates of the costs of the Scheme showed to be far too optimistic, especially within the pension sector. This is both due to higher pension earnings and a higher number of recipients than predicted. In 1982 the national insurance costs constituted more than 14% of the Gross National Product or approximately 30% of the total government spendings. Although there is a rather broad political agreement with respect to the role of the Scheme, it is important to note that there has been a certain political unrest due to the increase of costs. The implementation of on-line computer systems in the national insurance services has, in other words, been carried out in a situation when the possibilities of reducing costs, both with regards to benefits and administrative expenses have been in fo cus.
Outline of the local organization and tasks The National Insurance Schemc is administrated by the National Insurance Administration, 19 regional bodies and 451 local of fices, one in each municipality. The regional units deal almost only with disability pensions and rehabilitation aid. It is conse quently correct to say that case processing mainly is carried out on two levels; in the local offices and in the central administra tion. The large majority of local offices are extremely small. Approximately 50% of the offices have less than 6 employees, while approximately 93% have a staff of less than 21 persons. Even if some rather large offices exist in the largest towns, the major part of local case processing work is, in other words, characterized by a situation where the possibilities of specializa tion is very limited.
147 Compendium to the Erasmus Course
The work in the local offices are of a legal character in the sense that the employees in charge of case processing inter pret the provisions of the relevant Acts, subsidiary legislation and administrative guidelines and instructions. The employees in charge of case processing, however, have no juridical education, most of them have only attended the compulsory primary and secondary school. When taken on to an office, the employees are offered training courses within the various types of benefits. However, the number of courses per employee varies greatly, which means that practical experience is an important factor de ciding the skill level of the staff. The provisions, guidelines and instructions which apply to the case processing work are of a considerable volume, and become considerably more comprehensive as we move down wards in the legal hierarchy. A total of 100 provisions are made under the National Insurance Act and the total number of sec tions of the Act and subsidiary legislation exceeds 1000. Ap proximately 170 circular letters with instructions and guidelines, exceeding 2600 pages (A5) links to provisions in the Act and the subsidiary legislation. When regarding the organizational structure, educational level and the legal complexity of the case processing work, one could expect that this would lead to a unification of the smallest offices to make them stronger. Alternatively, one may expect that decision-making power be transferred from the local offices to the National Insurance Administration. It may therefore be surprising to hear that the Norwegian national insurance services has developed in the very opposite direction; emphasis is put on retaining the extremely decentral ized organizational structure and to delegate more decision power from the National Insurance Administration to the local offices. Furthermore, in connection with the introduction of on-line computer systems in the local offices, principles of job rotation have been implemented in the local organizations, Public Administration and Information Technology which implies that each employee could be set to execute every type of work within all sectors of benefit. This development requires an explanation of why such strategies are selected, i.e. which political objectives are the ba sis of these strategies. But it need also a clarification of how and by what means this increased emphasis on decentralized case processing is made possible
Background for the administrative choices The referred administrative choices could both be regarded on the basis of general political objectives as well as on the back ground of the special history of the national insurance services. The 1950'ies and 1960'ies were decades of centralization and strong growth in the public administration. In short, it was found necessary to change this situation and stop or at least re duce the growth of the sector. The strategy that was formed had two main elements: planning and delegation. Central authorities were found to be far too loaded with case processing work, with the result that too little time was spent on planning to achieve more efficient use of resources. In order to put more emphasis on the central planning functions, delegation of single cases to the local offices was regarded as important means. The argu ments in favor of delegation were not only related to increased efficiency, it was also stressed that decisions should be taken in as close relation to the public as possible, and that such a struc ture of the decision-making process would provide better service to the citizens and be more in accordance with ideals of democ racy. This emphasis on the importance of the local administra tion fitted well into traditions of the national insurance services. The strong position which the National Insurance Administra tion came into during the 1950'ies and 1960'ies, contrasted in many ways with the traditionally independent position of the lo cal offices. From the start in 1909 and formally to an amendment
149 Compendium to the Erasmus Course of 1971, the local offices were independent subjects of law with an independent economy. Their independence was gradually weakened as the government took over more financial respon sibility in the social insurance sector. However, the history of the local offices make them institutions with close links and much of their loyalty directed towards the municipalities, even though the offices are government institutions. It is not difficult to find examples of strong and open disagreements between the Na tional Insurance Administration on the one side, and one or sev eral local offices on the other.
The delegation process and the problem of equality Considerable parts of the decision power which previously were placed on the central authorities were delegated during the 1970'ies to the local offices. In the period from 1975 to 1980 the central handling of single cases was for example reduced with on an average approximately 58%. Central authorities have open regarded delegation prob lematic because it will always be difficult to retain national equality when many small units are processing cases. National formal equality has been a major concern for most political par ties in Norway, and has constituted one of the reasons for estab lishing a state responsibility within for example the social insur ance sector. The character of the social insurance legislation strongly illustrates this requirement for national equality; no lo cal independent considerations could - as a rule - legally be made. How has the National Insurance Administration handled the requirements for equality in a local administration with a wide scale of delegated authority?
The current computer systems - means to support formal equality The problem for the central administration has been to give sup port to the many small offices within a wide field of benefits and Public Administration and Information Technology within a framework of comprehensive and complex provisions that are under frequent change. Here, quite a lot of effort has been spent on training. In addition the central administration has built up the system of circular letters which contain detailed guidelines and instructions regarding interpretations of the Act and subsidiary legislation. More important, however, is the fact that the two on-line computer systems - Infotrygd and Nortrygd - have been intro duced at the offices. The prime objective with the introduction of these systems has been to cut down on the manning of the of fices and make the administration of the National Insurance Scheme cheaper. The possible reduction of staff is estimated to be approximately 20%, or about 800 posts, but only half of this will actually be withdrawn from the offices. The other half of the gain will be used to give the local offices more tasks and to up-grade the status of posts and the salaries of the remaining staff. The computer systems have important properties that may support the local decision-maker when processing single cases. Here, I will treat the two computer systems jointly, be cause in the eye of the user there are no major difference be tween them. The systems are built up with different routines for the different types of benefits, for example routines for pensions and routines for sickness benefits. Each routine consists of a certain number of query and registration functions. The pension routine consist for example of 7 soft copies for registration and 6 soft copies for query. The various routines are to a large extent inte grated, which implies that the system will collect certain types of information from other routines and automatically fill them into the open spaces of soft copies for registration. The registration functions are not unlike the use of tradi tional forms; the required information is filled in by the officers in charge. In this way the registration screens function as a
151 Compendium to the Erasmus Course checking list which instructs the user of the required inputs. The filling in of information is either mandatory or voluntary. This implies that if mandatory values are not filled in with a valid value, the person in charge will not be able to "produce a case", i.e. to make a final decision. This situation contrasts with the traditional case processing situation where the single employee received instructions of what was the required and relevant in formation, but where no practical obstacles existed if instruc tions were not followed. Only certain values are valid and accepted by the system. First, with respect to figures, certain limits cannot be exceeded, for example the limit for different types of amounts. These con trol functions may first and foremost prevent obvious mistakes such as error in writing. Other limits concern important letter in puts, for example the input for job category or type of injury in the sickness benefit routine. The person in charge cannot choose to use other letter codes than those selected for the system rou tine. This latter type of limit narrows the possibilities of the user to choose other results of these sub-decisions than those corre sponding to the accepted values of the system. The limits may therefore be claimed to function as a combination of formal in structions and practical necessities. When all the required information is filled in, the pro gram will assess the actual amount of benefit, tax and other amounts that may be part of the final decision. This is done automatically according to the computer programs which are made pursuant to the relevant legal provisions. The results are presented on the screen and the person in charge may choose to print a copy of the result and pay out the benefit. The fact that most legal rules for assessments of benefits, tax etc. are represented in the program, imply that inputs are equally handled once entered into the system, regardless of which office that processes a claim, and regardless of the skill level of the employee in charge. Experiences from the traditional Public Administration and Information Technology case processing showed that errors were open made in the course of manual assessment of benefits. There can be no doubt that these computation programs have eliminated a traditionally im portant source of error. On the basis of this brief description of the work situa tion of the computerized local offices it seems to be a likely conclusion that the current computer systems satisfy important political objectives that often are linked to the national insurance services: The possibilities for instruction and control with the local work are improved, the correctness of the assessments is improved and the work is carried out in a more rational way than before. On this background it is probably correct to say that the future basis for the retaining of a decentralized structure of the local offices is established, with on-line computer systems con stituting one the central conditions. T here are, on the other hand, both from the viewpoint of quality of services, social justice and job satisfaction for the employees, reasons to question certain aspects of the current situation. And I believe that these questions may point in the di rection of more advanced computer support in the local offices.
First, I will draw your attention towards the comprehensive ness of the current computer systems. The Nortrygd and Infotrygd systems cover the most important situations and alternatives within the various benefits. Nortrygd covers for example 9 different letter codes representing job cate gories. In the great majority of cases, these categories correspond to practical needs. However, if we regard the Infotrygd system, we will find 16 different job categories. This difference between the two computer systems give a good illustration of the imperfectness of the systems. Nortrygd obviously contains a much too limited number of alternatives for the registration of job categories. This will make it hard for the employees in charge of registration to find alternatives that ac-
153 Compendium to the Erasmus Course tually cover the facts of less typical single cases. On the other hand, the 9 categories in Nortrygd and the 16 categories in Info- trygd may also illustrate that it is hard to define a sharp limit of what will be a sufficient number of alternatives. From a practical point of view, such limitations and simplifying formalizations are understandable. The problem, however, is that the systems are designed as if they are complete representations of possible facts of single cases, without warn ings of their imperfections. I believe that when the computer systems only cover (let us say) 90% of the possible cases, and at the same time appear as if they cover every type of single cases, it is likely that the persons in charge of case processing will fail to see the less typical cases and instead try to interpret the facts of these cases so that they fit into the existing systems. In my opinion, the current computer systems should be developed so that they allow the user to make inquiries in the le gal systems to check out the provisions relevant for the less typi cal cases. They should also allow a satisfactory registration which always will make it possible to mirror the facts of these cases.
Second, I would like to focus upon problems concerning the level of support the current computer systems provide to the persons in charge of case processing, in relation to sub decisions. The computer programs handle, as mentioned, the larger struc tures of the legal provisions, for example the relation between the various variables of the rules. They also - to a large extent - tell the employees in charge which inputs are necessary in the production of cases. What the current computer systems don't do, however, is to support the person in charge when the various input values are to be established. For those persons operating the computer registering of the facts of cases, the challenge is first of all to make all the Public Administration and Information Technology many subdecisions that are required to feed the computer pro grams with the sufficient information. I believe it is correct to say that this registration is routine work in the majority of cases. Information like income figure, type of sickness etc. are often not questionable. On the other hand, several difficult cases occur when these subdecisions are causing doubts and require investigations in the many legal rules of the system. Such difficult problems will first and foremost necessitate knowledge of the contents of the subsidiary legislation and of guidelines and instructions in the circular letters. Regarding the comprehensiveness of these legal sources it is obvious that it is almost impossible to know and remember the contents of these documents. If, today, the person in charge discovers that an input criterion is questionable, she will have to interrupt the registering and consult the relevant legal documents in the traditional loose leaf files. I believe that the threshold towards using this kind of manual support has become higher in the situation with comput erized case processing. If the central legal support is to be re tained also on the level of the sub-decisions, I therefore believe it will be necessary to build a bridge between the current com puter systems and the traditional system of subsidiary legisla tion, guidelines and instructions. This could for example be done by allowing the person in charge to go directly from a field of registration to a system which could assist within the field of sub-decisions in question. Such a proposal does not necessarily mean that, for ex ample, the entire contents of the 2600 pages of the circular let ters should be represented. A possible introduction of such sys tems would obviously open for a principle discussion of the ex tent of central support and instruction power in a decentralized government administration. One alternative will be to include local authorized knowledge, another to limit the com prehen-
155 Compendium to the Erasmus Course siveness of the system and instead leave more considerations up the discretion of the single employees.
Third, I would like to question the capability of the employees in charge, to guide and explain decisions to the citizens More advanced tools are not only an idea which relates to the minority of national insurance applicants and recipients who will benefit from an improved support in difficult cases. As previ ously mentioned, the design of the current computer systems implies that the person in charge of registration of the facts of a case need not know the influence of inputs. In this situation, it will be difficult for the person in charge to explain to citizen actually how the decisions were made, i.e. of the relation be tween the various input variables of a case. This is probably not an acute problem because many of todayis1 employees have experienced the manual system and have therefore knowledge of the provisions which determine the assessment of benefits and taxes. On a longer term, however, as the turnover of employees creates a staff without this manual experience, this is likely to become a problem which will dete riorate the quality of the services provided. One may answer to these future perspectives that the most adequate way of meeting this problem would be to simplify the rules and to strengthen the legal training of the employees. The latter strategy will probably be required regardless of rule simplification. As for a possible simplifying of the provisions of the National Insurance System, I believe it is important to rec ognize a need for a comparatively high degree of rule complex ity, simply because socially just arrangements within such a comprehensive national scheme cannot be formulated in a sim ple way. I am not saying that the simplifications are impossible without a changed legal-material contents, but I do think that we always will have a too complex system of legal rules within this Public Administration and Information Technology
area, to make it possible for ordinary employees at the local of fices to survey them all. Parts of the information registered in the Infotrygd rou tines are accessed on-line by other public institutions such as the Social Welfare Authorities and the Maintenance Division. The collected information constitutes the basis for decisions of these other institutions. This arrangement is new and today limited to the region of Oslo, but plans of extending the routines to new districts and other public institutions exist, which is a tempting possibility, particularly from the viewpoint of cost-savings. Sev eral aspects of this development are well worth a thorough dis cussion, but here 1 will limit myself to comment the possibilities for the employees to survey the consequences of their registra tion of information into the computer system. When the registered information is used in external de cisionmaking, this is a good illustration of the work situation of the employees in charge at the local offices. They make a series of sub-decisions which are partly used inside the national insur ance sector and partly in other institutions. If the possibility of surveying consequences is limited within the national insurance sector due to rule complexity, it will usually be almost impossi ble to predict the effects of these sub-decisions being part of de cisions other institutions make. Both on the basis of the general rule complexity and be cause of the increasingly integrated character of the case proc essing in public institutions, it is, in my opinion, necessary to think of ways that can improve the possibilities of the employees to survey the effects of their activities. More advanced computer support in the decisionmaking process which could map and explain the relevant parts of the legal system, may be a means to attain an acceptable quality of services towards the citizens.
I will also focus on the need of securing an acceptable result
157 Compendium to the Erasmus Course
This is a point that is closely linked to the foregoing. To the de gree the person in charge has a liberty in the establishing of facts, she may make choices according to what will lead to rea sonable results. I believe it is generally accepted that the legal decision-making process is not a one-way process with the di rection from formulation of the problem, collection of relevant information etc. and finally leading to a result. Many will agree that in some cases, the decision-maker will try out the underly ing sub-decisions by testing the fairness of the final result the sub-decisions determine. In some situations it could be legally correct to alter the subdecisions and thereby also change the fi nal result. This is for example the case when more than one possible result of a subdecision is within the framework of a le gally correct interpretation of the relevant legal sources. If special precautions are not taken, it will not be possible for the employees to secure a reasonable result in such situa tions. Because of the integrated nature of the current computer systems it will be particularly difficult to predict the conse quences of sub-decisions and consequently equally difficult to exercise a control for a reasonable and fair result. This problem becomes even more obvious when we regard the task of survey ing consequences in routines in other institutions which collect information from the registers of the national insurance services. If such activities to achieve reasonable results are ac cepted in the case processing of the national insurance admini stration (and I think it should be accepted), this is obviously an argument in favor of employing more advanced computer sys tems which may support the persons in charge of case process ing and make it possible to survey the consequences of their le gal activity.
Finally, I will briefly mention the question of improving the job satisfaction for the employees of the local offices Public Administration and Information Technology
If we regard the job-situation of the employees, I will claim that the current Nortrygd and Infotrygd computer systems have turned the local office work into less skilled and partly routine tasks. I agree that it is necessary to arrange the local case proc essing work in such a way that the work can be done with an ac ceptable speed and quality, and that computer programs may contribute to this by guiding, controlling and computing. On the other hand, it will probably be in the interest of the employees to attain more knowledge of the decision-making process, including the parts the current computer programs han dle automatically. I am in other words assuming that the citizens and the employees of the local offices have common interests with respect to the development of more advanced support for the work at the local offices.
Conclusion In the foregoing, I have explained how the two computer sys tems in the local offices of the Norwegian national insurance services are focused on increased efficiency of case processing, and that important elements of the systems contribute to na- tion-wide equality despite a high degree of delegation of author ity to many small local units. Furthermore, I have pointed at as pects of the case processing work which the current computer systems do not cover and I have intimated possible future solu tions to the problems I have described. I believe it is correct to say that the Nortrygd and Info trygd systems are directed towards production of cases and that they only improve the quality of services to the extent that effi ciency is increased as well. My pointing at the possibilities of developing other and more advanced computer systems are more directed towards the improvement of the quality aspects than methods of increased efficiency. The first phase of the develop ment of the computer systems in the local offices is now termi nated. In my opinion, the next step should be design and imple-
159 Compendium to the Erasmus Course mentation of systems which first of all are directed towards im proving the legal quality of case processing. Public Administration and Information Technology
Representation of legal rules in computer programs - some implications on legal protection and service to citizens
Dag Wiese Schartum
This paper deals with the link between the computer programs as legal rules and the achievement of basic objectives o f a wel fare state founded on legal protection, using as a basis an in vestigation of three Norwegian computer systems. In the dis- passion the importance of legal knowledge of the officers in charge is stressed as a condition to provide service and obtain legal protection.
1. Background; Welfare Policy, Legal Complexity and Automation Automation of case processing in Norway has primarily taken place within the fields of public mass administration where tasks are regulated by comprehensive and complex legislation. Thus, the use of computers may be regarded partly as a tool for surviv ing with such legislation. The comprehensiveness and complex-
161 Compendium to the Erasmus Course
ity are often results of bad editing and a lack of ability to make rules understandable for others than those especially initiated. However, it should to a large extent also be judged as resulting from the policy of the welfare state and certain basic values in the Norwegian society. One of my basic assumptions is that computers may be regarded as a necessary aid to realize a com bination of agreed welfare policy and basic objectives of democ racy. The Norwegian welfare state has been marked by an active distribution and redistribution policy. The tax- and social insur ance schemes are two important examples of results of a modern detailed justice which has contributed to an increased volume and complexity of legislation. A marked trend in the welfare state policy has been a development from decisions made in the discretion of the administration towards rights oriented social schemes with contractual elements. The Social Insurance Scheme and the State Educational Fund Scheme are two impor tant examples of such development. The legislation within these areas is detailed and complex, as - to caricature the description - it may be seen as "contracts" written to fit every thinkable future person and situation. This contractual thinking and the need for predictability are closely related to the principles of "rettsikkerhet" (legal protection). Our welfare society has also introduced various kinds of re strictions and regulations to ensure prosperous development, health and security for the citizens. In themselves, such precau tions do not require a voluminous legislation. However, to avoid arbitrariness in the discharge of public authority and to secure "rettssikkerhet" and thereby create a certain degree of predict ability, there has been an unanimous demand for concrete and clearly formulated rules. The combination of the policy of the contractual oriented welfare state and the principle of Public Administration and Information Technology
"rettssikkerhet" is the first explanation of the volume and com plexity of legislation I will point out1. The second factor I will mention is first of all linked to the objective of effective case handling. The superior authorities' desire to instruct subordinate staff in order to produce quicker handling of cases, has pointed in the direction of detailed legal sources which could make the number of unsolved interpretation questions minimal. The strategy has been to establish a network of instructions and guidelines below the level of the formally binding rules of laws and statutes. In themselves, such instruc tions may not be required to achieve an effective handling of cases. But if efficiency and the objectives of "rettssikkerhet", in particular nationwide equality, should be reached, this would re quire documentation of detailed and binding interpretations as an aid for the officers in charge. Summing up, it is probably correct to say that a cost restric tive, just and democratic carrying out of the policy of the Nor wegian welfare state has required complex and comprehensive legislation. Claiming this, I am certainly not saying that the ob jectives of legal protection and efficiency have been achieved as intended. The present mass of legal rules is often seen as one of the greatest threats to "rettssikkerhet" and predictable legal im plications of legislation, and the wizards of rules represent in many ways the very symbol of an inefficient bureaucracy. Nev ertheless, "good" legislation, even the complex and voluminous, has a positive function as binding guidelines for public admini stration, through which citizens can be informed of their obliga tions and rights. This short outline of what may be comprehended as general legal and historical lines, is really of increasing current interest due to the computerization of the public administration. At least, with automated handling of cases, an extremely detailed system
1 Contractual-like social rights are of course closely related to "rettssikkerhet".
163 Compendium to the Erasmus Course
of rules is unavoidable. I will show how the computerization may be claimed to contribute both to a more comprehensive system of rules and to more complex legal structures.
2. Computer Programming and the Influence on Legislation If computers are to process cases and manipulate data according to a set of legal rules, the rules must of course be translated from the natural language of legislation into a formal language in the form of a programming language. Such translations may turn out to be quite troublesome to handle, at least if one tries to model every aspect of the rule, i.e. a "fully interpreted" rule. For obvi ous reasons, the programmers of most computer systems build the programs by means of manually established variables. After deciding which variables are relevant, the task is to establish the connections between the variables, for example in the form of a calculation routine or a decision tree. Instead of trying to auto mate the interpretation of "common law marriage" and "child support", the programs deal with the connections between the results of manual decisions regarding these facts of the case. This connection are typically arithmetical or logical operations. The fraction of a decision tree (below) indicates how the pro grams may be built up with structures of such manually estab lished facts, here with logical operations as the connections:
IF (APPL-IS-MARRIED OR COMMON-LAW-MARRIAGE} IF LIVES-WITH-SPOUSE = 'Y' IF SPOUSE-IS-STUDENT PERFORM SUBR-042 THRU SUBR-042-EXIT GO TO SUBR-Q4-NOT. Public Administration and Information Technology
I have picked out examples from my investigation of com puter systems in public administrations2, to present some exam ples of how quite straight forward programming may yield un expected results. In the LIS system of the State Educational Loan Fund deci sion trees (cf. the example above) decides whether calculations of certain benefits should be carried out by the computer or not. Thus, the value "true" of the facts in the structures of these pro grams, represents legal conditions determining the results of cases. The choice of which set of facts that must be true to carry out the calculation, is based on an interpretation of the legal rules of the domain. However, a legal analysis of decision trees in the us system, as well as examples from other systems, shows that central parts of its substantive, legal content differ from the traditional legal sources, or the correctness of the interpretations is doubtful. The first kind of occurrences may be illustrated by a change from the legal provisions stating the condition that the applicant must live together with his/her child to be eligible for supporter's supplement, to the program stating that the applicant must live together with his or her spouse. As a second example of faulty interpretations, I will mention a condition in the program, stating that foreign citizens must be married to be eligible for sup porter's supplement, while the legal provisions explicitly decides that common law marriage should be accepted as equal to formal marriage. The State Educational Loan Fund stresses that these devia tions from the legal sources need not express the final result of cases, but are restrictive (and cost saving) interpretations of the legal rules "to be on the safe side". Referring to the two exam ples above, this implies that an officer in charge will check if the applicant lives together with the supported child (according to
2 The investigation is carried out by Schartum. Results will be reported by autumn 1988 (in Norwegian) in the CompLex series of the NRCCL.
165 Compendium to the Erasmus Course the Act), in case he does not live together with his spouse. Sec ondly, foreigners will be especially checked if he claims to live in a common law marriage. The rules of the programs do, i.e. not result in a calculation of supporter's supplement to foreigners, but the case handlers have an opportunity to arrive at another decision. Such "rules of caution" in programs may be a wise procedure to introduce. The question is, however, if the actual decisions will be in accordance with the faulty programs or with acceptable interpretations of the legal rules? This question may of course not be satisfactorily answered without investigations of concrete decisions. However, the fact that the system does not identity critical points which lead to the faulty (interim) results, and leaves the officer in charge without traces of why the com puter processing arrived at no supporter's supplement, may lead to the assumption that the programs in these cases will open rep resent the rules which in fact determine the final decision. Another more straight-forward example may be picked from the NORTRYGD system's routine of daily sickness benefits used at the local insurance offices. According to the National In surance Act, applicants should as a main rule be 100% disabled due to sickness, to be eligible for benefits. There are, however, several exceptions from this basis which may result in payment of sickness benefits down to a degree of disability at 20%. In the programs of NORTRYGD, however, the degree of disability is always set to 100% for certain groups of job categories (for ex ample for military personnel and unemployed). This result may presumably not be arrived at on the basis of the existing legal sources. As indicated in these examples, faulty rules in programs need not be the results of difficult interpretations and problems deal ing with the legal texts in a correct way. Programming may also be used as an opportunity to "adjust" the handling of cases in ac cordance with administrative experience. However, one can in principle never with a 100% certainty claim that the program Public Administration and Information Technology
ming necessarily is intended to lead to a final and correct deci sion. The programs I have investigated never explicitly lay claim to be in 100% accordance with the legal sources, even if they may seem to be used to produce final results. The administrative organ may always claim that since the case handler have possi bilities of manipulating the system in order to devitate from pro grams, and since there always will be a skilled officer in charge who will give the out-print of decision from the system a second glance, the final decision will usually be correct irrespective of faulty or doubtful programming. Even if one accepts contentions that deviations in programs from the legal sources represent an acceptable programming strategy, this obviously places a great responsibility on the offi cers in charge, who, without explicit notions of doubtful pro gram contents, have the task to check the results from the sys tem. Moreover, such strategies really imply the introduction of two parallel sets of rules; the oversimplified or faulty rules of the programs, which are only correct in some groups of single cases, and the traditional legal sources which should be applied on the results from the computer to alter the out-put in special in stances. The next example of erroneous programs refers to problems of interpretation rather than to an intended result, i.e. the gov ernment administration in question admits the programs to be incorrect. The most typical way to use a computer in public case proc essing is to program the legal rules establishing the calculation of various benefits and taxes. The operations of computation de scribed in legislation are written in natural language and not in the unambiguous terms of arithmetic. This may entail interpre tation problems in the process of computer programming. An example from the system dealing with calculations of unem ployment benefits (DPS) may illustrate the point. The National Insurance Act establishes the maximum amount of previous
167 Compendium to the Erasmus Course earned income to be included in the final calculation of unem ployment benefits. The income figure is to be rounded to the nearest NOK 1000. The maximum limit equals 6 times a fixed figure in the scheme called "basic amount". In a lawyers mind, there can be little doubt that the maximum limit is absolute. The programming of the section leads, however, under certain con ditions to another result because the rounding of the sum is processed after the test of the maximum limit is executed. In periods when the product 6 times the basic amount equals a fig ure where the three last figures equal 500 or more, the program will round the product up to the nearest thousand, i.e. a result which is in poor accordance with the intended interpretation of the these rules. Finally, I will briefly mention two examples (among many) of non-trivial interpretations of legal rules in programs. In the DPS-system the programmer has introduced weeks of five working days, while in every other relevant relation, the week consists of six working days. The choice of five instead of six days leads to a change in the calculation of certain benefits, but is not documented other than in the computer program. When establishing the income basis for calculation of sick ness benefits, the Act allows to use the average of the three last pension point figures. In the Nortrygd system, the Act is inter preted to mean the average of the pension points of the three last years. It could obviously be questioned if this particular part of the program represents a correct interpretation of the Act.
3. New Rules, Increased Complexity These examples of faulty and non-trivial programming, are a taste of what could be found when programs and legislation are compared. Based on my investigation of three large government systems, I can conclude that there are occurrences of; Public Administration and Information Technology
1. Obviously incorrect interpretations of legal sources and 2. Questionable interpretations of legal sources3.
My guarded assumption is that examples from the two groups will be found in a great majority of systems and that this will have consequences for a considerable fraction of single cases handled by the systems. Systems design and computer programming are in most cases executed under a heavy time pressure, due in particular to politi cal and administrative as well as economic expectations. In this situation, the aim of the work is first and foremost to create a system as quickly as possible so that arrears can be reduced. One of the consequences of this is that written specifications from the legal expert to the programmer often lack. The reason is obvi ously that instructions and decisions through conversation are quicker than specifications in writing. Also, the systems are rarely and sparsely documented after programming is completed. On this background, the two groups of legal rules in computer programs (cf. above) may be regarded as "new" rules, because they are not documented in any other way than in the programs. Thus, they also represent an increased volume of legal rules4. Since they are different from the rules of the act, the provisions of the act and internal guidelines and instructions, they also constitute a complicating addition to the traditional rule system. The straightforward non-lawyer would maybe object: Since the rules of the programs are those that de facto come into use, and since they are unambiguous and detailed, they must repre sent a simplification rather than the opposite, because they rep resent the definite interpretation. It is true that in the large ma jority of cases the definite results are in full accordance with the
1 The number of sheer mistakes will probably decline with the time the sys tems have been in operation. 4 On the other hand, it is likely that some of the interpretations are in accor dance with practice of local offices in previous manual handling of cases.
169 Compendium to the Erasmus Course rules of the programs. The legal system, however, does not usually recognize the interpretations of the public administration as representing the ultimate decision. Thus, the courts may dis regard the decisions of the public administration and return a verdict of invalidity of particular interpretations represented in computer programs. Therefore, it is in principle always possible that other decisions than those following the results of the pro gram will be regarded as the legally correct.
4. The Legal Knowledge of Officer in Charge I have asked officers in charge in two social insurance offices if they felt that their knowledge concerning social insurance legis lation improved, remained unchanged or decreased after the in troduction of computerized case processing routines. The great majority of the responders answered that their knowledge was unchanged or improved5. These answers could partly be ex plained by focusing on two factors linked to computerization that may contribute to a learning process of the officers in charge. The screen pictures of the various routines of the systems show tables of all the defined relevant pieces of information in different types of cases. The overview that these tables present, would be quite demanding to obtain through studies of the tradi tional legal sources. The information of the soft copies will therefore probably represent pieces of new knowledge for the officers in charge. Messages from the system to the operator concerning excess of limits, special codes etc. defined in the programs, will also represent valuable knowledge, concerning the legal content of the systems and of the specific scheme. Moreover, introduction of on-line computer systems are often followed by an altered division of work, in the sense that a
5 Cf. Schartum ( 1987), page 202 and 212. Public Administration and Information Technology greater degree of flexibility is introduced0. This implies a change towards less specialized work and demands that the officers in charge should be able to handle several routines if work pres sure, absence from work or other circumstances make changes in the division of work desirable. Obviously, the role of the gen eralist will lead to acquisition of new legal knowledge. At the same time as these experiences constitute a basis to presume that new knowledge is gained, my description of the representation of legal rules in programs indicates that other types of legal knowledge will be difficult to maintain. To reca pitulate, the great majority of calculations will always be auto mated in a computer system for case handling. The persons in charge need therefore not know anything about the rules which establish the assessment of benefits, taxes, loans etc. These con nections between facts of cases are primarily a problem for the programmers and the computer. Moreover, an officer in charge without knowledge of questionable or anomalous program rules, will not be able to gain sufficient knowledge to predict or ex plain the result of the automatic processing of cases, no matter how trained she might be in relation to the traditional rules. A picture of the helpless officer in charge who only knows something about the facts of cases, isolated from the overall rule structure, is certainly an exaggeration. The experienced person does of course not forget all she knows about the connections between the facts of cases just because she changes from a man ual to a computerized job situation, and this knowledge will be of good help in many situations. However, when most employ ees with experience from manual routines are replaced due to staff turn-over, the case handlers will probably have knowledge of another type and quality than today. In my opinion, this change will in many ways represent a negative development, at least in relation to giving legally satisfactory advice. For such
6 The national insurance offices, the State Educational Loan Fund and the labour offices have reorganized in accordance with such strategies.
171 Compendium to the Erasmus Course service functions, detailed knowledge of computations is re quired, and a survey of inputs and mere results will be insuffi cient. My view is that it will take considerable effort to retain the legal knowledge of the employees, especially within complex legislation like tax and social insurance. On the other hand, I am not sure how important it is that officers in charge have a com plete knowledge of every legal rule that decides the automatic manipulation of the facts of cases. The important objective must be to create easy access to a complete documentation of the computerized operations, making it possible for a trained officer in charge - on the request of citizens - to access databases with explanations of computer operations.
5. Concluding Points In my opinion, the legal knowledge of the officers in charge is a central factor in the conduct of the principle of "rettssikkerhet", since such knowledge is a condition both for the possibility of explaining and justifying decisions, and to control and take re sponsibility for administrative actions. In conclusion, I will re turn to the point of departure and regard the relationship between the objectives of "rettssikkerhet" and efficiency. In my view a major problem lies with a renunciation of a common interest between the objectives of "rettssikkerhet" and efficiency. As I see it, the efficiency goals in a manual case processing routine will easily lead to a building of knowledge of the officers in charge. Here, efficiency requires detailed instruc tions to the people in charge of case processing, and thus legal knowledge is acquired by the same people. Provided that the division of work is not too specialized, this will educate case handlers and make them capable of giving correct answers to questions from citizens. In contrast, effective automated case handling requires detailed instructions to machines. These pro- Public Administration and Information Technology grams are, however, directed towards fast resolution and pro duction of cases and build only limited knowledge of the case handlers which can be returned as response to questions from citizens. As stated at the beginning of this paper, a combination of the implementation of the welfare policy and claims for "retssikkerhet", created a demand for passing and documenting of detailed and clearly formulated rules. One of the main objec tives was to create a possibility for people to read and under stand the legal rules, and thereby an ability to control the authority of public administration, to create predictability etc. Today, in the course of computerization and rationalization of the public organs administrating central parts of the welfare policy, we face a situation were such detailed rules are used in an automated production of cases, containing "secret" faulty and questionable programs. It is important that it is admitted that computer programming of legal rules regularly will lead to altered substantive, legal contents. However, it is impossible to tolerate the existence of a double rule system with traditional rules saying one thing and the rules in programs saying something else. It is equally hard to accept that the solution of legal questions are only documented in computer programs which only a handful of trained people can read. This new situation may seem to threat the possibilities of ensuring "retssikkerhet", predictability and legality control in a computerized public administration.
173 Compendium to the Erasmus Course
Literature
Arbeidsdirektoratet: System manual. Dagpengesystemet (DPS), OSLO 1987. Arbeidsdirekoratet: Dagpengesystemet, program code, VBEHAI, OSIO 1988. Rikstrygdeverket: NORTRYGD program code, AP036. Schartum, Dag Wiese: The Introduction of Computers in the Norwegian Local Insurance Offices, CompLex 9/87, Uni versitetsforlaget, Oslo. Schartum, Dag Wiese: "Delegation and Decentralization - Com puter Systems as Tools for Instruction and Improved Service to Clients", in Goranzon and Josefson (Eds) "knowledge. Skill and Artificial Intelligence ", Springer Verlag, 1988. Statens Lånekasse for utdanning: LIS program code, SLSAK2.
Summary Computer systems in public administration often imply pro gramming of legal rules which may result in a substantive legal content diverging from a legally correct or normally expected interpretation. On the basis of an investigation of three Norwe gian computer systems, it is claimed that the differences between the interpretation of traditional legal sources and the programs, create increased complexity and volume of the legislation in question. Also, the contents of programs are often badly docu mented, and thus some programs contain more or less "secret" rules. T his situation influence both the possible legal knowledge of the officers in charge of single cases and thus the service they will be able to offer citizens, as well as the ability to secure "rettssikkerhet" (legal protection) and legality control. In order to secure basic values in a welfare society founded on legal protec tion. it is concluded that a harmonizing of he legal rules in com puter programs and the traditional legal sources is required. Public Administration and Information Technology
Delegation and Decentralization - Computer Sys tems as Tools for Instruction and Improved Service to Clients
Dag Wiese Schartum
This chapter deals with issues such as those of delegation of de cision making, quality of service and competence, arising from the application of information technology in the domain of the Norwegian National Insurance Administration system. For ex ample, within the context of "national conformity" of practice, the quality of service will demand uniformity in interpretation regardless of the individual insured person's particular circum stances. The emphasis on efficiency, and time, and a degree of "tailor-made" information may lead to reduced access and com munication between the people who want to use the system and the service itself. Rationalization in complex domains (cases) of decision making may cause the case-handier either simply to ig nore the problematic aspects of a case, or to make an intuitive choice instead of conferring with the relevant legal sources. There is empirical evidence from the Norwegian case that in a large number of cases the use of data systems may have a posi tive effect as they can weed out elementary errors and omissions in calculations, but at a more complex level where legal exper tise and judgment is required, such systems may give no insight and support. It is important, therefore, to develop criteria for evaluating the non-time-related aspects of the quality of service.
175 Compendium to the Erasmus Course and thereby define aims for, and evaluate, public-oriented ac tivities. Public Administration and Information Technology
Introduction
Delegation and decentralization are two expressions which to many people have a positive content. In this discussion I will take a closer look at how on-line computer systems in the Nor wegian National Insurance Service can be said to affect the re alization of these two aims. I will focus in part especially on the use of information technology for information and instruction purposes. In addition I will discuss the effect of computer sys tems on certain organizational questions in the service. The various effects of information technology will then be evaluated in the light of aims of improved quality of service. The article is based on the Norwegian part of the joint Ger man/English/ Norwegian project. Naturally, I do not have the opportunity of going further into the details of the background materials this discussion is based on. For those who are inter ested. I would refer to Parts III and IV of the Norwegian report. Neither do I have the opportunity here of presenting an entire approximation of the question of quality of service, but an intro ductory effort to develop a method of defining aims and evalua tion of the quality of the services offered by the National Insur ance Service is found in Part V of the Norwegian report of the project. It has been discussed in Norwegian legal theory to what de gree an administrative organ, which by law or directive has been assigned tasks, can delegate these tasks to subordinate adminis trative levels. The most important arguments against such dele gation is that it can break with Parliament's and the government's intentions if establishments other than those laid down by regu lation carry out the work. It has been said that when the legisla tive powers assign tasks to a superior administrative organ, it is often due to regard for legal protection, because the quality of the work in a superior establishment normally must be regarded
177 Compendium to the Erasmus Course as being higher than in subordinate establishments. A second and related argument has been that it would be easier to manage and control the practices in a central establishment instead of several local offices. Important arguments against delegation of administrative authority is, in other words, attached to the regard for the quality of service in the case-handling work. Development in post-war years has been steadily in the direc tion of a more liberal attitude to delegation of administrative authority. These changes have been brought about by a strong practical need which has been created in particular by the con tinuous growth within public administration. Particularly since the 1970s, therefore, attempts have been made to transfer case work to the local administrative apparatus, while the central or gans first and foremost are responsible for planning, steering and control of the work of subordinate units. This division of work has been regarded, inter alia, as being a pre-requisite for suffi ciently expeditious case processing. In addition to the regard for efficiency, independent emphasis has been laid on the impor tance of closeness to the public, both because knowledge of lo cal conditions is regarded as being important to correct case-handling, and because geographic closeness has been re garded as important to people's use of, and attitudes to, the ad ministration. In other words, in addition to efficiency, regard for the quality of service has also been cited in favor of transfer of authority to local organs. On the other hand, delegation has not been justified by demands for greater local freedom in relation to questions of interpretation and practicing of the rules. I mentioned that traditional skepticism towards delegation of official authority has its background in the assumption that sub ordinate administrative levels do a less efficient job of case-handling than the superior organ. If one is to judge such possible differences in the quality of case-handling, it is in my opinion particularly important to direct attention to the superior organ's right to issue instructions to the organ to which it has Public Administration and Information Technology delegated authority. Subordinate administrative units (e.g., a lo cal national insurance office), in some connections can be said to be independent. At the same time, however, they are for a large part subject to the authority of the central administration, and must therefore follow the general instructions laid down con cerning organization, method of work, legal questions, etc. The effect of delegation must therefore be evaluated in relation to the degree of independence the decision-maker has with regard to the ramifications set by instructions and other conditions.
Delegation and Central Control in the Norwegian National Insurance Service During the 1970s and 1980s, a large part of the decision-making authority which previously lay with the central National Insur ance Administration was delegated to the 450 partly very small local national insurance offices.1 I have access only to the fig ures for the rate of delegation during the period 1975-1980 when the number of single cases in the National Insurance Admini stration was reduced by an average of approximately 60%.2 The figures for this 5-year period give an impression of a delegation process which has been taking place since 1970 and which is still taking place.3 I emphasized that delegation had to be seen in conjunction with the instructions given by central authorities. In the Norwe gian National Insurance Service there are especially two meth ods of control to consider: firstly, traditional instructions and
1 About half of the local national insurance offices have between I and 5 employees and about 90% have fewer than 20 employees. 2 See Habberstad: Organization of the National Insurance Service, June 1981, page 44. 1 Even though there have been great variations from benefit area to benefit area, the figures illustrate the extensive changes which the local national insurance offices have been through.
179 Compendium to the Erasmus Course guidelines concerning interpretation of rules, and secondly, use of on-line computer systems. I shall first say something about the written internal instructions and guidelines. As in all other legislation, the National Insurance Act with directives contains rules about which there can be questions of interpretation. To make national uniformity possible in this area, the National Insurance Administration has therefore developed a system of centrally laid down instructions and guidelines which take a stand on a number of interpretation questions. A loose-leaf system with a total of approximately 170 circulars (about 2600 A5 - pages) is distributed to all local offices and is updated by the National Insurance Administration. It stands to reason that such a paper-mill demands both good discipline and overview in order for it to be taken notice of and function as intended. The special thing with such instructions is that, in addition, they are kept on the shelf and are only activated by the desire of the employee. In a work situation under pressure where demands on time and efficiency are emphasized, there is therefore a danger that use will be more limited than the number of difficult cases and the existing interpretation problems should indicate. The possibility that delegation of authority could weaken the central control of the local administration worried the central authorities when the general decentralization and delegation policy was formed in the 1970s. In the government's long-term program of 19804, it was stated in this connection that informa tion technology makes possible easier access to information in the local environments and therefore weakens the arguments for centralized administrative solutions. At the same time it is maintained that the same technology offers the possibility of in creased control and coordination such that delegation does not necessarily have to lead to the creation of new, local, self-governed units.
4 1. Parliamentary Bill No. 79,198081, page 78. Public Administration and Information Technology
The on-line computer systems in the Norwegian National In surance Service are a good example of how information tech nology can be used as a tool of control of the organs to which authority is delegated.5 Instructions and guidelines which are a built-in function of the two data systems of the Service are of a totally different nature than the traditional instruction and guide line systems I have mentioned. The two most important func tions of the system are registration of case information and cal culation of benefits and tax. Where registration of information in individual cases is concerned, the system functions in the first instance as a reminder list of which details can be included in the case-handling. Next, the system indicates which details are mandatory and which can be obtained voluntarily. In addition there are certain marginal values which the registered figures cannot exceed, and certain lists of allowed values for represen tation of special types of facts. Some details are filled out be forehand by the system which transfers information from one routine to another. Moreover, the automatic calculation of benefits can only be activated when mandatory values are filled out properly. The computer system is therefore - in contrast to circulars - not a system which is activated by the individual employee when that individual is faced with a problem, but a tool she has to use to be able to do the job at all. The instructions and guidelines which are built into the computer system in the form of allowed values, indication of relevant facts, etc., are therefore something the employees will be confronted with in any case, and also rep resent partly the centrally laid down rules which they cannot ig nore.
5 In addition to use of information technology to instruct and guide the ex ternal system, various reporting functions have been established which facilitate a better overview for the National Insurance Administration, and thereby a strengthened potential for steering the local national insurance of fices. These tools of control, however, will not be discussed here.
181 Compendium to the Erasmus Course
Instructions and Quality of Service If the importance that instructions can have for the quality of service is to be evaluated, the question will in the first instance be what the relationship is between the demand for quality of service and the demand for national conformity in the applica tion of the rules by the National Insurance Administration. I am of the opinion that one must differentiate between the le- gal-political and legal-application aspects of this question. On the one side one is of course free to think that from the national insurance political aspect the best solution would be if each local national insurance office were given great freedom in the deci sion process. A service evaluation of the legal application under taken by the local offices must take place within the limits set by the law at all times. When legislation prepares for national con formity, it implies that a strong central control which leads to uniform practice as a basis has a positive effect on the quality of service. In my opinion it is clear that the Norwegian national in surance legislation does not aim at giving any room for local latitude. In the National Insurance Service, regard for the quality of service will demand uniformity in interpretation regardless of where the insured lives in the country. As I have pointed out previously, today's computer system instructs and guides the individual terminal operator. The sys tems establish, in addition, work routines which are the same for all offices and which therefore increase the probability of like results regardless of which local office is involved. All the same there are things in the work situation which have been created by the computer systems in the Service which in my opinion will necessitate a probable modification of this positive basis. With the system solutions chosen, the division of work be tween people and machine lessens the possibility for the case-handier to learn and understand the legal rules which steer Public Administration and Information Technology the results of her work. As an example, the system gives in structions on which information is relevant and mandatory, but does not give any guidance on what importance stipulation of the various factual details has for the final result. The case-handier can by all means be an expert of making minor de cisions with regard to whether the factual conditions are fulfilled or not, but she will have difficulty - without special measures - in gaining any particular insight into legal questions and knowl edge of the overlying structures in the set of regulations she works with. In general, instructions and guidance given by com puter programs correspond to correct results and deal with gen eral cases. In addition they are so self-instructing that good knowledge of the rules is not normally demanded in order to carry out case-handling in simple cases. The systems do not, however, give any support for solving particular problems. A more active use of traditional problem-oriented instructions and guidelines is therefore desirable to ensure a correct result in the problematical cases as well. In my opinion there is a danger that with today's computer routines the difficult cases will act as obstacles in otherwise well-oiled decision processes, both because the use of them breaks with the effective case procedure and because the tradi tional instructions and guidelines which are required in order to take a stand as regards the problems are very difficult to access. The result can both be that the case-handier simply chooses to ignore the problematical aspects of cases, or that she makes an intuitive choice instead of conferring with the relevant legal sources. The influence that information technology in the Norwegian National Insurance Service has had on the question of the quality of case-handling in the sense of regard for a correct result, points in other words not only in one direction. I think it is probable that the computer systems have a positive effect in the large majority of cases in that they first and foremost weed out ele-
183 Compendium to the Erasmus Course mentary errors and defects, pure omissions and errors in calcu lation. On a more advanced level, where legal insight is de manded, the systems give no insight and support, and at the same time the effective routines that the computer-based, case-handling creates, can elevate the threshold against the use of manual problem-oriented instructions and guidelines. For the sake of good order I emphasize that in the Norwegian part of the project I have only found indications of a develop ment such as I have described here. My interpretations of the employees' possibilities to view the decision process they are a part of is first and foremost based on knowledge of system solu tions, the employees' work situation and not least the employees' knowledge of the rules regulating award of benefits. Of course, these grounds are not sufficient to make a final decision in the matter.
Information Technology and the Decentralized Organiza tional Structure As mentioned, information technology has an influence on the possibilities of issuing instructions parallel to delegation of authority. This can, therefore, be said to have been an important support for the implementation of the changed division of work between the central and the local administrative units in the administration of the Norwegian National Insurance organ. At the same time, however, the on-line computer systems in the Service have had an effect on other important conditions for delegation of authority, and have contributed in addition to changing the content of the services that the local units offer the public. In this part of the discussion I will show how the com puter systems in the organ at the same time can be said to have weakened the arguments for today's decentralized administrative solutions. Public Administration and Information Technology
In 1982 there were approximately 150 service units in the National Insurance organ - called "attached offices" - in addition to the 450 local national insurance offices. The attached offices are national insurance offices which have lost their independent status because of the fusing of the municipalities in the 1950s and 1960s. The running of them is partly maintained and the of fices, as far as organization goes, are subordinate to the national insurance offices in the new large municipalities. The attached offices have therefore never been visible on official statistics of the number of national insurance offices. Neither were the at tached offices mentioned when Parliament deliberated appro priations for computer equipment for the national insurance of fices, and thus they were never given any such equipment.6 The transactions registered manually by an attached office must therefore be transferred later to the national insurance office's computer system, something which entails extra work. Introduction of on-line computer systems to the national in surance offices was carried out on the condition that - through natural departures - half of the goal of rationalization is with- •j drawn, i.e., 400 positions. At the same time as the introduction of data and the withdrawal of positions, all the local national in surance offices in addition went through a reorganizational proc ess.8 These circumstances partly created a very difficult work situation at many offices. A large number of local offices with responsibility for the running of attached offices were therefore forced to gather resources and applied to the National Insurance Administration for permission to close the attached offices. During the last few years more than half (or approximately 80)
6 Later five outside offices with almost full capacity were given computer equipment. 7 See Parliamentary Bill No. 116, 1983-1984. x Reorganization and introduction of data routines were carried out in paral lel to each other in the project "service, productivity and work environ ment" (the SPA project), which was completed at the turn of the year 1986/87.
185 Compendium to the Erasmus Course of the attached offices have been closed, most of them as a direct result of the circumstances I have mentioned. Introduction of the on-line systems in the National Insurance organ have, in other words, led to a less decentralized organ. One of the reasons that there has always been a local office in each municipality is that many payments of benefits have always taken place in the form of cash payments by the office. Even though payments of long-term benefits have taken place by giro for many years, many payments of short-term benefits and re funds of expenses still take place by the public visiting the office and receiving the payments in cash.4 As technological progress has made it possible since the 1950s to transfer long-term benefits by giro, so information technology makes it appropriate to transfer short-term benefits and refunds in the same way. The National Insurance Administration has thus recently begun the development towards such non-cash transfer (or "publicless") local offices.10 The Norwegian National Insurance organ also started to use information technology to inform recipients of benefits about their entitlements under national insurance without previous re quest. This is done by different groups of addressees being picked out with the aid of personal details which are accessible in the national insurance computer-registers. This personal in formation makes it possible to define smaller groups of address ees in order to dispatch a certain degree of "tailor-made” infor mation. The developments towards the so-called non-cash transfer offices and such active information measures make it appear less
This concerns 50% of all recipients of sick benefits and the majority of those receiving refunds of expenses which are covered by national insur ance. 10 The National Insurance organ has over a lengthy period of time requested the local offices to start to reduce cash payments. Molde and Lunner local national insurance office, as a trial project starting in the spring of 1987, has stopped cash payments completely. Public Administration and Information Technology necessary for people to visit the national insurance offices. This has occurred at a time when the organ, because of a lasting ra tionalization pressure, also uses other methods to deal with the volume of case-handling. At many offices, a public-free day a week has been introduced. As we see, the data systems have strengthened the central control of the local administration and thereby made it less criti cal to delegate authority to the local offices, while, at the same time, the same technology has contributed to a reduction of the local units, that is to say, made the structure of the organ less decentralized. In addition the use of information technology in the organ has made it less necessary for the public to visit the local offices, and has thereby probably weakened the basis for today's decentralized structure. In the following I will present a more in-depth treatment of this latter problem.
Quality of Service and the Importance of Direct Contact with the Public I mentioned examples which show that the Norwegian National Insurance organ - within today's ramifications - does not give priority to direct contact with the public. To make the point, one can say that the Norwegian National Insurance Service uses in formation technology to prevent the public from visiting the lo cal national insurance offices. By anticipating and carrying out certain tasks which the public usually visit the office for, the Service can pre-empt the situation such that the local offices reach out to the public instead of the public coming to the of fices. One could ask if this is not an example of the best type of service, because it does not anticipate any particular effort on the part of the insured and lays the work on the Service instead of the public. To a certain extent this has had a positive effect on the quality of service. When the answer still contains reserva-
187 Compendium to the Erasmus Course tions it is because one-sided efforts regarding such measures would in my opinion create imbalances in relation to other serv ice aspects. In the first instance I think that it is unwise to base assump tions of the public's needs on completely rational methods of observation. One should, for example, be reserved in believing that the people who today come to the local offices to receive cash payments, have only this rational errand and that it is there fore only a payment function which is affected when the local offices change to giro payments. In my opinion there are grounds to believe that many people who come primarily to receive cash payments also ask questions about other national insurance cir cumstances: they ask for themselves, for a neighbor, a relative or they take a brochure. Some questions will affect things which fall under the national insurance organ's area of responsibility, others will concern benefits under the social welfare system or other public services. We know little or nothing about how im portant other such errands are to the people who visit the offices. In my opinion, one should therefore be careful about introducing changes in the services offered without having investigated such circumstances further. The justifications used by the Service for transferring benefits to giro payments are service and efficiency. In the regard for ef ficiency lies a consideration that the public visiting the offices take up time and this time can better be used on other things. Of course, it can be reasonable to give priority to quicker case-handling. Efficiency and shorter case-handling time, how ever, have always been central aims in the National Insurance Service. The danger now is that when the time factor is included and given priority as a service aim, the work for improved serv ice will to a large extent coincide with the work for improved efficiency. In this way the work for better efficiency can be con tinued under the more popular name "service" without any change being made in the priorities. The time aspects of service Public Administration and Information Technology will in my opinion normally be sufficiently taken care of through rationalization work. I am of the opinion, therefore, that giving quick case-handling priority over using resources for direct contact with the public, leads to a distorted and one-sided focus on time as an aim of service. In my opinion it is therefore neces sary to direct attention towards the aspects of quality of service which are not directly linked to time and efficiency. The computer-based active information measures which the Service wishes to concentrate on will in my opinion only partly have positive effects if no supplementary measures are concen trated on at the same time. There are two things which character ize most of such information material. Firstly, it represents only an extract of the rules and is therefore incomplete. Secondly, it represents a simplification of the rules which regulate the area and is for that reason also insufficient. Information material containing a popularization of the rules will never give a com pletely correct legal picture, and therefore will not give suffi cient information to the whole group of recipients. The more extensive and complicated the legislation is, the more difficult it will be to attain a sufficient legal level. On the condition that dissemination of this material is only the first step in an infor mation campaign launched by an organ, such information is anyway positive. The condition is, however, that actively dis seminated information is combined with an offer of individual and concrete guidance. In the opposite case, such information activity will have the effect of guidance for some whilst others will be misguided and thus risk loss of rights. I am also of the opinion that for information to be of a suffi cient standard at the local information level, it demands concen tration on direct contact with the public. Such a demand makes it necessary to have qualified employees in the national insurance offices who can give guidance of sufficient quality. An impor tant condition for sufficient level of competence of the employ ees is that they participate in the decision-making processes and
189 Compendium to the Erasmus Course make use of the aids which give them insight into the legal ma terial about which they will render guidance. And here we are back to the question of instructions and their ability to give in sight into legal questions. If the National Insurance organ tries to keep the public at arm's length by the organ being active itself but at the same time not making direct contact with the public possible, there are further grounds to put a question mark by the basis for the whole of the present decentralized organizational structure. What then do we want with local offices operating near the public if the public does not have to visit the offices? Is it to save long-distance telephone calls? The necessity of direct contact with the public would seem therefore to be a condition for maintaining the principle of a local national insurance office in each municipality.
Conclusion In order to bind the parts of this discussion closer together, some of the factual contexts can be linked together in this way. A suf ficient level of information work demands direct contact with the public, direct contact with the public demands decentraliza tion, decentralized guidance demands sufficient legal compe tence which can best be obtained by practical experience, something which assumes delegation of authority. Execution of delegated authority to a sufficient degree in addition demands instructions and guidance which contribute to increased legal insight. As a brief summary I would put forward two clear formula tions connected to on-line computer systems in the Norwegian National Insurance Service. First, information technology is in use in the Norwegian Na tional Insurance Service in order to make decentralization and closeness to the public possible, while at the same time being applied in ways that lessen the contact with the same public. Public Administration and Information Technology
Second, the applications of information technology which are chosen give improved quality of service in cases which from a legal standpoint are not problematical, and unchanged or wors ened quality for atypical and difficult cases, both in regard to case-handling and information work. I think that in the future it will be important to direct attention towards the aspects of the quality of service which are not di rectly linked to time and efficiency. Here it is important to see that the Norwegian National Insurance Service has not utilized the information technology potential to assure other aspects of quality. Also, they have hardly maintained a carefully weighed attitude to the relationship between the quality of service and the application of computer systems chosen. Today we lack a method of evaluating many important as pects of service. I am of the opinion therefore that it is important to develop criteria for the non-time-related aspects of the quality of service thereby being able to define aims for, and to evaluate public-oriented activities. This to a large extent will be a condi tion for Norwegian authorities to be willing to assign priority to this work.
Acknowledgment. This article is based on: Dag Wiese Schartum (1987) The introduction of computers in the Norwegian local insurance offices. Corepoints and context. Complex 9/87. Nor wegian University Press, Oslo. The work is funded by the Stiftung Volkswagenwerk, West Germany.
191
Public Administration and Information Technology
Dirt in the Machinery of Government?
- Legal Challenges Connected to Computerized Case- Processing in Public Administration. 1
By Dag Wiese Schartum
1. Computers and the machinery of government
Computers are "general machines"; i.e. given the existence of a suitable computer programme, they are capable of executing operations within every field. One implication of this is that computers may be used for multiple purposes within the same organization. These purposes are often closely related: a system developed to support decision-making in single cases, will almost always also be an accounts system and at the same time often constitute an important element in the production of statistics and other planning information. Moreover, a computerized case-processing system will often have functions with respect to the keeping of case records, and thus be integrated in the administration of dossiers etc. Of course, the degree of integration between the functions of the computer varies. There may be one system with different modules or a bundle of systems. Notwithstanding this, the main point to note is the inter-connection of the many functions that
1 Thanks to Lee Bygrave, Kari Lie and Philip Opseth for valuable assistance with the English language
193 Compendium to the Erasmus Course computers within large organizations have. A description of computer systems in government administration, thus involves a description of structures and procedures that are important functions in the machinery of Government. Computers constitute a requisite component in a well- functioning public administration, not only in relation to "local functions", but also in relation to functions between organizational levels and between sectors. As for functions, a computerized case-processing system will usually give important support to several tasks. With respect to the computer systems' influence on the relationship between organizational levels, it should here be sufficient to point to the fact that use of systems with case-processing and account functions, will imply the implementation of directives to subordinated levels as to how these functions should be executed. The collection of statistical information etc. by means of systems also represents an important feed-back from street level or other subordinated parts of the public bureaucracy to the central bodies of government. In addition, modern Norwegian computer systems, are in the process of being developed to fill inter-sectorial functions, especially with regard to information How and the sharing of data between different public sectors (including central and local government) and between the private and the public sector. The objective is, as indicated, to exchange information by way of case-processing, statistics production and to facilitate the accessibility of public information, as a step towards, for example, establishing public information services on a pay-basis. In the Norwegian legal system, a general code of administrative procedure is embodied in legislation. Important parts of this procedure concern the handling of single cases. General administrative procedure rules and similar rules in laws and regulations governing particular functions of government, safeguard the rights of citizens, and contribute to decision- Public Administration and Information Technology making in single cases in accordance with legislation. The general code of administrative procedure also contains provisions regulating the execution of executive authorities' legislative powers. For example, rules are formulated in order to ensure that regulations issued by the administration do not exceed delegated powers. Statutory directions addressed to public authorities which exercise power in single cases and in the process of bureaucratic law-making, are necessary as legal safeguards over public administration. However, this statutory protection of citizens does not embrace the fact that the administrative machinery of the 1990s to a large extent is run by computers. This article is based on the view that legal safeguards concerning bureaucratic lawmaking and the handling of single cases, are obviously inadequate when the execution of public powers is computerized. However, my perspective is that the interwoven fabric of computerized functions within the administrative machinery does not allow us to see legal questions isolated from other questions concerning the tasks and functions of public bureaucracy. Therefore, although emphasis is put on legal aspects in the first part of this article, towards the conclusion, interests and perspectives other than that of the legal protection of citizens, shall be taken up. The observations and examples presented in this article are based on my investigations of legal contents of case-processing systems in public administration, as well as investigations in the development of such systems.2 Although these investigations have been of systems of vital importance to a great number of citizens, it is important to stress that 1 do not base my views on any representative material. Moreover, sensible measurements
2 See Dag Wiese Schartum: En rettslig undersøkelse av tre edb-systemer i offentlig forvaltning, Complex 1/89, Tano 1989 and Dag Wiese Schartum: Rettssikkerhet og systemutvikling i offentlig forvaltning, Universitets forlaget 1993.
195 Compendium to the Erasmus Course and average figures are, in my view, impossible to work out in order to state a figure that may quantify the legal problems addressed. It should thus be sufficient to show that certain problems occur in administrative bodies, which make a large number of single case decisions every year. The fact that the results of my investigations of Norwegian systems have been confirmed by selected systems in Sweden and Denmark,3 may indicate that the results have an interest beyond the Norwegian reality. After elaborating upon some of this article's main points, I will explain why and how important parts of system development in public administration must be regarded as part of the public authorities' execution of legal powers. In this connection, I will identify and present some major legal challenges and problems and indicate some solutions. The need for new legal safeguards is discussed on the basis of an examination of the relationship between different citizens' interests and other interests with relevance to the process of computerization in government. In conclusion, I argue in favour of certain principles and measures which should be laid down in order to improve legal safeguards for citizens.
2. Limitations, basic views and definitions.
Before going into the different legal aspects of case-processing systems, it is necessary to present some basic views, explain some limitations and clarify some basic concepts that are applied in this article. First, I regard computerized case-handling in government mass administration as synonymous with case
1 See the descriptions of computer systems within social insurance administration of Sweden and Danish computer systems within the tax administration in Dag Wiese Schartum: Rettssikkerhet og systemutvikling i offentlig forvaltning, Universitetsforlaget 1993, pp. 402 - 420 Public Administration and Information Technology
handling; i.e. I see computerized case-processing as a normal element in the handling of individual cases. In Norway and - I believe - in most western countries, computerized case- processing is an inevitable and established element in the decision-making of public administration, which has the task of distributing money in the welfare state. Thus, this article treats computerized case-handling as a standard occurrence. Secondly, accepting computerized case-handling as being synonymous with case-handling, I see little sense in comparing totally manual routines with computerized ones. There is no room for conducting sheer manual case-handling in mass- administrative decision-making. Thus, there is little sense in discussing "what is worse": manual or computerized case- handling. When, in this article, legal problems concerning computerized case- processing are placed on the agenda, it is not, a demonstration of a "hostile" attitude towards information technology. The aim of this article is rather to formulate and discuss special problems concerning computerized case- processing in a way which may lead to practical approaches and solutions. "Problems" are in other words thought of as challenges rather than obstacles. A third important element is the values and interests which I identify myself as having and which, thus, may influence the way in which I deal with issues. Being a lawyer, the ties to basic legal principles are important for me, in particular the principle of "rettssikkerhet", which may be translated (somewhat inaccu rately) as the principle of rule of law. This principle may to a large extent be explained as an objective of achieving legally correct results, which again may be supported by the application of legal method. Even if I primarily advocate values and interests which may be associated with the principle of rule of law, I do not, of course, deny the fact that other interests are relevant and may lead to a weighting of interests and thus a reduced significance of the "ideal" values. In this article,
197 Compendium to the Erasmus Course however, attention is drawn more towards possible syncretisms, rather than towards conflicts. Here, I denote the computer systems in public mass- administration as "case-processing systems". Case-processing systems constitute integrated parts of the total case-handling, i.e. manual and computerized routines are closely linked. This close linkage makes it hard to discuss case-processing systems as distinct from the totality of case-handling. Nevertheless, in this article emphasis is put on the computer aspects of case-handling. Case-processing systems are common and important in the parts of public administration which have the task of distributing money in the welfare state. The local offices in tax and social insurance administration are central examples of services which make comprehensive use of such computer systems. Case- handling in this part of the public administration of the welfare state, is characterized by a large number of cases, strongly routinized work and relatively little individualized decision making (compared for example to the handling of administrative appeals). In my discussions, I mainly have in mind case- processing systems within such types of mass administrative organizations. Authorities that come under this category of public administration encompass very important parts of public administration which decide a great number of cases every year, with decisive influence on the economic situation of the individual citizens. The second expression which needs amplification is "legal aspects". I have no ambition of discussing every legal aspect relating to case-processing systems in public administration. Systematically, it is appropriate to distinguish between two groups of legal issues: those concerning the description of system development as a legal decision-making process and those concerning the legal regulation of system development. The latter group may be divided into two categories: (i) questions concerning government regulation of system
198 Public Administration and Information Technology development, and (ii) questions concerning contractual regulation in case-processing systems. In this article, I discuss many of the issues within the first group (description of system development as a legal decision-making process) as well as certain issues concerning government regulation of system development. Contractual and other private law questions are not treated. The third element of this presentation which needs to be clarified is "system development". This expression is, of course, far too comprehensive for my purpose. I could have chosen to write about the development of computer programmes, but even this expression fails to give a precise description of the fraction of system development work I have in mind. My discussion only takes up questions connected to the transformation of legal sources to computer programmes. Furthermore, when I talk about "system development", I think of development during a life cycle; i.e. I am concerned not only with the first and major effort to establish a computer system, but also with the continuous maintenance of systems occasioned by amendments of legislation etc. I use the term "transformation" to designate the formalization process in which legal sources are selected, interpreted and expressed as instructions written in a computer language. This transformation is required in the development of case-processing systems supporting the handling of single cases. If measured in number of lines of programme code, I suggest that I only speak about the development of 5 - 10 % of the programme code. On the other hand, much of the information analysis work precedent to the programming would be linked up to the interpretation of relevant legal sources.
199 Compendium to the Erasmus Course
3. Traditional procedural systems versus expert systems
In recent years, much effort has been directed towards discussing legal expert systems and knowledge-based systems. These are computer systems which have been regarded by some people as advanced tools for government administration, lawyers and others, representing the wave of the future. It may thus be surprising for some readers to learn that questions concerning such systems are not the subject for my discussion. Although legal expert systems have received a lot of attention, they have for a long time mainly been at an experimental stage with prototypes rather than utilitarian system solutions put in regular operation. This is not to say that work within deontic logic and expert systems in law are of little general interest. Other contributions in this book may illustrate the usefulness of logic, both as a tool for analysis and as a future basis for automatation of decisions. However, the whole money-distributive welfare administration in our society is totally reliant on the existence and smooth functioning of "old- fashioned" computerized case- processing systems. In order to handle the challenges of modern mass administration, there has been a rapid development within public administration of traditional, procedural computer systems. Since the first computer systems of this kind were worked out in the early 1960s, there has been several alternations of technical stages towards the development of the traditional but advanced and powerful systems of the 1990s. My assumption is that although new methods, computer languages and tools certainly will be introduced, at least two characteristics will remain. First, mass-administrative systems in public administration will mainly be based on explicit and fixed procedures in which individual characteristics will seldom be found. Secondly, and in extension of the first point, computer systems in mass administration will mainly comprise procedures directly
200 Public Administration and Information Technology
extracted from legislation, and will not contain representations in which vague and discretionary concepts are "dissected" and detailed.4 Other strategies will obviously create serious delays and will thus not be feasible without larger and more expensive bureaucracies. Besides, the mass of individual decisions made according to the detailed legislation on welfare money- distribution schemes are trivial. If only 10-15% of these cases may be judged as "hard cases" which deserve detailed individual judgements, it is likely that the computer systems in local administration will be geared first and foremost to the requirements of the great majority of trivial cases. Computer support of "hard cases" is most likely to be developed as attached modules, with the main purpose of assisting in the handling of appeals. Notwithstanding future technical developments, it is a fact that our societies are - and will be - heavily reliant on the existence of stable and effective case- processing systems. This observation again indicates further reforms of such systems rather than revolutions. Thus, the topics taken up in this article should be expected to be of great importance, not only tomorrow, but in the foreseeable future.
4. System development seen as a legal decision-making process
Above, I have mentioned how an important group of legal questions are connected to the transformation of legal sources to computer programmes. The situation I presuppose is that the task of the public administration is to work out computer programmes which, when employed, yield legally correct results in single cases. The programmes must, in other words, contain rules which are in accordance with the picture drawn by the total
4 For example vague concepts like "supporter" and discretionary concepts like "if reasonable".
201 Compendium to the Erasmus Course collection of legal sources relevant to the interpretation problem at hand. To be more concrete; in most such transformation tasks, the challenge is to carry out legally correct interpretation of statute law (for example provisions in Norway's National Insurance Act or Inland Revenue Act), in the light of other relevant sources, such as preparatory works, case law and administrative practice. In the introduction, I have mentioned single case decisions and bureaucratic lawmaking as two main groups of powers in public administration. Within public administration law we are familiar with a distinction between administrative decisions of single cases and general decisions, such as regulations and general instructions. The first category of decisions is directed to specified individuals, while regulations are directed towards a non-individualized circle of citizens. Transforming legal sources to computer programmes also represents a type of decision making process, but may hardly be categorized under one of these two groups. Instead, judgements made in the process of transforming legal rules to programme code should be described as something in between single and general decisions. A general administrative decision in the form of bureaucratic lawmaking (e.g. a set of regulations) will often be much more detailed than Acts of Parliaments. Nevertheless, when confronted with "real cases", there will usually be a number of problems in interpreting elements of these decisions as well. In the course of traditional processing of single cases, the interpretation is case-driven, i.e. each specific case occasions interpretive questions and answers. Problems of interpretation may, in other words, lie untouched and will remain unsolved unless a case gives rise to the problem. The development of case-processing systems implies another and a much more difficult approach. In the course of transforming legal sources to computer programmes, every question of interpretation of the Public Administration and Information Technology source should, in principle, be posed and answered; i.e. without waiting for an individual case to bring up the problem. When I describe decisions made in the course of transforming legal sources to computer programmes as "transformation decisions" representing something in between decisions in single and general cases, it is because the transformation process will yield resolutions to a great number of such interpretation problems, in the form of very detailed general decisions. Therefore, the computer programme should contain direct applicable solutions to almost all single cases within the domain. Transformation decisions should thus, first and foremost, be placed in a continuum between single case decisions and general decisions, but with the recognition that when placed in this continuum, there will often exist an affinity towards the side of single case decisions. In such cases, it may be appropriate to talk about establishment of the transformation as embodying quasi individual decisions.
5. About the legal basis of the transformation process
I have described the transformation process as a process of interpreting statutory texts and other relevant legal sources. The first problem I want to point to in this connection is the fact that it often seems to remain a secret which legal sources are chosen as the basis for the programming.5 Of course, in many incidents this question has obvious answers: it will often, for example, be obvious that a certain calculation has its legal basis in a particular section/paragraph of an Act, and that only can tankerous people could find the opportunity to argue in favour of
5 The question of referring to legal bases for transformation decision is closely linked with questions of a legal documentation of the computer programmes. Below, I will return to this documentation and transparency problem. Compendium to the Erasmus Course another interpretation of the provision in question. However, transformation decisions will often contain much more detailed procedures than those which one could claim that the statutory texts dictate, c.f. below. Thus, in order to facilitate legality control, it will in many cases be advantageous to have references to the legal sources which constitute the legal basis for the transformation decisions. This is particularly desirable in cases where no legal basis appears to exist; i.e. in situations when the persons in charge of system development are in the position of law-makers rather than of interpreters. References to the legal basis for the transformation may be regarded as a measure to facilitate legality control of the transformation, i.e. of the legally based sequences of computer programmes in case-processing systems. However, such references also constitute a starting point for checking the legal basis itself. In other words, one thing is to check that the interpretation resulting from the transformation is correct, or that the lawmaking in course of the transformation is within delegated powers etc. Another thing is to check if form and content of the legal basis itself is acceptable. Examples from my investigations of Norwegian computerized case-processing systems, for example, show that provisions of the National Insurance Act are without a sufficient legal basis: There was, for example, no sufficient legal basis for "negative" provisions laying burdens on citizens, restricting rights etc. If there is doubt, steps should either be taken to change and reinforce the legal basis or to amend the regulations so that they clearly have a sufficient legislative basis. An example of problems of a more formal type, are provisions with basis in an Act, but where the statute delegating statutory powers has been repealed by mistake. For instance, regulations concerning calculation of fishermens' pensions were given according to s 3-15 (2) of Norway's National Insurance Act. This paragraph was repealed, however, without a new legal basis Public Administration and Information Technology being established. In other incidents, there are incorrect references in a provision of the Act to its legal basis. A special situation occurs when a provision undoubtedly has a content describing operations which should be the subject for transformation, but where this content is of so little current interest that it is skipped in the transformation process.6 In the Norwegian Administrative Procedure Act a certain type of general decision is defined - "forskrift", which may be translated as "regulation". If a decision comes under the statutory definition, certain procedures must be followed in accordance with the Act. One effect is that every such decision is published and is consequently made easier to access than most other general decisions (for example instructions) made by public administration. Moreover, the question of whether the decision is a "forskrift" or not, may influence the question of the legal power of general decisions, for example, to what extent citizens may base rights on such decisions. Generally, I regard it as a task for public administration to clarify the relationship between their general decisions and the definition of "forskrift". In my view, it is particularly important that this problem is handled when general decisions are included as a basis for computer programming. System developers should know whether they base their work on a "forskrift" or on another general decision. If, for example, one of the statutory texts is found to be a "forskrift", but without the special procedures being followed, then this omission should be "repaired" as a result of the transformation work. A "forskrift" which is not published, should be published, for instance, on the occasion of the transformation. The intention behind these examples is to show that the transformation process also represents a good opportunity to carry out a legal control on more than one level. One of the
6 For example a minimum limit which has not been adjusted and thus does not match with the figures of the real world.
205 Compendium to the Erasmus Course simple thoughts behind the advice to exercise such control, is that computerized case- processing systems should only be built on solid legal ground. This implies that quality control of these legal aspects of system development should be taken more seriously than what is usually the case in relation to traditional statutory texts. Transformation work should, in other words, be based on legal sources which do not suffer from obvious or unnecessary errors which again may lead to a questioning of the legality of the case- processing system and consequently, of the decisions in single cases.
6. The extent of the transformation
When legal sources, i.e. typically Acts and regulations, are transformed into computer programmes, only a fraction of the meaning represented. In short, only those fractions of the texts which describe arithmetical and/or logical operations are selected. This implies that statements expressing calculations and trial of conditions constitute a considerable part of the legality of a relevant programme code. If case relevant information is accessible in machine-readable form, then also the establishment of facts in cases may be integrated in the operations capable of being computerized. However, despite the Norwegian government's strong aspiration to make much better use of already stored data, the general picture is still that the facts of cases are established on the basis of human considerations brought about by the current case. The main picture is, in other words, that computer programmes only contain a representation of limited fractions of the legal sources, and thus many legal questions remain for people to consider in single cases. Nonetheless, the fractions which do undergo transformation are indeed important parts of the total body of legal sources, in particular because arithmetical nnerations are of core importance when deciding amounts of Public Administration and Information Technology
excise, tax, social insurance, benefits, loans etc. Furthermore, if the policy of the government prove to be a success, still larger parts of the legal sources will be subject to transformation, i.e. an increasing number of information types will be collected by means of linking registers, controlled by operations executed by computers.
7. The degree of the transformation
Transformation results in a programme code whose adopted content is changed in relation to the natural meaning of the legal sources constituting the bases for the process. The degree of required legal interpretation does, however, vary quite a lot. One group of transformation incidents is characterized by the fact that the legal source itself use mathematical or logical notions which allow a direct transfer from one formal expression to another. This will imply that no legal interpretation is required, only the application of the rules of the two systems of formal notions. The second step, representing a slightly higher degree of transformation, may be exemplified by legal texts written in natural language, which most legally educated people will find trivial to interpret. There are texts which may contain linguistic uncertainties, but without these representing a legal interpretation problem. If, for example, the terms "income" and "annual income" used in a provision, case law, preparatory works etc. clearly show that these terms must be read as synonyms, then the interpretation should be seen as one merely of linguistic interest. If, on the other hand, other legal sources do not contribute to a clear interpretation of the two terms, and grounds exist to regard the terms as distinct from each other, the interpretation should be classified as "legal". The third and last step of my simple division according to the degree of transformation, is represented by cases when a
207 Compendium to the Erasmus Course qualified legal interpretation is needed in order to decide what is the exact meaning of the legal sources. Closely related to this group, but outside what should be designated "transformation", are incidents where the legal sources do not give any basis for the solutions required by the planned computerized routines. Obviously, it is transformation of this "third degree", as well as cases where a legal basis is missing, which are of the highest interest in a legal discussion. In the Norwegian system for unemployment allowance, for example, the relevant Act contains two alternative income bases for calculating allowances. Either the basis is the income during the last whole year, or the average income of the three last years. In the programme code of the case-processing system, the Act is supplemented so that if only income figures from the two last years are registered as income basis, then the system calculates an "average" by dividing by three, as if the income figures referred to three years.
8. Selecting among the operations capable of being computerized
On the basis of my investigations, it may be stated that not everyone of those legal sources which may be subject to transformation, will actually undergo transformation. This implies that often the representation of legal sources in computer programmes are incomplete. Some operations which are capable of being computerized are, for example, thought to be of so little practical importance and/or so complicated that they are omitted. One example of this is a "hole" in the case processing system for calculating old age pensions, resulting in manual calculations of deductions in pensions due to stays at public homes for the aged Public Administration and Information Technology
and nursing homes. A control showed that these manual operations resulted in an error rate of 22,6 per cent.7 In other cases, it is expected that statutes will be repealed, and these are thus not comprised by the transformation work. The most important examples of types of operations that are left out of transformation, are incidents where decisions requiring an overall solution, are of a discretionary nature. To receive a supporters' supplement according to the State Educational Loan Scheme, a condition is that you support a child under 18 years old. For some groups of applicants, this condition was dissolved in a set of simpler sub- conditions, and made subject to a computerized assessment of the conditions. As for other groups, the conditions were assessed by the computer in order to arrange a basis for an officer to make a comprehensive decision, these incidents were not comprised by the representation of the legal sources in the programme code. The fact that case-processing systems are likely to contain an incomplete representation of the underlying legal rules seems to represent understandable choices made on the basis of cost- benefit analyses and other administrative considerations. This incompleteness should in itself not imply legal problems for citizens. However, in the Norwegian systems which were investigated, there was no information on the fact that some elements of the statutory law were not part of the case- processing system. In my view, it is obvious that in a situation were the officers using the system are unaware of omissions in the computer programmes, erroneous case handling may occur. If, for example, two figures are calculatcd according to the same formula, but with different maximum limits persuant to legislation, and only one of these limits is integrated in the system, there will be a danger that cases to which the other limit apply are also processed in the routines of the system. In
7 See "Vederlag for opphald i kommunale alders- og sjukeheimar. Kontroll av trekk i trygdeytelser." Trygderevisjonen, 27 July 1992
209 Compendium to the Erasmus Course addition, there is a chance that a combination of ignorant officers and un-documented omissions of the case-processing system may lead to incorrect results because the case handling operations in question are dropped. This situation points to the problem of documentation and system transparency, see below.
9. Supplement to operations which are capable of being computerized
At the same time as parts of the legal representation are incomplete, other parts of the programme code are added, a content which may not be directly read from the legal sources consituting the basis of the transformation. This occurs when the legal sources must be interpreted and the interpretation yields a result which represent one of several possibilities. In such situations, the programme code will express a content which differs from the content of the natural language being subjected to interpretation, because the programme code is made more precise. One example could be taken from the unemployment allowance system. During system development, the project group needed to decide how many days there should be in a week, and defined this as five days. This was not only a new definition (hidden in the code), but also a quite surprising choice, since "a week" in closely related cases could be four alternatively six days. Operations that may be transformed may also be supplemented by means of replacing a non-transformable operation with a transformable one. A statutory provision refers, for example, to a piece of information which is not accessible in machine-readable form, the operation of collecting that information is thus not transformable. However, in such a situation, one possibility will be to replace this factual information by another piece of (hopefully) synonymous information or by a combination of facts which are believed to Public Administration and Information Technology express the same meaning the statutory expression refers to. In the system of calculating loans and supporter benefits under the Norwegian State Educational Loan Scheme, for example, the information about "common household with the supported child", was replaced by information concerning common household with the spouse with whom the applicant had a child. The contents of the parts of computer programmes with a legal basis, need not merely have this basis in accepted legal sources. In addition, the content of programmes may be formed in order to facilitate reasonable and acceptable case-handling routines. There are examples of programmes with deliberate "incorrect" content; for example, a content yielding too restrictive results for citizens. The intention is, however, to let such cases undergo an individual judgement by an officer on top of the result from the computer. Another important type of content which is often added to that content which is based on legal sources, refer to different limits and input-controls. These are made part of the case-processing system in order to detect mistakes due to writing errors and registration of incorrect information due to unlawful actions, ignorance, misunderstandings etc. My investigations8 showed examples of in-put controls which stopped the registration of lawful facts, as well as controls which seemed to be quite casually defined and thus without full effect as a quality control.
10. Incorrect and doubtful legal representation
Above, I have described the transformation task. Here, I will briefly address two types of legal problems in this connection, concerning the legal content of the case- processing system; (i) incidents when the representation which the transformation
8 See Dag Wiese Schartum: En rettslig undersøkelse av tre edb-systemer i offentlig forvaltning, Complex 1/89, Tano 1989. pp. 26 - 33.
211 Compendium to the Erasmus Course process produces may be claimed to be incorrect or of doubtful correctness;9 (ii) errors connected to the interlocking of the computerized and the manual case-processing routines. Although the objective of producing legally correct computer programmes is a central point in this article, it would be taking things too far to discuss in any great detail what "legally correct" should imply. For my purpose, a brief definition of a legally correct computer programme is sufficient. When I speak in the following about legally correct computer programmes, I refer to computer programmes which are the result of a transformation process conducted in accordance with generally accepted legal methods. Such methods encompass, for instance, demands on the selection of legal sources, weighting of different sources and principles of interpretation. This definition of legal correctness is linked to process rather than to results. Nevertheless, the correctness may be judged, to a large extent, on the basis of the results of the computerized processing of single cases. There are always a certain number of errors in new programmes, even after machine testing. Figures from the U.S.A. indicate from 3 to 5 errors per 100 lines of programme code in Cobol programmes, which is a considerable figure when we take into consideration that each programme may contain several thousand lines of programme code.10 On the other hand, errors in the representation of legal rules are not necessarily of juridical nature. Often errors are due to bad programming and do not refer to legal misinterpretations. Nevertheless, my
9 A complete list of problems may be found in Dag Wiese Schartum: Rettssikkerhet og systemutvikling i offentlig forvaltning, Tano 1993, p. 221. l0See Edward Yourdon: A Nation at Risk, conference note, fifteenth annual computer security conference, 1988. Public Administration and Information Technology
investigation showed that still after more than two years of use, the programmes may contain errors of juridical significance."
An example could be the Norwegian system for calculation of unemployment benefits, where the system contained two serious errors in a special routine for benefits to people working in the fishing industries. In a part of the programme code which affected most applicants, the rounding to the nearest thousand NOK was executed after, instead of before, a deduction to a maximum amount limit.
I also found that examples of programmes containing doubtful interpretations were more usual than occurrences of errors. Some of these interpretations had a basis in circulars, while other questions of interpretation had previously not been subjected to commentaries from superior authorities.
In the programme which calculates sickness benefits, the income figures from the three last years were selected as a basis for calculation. According to the wording of the existing provisions, the basis should be the three last income figures. Another interpretation in the programme code made it impossible for farmers to have a reduced rate of voluntary sickness benefits during the first 14 days of sickness. 100% benefits would be paid from the 15th day according to a treaty with the farmers' organization. In the programme, the fact that rates were 100% from 15th day, implied also 100% from the first day.
There are also examples of "new rules", i.e. rules only codified in the computer programme. However, I did not investigate to what extent the officers actually had knowledge of them. The first example (below) probably represents a totally new rule, while the second rule I mention was probably known to the officers.
In connection with a particular calculation in the programme processing unemployment benefits, five days were defined as a work week. However,
1 *See Dag Wiese Schartum: En rettslig undersøkelse av tre edb-systemer i offentlig forvaltning, Complex 1/89, Tano 1989, pp. 143 - 44.
213 Compendium to the Erasmus Course
in every other respect in the Act upon which the programme was based, a week contains six work days. In a system for calculation of state educational loans etc, a condition was changed from "living together with the supported child" to "living together with spouse or partner in common law marriage relationship", which is obviously not the same.12
Computer programmes may give correct results in the sense that no misinterpretation is made in the process of transformation. Nevertheless, legal questions arise if the manual routines are not sufficiently integrated with the computerized routines, giving incorrect results. An example of this last category is the fact that the Directorate of Taxation, throughout more than two decades, has omitted to use the results of the final tax assessment as a basis for the computerized calculation of pension-earning income. The computer programmes were correct, but a gap existed between the computerized and the manual routines, which again lead to calculations on an incorrect basis, and which affected a large number of pensioners.
11. The handling of factual information
Most of the facts of the single cases are registered into the case- processing system by an officer, on the basis of forms and other documents in the dossier of each case. In the course of this registration work, factual information is transformed and represented as figures and letter codes with a specific significance in the context of particular programme codes. The employment of case-processing systems implies a two-step transformation: First, the operations described in the legal sources which could be computerized, are transformed into programme code. This is carried out as part of system development; e.g. prior to the use of the system. Afterwards,
12Both examples could be classified as "errors", but this category was reserved those rules which expressed unintended results. Public Administration and Information Technology when the system is put in operation, the factual information on which the programme code operates, is described by means of the registration of in-put values, expressed in a well-defined coding. These in-put values may be regarded as pointers both to the programme code and to the relevant legal source which defines the categories of case relevant facts and their allowed values. However, in my investigation of case-processing systems in Norway, Sweden and Denmark, no such direct references were detected. When these references are omitted the result is a problematic division between the statute law forming the basis of the case-processing system and the legal content of the system. An example from the sickness benefit system is a field on a registration screen with the category "earned income", which is "explained" as "income" in the users' manual. This income information constitutes a reference to "salary the insured person has earned from work in private or public employment [...]". Three different terms are, in other words, used to represent the same type of information. The designations in the system were chosen differently and without reference to the wording of the statutory text. In my view, this is not acceptable. The computer system is closely linked to its legal basis and by no means independent of these legal sources.
12. Documentation and system transparency
The decisively most important issue I highlight, which many of the above mentioned problems lead up to, concerns legal documentation of programme instructions, i.e. the working out of explicit links between legal sources and the computer system, as well as clarifications of interpretive choices and transformation decisions which complement the legal basis. In a wider perspective, this is part of the question of system transparency which I address towards the end of this section.
215 Compendium to the Erasmus Course
Insufficient documentation of legal sources clearly has serious implications, particularly for control of its' legality control and for officers' possibilities to give guidance to citizens regarding the computerized handling of cases. Although the case-processing system is closely linked to legal sources, one rarely sees systems designed and documented in a way which show this link. Programme codes, documentation, user manuals and other relevant material very seldom contain information about the legal basis of the computer system. This does not imply that the system documentation is entirely missing. My investigations showed both systems with relatively good documentation - for purposes of "technical" system maintenance - and systems with (almost) no documentation whatsoever. In view of my explanation of the transformation process as legal decision-making, one may speak of the latter type of systems as harbouring "secret" legal decisions. In the case of programmes calculating sickness benefits, for instance, only two people knew the programme code, but they were both unable to explain several details without prior in-depth and time- consuming analyses.13 When I stress the need for comprehensive system documentation of high quality, I speak first and foremost about the documentation of those fractions of the systems constituting transformed legal sources; i.e. a limited part of the total programme code. Because I see this transformation as a type of legal decision-making, I talk about putting legal decisions down in writing in a language which is developed for "ordinary people" to read. At the same time, what finally ends up as documentation should, during the transformation work, function as specification at the stage of system realization. It will often be difficult to satisfy both requirements - general readability and specification function - if documentation is
,3See Dag Wiese Schartum: En rettslig undersøkelse av tre edb-systemer i offentlig forvaltning, Complex 1/89, Tano 1989, pp. 65 - 66. Public Administration and Information Technology
merely worked out on one level. In some cases, documentation should therefore be worked out on two different levels of formalization, with difference between levels so little that no major doubt exists about the transition from one level to another. The most formalized level will be suitable for technical implementations of changes and maintenance, while the less formalized version will cover needs of the specialized departments of central administration and of officers serving behind the counters in local administration. This less formalized version may be integrated into traditional instructions concerning interpretation of legislation and other legal sources. Legal decisions in the course of system development, should not only be made visible in the form of documentation, but, in addition, form the basis for amendments of the legal sources which are the subject for transformation. The idea is that in addition to regarding legal decisions as part of system development, we should regard system development as part of the effort to improve rules and regulations. Then transformation from legal sources to computer programmes is seen as an analysis illuminating errors, inaccuracies and relations for the purposes of effective legal regulation. Can sufficient legal documentation of case-processing systems create system transparency? Of course, with a detailed and intelligible documentation it will become possible for citizens to find out and understand how their cases have been processed - at least with some help from officers. Such a documentation may be sufficient in cases where the citizen wishes to investigate whether or not a foundation exists for an appeal over a single decision. On the other hand, when citizens are offered a careful explanation of computer operation, this is by far a much less desirable situation to be in as compared with the situation of dealing with public officials: The computerization of public administration has meant that the officials in the administration have become more independent from the statutory texts. No
217 Compendium to the Erasmus Course
street-level officer needs to read and understand how benefits, taxes etc are calculated. Calculations are the work of government computers, not government officials. On the other hand, calculations are still the task of citizens who are so foolhardy as to try to penetrate the often very complicated statutory law. Therefore, if we compare the situation of the officers and the citizens throughout the period of computerization of public administration, the picture we gain is of an administration which has more powerful tools to execute laws. On the other hand, most citizens are left in more or less the same situation as in the childhood of the computers: Before calculations can be executed, the many fragments of related statutory provisions must be identified, interpreted and put together. Thus, computerization of public administration may be regarded as representing a displacement of the relationship of strength between the administration and the citizens. The question should be asked, have computers made the public administration too powerful vis-a-vis citizens when it comes to tackling complicated statute law? In the case of Norway, my opinion is that the answer should be yes. In such a situation, system documentation should not be the only means to create system transparency. In addition, there is a need for public access to case-processing systems as a supplement to the traditional safeguard of public access to statutory law. Such an improved publicity of legislation could be carried out, for example, through giving access of case-processing systems to banks, union offices, citizens advice bureaus and to other organizations which are able to specialize in operating the system, and at the same time identify themselves with the interests of the single citizen. Public Administration and Information Technology
13. System development and work division in the project group
Various expert groups are involved in the creation and maintenance of computer systems. The job of a jurist vis-a-vis the computer is first of all to ensure that the law describing calculations etc is correctly implemented in the computer system. This is also partly the challenge for the accountants within the field of money- distribution. Moreover, the choices of jurists and accountants will deeply affect the jobs of the system- designers, programmers and other "technical staff'. Thus, the computer makes system development and maintenance a common arena for several expert groups of an organization. This makes it necessary to implement processes based on parallel decision-making, processes which are specific for each subject, but which have the goal of implementing results in the same computer system. The meeting between different expert groups surrounding the computer, as described above, creates a need for close co operation. Ability to co-operate does not only require willing ness to co-operate, but knowledge of other expert groups' perspectives and approaches is of key importance. Today, such knowledge often seems to be lacking. One of the future tasks is thus to build bridges; i.e. to lay the groundwork for fruitful and close contacts between the involved professionals.
14. Lawyers in the project groups?
In the following, I will present some characteristics of the organization and division of work in selected system development projects. The description is based on a detailed investigation of three projects in three different branches of Norwegian government administration. Here, I will concentrate on system development within the sickness benefit scheme in
219 Compendium to the Erasmus Course the National Insurance Administration, and the development of a tax deduction system in the Directorate of Taxation. Both investigations were carried out in 1991/92. In addition, I base my views on less detailed knowledge of three Norwegian and three foreign projects. However, since the research basis first of all concerns Norwegian system development projects, I will stress that the results primarily describe a Norwegian reality. Results from the foreign projects may, on the other hand, support the assumption that these questions have a more general relevance. I assume that no-one will be surprised to hear that project groups seem to be an usual choice in the organization of system development. In Norway, project organization was recommend ed already in a white paper from 1968 concerning government computer policy for the 1970's. Today, the Directorate of Organization and Management (Statskonsult) generally regards project organization as suitable for many purposes, and the general picture is that project groups constitute an increasingly important way of organizing tasks which are to be carried out within a certain time and budget. All project groups I have investigated could be characterized by their inter-disciplinary composition of people, with partici pation from different departments of the institution in charge, local administration and often with people participating from external, private, consulting firms. However, the organizational context of the project groups seems to vary in that no standardized organization is built around these groups. One common feature is a relatively loose connection between the project groups and the specialized department having authority within the legal domain in question. The connections to computer departments were, in general, much stronger than to the specialized departments. In the National Insurance Administration, local office experts on social insurance regulations held a very central position in Public Administration and Information Technology loosely organized project groups. These experts' employment in the social insurance organization was partly linked to the System Department of the National Insurance Administration and partly to the local offices. In addition, people from private consultant firms participated, mainly as programmers. The group had no steering committee or superior group to follow and control their work, but reported to the appropriate specialized department of the National Insurance Administration; i.e. the department in charge of the special benefit being subject to system develop ment and transformation. This contact consisted of introductory talks with the department and meetings, which were called by local office experts if they found it necessary to discuss with the specialist departmental interpretations concerning the transfor mation. In the project of the Directorate of Taxation, the group was manned with people from both the specialized department and the computer department, and with participation from a private consulting firm. The project group was placed under a "project director" who reported to a board of directors which executed control over a scheme of computer projects. Two representatives from the project group regularly met with the "project director", but here discussion was limited to administrative matters. The board of directors primarily had the function of deciding the implementation of pilot- and main projects. The local office experts on legislation participating in the project of the National Insurance Administration, had their background in local office work. They had no legal education, but were highly competent within the field of the Social Insurance Act and attached regulations. The professional background of the project group representative from the specialized department of the Directorate of Taxes, was similar to the background of the local office experts of the National Insurance Administration, i.e. she was a specialist in the rules and regulations in question, but lacked formal legal education. In
221 Compendium to the Erasmus Course
both examples, lawyers of the specialized department did not actively participate in the transformation of legal sources. Why? Explanations of the passivity of the lawyers must generally be based on the fact that this group of officers are almost solely employed in central administration. Their main tasks are the handling of appeal cases and preparation of rules and regulations, including the drawing up of general instructions regarding interpretation of legislation. This reflects the division of work between central and local administration implemented from the 1970s: central administration should mainly have functions such as policy-making, planning, control and instruction; the local administration, on the other hand, should be decentralized and execute processing of single cases, often within very wide fields of legislation. This work division has lead to heavy work loads for local administration and to a computerization directed, first and foremost, towards managing the local office situation. 1 am not saying that the process of computerization has not had any influence on central administration. Several central computer systems have been established and run during the last decade. Computerization of central administration has, however, primarily affected computer departments, while the majority of the central administration has had very little experience with case-processing systems. An effective participation in system development work demands a certain minimum knowledge of the functions of the case-processing systems and of relevant case-processing routines etc. Up until now, most lawyers have not had the experience that has made them sought after when project groups have been composed. The officers of local administration have been in the opposite situation because they have possessed a combination of knowledge of rules and regulations, of the case-handling process where the computer system is to be introduced and they have had practical experience with the use of functioning computer Public Administration and Information Technology systems. They have been, therefore, useful participants in system development projects. Lawyers' lack of experience with use and development of computer systems may also have prevented them from developing a legal-based view on the significance of transformation of legal sources to computer programmes and, generally, of computer supported case-processing. What 1 am indicating is that lawyers may lack insight on the legal implications of computerized case-processing.
15. Why give the lawyers influence?
To create case-processing systems which may better contribute to legally correct results, and, in particular, to solve the two types of legal problems I have mentioned, it is essential that lawyers are actively integrated in the transformation of legal sources. Both lawyers in central administration and the legal experts of local administration, have knowledge of rules and regulations. Lawyers differ from the great majority of local experts in that they have education and experience in general legal method. Lawyers do not merely have knowledge of a sector of legislation (e.g. social security or tax legislation) they also have knowledge of related legislation and of general legal principles. Legal education gives lawyers competence to have a wide and open in put in decision-making processes; i.e. to interpret legislation in light of a variety of legal sources. This competence constitutes one important reason for employing lawyers in central admini stration and a reason to give them the task of deciding appeals and giving general instructions on the interpretation of legislation. 1 have previously concluded that the transformation of legal sources to computer programmes should be regarded as quasi-individual legal decision-making and something that systematically may be placed between the general and individual
223 Compendium to the Erasmus Course decisions of public administration. The fact that these quasi individual decisions have a direct and decisive effect on the result of each case, brings about the question: why should we have a different quality standard for the legal work connected with these decisions than for other general decisions in public administration? Does the fact that instructions are issued in the form of instructions in a computer programme imply that the legal problems in question should be less accurately handled than traditional instructions? In my opinion, the standard of legal investigations and discussions should be the same for general instructions regarding interpretation of legislation on the one hand, and transformation of legal sources resulting in instructions in computer programmes on the other. In almost every project I have investigated, representatives of computer departments have stated that individuals employed in these departments should be regarded as top experts on those parts of legislation which are implemented in the case- processing systems. In light of the previously presented picture of rather passive, specialized departments, I assume that this is often a correct description of the situation. My investigation gives an impression of the introduction of a schism between the legally oriented work of specialized departments and of the legally oriented work of the computer departments. Computer departments are, to a large extent, allowed to decide on the handling of legal questions regarding transformation of legal sources; i.e. first and foremost, questions concerning algorithms controlling calculations and the testing of conditions. The specialized departments, on the other hand, decide on what may be seen as "true legal problems"; i.e. problems of interpretation of vague and discretionary concepts, such as "supporter" and "if reasonable". Such a division, however, is not acceptable because it implies a division between legal problems which are highly integrated with each other. Public Administration and Information Technology
According to present computer policy in public administration, it is an objective to re-use information on individuals once it is collected and registered in machine- readable form. Such an "ecological" view of information requires an increased linking of registers. This again implies that definitions of case-relevant facts will be introduced to establish the necessary degree of equivalence between types of information in linked computer systems. To the extent that these definitions are established at an administrative level, work on them will involve increasingly important types of legal considerations in the course of system development. The situation may be, for example, that the method of collecting a certain type of information is changed in order to cut down on expenses. In the course of system development, it would have to be considered whether or not information, which already is collected by another body, could be used instead. This would involve, in turn, examining whether or not that other information has definitions such that it can be used for the purpose for which the former type of information was collected. In this connection, it should also be considered whether or not manual routines are required to check certain categories of the automatically collected information. These two questions are of a legal nature. If the policy of the government is carried out it will be, therefore, even less satisfactory not to involve lawyers and other experts from specialized departments in the work on system development. I emphasize that my argumentation for the necessity of lawyers in parts of the work on system development, does not imply that there is less need for local office experts. I regard local experts as especially important because of their knowledge of local interpretations and views on legal questions, their experience and knowledge of the practical functions of computer systems in operation and their ability to pin-point problems regarding work environment. As I will soon return to, I prefer a
225 Compendium to the Erasmus Course solution where both lawyers from central administration and local office experts participate in the transformation work.
16. A sketch of an organizational structure
Above I have discussed the work division between lawyers and local experts, and between the specialized departments and the computer department. My next step is to suggest some possible changes in organizational structures. The startingpoint is system development work organized in project groups. Since system development consists of activities which require a number of skills, there is, in my view, no alternative to composing project groups along inter-disciplinary lines. However, there are no definite answers to the question of how project groups may be integrated into the rest of the organization. Furthermore, there is no antagonism between project groups and an hierarchical organization. In an hierarchical organization, cases are often handled in two steps. The first step may be the preparation of the case by an officer; the next step, the evaluation and decision-making by an office manager. This structure has certain advantages that should be recognized. First, the second phase of a two-step case- handling may have the function of quality control. The preparation of the case provides a structure, formulates the problems and suggests solutions. On this basis, it is possible for the office manager to concentrate on other aspects of the case; for example, to assess the coherence of the prepared text. It is equally important that persons involved in an hierarchical case-handling, may have different views on the solutions to the problem at hand. Thus, two-step case-handling may make the evaluation of a case more comprehensive. If the person in charge of the second step is an office manager with the responsibility for several related problem areas, the second phase of the case- handling may also imply inter-specialist considerations which Public Administration and Information Technology
the specialized officer, who carried out the preparatory work, does not have the qualifications to perform. In other words, an hierarchical case-handling routine of legal questions contained in system development, has features which facilitate quality control of the transformation of legal sources making it possible to consider questions in the light of related legal issues. At the same time, it is important to recognize similar advantages linked to project groups. I assume that discussions concerning legislation and its interpretation will be more thorough in groups where both lawyers and local office experts participate than if only one of these parties contribute. Furthermore, the participation of system analysists and programmers may raise legal questions not considered by the legal experts. System analysis may, for example, reveal logical inconsistencies in the model drawn up by legal experts. In Norwegian government administration, it may also be relevant to have participation from people dealing with accounting, auditing and statistics. These aspects may also raise legal questions. System development constitutes - as indicated - an inter disciplinary task. In order to be able to cope with each of the activities in an adequate way, it is in my view necessary to regard each activity as part of a closely related but separate decision-making process. A legal decision-making process constitutes one important process when case-processing systems are developed. Today, this process is to a large extent not made explicit. The legal questions are first and foremost handled within project organizations without major interference from the hierarchical organization of specialized departments. My vision of an improved organizational solution for system development consists of three levels, of which the first two are the most important. The project group, i.e. an inter-disciplinary and inter-departmental group constitutes the bottom level. Here, preparations arc made and proposals to solutions worked out.
227 Compendium to the Erasmus Course
The second level is that of specialized departments, typically the department in charge of the benefit or duty in question (sickness benefit, VAT, etc.). Here, decisions are made if the question involved comes under the authority to the particular department. Amendments of the proposals from the project group should be considered by the project group before a final decision is made. The total system solution is decided by the top management of the agency on the basis of recommendations from the departments involved (3rd level). In this simple model, the project group exists without a superior project organization. Instead, the idea is to combine project groups with a traditional hierarchical organization by regarding the group as the first level in a two-step case-handling, preparing decisions in different departments. One important difference compared to the common two-step decision-making, is that the preparation of cases is inter-disciplinary, which implies, for example, that legal questions are actively discussed and adapted to the approaches of other disciplines. I expect that some of the readers will object and claim that this way of organizing system development will not work, partly because the specialized departments do not have the competence required, and partly because such an organization creates increased bureaucracy which may create rigidity with the danger of increased time-consumption for each system development project. I agree that these are possible effects of my proposal. However, in my opinion, public administration too often has neglected the special perspectives of their role as bodies of Civil Service with power to decide in cases concerning rights and obligations of great importance to those affected. In my view it is highly relevant to emphasize this special feature of public administration. One major objective should be processing single cases in accordance with the principle of rule of law. Another major objective should be that of modernizing and changing public administration in ways that make citizens retain Public Administration and Information Technology confidence in it. If these objectives are adopted, the consequences are - inevitably - less emphasis on the speed and costs of single, system development projects and more attention directed towards creating systems that contribute to legally correct results and to the securing of connections and consistency between computer systems and the relevant legal environment. However, assessed over the life cycle of a computer system, the investment in higher legal quality of system development work may pay. Such an economical effect could, for example, be due to cost savings because of improved legal documentation of the programme code, and thus easier and quicker system maintenance in connection to statutory amendments. It is also likely that an increased legal contribution may lead to a reduction of errors in the programme code, and thereby to a reduction of the number of questions and complaints concerning single case decisions. If so, both a decrease of the number of altered decisions in single cases, and fewer changes of mal-functions of the case- processing systems may yield cost reductions.
17. Interests and values - conflicts and mergings
In the introduction of this article I emphasized the multiple functions that case-processing systems have. Keeping of accounts and case records and producing statistics, are among functions which are often closely linked to the processing of single cases. Up to now I have only been occupied with challenges and problems within this last mentioned - and to a lawyer - the most important function. This engagement has its' background in the recognition of inadequate measures to safeguard the rights of citizens being subject to computerized case-processing. Above, I have illustrated various needs for improvements within the Norwegian public administration. These changes may
229 Compendium to the Erasmus Course be motivated on the basis of the principle of rule o f law. However, they may partly also be seen as reflections of data protection as a fundamental right. Case-processing routines regarding collection and in-put control of data, for example, may be seen as concerning data protection, since routines which are not satisfactory may lead to incorrect or inadequate personal data. Likewise, the case-processing system produces new pieces of personal information when calculations are executed. Incorrect computer programmes yielding incorrect results are thus threatening the citizens interests of data protection. Some of the challenges I have pointed out may even be regarded in the perspective of "service quality". The questions of a legal system's documentation and of system transparency may certainly be viewed in this way. It is, for example, almost impossible for officers to give accurate guidance to citizens concerning the content of automatic case-processing procedures if they do not have access to intelligible documentation of the legal based procedures. In my opinion, the most important thing is not how we argue in favour of appropriate measures to secure the rights and needs of citizens. Rule of law, data protection and service are terms of perspectives which overlap each other. Partly, they may thus all be applied to describe the same legal aspects. Measures directed towards reducing the amount of errors in legally significant parts of computer programmes, improve rule of law because the risk of erroneous decisions is reduced. At the same time, it improves the data protection, because the chances of producing incorrect personal data are lessened. A strengthening of the legal competence in the transformation work probably will give positive contributions in respect to all referred statements to the problem. In other words, it is not the terms we use that counts, but the factual effects of the measures we apply in order to safeguard the interests of citizens. Public Administration and Information Technology
Often, we think of citizens' legal interests in respect to public administration as conflicting with other interests linked to public administration, for example the governments' interest of cost- efficiency and political control. Surely, it is not difficult to see that an objective of political control may give grounds for collecting personal information as basis for statistics etc, to a degree exceeding what should be accepted if privacy and data protection are to be put in the forefront. Furthermore, it is easy to recognize that some measures established to secure the legal protection of citizens will constitute a cost for the public sector. On the other hand it is in my view probable that quite a few measures may both serve citizens' and governments' interests. Emphasis in this article has been placed in objectively working out legally correct computer programmes. The role of lawyers in the transformation from legal sources to programme code, was one of the measures I have argued in favour of. There may, of course, exist different views with regard to if this is an effective remedy or not to improve the legal quality of case- processing systems. However, under the condition that it is believed to have a positive effect, it should be recognized that such a measure is likely to have a positive effect on government interests. If faults with legal significance occur in a case- processing system, we can be quite sure that these mistakes will lead to a high number of incorrect decisions. Given a reasonable degree of legal protection, this will lead to a number of appeals and converted single decisions, maybe with consequences for several thousand previous incorrect decisions. It will also necessitate correction of the programme code in question, with eventual chain effects on other parts of the computer system. In other words, a badly executed legal work is an effective way to deposit bombs of administrative costs. Unless case-processing systems are held totally intransparent and/or appeals are denied, proper juridical craftsmanship may prevent considerable administrative costs. In a similar way, possible mergings of
231 Compendium to the Erasmus Course citizens' interests and the government's interests of developing statistics and keeping of correct accounts, should be notified. Correct data is a condition for the production of acceptable statistics and correct accounts, as well as for the safeguarding of data protection and of rule of law. These examples of possible associations between citizens' interests and interests of the government, may illustrate how it is possible that the security of an acceptable statistical material and the safeguarding of correct account keeping, also may serve the more ideal interests of citizens. This opportunity opens for investments in measures serving more than one objective. Routines for working out documentation standards for case- processing systems may show important coordination potentials: Experts of law are not the only group who wish to formulate requirements for a system documentation. Staff working with statistics need to have the various pieces of factual information defined and documented, and the audit needs to define requirements to system documentation in order to execute an effective revision. The situation is that lawyers, statisticians and auditors to a large extent have a common interest in securing a sufficient documentation concerning the same elements of the case-processing systems. This is not to say that one can expect a total overlap of interests. My point is that the costs, for example in relation to the establishment of measures safeguarding legal protection in connection with computerized case-processing, will not necessarily be as high as if citizens' interests were the only concern. Within the Norwegian public administration, special requirements for system documentation have been established on the basis of statistics and auditing, while no requirements have been made on the basis of legal protection of citizens. When this being the case, it is quite easy to see that a basis exists for creating a fundament for a relatively inexpensive safekeeping of citizens' interests. Public Administration and Information Technology
Documentation and system transparency is furthermore not only for the good of citizens, lawyers, statisticians and auditors. Maintenance of systems will often create problems because it is hard to survey all effects of changes in the computer programme resulting from amendments etc. A reasonable penetrable computer system, i.e. a sufficiently documented system, will in such cases constitute a condition for inexpensive system maintenance. There is in my view, at least within the Norwegian public administration, promising prospects of being able to coordinate and merge various important interests with respect to case- processing. This may open for an adoption and strengthening in the safeguarding of legal protection and other citizens' interests.
18. More effective measures to safeguard legal protection of citizens
In a traditional, manual case handling situation, the responsibility for quality of the decisions in single cases rest to a large degree on all the single officers in charge. In sectors like social security and inland revenue, this implies that each of several thousands of officers have great impact on the results in "their" cases. If measures to improve the quality of case processing and to secure correct results are to be established in such a manual work situation, steps must be taken - at least to a large extent - in relation to each and every one of these officers. Training, instruction and supervision are examples of possible and usual steps to secure the quality of case processing. Computerization of case processing implies that certain operations are moved from the desk of the many officers in charge and into the hands of a rather limited group of experts being responsible for system development and maintenance. Instead of going through the same arithmetic and logical operations over and over again on a manual basis, the operations
233 Compendium to the Erasmus Course are established "once and for all", or at least for the purpose of millions of automated recurrences. Thus, measures to improve the correctness and quality of the system development work of the small project group, can be much more effective than measures directed towards improving the work of the thousands. This should give room for spending considerable time and resources improving the quality of case-processing through the groups in charge of system development, in particular the fractions of such groups dealing with transformation and legal issues.
19. Conclusions
In the title of this article I ask if legal challenges in connection with case-processing systems in public administration should be regarded as "dirt" in the machinery of government. The question is posed because these challenges create problems which until now have not drawn much attention and thus, if taken seriously in the future, may create new disturbances in apparently well functioning case-processing systems. The answer to my somewhat rhetorical question is - of course - no. The first reason for this answer is that it should in principle not be difficult to create case-processing systems which are well integrated with legal requirements. The fulfillment of these requirements is thus possible with a well functioning administrative machinery. The second reason for my answer is connected to the discussion of relations between citizens' and other interests regarding computerized case-processing. I have tried to argue that the legal perspective may help to prevent problems within other functions of government administration. The objective of "legal quality" is closely connected to other quality goals of modern administration. Thus, the perspective of the law may show to be fruitful for other professional groups as well as for jurists. Public Administration and Information Technology
Organization of Systems Development - How Do We Secure Legally Correct Computer Programs ?
Dag Wiese Sc hart uni
1. Introduction One of the areas that will be given special attention in the new subject of "information technology and administrative systems", at the University of Oslo, concerns the inclusion and representation of legal rules in computer programs. Several aspects concerning this formalization process may be analysed and discussed. One very central aspect is the legal, which again may be divided into two categories, first, questions concerning the correctness of the representations, the methods of formalization etc and, secondly, questions concerning the legal regulation of the formalization process and of related activities to do with systems development. Today, I will take up the first-mentioned question concerning the correctness of representation of legal rules as computer programmes and how- organizational measures may contribute to this correctness. The first term used in this presentation which I will clarify, is "systems development". This expression is, of course, far too comprehensive for my purpose. I could have chosen to speak about the development of computer programs.
235 Compendium to the Erasmus Course
but even this expression fails to give a precise description of the fraction of systems development work I have in mind. My issue only implies discussion of questions connected to the transformation of legal sources to computer programs. If measured in number of lines of programme code, I suggest that I speak about the development of 5 - 10 % of the programme code. On the other hand, much of the information analysis work precedent to the programming would be linked up to the interpretation of relevant legal sources. For lack of a better term, and with this explanation in mind, I will continue to use the expression "systems development". When I talk about "systems development" I think of development during a life cycle, i.e. I am concerned not only with the first and major effort to establish a computer system, but also with the continuous maintenance of systems occasioned by amendments of legislation etc. In the following, I will use the term "transformation" to designate the formalization process in which legal sources are selected, interpreted and expressed as instructions written in a computer language. This transformation is required in the development of computer systems to support the processing of single cases. Here, I denote these systems "case-processing systems". Case-processing systems are common and important in these parts of public administration which have as their task the distribution of money in the welfare state. The tax administration and social insurance administration are central examples of services which make comprehensive use of such computer systems. The objective of produceing legally correct computer programs is a central point in my presentation. However, in this context time does not allow us to discuss in greate detail what "legally correct" should imply. For my purpose, a brief definition of legally correct computer program is sufficient. When I speak in the following about legally correct computer Public Administration and Information Technology programs, I refer to computer programs which are the result of a transformation process conducted in accordance with generally accepted legal methods. This definition is, in other words, linked to proccss rather than to results. Nevertheless, the correctness may to a large extent, be judged on the basis of the results of the computerized processing of single cases. In addition, I will highlight, in part, the relationship between computerized and manual elements of case-processing. In this perspective, the computer programs may give correct results in the sense that no misinterpretation is made in the process of transformation. Nevertheless, legal questions arise if the manual routines are not sufficiently integrated with the computerized routines, giving incorrect results. An example of this last category is the fact that the Directorate of Taxation, throughout approximately two decades, has omitted to use the results of the supplementary tax assessment as a basis for the calculation of pensionable income. The computer programs were sufficiently correct, but a gap existed between the computerized and the manual routines, which again lead to incorrect calculations affecting a large number of pensioners. The second central element in this presentation concerns questions of organization. In the following, I will not restrict myself to a sharp definition of what should be regarded as organizational questions. With my approach, there is no strict boundary between questions of organization and questions of work division. In this presentation, the solutions to legal problems partly lie in measures concerning organization and work division. Before I arrive at organizational issues, it is necessary to discuss problems which give rise to these solutions. First I will briefly explain the legal nature of some of the decisions arrived at in the course of systems development. The objective is partly to create a basis for a comparison between these decisions and other well-known and well-defined types of legal decisions. I have
237 Compendium to the Erasmus Course already commented on the question of what should be understood by "legally correct". The second topic I will address concerns categories of legal problems which should be met by organizational measures. On this basis, I concentrate on questions concerning organization and work division. First, I describe some common elements of today's organization of systems development and discuss particularly the role of lawyers employed in the relevant specialized departments. Finally, I will give a sketch of how parts of systems development work which influence the transformation of legal sources could be organized in order to meet legal problems in a better way.
2. A "New" Type of Legal Decision-Making Process In public administration law we are familiar with a distinction between the decisions of the administration in single cases and general decisions, such as regulations and general instructions. The first category of decisions is directed to specified individuals, while general decisions are directed towards a non-individualized circle of citizens. When legal sources, i.e. typically acts and regulations, are transformed into computer programmes, a fraction of the meaning of the legal sources is represented. Interpretation is carried out with the aim of covering the majority of relevant individual cases. The task is - so to speak - to produce a semi-manufactured procedure that could be applied in most single cases. The decisions, in the form of instructions to the computer, are non-individualized, similar to traditional general decisions. However, the decisions result in a limited number of procedural alternatives to cover most of the possible individual cases. The choice of alternatives, i.e. the individual route through the total computerized routine, is usually not controlled by the officer in charge when cases are processed. The relevant fractions of the legal sources are, in other words, made ?? Public Administration and Information Technology redundant in the context of single ease-handling. These distinctive features may make us regard the transformation of legal sources to computer programs as a new kind of legal decision, which may be categorized as quasi-individual. In single case-processing the interpretation is case-driven, i.e. it is each specific case which occasions the interpretive questions and answers. Problems of interpretation may, in other words, lie untouched and unsolved unless a case gives rise to the problem. In the course of transformation from legal sources to computer programmes, every question of interpretation of the source should, in principle, be posed and answered, i.e. without waiting for an individual case to bring up the problem. This distinctive mark of transformation work points to a need for a supplementary approach to the making of individual decisions. The case-driven approach should be supplemented by linguistic and logic analysis of the texts in question. If so, the traditional lawyer may only contribute with a part of what is necessary to reach the quasi-individual legal decision in form of a computer programme. As I will return to later, this must obviously have consequences for the manning, division of work and organization of systems development. To conclude this section, it should be recognized that decisions which are results of the transformation of legal sources may be described as quasi-individual legal decisions distinct from general legal decisions and legal decisions in single cases. Furthermore, these quasi-individual legal decisions require perspectives and abilities which are new to traditional legal decision-making. However, while individual and general decisions are regulated by provisions regulating the case-handling, the transformation is hardly recognized as a category of legal decision-making, and is not regulated by any special provisions, unlike the case for individual and general decisions.
239 Compendium to the Erasmus Course
3 LEGALLY CORRECT COMPUTER PROGRAMS - WHAT IS THE PROBLEM? Before describing a small selection of legal problems related to case-processing systems, I will underline that I do not regard legal problems related to computer-based case-processing systems as any worse than legal problems concerning manual decision making processes. The reason for my interest in computerized routines is that these give rise to "new" problems with serious implications for individuals. Furthermore, the question of future confidence in public administration in times of radically and rapidly changing public sector, is closely linked to questions concerning computerization. I will briefly address two types of legal problems linked to the quasi-individual decision making process of systems development. The selection is limited to problems that may give arguments for organizational measures. A complete list of problems may be found in my report nb. 1/89 of the Complex-series. Sensible measurements and average figures are, in my view, impossible to work out in order to state a figure that may quantify legal problems. In my view, it is sufficient to show that certain problems occur in administrative bodies, which make a large number of single case decisions every year. Now to the first problem. Let us regard the computer system as a building on a foundation wall. In this metaphor the foundation wall consists of legal sources. My view is that it is important to ensure that this wall does not suffer from unnecessary weaknesses in the form of uncertainties and errors; i.e. it must be important to secure that the legality of the computer system is not questioned. To abolish completely uncertainties linked to legislation is, of course, impossible. What I have in mind is to avoid obvious formal errors in decisions constituting the basis of the case-processing system when these are constructed. Here I will mention three types of Public Administration and Information Technology errors or unnecessary uncertainties linked to the legal basis of systems. In the Norwegian Administrative Procedure Act, a certain kind of general decision is defined - "forskrift", which may be translated as "regulation". This type of decision is often, but not always, the result of the execution of delegated legislative power. If a decision comes under the definition, certain procedures must be followed in accordance with the Act. One effect is that every such decision is published and consequently more easy to access than most other general decisions made by public administration. Moreover, the question of the decision being a "forskrift" or not, may have influence on the question of the legal power of the decision, for example, to what extent citizens may base rights on the decision. Generally, I regard it a task for public administration to clarify the relation between their general decisions and the definition of "forskrift". In my view, it is particularly important that this problem is handled when general decisions are included as a basis for computer programming. The systems developers should know whether they base their work on solid rock or on sand. The next problem I will mention which is linked to the legal sources constituting the basis of the system, concern incidents of wrong references from or to the provision in question, inadvertent repeal of subsections and other obvious mistakes connected to the text of the transformed legal source. Every such mistake should be detected and corrected during the transformation work. One example is regulations concerning calculation of fishermens' pensions which were given according to section 3-15, 2nd paragraph of the National Insurance Act. This paragraph is, however, repealed without a new legal basis being established. The last question concerning the legal sources is whether or not a sufficient legal basis exists for "negative" provisions
241 Compendium to the Erasmus Course laying burdens on citizens, restricting rights etc. In incidents of execution of delegated legislative power, the question may for example be whether or not the provisions of the Act delegate a power in accordance with the issued regulations. If there is doubt, steps should either be taken to change and reinforce the legal basis or to amend the regulations so that they clearly have a sufficient legislative basis. Summing up this first legal challenge, it is my view that systems development should be regarded as a step in a process of the development of rules and regulations. Transformation work should, in other words, be based on legal sources which do not suffer from obvious or unnecessary errors which again may lead to a questioning of the legality of the system. The second and most important issue I will highlight, concerns documentation of programme instructions with their legal basis, including creation of explicit links between legal sources and the computer system. Although the case-processing system is closely linked to the legal sources, one rarely sees systems designed and documented in a way which shows this link. Programme codes, documentation, user manuals and other relevant material very seldom contain information about the legal basis of the computer system. In-put values should, for example, be regarded as pointers both to the programme code and to the relevant legal source which defines the categories of case relevant facts and their allowed values. When these pointers are omitted the result is a division between the law and the computer system with legal content. An example is a field on a registration screen with the category "earned income", which is explained as "income" in the users' manual. This income information constitutes a reference to "salary the insured person has earned from work in private or public employment [...]". Three different terms are, in other words, used to represent the same type of information. The designations in the system were chosen differently and without Public Administration and Information Technology reference to the wording of the statutory text. To me, this is as unacceptable as issuing traditional general instructions without clear reference to the provisions of the act which they discuss. The computer system is nothing of its own and is not superior to the legal sources. This problem of links between computer programmes and their legal sources is really a special feature of the general problem of documentation of legally significant decisions and of coherence within the total administrative system. Insufficient documentation of legal sources clearly has serious implications, in particular for possibilities of legality control and of officers' possibilities to give guidance to citizens regarding the computerized handling of cases. Investigations I have made both showed systems with relatively good documentation - at least for purposes of systems maintenance - and systems with (almost) no documentation controlling hundreds of thousands of decisions in single cases every year. In view of my explanation of the transformation process as legal decision-making, one may speak of the latter type of systems as harbouring "secret" legal decisions. When I stress the need for a comprehensive and high quality systems documentation, I speak first and foremost speak about the documentation of those fractions of the systems constituting transformed legal sources, i.e. a limited part of the total programme code. Because I see this transformation as a type of legal decision-making, I talk about putting down legal decisions in writing in a language which is developed for people to read. At the same time, what finally ends up as documentation should, during the transformation work, function as specification at the stage of systems realization. It will often be difficult to satisfy both requirements - readability and specification function - if documentation merely is worked out on one level. In some incidents, documentation should therefore be worked out on two different levels of formalization, with a
243 Compendium to the Erasmus Course height of the steps between levels so little that no major doubt exists about the transition from one level to another. The most formalized level will be suitable for technical implementations of changes and maintenance, while the less formalized version will cover needs of the specialized departments of central administration and of counter officers in local administration. This less formalized version will have much of the same function, and may very well be integrated into traditional instructions concerning interpretation of legislation and other legal sources. Moreover, legal decisions in the course of systems development, should not only be made visible in the form of documentation, but, in addition, form the basis for amendments of the legal sources which are the subject for transformation. The idea is that instead of regarding legal decisions as part of systems development, we should regard systems development as part of the efforts to improve rules and regulations. Then transformation from legal sources to computer programmes is seen as an analysis illuminating errors, inaccuracies and relations not serviceable to the purpose of effective legal regulation. Here, I have described legal problems connected to systems development. This description leads us to questions of organization and division of work, and, in particular, to the role of lawyers in the central administration.
4. Lawyers in the Project Groups In the following I will present some characteristics of the organization and work division in selected systems development projects. The description is based on a detailed investigation of three projects in three different branches of Norwegian government administration. Here, I will concentrate on systems development within the sickness benefit scheme in the National Insurance Administration, and the development of a tax Public Administration and Information Technology deduction system in the Directorate of Taxation. The investigation of the National Insurance Administration does not include projects within TRESS-90. Both investigations were carried out in 1991/92, which implies that I do not base my description on systems development within the TRESS-project. In addition, I base my views on less detailed knowledge of three Norwegian and three foreign projects. Generally, it should be commented that there are relatively significant variations between the organization of the different projects. I will therefore limit myself to factors and descriptions that may be claimed valid for the Norwegian projects. I assume that no-one will be surprised to hear that project groups seem to be an usual choice in the organization of systems development. Project organization was recommended already in a white paper from 1968 concerning government computer policy for the 1970's. Today, the Directorate of Organization and Management generally regards project organization as suitable for many purposes, and the general picture is that project groups constitute an increasingly important way of organizing tasks which are to be carried out within a certain time and budget. All project groups I have investigated could be characterized by their inter-disciplinary composition of people, with participation from different departments of the institution in charge, local administration and often with people participating from external private consultant firms. However, the organizational context of the project groups seems to vary in that no standardized organization is built around these groups. One common feature is a relatively loose connection between the project groups and the specialized department having authority within the legal domain in question. The connections to computer departments were, in general, much stronger than to the specialized departments.
245 Compendium to the Erasmus Course
In the National Insurance Administration, local office experts on social insurance regulations held a very central position in loosely organized project groups. These experts' employment in the social insurance organization was partly linked to the Systems Department of the National Insurance Administration and partly to the local offices. In addition, people from private consultant firms participated, mainly as programmers. The group had no steering committee or similar superior group to follow and control their work, but reported to the relevant specialized department of the National Insurance Administration, i.e. the department in charge of the special benefit being subject to systems development and transformation. This contact consisted of introductory talks with the department and meetings which were called by the local office experts if they found it necessary to discuss with the specialist department interpretations concerning the transformation. In the project of the Directorate of Taxation, the group was manned with people from both the specialized department and the computer department, and with participation from a private consultant firm. The project group was subordinated to a responsible "project director" who reported to a board of directors which executed overall control of a scheme of computer projects. Two representatives from the project group regularly met with the "project director", but here discussion was limited to administrative matters. The board of directors primarily had the function of deciding the implementation of pilot- and main projects. The local office experts on legislation participating in the project of the National Insurance Administration, had their background in local office work. They had no legal education, but were highly competent within the field of the Social Insurance Act and attached regulations. The professional background of the project group representative from the Public Administration and Information Technology
specialized department of the Directorate of Taxes, was similar to the background of the local office experts of the National Insurance Administration, i.e. she was a specialist in the rules and regulations in question, but lacked formal legal education. In both examples, lawyers of the specialized department did not actively participate in the transformation of legal sources. Why? Explanations of the passivity of the lawyers must generally be based on the fact that this group of officers are almost solely employed in central administration. Their main tasks are the handling of appeal cases and preparation of rules and regulations, including the drawing up general instructions regarding interpretation of legislation. This mirrors the division of work between central and local administration implemented from the 1970's: central administration should mainly have functions such as policy-making, planning, control and instruction; the local administration, on the other hand, should be decentralized and execute processing of single cases, often within very wide fields of legislation. This work division has lead to heavy work loads for local administration and to a computerization first and foremost directed towards managing the local office situation. I am not saying that the process of computerization has not had any influence on central administration. Several central computer systems have been established and run during the last decade. Computerization of central administration has, however, first and foremost affected computer departments, while the majority of the central administration have had very little experience with case-processing systems. An effective participation in systems development work demands a certain minimum of knowledge of the functions of the case-processing systems and of relevant case-processing routines etc. Up until now, most lawyers have not had the experience that have made them sought after when project groups have been composed. The officers of local
247 Compendium to the Erasmus Course
administration have been in the opposite situation because they have possessed a combination of knowledge of rules and regulations, of the case-handling process where the computer system is to be introduced and they have had practical experience with the use of functioning computer systems. They have, therefore, been useful participants in the systems development projects. Lawyers' lack of experience with use and development of computer systems may also have prevented them from developing a legal-based view on the significance of transformation of legal sources to computer programs and - generally - of computer supported case-processing. What I am indicating is a possible lack of insight on the legal implications of computerized case-processing.
5. Why Give the Lawyers Influence? To create case-processing systems which may better contribute to legally correct results, and, in particular to solve the two types of legal problems I have mentioned, it is essential that lawyers are actively integrated in the transformation of legal sources. Here, I will point out two important reasons why this is necessary. Lawyers in central administration and the legal experts of local administration, both have knowledge of rules and regulations. Lawyers differ from the great majority of local experts in that they have education and experience in general legal method. Lawyers do not merely have knowledge of a sector of legislation, for example, social security or tax legislation, they also have knowledge of related legislation and of general legal principles. Legal education gives lawyers competence to have a wide and open in-put in decision-making processes, i.e. to interpret legislation in the light of a variety of legal sources. This competence constitutes one important reason Public Administration and Information Technology for employing lawyers in central administration and a reason to give them the task to decide appeals and give general instructions on the interpretation of legislation. I have previously concluded that the transformation of legal sources to computer programs should be regarded as quasi-individual legal decisions and something that systematically may be placed between the general and individual decisions of public administration. The fact that these quasi-individual decisions have a direct and decisive effect on the result of each case, brings about the question: why should we have a different quality standard for the legal work connected with these decisions than for other general decisions in public administration? Does the fact that instructions are issued in the form of instructions in a computer program imply that the legal problems in question should be less accurately handled than traditional instructions? In my opinion, the standard of legal investigations and discussion should be the same for - on the one hand - general instructions regarding interpretation of legislation and - on the other hand - transformation of legal sources resulting in instructions in computer programs. In almost every project I have investigated, representatives of computer departments have stated that individuals employed in these departments should be regarded as top experts on those parts of legislation which are implemented in the case-processing systems. In light of the previously presented picture of rather passive, specialized departments, I assume that this often is a correct description of the situation. My investigation gives an impression of an introduction of a schism between the legally orientated work of specialized departments and of the legal orientated work of the computer departments. Computer departments are, to a large extent, allowed to decide on the handling of legal questions regarding transformation of legal sources; i.e. first and foremost, questions concerning algorithms controlling calculations and the testing of
249 Compendium to the Erasmus Course conditions. The specialized departments, on the other hand, decide on what may be seen as "true" legal problems; i.e. problems of interpretation of vague and discretionary concepts such as "supporter" and "if reasonable". Such a division is, however, not acceptable because it implies a division between legal problems which are highly integrated with each other. According to the present computer policy in public administration, it is an objective to re-use information on individuals once it is collected and registered in machine-readable form. Such an "ecological" view of information will lead to an increased linking of registers. This again implies that definitions will be introduced to establish the necessary degree of equality between types of information in the linked computer systems. To the extent that these definitions are established at an administrative level, work on them will involve increasingly important types of legal considerations in the course of systems development. The situation maybe, for example, that the method of collecting a certain type of information is changed in order to cut down on expenses. In the course of systems development, it would have to be considered whether or not information, which already is collected in another body, could be used instead. This would involve, in turn, examining whether or not that other information has definitions such that it can be used for the purpose for which the former type of information was collected. These questions are of a legal nature. It will, therefore, be increasingly unsatisfactory not to involve lawyers and other experts from specialized departments in the work on systems development. I emphasize that my argumentation for the necessity of lawyers in parts of the work on systems development, does not imply that there is less need for local office experts. I regard local experts as especially important because of their knowledge of local interpretations and views on legal questions, their experience and knowledge of the practical functions of computer Public Administration and Information Technology systems in operation and their ability to pin-point problems regarding work environment. As I will return to soon; I prefer a solution where both lawyers from central administration and local office experts participate in the transformation work.
6. A Sketch Of An Organizational Structure I have discussed above the work division between lawyers and local experts and between the specialized departments and the computer department. My next step is to suggest some possible changes to organizational structures. My point of departure is systems development work organized in project groups. Since systems development consists of activities which require a number of skills, there is, in my view, no alternative to composing project groups along inter-disciplinary lines. On the other hand, there arc no definite answers to the question of how project groups may be integrated into the rest of the organization. Furthermore, there is no antagonism between project groups and an hierarchical organization. In an hierarchical organization, cases are often handled in two steps. The first step could be preparation of the case by an officer; the next step, evaluation and decision by an office manager. This structure has certain advantages that should be recognized. First, the second phase of a two-step case-handling may have the function of quality control. The preparation of the case provides a structure, formulates the problems and suggests solutions. On this basis, it may be for the office manager to concentrate on other aspects of the case; for example, to assess the coherence of the prepared text. Equally important is that the persons involved in an hierarichical case-handling, may have different views on the solutions to the problem in question. Thus, two-step case-handling may make the evaluation of a ease more
251 Compendium to the Erasmus Course comprehensive than without a second step. If the person in charge of the second step is an office manager with the responsibility for several problem areas, the second phase of the case-handling may also imply inter-specialist considerations which the specialized officer who carried out the preparatory work did not have the qualifications to perform. In other words, an hierarchical case-handling routine of legal questions has features which facilitate quality control of the transformation of legal sources and which make it possible to consider questions in the light of related legal issues. At the same time, it is important to recognize similar advantages linked to project groups. It is likely that discussions concerning legislation and its interpretation will be more thorough in groups where both lawyers and local office experts participate than if only one of these parties contribute. Furthermore, the participation of systems analysists and programmers may give rise to legal questions of which the legal experts would not think. Systems analysis may, for example, reveal logical inconsistencies in the model drawn up by legal experts. In Norwegian government administration, it may also be relevant to have participation from people dealing with accounting, auditing and statistics. These aspects may also give rise to legal questions. Systems development constitutes - as indicated - an inter-disciplinary task. In order to be able to cope with each of the activities in an adequate way, it is in my view necessary to regard each activity as part of closely related but separate decision-making processes. A legal decision-making process constitutes one important process when case-processing systems are developed. Today, this process is to a large extent not made explicit. The legal questions are first and foremost handled within project organizations without major interference from the hierarchical organization of specialized departments. Public Administration and Information Technology
My picture of an improved organizational solution for work on systems development consists of three levels, of which the two first is the most important. The project group, i.e. an inter-disciplinary and inter-departmental group constitutes the bottom level. Here, preparations are made and proposals for decisions worked out. The second level is that of specialized departments, typically the department in charge of the scheme in question (sickness benefit, VAT, etc.). Here, decisions are made if the question involved comes under the authority of the particular department. Amendments of proposals from the project group should be considered by the project group before a final decision is made. The total system solution is decided by the top management of the agency on the basis of recommendations from every involved department (3rd level). In this simple model, the project group exists without a superior project organization. Instead, the idea is to combine project groups with traditional hierarchical organization by regarding the group as the first level in a two-step case-handling, preparing decisions in different departments. One important difference compared to the common two-step decision-making, is that preparation of cases are inter-disciplinary, which implies, for example, that legal questions are actively discussed and adapted to the approaches of other disciplines.
7. Conclusion I expect that some of you will object and claim that this way of organizing systems development will not work, partly because the specialized departments do not have the competence required, and partly because such an organization creates increased bureaucracy which may create rigidity with danger of increased time-consumption for each systems development project. I agree that these are possible effects of my proposal. However, I think public administration too often has neglected
253 Compendium to the Erasmus Course the special perspectives of their role as bodies of Civil Service with power to decide in cases concerning rights and obligations of great importance to those affected. In my view it is highly relevant to highlight this special feature of public administration. One major objective should be that of processing single cases in accordance with the principle of rule of law ("rettssikkerhet"). Another major objective should be that of modernizing and changing public administration in ways which make citizens retain confidence in it. If these objectives are adopted, the consequences are - inevitably - less emphasis on the speed and costs of single, systems development projects and more attention directed towards creating systems which contribute to legally correct results and to the securing of connections and consistency between computer systems and the relevant legal environment. However, assessed over the life cycle of a computer system, the investment in higher legal quality of systems development work may pay, due primarily to easier maintenance in connection with amendments. Public Administration and Information Technology
Regulatory Co-Operation through computer assisted solutions*
Version 6 February 1994
By Jon Bing
1. Introduction: Computerised Information Systems for Regulatory Co-operation
This paper will discuss legal information systems as a possible tool for improving co-operation between regulatory managers. By "regulatory managers” are understood a range of persons in different roles. First, the legislators. not only the members of the parliament that enacts statutes, but also the government, or
* This report is contracted by the Organisation for Economic Co-operation and Management, Public Management and Service (HR/U3/9I/1545/AN/NMC/). Parts off this report have been developed in co-operation with Mark G Schoenberg of the Regulatory Information Service Center. Washington DC. United States of Amerika. The work has been orgnised at the Norwegian Research Center for Computers and Law under the projekt identification NORIS (96) 11-2.
255 Compendium to the Erasmus Course government agencies, which on the basis of authority derived from the constitution or from specific statutes, issue regulations.1 Second, the evaluators, those charged with evaluating whether a certain regulation is appropriate or optimal on its objective, changes in policies, society or technology, etc. Third, the reporters, who on the basis on a policy decision, develop a report containing a proposal for a new or amended regulation. Fourth, the reviewers, who evaluates the proposal about impact on society, and about administrative or economical consequences. Fifth, the drafters, who - perhaps on the basis of hearings or comments - complete the formal drafting of the proposal. This is a rather large group of persons, and one may also include those outside the government that take an interest in the regulations on a specific subject, for evaluation of current regulation and review of proposals - this may be organisations for trade, interest groups, political parties, lobbyists, etc. This group has one thing in common, that they are not working with regulations to solve legal problems - like a lawyer, a judge or a case handler in a public agency. Their interest is of a more general nature, they view regulations as tools for organising society, stimulate or restrain certain activities, solve conflicts, etc. In a sense they are social engineers.
1 In this paper, "regulations" will be used as a term that includes both acts of parliament and regulations by the government. Within jurisdictions, regulations may be further classified into different categories - like presidential decrees, guidelines, resolutions, etc. These vary, however, quite much, though the distinction between acts of parliament and regulations issued by the government seem to be universal. It will be sufficient in our context to use one term - regulations - though the context may require the term to be qualified. Public Administration and Information Technology
In our context, it should also be emphasised that this is a dynamic process, which is emphasised by the circle in the small illustration. This is not a process that ends once a regulation has been adopted - it only enters a new phase in its life cycle. Regulatory managers are contained within one jurisdiction. A jurisdiction is defined by the legal system. The primary point of departure is the sovereign state. This is defined by its own constitution, and recognised by public international law. But this state may itself have an internal, rather complex structure, and it will be part of other structures through international agreements. The result will be a rather complex structure of interlocking legal systems. Internally, the state typically is divided geographically into smaller jurisdictions - regions and municipalities. These smaller jurisdictions will have authority to issue regulations for certain matters - typically, a town may issue by-laws2 on its own affairs. There are also federal states,3 which are organised rather
- Actually, "by-law" literally means "the law of the town", the prefix "by" originating from the Norse word for town, still used in the Nordic languages, and part of the name of many British towns (for instance Grimsby). 3 In the small illustration following below, the "federation" is pictured as the top of an implied hierarchy of regulations within the nation. This is not Continues nest page
257 Compendium to the Erasmus Course differently (one may think of the constitutional basis for nations as different as the United States of America, United Kingdom,4 and Switzerland). Externally, the state will be party to a number of agreements. A definition of a sovereign state is that another state or organisation cannot impose legally binding regulation. The sovereign state can, however, freely make an agreement with one or several other states and in this way have obligations under international law. In a sense, it may be misleading to talk about the public international law as "one legal system". Though there are international rules that are commonly held to apply to all states, primarily of customary nature, the regulation applying to a certain state will depend on the agreements to which this state is party. The combinations of all these agreements constitute the international public law for that state, and this will probably - at least in detail - be different from the international public law affecting any other state. The relation between the international agreements, treaties or conventions in force and the national legal system has two major versions. For some states, the legal system includes these international legal instruments, and makes their rules directly applicable within the jurisdiction, for instance at the national courts (monistic principle). For others, an explicit transformation has to take place, typically by drafting a national regulation containing the substantive rules of the international legal instrument (dualistic principle). In the latter case, one will
Continues from last page
necessary so, as the federation can be created by states, and derive its authority from the states - rather than the states deriving their degree of autonomy from the federal constitution and law. ^ Which it may be contested properly can be classified as a federal state, though in our context it shares some of the same characteristics. Public Administration and Information Technology have to make a distinction between the rules of public international law, which apply to the state, and the national rules flowing from national regulations that are based on the international legal instruments. In this case, it is quite possible that there is a discrepancy between the two sets of rules; the state may have failed to comply with an obligation under international law and not have transformed the appropriate international legal instrument to national law, or an inappropriate interpretation may have occurred in the transformation process. This creates a rather complex situation, flowing from the fact that there does not exist a single, supranational regulatory authority. But in certain cases, nations have formed groups - unions5 - through treaties, which also have created supranational agencies which authority and decisions the sovereign state has agreed to follow. A very strong version of such a union is the European Communities, where the Community Court in Luxembourg will decide matters of Community law that the will be applied at national courts. The union resulting from the European Economic Space Agreement will be similar, though somewhat weaker. But there are more limited examples of the same nature, like the countries that have agreed to accept the dispute resolution mechanism of the International Court in the Hague, the decisions of the Human Rights Commission in Strasbourg, etc down to the rather weak forms of dispute resolution that is part of such organisations at the GATT.6
5 In this context, "union" is used as a technical term, not indicating the policy issue related to the reformation of the European Communities into a more integrated, political organisation similar to a federal state. It is quite common to name those countries parties to the same treaty for a "union" under that treaty, cf for instance the Berne Union based on the Berne Convention on Copyright. 6 The General Agreement on Trade and Tariffs, which often is said not to be an organisation.
259 Compendium to the Erasmus Course
This gives some indications of the complexities of interlocking jurisdictions, and the necessity for regulatory managers to co operate. There are at least three distinct reasons for managers to co-operate in the sense that they need information on the regulations existing on different levels in the hierarchy of regulations: (1) Consistency and compliance. Regulations within one jurisdiction should be consistent. This requires managers to be aware of regulations on higher levels of the hierarchy and on the same level. A regulation issued under the authority of a statute should obviously not be in conflict with the statute itself - or other statutes. And regulations issued by one agency should not be in conflict with those issued by other agencies. To ensure consistency and compliance with superior regulations. The regulatory managers must have access to the regulations. There should be systems set up to access the regulations in force (this is generally solved by national legal information services), and there should be routines for notify that reform is considered in certain areas to co-ordinate efforts at the earliest stage (this is Public Administration and Information Technology generally solved within a jurisdiction by a regulatory agenda or similar system). (2) Learning front the experience of others. There are several reasons for different regulatory managers being faced with the task of developing regulations of a similar nature. The most basic reason is that our societies - though contained in sovereign national states - nevertheless share many of the same characteristics, and develop under the influence of many of the same forces. Therefore regulatory managers in different countries may be faced with the task, for instance, of developing a regime for the protection of integrated circuits, or setting standards for traffic safety. Another reason may be that a decision is made on a higher level in the regulatory hierarchy, and this has to be implemented in parallel on the lower levels. For instance, a directive by the Council of the European Communities has to be implemented in the national legislation of the member countries, or the provision of a new treaty on reduction of the use of freon has to be implemented in the national environmental regulations - perhaps a large number of regulation governing different industries. In such cases, the regulatory manager may want to look at already existing regulations in other jurisdictions. The reason for this may be to facilitate developing and drafting the regulation, or to look at the experiences made in that jurisdiction. This may be combined by comparing the regulatory strategies of different jurisdictions in addressing the same issue for better being able to choose the strategy for one's own jurisdiction. Looked upon in this sense, the world becomes a laboratory of regulatory experiments yielding information and results for the benefit of regulatory managers. There are many obstacles for being able to benefit from this wealth of information: There is language barriers, there are differences in the basic principles on which jurisdictions rest,
261 Compendium to the Erasmus Course etc. But clearly, without sharing information, none of these benefits can be reaped. (3) Control. Finally, the issue of control also should be mentioned. This also has several aspects. The loyal regulatory manager has, as mentioned above at (1), a need for information to ensure that the regulations being developed comply with requirements originating from the higher levels in the hierarchy, and that they are consistent with regulations both on higher levels and the same level in the hierarchy. The counterpart to this need, is the regulatory manager on one level checking that the regulations on the lower levels actually comply with the requirements decided, and that there is consistency. This also is a need for regulatory managers in international organisations, who would like to keep track of the national implementation of the provisions of an international agreement by the members of a union. But there is also a need for regulatory managers in one jurisdiction checking that the regulations of another jurisdiction actually comply with the requirements. A nation will be concerned that a balance is kept in the international community. For instance, if actions should be taken to improve environmental control, these actions will have a cost. This may increase the price of the goods or services offered on the international market. The regulatory managers within one jurisdiction loyally implementing those measures which flow from an international agreement, will - quite reasonably - have an interest to check that the other members of the union likewise implement the provisions agreed upon. (4) International co-operation and trade. It may be argued that a fourth aspect ought to be identified. Regulatory managers are to implement policies which in many cases emphasise international co-operation and trade. To develop national regulation that promotes such policies, it would be desirable to Public Administration and Information Technology
have knowledge of the national regulations of trade partners and others. This is, however, an interest closely related to that of compliance and consistence. This aspect will therefore not be singled out for discussion in this report. But it is rather obvious that for government officials in charge of developing policies, or for lawyers advising clients, the main interest in foreign national regulation will be that they are part of the legal framework in which international trade in goods and services take place, and which those engaged in these activities need to have information to comply with, or take advantage of the law of the other country.
In this paper, we will briefly discuss some ways in which ways computerised information systems may help to satisfy the information needs of regulatory mangers, and provide a basis for co-operation, with an emphasis on the relation between a national and the intra- or supranational obligations of this nation. This will be a discussion of the relation between to levels in the hierarchy, and the discussion may be analogous to the relation between hierarchical levels within a jurisdiction. However, within a jurisdiction co-ordiantion is always possible through the government, and therefore the solution may be more simple than those which are in the focus for the discussion in this paper. One will note that the two perspectives - access to international regulations for national managers, and access to the national implementation of such regulations for international managers - are but two sides of the same coin. In the following we will exemplify the two main perspectives. After this exemplification, we will mention two areas of law where international co-operation would seem to be rather strong.
263 Compendium to the Erasmus Course
2. National Access to International Regulations
2.1. Introduction: Early initiatives
There are special problems with the documentation of international agreements. For instance, the agreement may have several versions, and a state may not be party to the treaty in its latest revision, while other states are. Keeping track of the obligations regarding a certain other state may therefore create a problem. Also the agreements often have an authentic language different from the official language of the nation. In this case, the agreements may be documented both in the official version and in a translation - creating a situation of multi-lingualism much similar to that discussed below. The complexity of the problem of determining which agreements are in force between a state and any other state has been attractive to those concerned with legal information services and played a rather major role in the late 1960s as a reason for exploring computerised systems.
The major example may be Hugh Lawford's initiative at Queen's University, Kingston, Canada. Since 1961, the university has been engaged in a Treaty Project, collecting and annotating the treaties of the British Commonwealth. In 1967 word processing was introduced. The Treaty Project was used in preparing the treaties of a number of developing countries. In 1968 the Queen's University Institute for Computers and Law (QUIC/LAW) was funded based on this project. The initiative is basic to what is today known as QL Systems Ltd, a computer and communication service primarily offering legal information services for the whole of Canada. The initial relation to the Treaty Project has not, however, led to an emphasis of international legal instruments. Public Administration and Information Technology
Also in 1968, the Committee of Experts on the Publication of National State Practices in the Field of Public International Law at the Council of Europe recommended to the European Committee on Legal Co-Operation (CJJ) that another committee of experts should be appointed to study "the question of harmonisation of technical means of programming international treaties into computers", which led to the establishment in 1969 of the committee which today is known as the Committee on Legal Data Processing in Europe(CI-IJ). Again, in spite of the initial relation to treaties, the activities of the committee have not really been concerned with the special challenges of international legal instruments, but rather more general issues. It led to the computerisation of the Council of Europe Conventions (which are made available to interested member countries in computerised form), and to an interest in treaties among some of the member countries represented at the Committee. A prime example was the establishment of the now discontinued system IBERTRAT (1972) by the Spanish Ministry of Foreign affairs.7 Today, treaties are data bases within a large number of national legal information services.8
^ These examples are from Jon Bing et al Handbook for Legal Information Retrieval, North-Holland, Amsterdam 1984. 8 A rather special example was the Bulgarian service offered as part of the NORMA system. This is developed as a stand-alone system for PRAVETZ-16 computers under MS-DOS. The data base is developed on the basis of the work of Committee for Mutual Economic Assistance, established in 1987 for multi- and bilateral agreements between socialist countries as a preparation for economic integration and joint ventures with Western countries. Documents from Bulgaria, Hungary, Poland, Romania, the Soviet Union, and Czechoslovakia were included with bilateral agreements with Cuba, Mongolia, Vietnam, etc. The data base was distributed to subscribers every three month as diskettes, key words in the original language, Russian and English. In 1988 the work of translating all Continues next page
265 Compendium to the Erasmus Course
Some international organisations, such as the Council of Europe, have computerised the international agreements for which they are responsible. Generally, the volume is quite limited, and the organisation will not organise any computerised service for outside users, though the treaties may be available to interested services. It is also be of interest to determine which other countries are party to a certain treaty. This is a task that is hardly appropriate for any national legal information service. Traditionally it is solved by the treaty naming a depository that is charged with the task of keeping track of which countries are parties to the treaty. Such a depository may be an international organisation, and this may again use a computerised system. These are, however, only partial solutions. A more general solution may emerge out of the United Nations register of treaties. At the 28th session of the General Assembly (1973) a proposal for the establishment of a United Nation Treaty System was adopted.
Continues from last page
texts to Russian was begun. The example is from Alexander Manov "Computer Technology and Legal Information Processing", 6th Student Pugwash USA International Conference, University of Colorado at Boulder, 1990. The NORMA system is still offered to the market in Bulgaria, but obviously this international part of the service now only has historical interest - though the example still is valid as an illustration of a straightforward application of information technology to the problem of distributing information on international agreements. I Public Administration and Information Technology
2.2. Access to supra-national regulation: The European Communities and their CELEX service
There is one major exception to the lack of organised access to international agreements, etc. The European Communities took the initiative to create a legal information service in 1967, which first became operational in 1970, known as the CELEX (from Communitatis Europaeæ Lex). This is today a major service, covering all aspects of Community law.
A few examples will be given in the following, where appropriate. The CELEX service is available on-line to subscribers and is a rather conventional service based on text retrieval.9 The data bases are in English and French, and bases for the other languages of the Community are in preparation.
CELEX offers its data base to other services and is quite happy to have such subcontractors distribute their material through their own networks. The British Context service has produced a version of CELEX in the form of CD-ROM. Several national information services10 have had the whole CELEX data base transferred and offer this with the national material and under its own software.11 Obviously the CELEX service is a prime example of the value which regulatory managers will find in having on-line access to the directives and other instruments to which their national regulations have to conform.
^ The Communities are, however, in a process of evaluating the service, and it may be re-design as a result. '0 At least the German JURIS system and the Norwegian Lawdata service. I * The examples of CELEX documents given below have been downloaded from the data bases of Lawdata.
267 Compendium to (he Erasmus Course
The CELEX system not only documents community law as such, but also the international agreements to which the community is a party. The CELEX sector 1 documents treaties, and sector 2 international agreements. An example of an entry in the sector 2 data base follows:
Example of document o f international agreement______DOC.NUM: 281 A0122(01) TITLE: COMMUNITY-COST CONCENTRATION AGREEMENT ON A CONCERTED ACTION PROJECT IN THE FIELD OF TELEINFORMATICS (COST PROJECT 11 BIS) PUB.REF: OFFICIAL JOURNAL NO. L 350, 23/12/80 P. 0046 GREEK SPECIAL EDITION . CHAPTER: 16 VOLUME 02 P. 46 AUTHOR: EUROPEAN ECONOMIC COMMUNITY; FINLAND; SWEDEN; FORM: AGREEMENT TREATY: EUROPEAN ECONOMIC COMMUNITY; DATES: OF DOCUMENT ..04/12/1980 OF END OF VALIDITY: 11/09/1983; PUB: 1980/12/23 DOC: 1980/12/04; LEG.CIT: 157E...... 379D0783 ...... 380 D 0317...... TEXT: ++++ The text follows, but is excluded here _____
Most international agreements do not have associated case law on an international level. The exceptions are few, but one such I
Public Administration and Information Technology
exception is the Commission and Court of Justice created by the European Convention of Human Rights.12 The Council of Europe has converted the decisions of the Commission and Court to computerised form, and it is expected that the data base will be made available on-line to outside users. The value of such a system for the co-operation of regulatory managers become especially evident if a country joins the Communities at a later stage. In 1981 Portugal requested entrance to the Communities. At the same time, the Gabinete de Documenta9ao e Direito Comparado da Procuradoria-Geral da Republica started its operations. The Gabinete was linked to CELEX in 1984 and served as a consultant for the public administration, working for the Ministry of Justice, the Law Reform Commission, the Prime Minister's Office, the President, and the Parliament. Today, the Gabinete is specialised in comparative law and use of a variety of foreign information services, which are consulted as part of the national regulatory management. A similar situation exists today for the countries that have negotiated the European Economic Space Agreement13 that will, when (and if) the agreement takes effect, require that national legislation comply with the Community requirements. Obviously the access to CELEX on-line, by purchase of the data base for integration in a national service, or by the available CD-ROM is very useful. However, a special solution to facilitate the work facing national regulatory managers has been implemented in Norway. Here the CELEX data base has been imported into the national system, Lawdata. There is a secretariat within central
Another is, of course, the European Communities, which has a Community court in Luxembourg, and whose case law is documented through the CELEX system, sector 6. Austria, Finland, Sweden, and Norway.
269 I Compendium to the Ems mus Course government responsible for national co-ordination of the amendments that have to be made to national regulations. This secretariat had developed a number of notes in which the necessary amendments are discussed. These notes are themselves documented as a data base in the Lawdata system, and citations of both national and community regulations are activated as hyperlinks. In this way, a regulatory manager may read the note, and consider the changes to be made. He or she may be a simple keystroke jump into the national regulation under discussion, then back to the note, and onwards to the cited Community regulation.
Despite considerable interest in the 1960s and 1970s, information technology has not yet delivered on the hopes that it would provide easier access to international treaties. Yet, given the proliferation of international treaties and agreements since the 1980s and advances in information technology, its potential for bringing order, openness, and accessability to the international legal system is greater than ever.
3. Access to Information on National Implementation
Obviously it may be of interest for a regulatory manager in one country to study how another country has implemented a certain I Public Administration and Information Technology
international agreement. There may be several reasons for such interest. It may be useful to have models for drafting new, national regulations, or it may be of interest to explore how a certain other country interprets an international legal instrument, perhaps to determine whether that country complies with the international rules. But this interest is shared by regulatory managers in international organisations, who would like to keep track of the national regulations implementing the agreements for which their organisations are responsible, as well as the national decisions based on such regulations. Traditionally this type of information has been collected and distributed through specialised journals such as the journal Copyright published by the World Intellectual Property Organisation, which is responsible for the Berne Convention on Copyright. In this journal, relevant developments within member countries are discussed, and through comparative studies, national differences are analysed. There may also be academic research centres for certain areas of law that attempt to keep track of developments in many countries and which may be publishing digests, analyses, encyclopaedias, etc. In this paper, we will not be further concerned with these conventional efforts, though we should note that they are still the main tools for serving those interested in the national implementation. Some of the international organisations have supplemented their traditional systems with computerised facilities. For instance, the World Health Organisation at their regional office in Europe started in 1983 the establishment of a data base containing records of national regulations implementing WHO directives or recommendations. Again, the European Communities emerge as the major example of an international organisation keeping track of national implementation through the computerised CELEX system, which in its sector 7 documents national
271 Compendium to the Erasmus Course implementation. An example follows. As will be evident from the example, only bibliographical data on the national legislation is recorded, and the national title of the legal instrument is not retained.
Example o f document on national implementation______DOC.NUM: 761I1513GR TITLE: GREEK PROVISIONS RELATING TO: COUNCIL DIRECTIVE ON ADMINISTRATIVE PRACTICES AND PROCEDURES CONCERNING SETTLEMENT, EMPLOYMENT AND RESIDENCE IN A MEMBER STATE OF THE COMMUNITY OF WORKERS AND THEIR FAMILIES FROM ANOTHER MEMBER STATE OFFICIAL JOURNAL NO P 080, 13/12/1961 PAGE 1513 AUTHOR: GREECE FORM: NATIONAL IMPLEMENTATION MEASURE TREATY: EUROPEAN ECONOMICCOMMUNITY; TYP.DOC: 7; NATIONAL MEASURES FOR IMPLEMENTING DIRECTIVES: 1961; L; GREECE SUB: FREE MOVEMENT OF WORKERS; FREEDOM OF ESTABLISHMENT AND SERVICES; REGISTER: 05100000; ______
So far, only national regulations are included, but it is planned that the CELEX sector 8 data base will contain national case law relating to the Community regulations.
4. Examples of computerised systems for three areas
Above, we have looked at computerised information systems from two perspectives: The perspective of the national regulatory manager who needs access to international regulation to ensure compliance and consistency when developing the national regulation. And the perspective of the international regulatory manager who needs to track national implementation of international agreements. At the same time this will serve the national regulatory manager who needs access to the national Public Administration and Information Technology
regulation to ensure that the international obligations of that other country have been properly implemented. And they also serve - within the area documented - the national regulatory manager who seeks to learn from the experiences of others, though this is limited to the drafting of the text. We see that the different ways of co-operation between regulatory managers all are facilitated to a certain degree by these international information systems. There may therefore be something to be learned by briefly look at the information systems created within three areas where co-operation has been perceived as quite important - health law, environmental law, and labour law.
4.1. Health Legislation
The Regional Office for Europe14 of the World Health Organisation launched in the early 1980s a program for health legislation. On this basis, a system was set up which included indexed entries of the health legislation of member countries and a specially developed form document.15 It was originally set up in co-operation with the Uppsala University (Sweden), using a rather powerful data base system.16 In our context, the more interesting aspect of this project is the arguments put forward by the WHO for stronger co-operation between national regulatory managers. At the 13th World Health Assembly as long ago as 1977, the organisation expressed a concern for national health legislation, emphasising that legislation itself was a strategy towards protecting and
This office is situated in Copenhagen (8, Scherfigsvej, DK-2100 Copenhagen). * 5 "Euro-Health Legislation: Notification of new legislation", printed as Annex IV to ICP/HLE 002, WHO, Copenhagen 1982 16 The system MIMER-IR, this was also developed by UDAC at Uppsala University.
273 Compendium to the Erasmus Course improving the health of the individual and of the community.17 In 1981, the regional office established an advisory committee for Europe, which was to provide guidance "on the major direction of development of the health legislation in the European Region".18 In a note to the committee, several issues are addressed. As problem 4, the need for co-operation between regulatory managers is addressed:
"... increased need of access at national level to international exchange of information on health legislation.
Situation analysis
The solidarity and interdependence of European countries create a current need to obtain rapidly form other countries available information on health legislation for comparison and decision-making."19
The emphasis on communicating information on health legislation also lead to a certain re-defining of the role of the International digest o f health legislation, which is a conventional journal, but an obvious response for the increased need for co-operation between regulatory managers in health law on an international scale.
Cf resolution WHA30.44. * ^ C f Health Legislation: European Programme, ICP/HLE 002, WHO, Copenhagen 1982:3. G Pinet "Health Legislation - Medium-term Programme - Outline", Annex III to ICP/HLE 002, WHO, Copenhagen 1982. Public Administration and Information Technology
4.2. Environmental Law
Environmental regulation is by its nature international, as waste is carried by a river through the territories of several countries, and fumes from factories are borne by the wind across boundaries. There are therefore obvious reasons for taking an interest in establishing some sort of international legal information service to make it possible to form a coherent and comprehensive understanding of the law in force on an international scale.
4.2.1. The United Nations Environment Programme
The major international organisation working for better regulatory co-operation is the United Nation. Its Environmental Programme (UNEP) has a mandate from the Governing Council20
"To collect and disseminate information on national environmental legislation and maintain a register of International Treaties and Other Agreements in the Field of the Environment; to strengthen and co-ordinate the use of existing information sources and databases".
The long-term goal is to establish an "operational comprehensive database on national and international environmental law". Today, UNEP systematically collects information from countries, including on legislation, for its Country Fact Sheets' database and the Environmental Law and Institutions, Programme Activity Centre (ELI/PAC) in Nairobi21 maintains country files for those countries that have been
20 GC 16/25. 21 PO Box 30552, NAIROBI, Kenya.
275 Compendium lo the Erasmus Course assisted through its technical assistance programme. ELI/PAC also publishes the Register of International Treaties and Other Agreements22 and has published two volumes of selected multilateral treaties. It has been reported23 that ELI/PAC is currently reviewing its policy, strategy and future work programme concerning a data base on environmental law, moreover considering possible co operative partners.24 The UNEP's Regional Office for Latin America and the Caribbean (ROLAC)25 has developed a regional data base containing considerable information on environmental law and institutions in the region. This data base also includes information on environmental conventions that the countries in the region have ratified.
4.2.2. The Environmental Law Information System (ELIS)
Though the United Nations has taken an initiative for regulatory co-operation, within environmental law there are also two other major initiatives. The International Union for Conservation of Nature and Natural Resources - IUCN - is an organisation with consultative status with the United Nations, and whose members include 60 sovereign states and some 560 governmental and non-governmental organisations from approximately 120
“ Grotius Publications Ltd, Cambridge, United Kingdom. 23 Letter of March 29, 1993, (ref LK/KH/290393) from Sun Lin, director for the ELI/PAC as a response made to a request based on the work basic to this report. 24 Confirmed in letter of April 23, 1993 (ref SO/vm) from Sun Lin, director for the ELI/PAC. Boulevard de los Virreyes No 155, Lomas Virreyes, 1000 MEXICO DF, Mexico. Public Administration and Information Technology countries. The IUCN activities are planned and co-ordinated by a Secretariat and by commissions for ecology, species of plants and animals, protected areas, etc. In 1968, the IUCN Commission on Legislation began to investigate computerised information services.26 In March 1972, the then new IBM retrieval program STAIRS (Storage and Information retrieval System) was made the basis for the system, and this version was demonstrated in June 1972 at the United Nation Stockholm Conference and the Second International Parliamentary Conference on the Environment. The current system is based on a customised software known as ROMULUS (Retrieval Oriented Multilingual Updating System). The system is today operated by the IUCN Environmental Law Centre (ELC), which is part of the IUCN secretariat in Bonn.27 It co-operates closely with the IUCN Commissions, particularly the Commission on Environmental Policy Law and Administration (CEPLA). The system is run in co-operation with the International Council of Environmental Law (ICEL.), which is an international non-governmental organisation with individual and organisation as members, approximately 290 members from 60 nations. Its sole purpose is the promotion of contacts and exchange of information. ICEL shares facilities with the ELC. The CEPLA and the ELC maintain a collection of legal provisions relating to environmental issues in different countries, as well as international instruments. Currently, regulatory information of more than 150 different states has been documented, with addition of bi-lateral and multi-lateral agreements and binding international legal instruments. The
-6 Cf Wolfgang E Burhenne "An Information System for Lawyers"; Law and Computer Technology, August 1968:17.
27 214 Adenauerallee, 53 BONN, Federal Republic of Germany.
277 Compendium to the Erasmus Course collection contains currently some 32,000 documents. The annual increase is estimated to 1,500 documents. ICEL maintains a collection of literature relating to environmental policy issues, law and administration, (approximatly 43,000 units with an annual increase of 2,000-2.500 units. A selection of court cases is also maintained, emphasising the Federal Republic of Germany, the United States, and France. The majority of the cases (approximately 2,300) are German. The combined documentary resources of the two organisations are the basis of the joint information service, ELIS, which is organised in eight data bases: national legislation, international legal instruments, treaties, documents from the European Community,28 court decisions, literature, fauna species and flora species.
National legislation 32,156 documents Treaties 1,035 documents EEC 527 documents Literature 43,464 documents Court decisions 2,318 documents Flora species 3,680 documents Fauna species 11,775 documents Total 94,955 documents
Abstracts are only prepared for special projects. One of these is ENLEX (see below), the others are a project to index species mentioned in national regulations, protected areas of the Mediterranean and wetland legislation. Abstracts are developed mainly for helping the user to determine the relevance of a
28 These seem to be termed "soft laws" in the terminology of the system. Public Administration and Information Technology document. Relationships are specified, especially vertical relationships as between a statute and its subsidiary regulation. This system is clearly more comprehensive than the one on health law. Compared to CELEX, one may also perceive differences. It may be fair to say that ELIS have been developed not only as a legal information system, supporting decisions in specific cases, but more as a policy information system. Its inclusion of case law and literature within its domain will make it possible to assess and compare different regulatory strategies. It would seem that the system can be an indication of what can be achieved by conventional computerised system in terms of regulatory co-operation.
4.2.3. The EN LEX system of/TALGIURE
In Paris 1972, the European Communities at a summit meeting decided that it would be important to emphasise the work for the protection of the environment.29 A decision by the European Communities was made in 1977 to establish an information service for environment law.30 This has as its objective to supply bibliographical material on all legal sources relating to the protection of the environment. The objective is to furnish small and medium-sized businesses with a comprehensive legal information to facilitate their operation. In December 1981, the Commission gave financial support to two organisations to
Cf Enzo Meriggiola "Interdisciplinary research in legal data bases, in particular concerning environmental law data bases", Proceedings of the 9th Colloquy on Legal Data Processing in Europe, Council of Europe Press, Strasbourg 1992:25-43, and Amedeo Postiglione "National report for Italy", Proceedings o f the 9th Colloquy on Legal Data Processing in Europe, Council of Europe Press, Strasbourg 1992:121-131.
30 C f Official Journal series C 139, June 13, 1977.
279 Compendium to the Erasmus Course develop the system, which was named ENLEX (ENvironment LEX).31 The IUCN should develop a data base for legislation and literature. This is done within the framework of ELIS (see above). Of the approximately 95,000 documents of the ENLEX data base, approximately 9,000 has been made subject to the special analysis required for ENLEX. The Centro Elettronico di Documentazione (CED) of the Corte Suprema di Cassazione has developed a data base of court decisions. Though the origin of the ENLEX project is closely associated to the European Communities, the project must be regarded as Italian, and currently (October 1991) it is operated as a part of the ITALGIURE system without any financial support from the European Communities. The work is carried out within the framework of the national legal information service of Italy, ITALGIURE. The CED has established co-operation with experts in other countries to ensure the correct translations of abstracts of decisions. It is estimated that the cost of preparing one document is approximately 100,000 Italian lire.32 Currently, 12,262 documents (June 1, 1991) have been prepared which give a coherent view of the jurisprudence and doctrine of the law of the European communities. But the system is still incomplete, the areas of energy, flora, fauna, pollution of the sea, etc have not yet been documented. The systems currently in an experimental phase, and it is expected soon to have mastered the last difficulties for a unification of the German and Italian material.
3 * Cf Amedeo Postiglione "National report for Italy", Proceedings o f the 9th Colloquy on Legal Data Processing in Europe, Council of Europe Press, Strasbourg 1992:121-131 The project was also known as "Project 80", referring to the date of the important decision by the Communities.
32 Approximately 100 ECU. Public Administration and Information Technology
The material originating in Bonn is as an experiment made available under the ITALGIURE system as a separate data base, ROMSL. This includes currently 6,290 documents (June 1, 1991). The unification makes it possible to inspect country by country the legislation and the jurisprudence in an English version. Also the original language of the regulation or statute is available, and the two versions can be inspected and compared.33
4.2.4. The OECD System on Chemical Hazard Laws and Regulations
The OECD maintains a data base on regulations related to chimical hazards within its Environment Directorate.
4.3. Labour Law
At the International Labour Organisation34 the Labour Law Information Branch follows national developments in labour, social security and related human rights legislation. This implies that it receives a large number of publications, consults national data bases, and has access to the information collected by the ILO regional offices and national correspondents. In addition, it maintains contacts with public agencies and research institutions. From this material, a team of lawyers selects the documents that are to be included in the computerised service NATLEX.35
33 Belgium is given special consideration, as both the equally authentic French and Flamish languages of the original regulations are available.
34 Routes des Morillons 4, CH-1211 Genève 22, Switzerland. 35 The system is based on a Hewlett-Packard 3000 computer using the M1N1S1S, a data base management system designed by the International Development Research Centre, Canada. NATLEX is accessible on-line through the International Labour Information System, Referral System, Continues next page Compendium to the Erasmus Course
Each legislative text is represented by a record containing specific fields. Documents in more than 40 languages are included, and the analytical summaries for the indexes are carried out in the three working languages of ILO (English, French, and Spanish). More than 26,000 records are now available, with an annual growth of approximately 3,000 records. Approximatly one third of the records relates to social security legislation, and more than 500 legislative records deals with migrant workers. The NATLEX is part of LABORLEX, which is maintained by the Labour Law Information Branch (INFLEG) of the ILO International Labour Standard Department36. This also includes the data base ILOLEX, which contains information on international labour standards, including the International Labour Conventions and Recommendations, the Reports of the sypervisory bodies on the application of standards, and the ratification of member countries. The tri-lingual data base of ILOLEX is also available as a CD- ROM,37 and by the way of illustration, the content is listed below:
Continues from last page
which include a program for user support for those not too familiar with the MINISIS search language. This Branch also publishes the Labour Law Documents three times annually, and the bulletin Legislative Information on a monthly basis. ^ Published by Martinus Nijhoff Publishers, PO Box 163, NL-3300 AD Dordrecht, Holland (US$850). Public Administration and Information Technology
• ILO Conventions • ILO Recommendations • Triennial Report of the Committee on Freedom of Association (1985) • Comments of the Committee of Experts on the Application of Conventions and Recommendations (1987) • Annual Report of the Conference Committee on the Application of Standards (1987) • Reports of Committees and Commissions established under art 24 and 26 under the ILO Constitution to investigate representations and complaints • Ratification lists by Convention and by country • ILO Constitution
This example illustrates a computerised service more clearly used to monitor national implementation. Obviously, with respect to labour law, not only national regulatory managers in the narrow sense take an interest in this, but also those interest organisations in private sector which in many countries have a strong policy interaction with the government. With the many controversial policy issues in the area, it is easy to imagine that the ILO systems contribute to regulatory co-operation: Comparison of regulatory strategies, monitoring implementation of international obligations in other countries, etc.
4.4. Other examples
Above, examples have been selected to illustrate some of the reasons for using computerised systems to promote regulatory co-operation. The three areas are all of an international character, and the need for international co-operation is certainly a major characteristic of them.
283 Compendium to the Erasmus Course
But there are many other possible examples. No survey in existence lists the computerised legal information services of international organisations. And - as mentioned in the introduction to this chapter - no supranational organisation has emerged which offers a truly international service. In preparing this report, an effort has been made to collect information on activities in international organisations, and several other examples than those mentioned above, are known. The OECD itself maintains an information system under the regulations on genetic modified organisms that monitor those released into the environment in member countries. The European Patent Office maintains an information system on decisions of its Board of Appeal,38 Convention, Treaties and Guidelines,39 and forms.40 We are sure there are numerous such examples, and a type of compilation describing them would be welcome. But this is not the purpose of this paper, though it is with some regret we limit the discussion to those examples offered.
5. Coping with multi-Iinguism
5.1. Introduction
There are a surprisingly large number of states that have more than one official language. Among them are:41
38 All decisions from 1980. European Patent Convention, Patent Cooperating Treaty, Harmonisation Treaty, and Guidelines for Examination. Standard EPO forms that can be displayed, but not printed. 4 1 Before the unrest in the summer of 1991, the former federal republic of Yugoslavia had six official languages, one for each of the states, though three are quite closely related, there were also three alphabets for these languages, two of which were Cyrillic Public Administration and Information Technology
• Belgium Dutch, French • Canada English, French • Cyprus Greek, Turkish • Finland Finnish, Swedish • Ireland English, Gaelic • Malta English, Maltese • Switzerland French, German, Italian
The European Communities does, of course, have as many official languages as there are official languages in their member countries - at the moment 9 languages (Danish, Dutch, English, French, German, Greek, Italian, Portuguese, Spanish).42 In some states, there are also distinct versions of the same language in use at the same time, for instance Greece (Demotic, Katharevousa) and Norway (Bokmcil, Nynorsk). In addition, minority languages are recognised, spoken and used, for instance, in court proceedings in a large number of countries.43
5.2. Drafting.
In jurisdictions with more than one official language, drafting becomes even more complex. One has looked to computerised systems for assistance. The dream of machine translation is yet to be realised by sophisticated knowledge-based methods in a possible future. But different methods for computer-assisted translation (CAT) are already in use.
42 The Gaelic of Ireland is not an official language of the Communities, neither is the Luxembourgois of Luxembourg. 4-* Again, the most complex situation may have been the former federal republic of Yugoslavia, where as much as 14 different languages are recognised for use in the courts. ■
Compendium to the Er
Bi-lingualism is % and the government has had experience with different ->uter-assisted methods since 1977. In the early 1980s the Tn slation Bureau of the Department of the Secretary of State s it up a project to determine whether current technology meets the demand for translations. Test sites have been set up in different translation sections, which have available the Logos translation software, specialised tools including LOTUS 1-2-3 for data collection, WordPerfect for checking of spelling and for word processing, Kurzweil for entry of texts on machine-readable media, and Keyword and Pride Local for converting word processing documents. The work is divided in four main steps: (1) Pre-editing: The translator eliminates any difficulties in the text that the system will not be able to handle; (2) Terminology research: The system produces a list of new words not found in its automated dictionary. When the translator has manually found equivalents, they are added to the terminology data base, coded by subject matter, and the necessary semantic rules are added to the data base; (3) Translation: This is completely automatic, and results in the form of a word processing document; (4) Post-editing: The translator revises the raw translation on the screen before it is delivered to the client. It is suggested that a new technological platform will emerge, combining the translator's work station with a terminology bank and translation software.44
It would, however, seem that the accuracy necessary for the translation of regulation is rather far beyond the Capabilities of
The description is mainly based on material made available by CAT Project of the Department of the Secretary of State of Canada, enclosed in a letter of December 5, 1990. Public Administration and Information Technology current systems. A "raw translation" is perhaps not too helpful for the drafter, though it may indeed be very helpful for a translator of a lengthy report. Perhaps computer-assisted translations may prove more useful for access to case law in the short term, but these possibilities will not be pursued here. One should mention that it is possible to draft a regulation in a language-independent formal notation and have a system that uses the information provided by this form to construct texts in several languages.
An early example of wliat is called a "canonic input formalism" is provided by Hélène Bauer-Bernet, who was very influential within the CELEX system for a long period.45 In the example, a half-legible canonic English or French input format is automatically translated into proper English or French.
Canonic input formalism and translations
INPUT
E SIM C * HARMONISATION / ‘LEGISLATION+/MEMBER+ STATE+ E TURNOVER TAX+ F BUT STRUCTURE & PROCEDURE+, APPLICATION/ SYSTEME COMMUN/ F TAXE SUR LA VALEUR AJOUTEE F CHM TAXE SUR LA VALEUR AJOUTEE, (A SI A1) ET B F SIM (A) LIVRAISON+/BIEN & PRESTATION+/SERVICE F PAR (A1) EFFECTUE*,,ASSUJETTI, A TITRE______
45 The example is from Hélène Bauer-Bernet "Drafting, Retrieval and Interpretation of Computerised Community Law", Bryan Niblett (ed) Computer Science and Law, Cambridge University Press, Cambridge 1980:101-1 15. Compendium to the Erasmus Course
OUTPUT FRENCH
Harmonisation des légaslation's des Etats membres concernant les taxes sur le chiffre d'affaires - Structure et procédure pore l'application du système commun de taxe sur la valeur ajoutée. Sont soumises à la taxe sur la valeur ajoutée: (A) les livraisons de biens et les prestations de services si (A1) elles sont effectuées par un assujetti à titre onéreux à l’intérieur du pays (B)les importations de biens.______
OUTPUT ENGLISH
Harmonisation of legislations of the Member States concerning turnover taxes - Structure and procedures for the application of the common system of value added tax. The following shall be subject to value added tax: (A)the supply of goods and the provision of services if they are carried out (A1) by a taxable person for payment within the territory of the country (B)the importation of goods.______
The European Communities have used large resources to develop means for machine translation. For the drafting of regulation it is understood, however, that a canonic form is still used when computer-assisted methods are applied, besides such methods as exemplified above on the basis of Canadian experiences. There are, however, other ways in which computerised systems may be used to assist the drafting of bi- or multilingual regulations. One interesting possibility is methods for checking the consistency, where the multi-lingualism can be turned to advantage, using the terms of one language as some sort of control of the other, assuring that there are no undesired deviations. Public Administration and Information Technology
Example of output from term-tuning^
Förutse
révoir
►disposition
► disposer
tablir
► énocer
exprim er, {expression
► dresser
The Finnish project of term-tuning may serve as an example.46 The computer system is designed to check and evaluate the translation, direct the attention of the translator
46 The discussion is mainly based on "Term-tuning: A method for computer-aided revision ofjmilti-lingual texts", SYMP/Informatique Jur (77)8, Council of Europe, Strasbourg 1977.
289 Compendium to the Erasmus Course
to possible inadequate passages, offer a range of term translations to a translator or drafter, and uncover incongruities or vagueness in the formulation or the logical structure of the text. The last function is seen as the most important. The first assumption is that there is a 1:1 correspondence between term occurrences in one text and those in a parallel text, though "term" is often interpreted as a few consecutive typographical words as one unit. This justifies the approach of studying congruency between the parallel texts.47 The second assumption is that one term should have the same meaning every time it occurs in the text, though it is emphasised that the "empirical incorrectness is conspicuous to anybody with some experience of analysis of actual texts in any field or language". Nevertheless, it is generally useful. It assumed that the regulation does not contain homonyms or synonyms. The system assigns term occurrences in one text to those in the parallel text, finding all cases where there is not a 1:1 correspondence, and presents these results to the drafter.
5.3. Access to Documents in Another Language.
Within a multi-lingual jurisdiction, the regulatory manager may want to access documents that are in another language than the one most familiar to him or her. This represents problems, from the rather trivial - but troublesome - problems of national
^ The project was carried out for Finnish and Swedish, which in contrast to English (or French) rely heavily on compounded words, making this perhaps less of a problem. One should note, however, that Finnish and Swedish belong to widely different language traditions and much are less related than, for instance, English and French. Public Administration and Information Technology characters not supported by the system available to the user, to the problems of formulating a search request or understanding a retrieved document in another language. There have been some efforts in trying to extend help supported by information technology though the solutions adopted are frequently based on a language-independent indexing scheme. Search requests can be formulated in the formal language of the index and will retrieve documents in any natural language. This does, of course, presume an intellectual indexing of all documents and require that the user have knowledge of the indexing scheme, or is offered help by the system in using this scheme.
The oldest European legal information service,48 the Belgian CREDOC system, created by the notaries in 1966, provides an example of this approach. The main component of the indexing language is the descriptors, defined by a four digit numeric code in a bi-lingual thesaurus. These may be modified by ante- and post-descriptors ('[/'cicelies" and "specificateurs"). There are also defined hierarchical structures between indexing terms. Combining these elements, it is maintained that more than 60.000 concepts can be specified.
There have also been attempts at providing tools for specifying a search request in one language and having the system transform this into a request in another language, retrieving documents in both languages. This could have been realised if a sufficiently efficient general method for machine translation had been available. However there are, to the knowledge of the author, no such systems in operation, though there have been attempts to achieve such functionality using simpler methods.
48 There may actually be an Estonian service that is as old, or older - but this did not become operational.
291 Compendium to the Erasmus Course
The major example is the Canadian DATUM system, based on a project initiated in 1970 in the French-speaking province of Quebec. To support retrieval, a novel thesaurus structure was designed. Two thesauri were developed, the s-thesaurus and the g-thesaurus. The latter supplied grammatical expansions of a word included in a search request. The s-thesaurus was developed to expand a certain word into a series of equivalent words or phrases, in both English and French. It adopted a method suggested by Irving Keyton (1966). Selected passages from the documents to become parts of the data base were examined, and important words were replaced by a synonym in the context where the word occurred. A great number of "source lists" was produced, consisting of the original word and the assigned synonyms. These were processed by a statistical program that decided when words were synonyms or homonyms. In this way, lists of synonymity in each language were produced. The source words were translated into the other language, and this was included in the first list, creating a unified thesaurus with entries in both languages, and synonyms in both languages for any entry. The user could specify a search word and, using the thesaurus, would retrieve documents in both languages. The system provided controls for the search that are not detailed above. The DATUM service was discontinued on June 19, 1979, but the bilingual thesaurus remains a major example of a rather innovative attempt to solve the problem of retrieving documents from a multilingual data base.
A more simple support is the generation of language-relative templates for displaying documents. Documents such as those above from the European Communities CELEX system have a large number of terms defining what data are found in the field. Public Administration and Information Technology
These terms may easily be provided in the language of the user, though the content of the field still will be identical for all users. Often the content has a form that is not very language sensitive (typically dates, citations, names of authors, etc), and the help provided by such a language-sensitive template is considerable.
6. Communication between regulatory managers
Obviously, the exchange of information on regulations is not only a question of access to data bases, but also of communication between regulatory managers. This small section of the papers emphasises this traditional and indispensable strategy for learning of and from other's experiences. On the basis of this realisation, in 1991 the International Legal Information Network was initiated. The institutions behind this initiative are, among others. Centre d'information juridique internationale, Computer Center for Information Dissemination of the European Communities, Council of Europe, Harvard Law School, International Labour Organisation, Library of Congress (USA), Pan American Health Organisation, and World Health Organisation. The objective of this initiative is to improve availability of national legislation. There is under preparation a conference system (IL1N) under administration of Cornell Law School, and a data base of libraries in different countries documenting regulations will be established by Harvard Law School.49 This is, however, only one of many initiatives where regulatory managers set up some sort of information system
1L1N-91 is already available in book form from UN-IFO Publishers, Sarasota, USA and Legal Library Publicising Service, Yeovil, UK. The ILIN-92 is under publication, and it is reported that a conference (ILIN- 93) is planned for France.
293 Compendium to the Erasmus Course
between themselves. The computer-assisted communication systems allow several forms of exchange of information. There are electronic mail systems based on list servers: The subscriber to a list will have a copy of any message mailed to the list. This makes it possible to "broadcast" requests for information. As the list may have many thousand subscribers, this may be almost as asking an oracle for an answer. The lists will be specialised to some extent to make its domain appropriate for users. There are also the bulletin boards where notices are pinned, and where comments may be attached. This allows for a communication similar to conferences, where discussions are going on. Bulletin board systems (BBS) are also generally specified concerning interests, and some are designed to attract lawyers and regulatory managers. Obviously, not only brief messages may be conveyed in this manner. The text of a regulation may be communicated through the computer-communication network, often more easily that transferring files by a physical medium like a diskette as the communication protocol provides some sort of compatible format. Regulatory managers rely on communication, and computer- assisted communication methods are efficient and convenient. One will therefore in the future expect regulatory managers to be one user group that will appreciate the potential for improved performance.
7. Conclusion
This paper has looked at some aspects of communication between regulatory managers facilitated by information technology. Access to the regulations of international organisations and countries is today facilitated by computerised systems. But these systems have not really been designed to Public Administration and Information Technology meet the requirements of regulatory managers, rather the end user is seen as a lawyer advising a client on a legal problem, or a librarian locating a certain document. The regulatory manger would have different needs. Perhaps there are two aspects related to a certain regulation on which the regulatory manager would like to have information. First, the policy issue addressed when the regulation was developed: Its objective, the comments expressed by those involved in reviewing the regulation before its was passed, etc. This may be of interest to a regulatory manager within another jurisdiction trying to develop regulations within the same domain. Second, evaluations of a certain regulation. This may be all types of evaluations in the form of statistics, legal literature or any other item of information that will help a regulatory manager to learn whether a certain regulation achieved its objective - or, if it failed, what was the probable cause. One may easily see a possibility of initiatives - like the ILIN mentioned above - growing into an international service for regulatory managers exploiting the powerful tools of information technology better to communicate and co-operate. It way be possible to experiment with different types of regulations, run simulations of the different types against a model of the society in which the manager works, etc. Without pursuing this perspective, one may - perhaps - see the outlines of a new and exciting way in which regulatory managers can profit from the experience of others, and virtually turn the world into a regulatory experimental laboratory.
295
Tidligere utgitt i skriftserien CoinpLex
All reports in the CompLex series may be purcahased through the Norwegian Association for Computers and Law, PO Box 7557 Skillebekk, N-0205 OSLO 2. The Assosiation also maintains a subscription service. Subscriptions may be limited to English language publications, or to «flyers» announsing new reports. In the list below, English language repports are printed in bold face.
Com pLex 1/81 Johs Hansen Et edb-system for analyse av rettslige avgjørelser NOK 64,-
CompLex 2/81 Johs Hansen (red) Notater om deontiske systemer NOK 47,-
CompLex 3/81 Vidar Sørensen Informasjonssystem mm for Norges Geotekniske Institutt NOK 83,-
CompLex 4/81 Kjetil Johnsen Systemtekniske konsekvenser av persondatalovgivningen NOK III,-
CompLex 5/81 Marit Thorvaldsen Teledata og rettsinformasjon NOK 65,-
CompLex 6/81 Anne Kirsti Brække Postmonopolet NOK 27,-
CompLex 7/81 Knut S Selm er (ed) The LAWDATA Papers NOK 28,-
CompLex 1/82 Norsk forening for jus og edb & Svenska foreningen for adb och juridik Nordiske personregisterlover, Council of Europe Convention og OECD Guidelines NOK 31,50 CompLex 2/82 Harald Brænd og Vidar Sørensen Oljevern: Brukerbehov og kildemateriale NOK 55,
CompLex 3/82 Jon Bing og Dag Frøy stad Rettskildebruk N O K 90,
CompLex 4/82 Thomas Prebensen Steen Post- og televerkets regulerte ansvar NO K 77,
CompLex 5/82 Datatilsynet Årsmelding 1981 NO K 44,
CompLex 6/82 Birger Eckhoff in cooperation with Jon Bing, Dag Frøystad and Anja Oskamp CATTS: Computerized program for teaching text retrieval systems NOK 61,
CompLex 7/82 Jon Bing Informasjonssystemer for Trygde rettens kjennelser NO K 126,
CompLex 8/82 de Mulder, Oskamp, van der Heyden and Gubby Sentencing by Computer: An Experiment NOK 123,
CompLex 9/82 Colin Tapper An Experiment in the Use of Citation Vectors in the Area of Legal Data NOK 56,
CompLex 1/83 Arve Føyen Utredning om endringer i pensjonregisterloven NOK 70,
CompLex 2/83 Stein Schjølberg Computers and Penal Legislation NOK 140, CompLex 3/83 John S Gulbrandsen og Terje Hoffmann Rettigheter i idrettsarrangementer NOK 64,-
CompLex 4/83 Sally Moon and Anja Oskamp The Law of Legal Information Systems: Two Essays NOK 95,-
CompLex 5/83 Datatilsynet Årsmelding 1982 utsolgt
CompLex 6/83 Olav Torvund Betalingsformidling NOK 146,-
CompLex 7/83 Else Ryen Lov og lovmottaker NOK 103,-
CompLex 8/83 Daniel Stripinis Probability Theory and Circumsantial Evidence NOK 23,-
CompLex 9/83 Frede Cappelen Edb-basert informasjonssystem for forvaltningens praksis NOK 36,-
CompLex 10/83 NORDIPRO Legal Acceptance of international Trade Data utsolgt
Com pLex 11/83 Jørgen Hafstad og Thomas Prebensen Steen Teleksrett og merverdiavgift på programvare N O K 42,-
CompLex 12/83 Kristin Kjelland-Mørdre Om forenkling av regler utsolgt
CompLex 13/83 Jon Bing Edb: Mulighet og problem ved forenkling av regelverk NOK 92,- CompLex 14/83 Tarjei Stensaasen Utvalgte emner i jus og edb (3. utgave) N O K 1 9 0 -
CompLex 15/83 Anette Klafstad og Ulf Alex Samer Plan for et forsikringsrettslig informasjonssystem N O K 60,-
CompLex 1/84 Jon Erling Skjørshammer Kabelnett: Bygnings- og ekspropriasjonslov utsolgt
CompLex 2/84 Tore Andreas Hauglie og Dag Wiese Scharlum Forslag til et helserettslig informasjonssystem N O K 120,-
CompLex 3/84 Justisdepartementet Den elektroniske grunnbok utsolgt
CompLex 4/84 Datatilsynet Årsmelding 1983 utsolgt
CompLex 5/84 Tove Fjeldvig TekstsØking: Teori, metoder og systemer N O K 90,-
CompLex 6/84 Jon Bing Offentlighetsloven og edb NO K 94,-
CompLex 7/84 Gunnar Bach, Beate Jacobsen and Vidar Stensland The National Social Insurance System of Norway NOK 51,-
CompLex 8/84 Dag Frøystad Data Protection in Practice I: ldentifying and Matching Klements NOK 80,-
CompLex 9/84 Tove Fjeldvig og Anne Golden Automatisk rotlemmat ise ring - et lingvistisk hielnem idsJpl fn r tpkstnrilrino MHk' on . CompLex 10/84 Elling Øyehaug Ose Retstidene som informasjonssystem NOK 120,-
CompLex 1/85 Jon Bing Data Protection in Practise II: International Service Bueraux and Transnational Data Flow NOK 47,-
CompLex 2/85 Jon Bing Opphavsrett og edb NO K 176,-
CompLex 3/85 Dag Wiese Schartum Codex, Calculation and Computers utsolgt
CompLex 4/85 Olav Torvund To informasjonsrettslige arbeider NO K 52,-
CompLex 5/85 Datatilsynet Årsmelding 1984 utsolgt
CompLex 6/85 Hans Chr Aakre (red) Utvalgte artikler i rettsinformatikk utsolgt
CompLex 7/85 Johannes Hansen SARA: Brukerveiledning og programdokumentasjon NOK 150,- CompLex 8/85 Johannes Hansen Modelling Knowledge, Action, Logic and Norms NOK 90,-
CompLex 9/85 Thomas Prebensen Steen (red) Kompendium i informasjonsrett utsolgt
CompLex 10/85 Tarjei Stensaasen Opphavsrettslovens § 43 («katalogregelen») NOK 97,- Com pLex 11/85 Jon Bing Straffelovens definisjon av «tiykt skrift» anvendt på datamaskinbaserte informasjonssystemer NO K 65,-
CompLex 12/85 Magnus Stray Vyrje Vanhjemmel, opphavsrett og datamaskinprogrammer NO K 161,-
Com p Lex 1/86 Anne Kirsti Brække Formidlingsplikt for kabele ie r N O K 80,-
CompLex 2/86 Dag Wiese Schartum Stans av edb-tjenester i.-igs- og,-isesituasjoner N O K 80,-
CompLex 3/86 Ingvild Mestad «Elektroniske spor» - nye perspektiv på personvernet NO K 100,-
CompLex 4/86 Jon Skjørshammer Opphavsrett, databaser og datamaskinprogrammer: Kontraktrettslige aspekter NO K 205,-
CompLex 5/86 Datatilsynet Å rsmelding 1985 NO K 73,-
CompLex 6/86 Johannes Hansen Simulation and automation of legal decisions NOK 132,-
CompLex 7/86 Stein Schjølberg Datakriminalitet og forsikring NO K 54,-
CompLex 8/86 Thomas Prebensen Steen (red) Kompendium i informasjonsrett (2.utg) NOK 70,-
CompLex 9/86 Peter Blume CompLex 1/87 Joseph A Cannaiaci Privacy and data protection law: International Development and Maltese Perspectives NOK 170,-
CompLex 2/87 Jon Bing og Jon Bonnevie Høyer Publisering av rettsavgjørelser NOK 170,-
CompLex 3/87 Hagen Kuehn The Social Security System in the Federal Republic of Germany NOK 77,-
CompLex 4/87 Hanne Plathe Maartmann Personvern i sykepengerutinen NO K 70,-
CompLex 6/87 Jon Bing Electronic Publishing: Data Bases and Computer Programs NOK 55,-
CompLex 7/87 Ande Somby Selektiv gjenfinning av bestemmelser i bygningslovgivningen NO K 119,-
CompLex 8/87 Jon Bing FOKUS: Knowledge based systems for publie administration NOK 125,-
CompLex 9/87 Dag Wiese Schamim The introduction of computers in the Norwegian local insurance offices NOK 230,-
CompLex 10/87 Knut Kaspersen K re di ttopplysn ing N O K 75,-
CompLex 11/87 Ernst Buchberger, Bo Göranzon and Kristen Nygaard (ed) Artificial Intelligence: Perspectives and Implications NO K 120,- CompLex 12/87 Jon Bing, Kristine M Madsen og Kjell Myrland Strafferettslig vern av materielle goder NOK 118,-
CompLex 13/87 Tove Fjeldvig Effektivisering av tekstsøkesystemer: Utvikling av språkbaserte metoder NOK 175,-
CompLex 14/87 Jon Bing Journalister, aviser og databaser NOK 66.-
CompLex 15/87 Andreas Galtung Skatterettslig ekspertsystem: Et eksempel basert på skattelovens § 77 NOK 82,-
CompLex 1/88 Robin Thrap-Meyer Utvalgte emner i rettsinformatikk (4. utgave) NOK 228,-
CompLex 2/88 Ernst Buchberger, Bo Göranzon and Kristen Nygaard (ed) Artificial Intelligence: Perspectives of A I as a social technology NOK 178,-
CompLex 3/88 Jon Bing og Anne Kirsti Brække Satellittfjernsyn NOK 128,-
CompLex 4/88 Jon Bing Journalists, Databases and Newspapers NOK 68,-
CompLex 5/88 Joseph Cannataci Liability and responsibility for expert systems NOK 108,-
CompLex 6/88 Datatilsynet Årsmelding 1987 NOK 58,- CompLex 7/88 Henrik Sinding-Larsen (ed) Artificial Intelligence and Language Old questions in a new key NOK 290,-
CompLex 8/88 Jon Bing Database for offentlige publikasjoner (DOP): Fremtidig organisering NORIS (87) NOK 178.-
CompLex 9/88 Jon Bing Conceptual Text Retrieval N O R IS (77) NOK 118,-
CompLex 10/88 Henning Herrestad Alderstrygden i et nøtteskall, eller skall, skal ikke, en studie av anvendelsen av ekspertsystemskall for PC innen det juridiske domene NOK I0H,-
CompLex 11/88 Ingvild Mestad L'Informatique, la Liberté Individuelle et la Recherche Epidéniiologique Data Protection and Epideminologic Research NOK 98,-
CompLex 12/88 Johs Hansen SAFE P: Sikring av foretak, edb-anlegg og personverninteresser etter personregisterloven TERESA (50) NOK 178,-
CompLex 13/88 Henning Herrestad & Dag Sy vert Mæsel (red) Five Articles on Artificial Intelligence and Legal Expert Systems NORIS (83) FKIJS NOK 108,- CompLex 1/89 Dag Wiese Schartum En rettslig undersøkelse av tre edb- systemer i offentlig forvaltning NORIS (85) NOK 178,-
CompLex 2/89 Datatilsynet Årsmelding 1988 NOK 108,-
CompLex 3/89 Dag Syvert Mæsel og Andreas Galtung XCITE NORIS(91) NOK 238,-
CompLex 4/89 Robin Thrap-Meyer Pantsettelse av datamaskinprogrammer og databaser TERESA(70) NOK 118,-
CompLex 5/89 Ellen-Kathrine Thrap-Meyer Forbrukerkjøp og EDB TERESA(58) NOK 178.-
CompLex 6/89 Andreas Galtung Rettslige aspekter ved miljøinformasjonssystem TERESA(72) /VOK 118,-
CompLex 1/90 Dag Wiese Shartum IRIs SPECTRUM 20 artikler i anledning Institutt for rettsinformatikks 20 års jubileum NOK 298,-
CompLex 2/90 Finn-Erik Vinje (red.) Språket i lover og annet regelverk NOK 188,- CompLex 3/90 Datatilsynets årsmelding 1989 NOK 108,-
CompLex 4/90 Eirik Djønne (red.) Datatilsynet - 10 år som personvernets vokter NOK 108,-
CompLex 5/90 Rolf Riisnæs (red.) Elektronisk betalingsformidling og forbrukerinteresser I NOK 238,-
CompLex 6/90 Beate Heidenstrøm Hasseltvedt Rettslige normer og juridiske ekspertsystemer En vurdering av enkelte sider ved TAXMAN NOK 118,-
CompLex 7/90 Anne Oline Haugen Offentlig informasjon NOK 138,-
CompLex 8/90 Frank Iversen (red.) Nordisk rettsinformatikk - Foredrag fra Oslo-konferansen 21. og 22. september 1989 NOK 228,-
CompLex 9/90 Jon Bing Merverdiavgift og Edb TERESA (63) NOK 98,-
CompLex 10/90 Jon Bing Electronic Copying NOK 98,-
CompLex 11/90 Karl Fr. Jarbo Kabelnett i Oslo-området - Rettslige aspekter NOK 98,-
CompLex 12/90 Henning Berger Herrestad Norms and Formalization NOK 158,- CompLex 1/91 Marek Sergot The Representation of Law in Computer Programs: A survey and comparison NOK 138,-
CompLex 2/91 Jon Bing The Portuguese Infojur Project and Related Issues NORIS (96) III NOK 128,-
CompLex 3/91 Ola-Kristian Hoff (red.) 20 år med rettsinformatikk - 20 år i forvandlingens tegn Institutt for rettsinformatikks jubilcumsseminar NOK 148.-
CompLex 4/91 Andreas Galtung Paperless systems and ED I NOK 198,-
CompLex 5/91 Rolf Riisnæs (red.) Elektronisk betalingsformidling og forbrukerinteresser II Rapport fra den nordiske konferanse 1990 NOK 278,-
CompLex 6/91 Knut Rognlien Forutberegnelighet og teknologisk utvikling som juridiske argumenter NORIS (89) NOK 198,-
CompLex 7/91 Morten Daae Lovdata - Historie, Lov og forskriftssystem NOK 138,-
CompLex 8/91 Anne Kirsti Brække Fjernovervåkning og informasjonstjenester i telenettet TERESA (83) NOK 238,-
CompLex 1/92 Beate Hasseltvedt Avskrivninger av mikrodatamaskiner med tilbehør - en nordisk studie TFRFSA fX7> CompLex 2/92 Nils Kr. Einstabland Kri ngkasl i ngsbegrcpet TERESA (78) NOK 208,-
CompLex 3/92 Maria Strøm Rettskilderegistre i Helsedirektoratet NORIS (94) I & II NOK 228,-
CompLex 4/92 Ditlev SchwanenfUigel Softwarepatentet - Imaterialrettens enfant terrible En redegørelse for patenteringen at softwarerelaterede opfindelser i amerikansk og europæisk ret NOK 158,-
CompLex 5/92 Lars Borchgrevink Grindal Abonnementskontrakter for kabelfjernsyn TERESA (78 II) NOK 248,-
CompLex 6/92 Kolf Kiisnæs Implementing ED I - a proposal for regulatory form NOK 118,-
CompLex 7/92 Morten S. Hagedal Deponering av kildekode «escrow»-klausuler TERESA (79) NOK 128,-
CompLex 8/92 Ola-Kristian Hoff EDB i juridisk undervisning - med en reiserapport fra England og USA NOK 228,-
CompLex 9/92 Jon Bing og Dag Elgesem Universiteters ansvar for bruk av datanett TERESA (94) NOK 198,-
CompLex 10/92 Andreas Galtung Rettslige sider ved teletorg NOK 148,- CompLex 1/93 Giovanni Sartor Artificial Intelligence and Law Legal Philosophy and Legal Theory NOK 148,-
CompLex 2/93 Connie Smidt I nformationsansvaret Erstatningsansvar for informationstjenester, særlig ved databaseydelser NOK 138,-
CompLex 3/93 Ingvild Hanssen-Bauer Personvern i digitale telenett NOK 178,-
CompLex 4/93 Joachim Benno Consumers Purchases through Telecommunications in Europe - application of private international law to cross-border contractual disputes NOK 198,-
CompLex 5/93 Morten S. Hagedal Four essays on: Computers and Information Technology Law NOK 218,-
CompLex 6/93 Marianne Rytter Evensen Sendetidsfordeling i nærradio MERETE (3) III NOK 148,-
CompLex 7/93 Richard Susskind Essays on Law and Artificial Intelligence NOK 158,-
CompLex 1/94 Andrew J. I. Jones & March Sergot (edlitors) Deon’94 Second International Workshop on Deontic Logic in Computer Science NOK 358,-
CompLex 2/94 Beate Jacobsen Film og videogramrett TCDCC A /£A\ uni/1IO CompLex 3/94 Rolf Risnæs Elektronisk datautveksling i tollforvaltningen - Rettslige spørsmål knyttet til TVINN NOK 225,-
CompLex 4/94 Mari Bø Haugstad Sykepenger og personvern - Noen problemstillinger knyttet til behandlingen av sykepenger i Infotrygd NOK 148,-
CompLex 5/94 Mads Andenæs, Rolf Høyer og Nils Risvand EØS, medier og offentlighet TERESA (103) NOK 148,-
CompLex 6/94 Jon Bing Offentlige informasjonstjenester: Rettslige aspekter NOK 148.-
CompLex 7/94 Nils Eivind Risvand Satellittfjernsyn og norsk rett. MERETE (3) IV NOK 138,-
CompLex 8/94 Beate Jacobsen (red.) Videogram på forespørsel. MERETE (14) IV NOK 158,-
CompLex 9/94 Bjørn Bjerke «Reverse engineering» av datamaskinprogrammer. TERESA (92) IV NOK 198,-
CompLex 10/94 Gjert Melsom Skattemessig behandling av utgifter til anskaffelse av datamaskinprogrammer TERESA (75) NOK 198,-
CompLex 1/95 Jon Bing Rettslige konsekvenser av digitalisering: Rcttighetsadministrasjon og redaktøransvar i digitale nett NOK 368,-
CompLex 2/95 Sverre Sandvik Rettslige spørsmål i forbindelse med utvikling og bruk av standarder innen telekommunikasjon NOK 178,- CompLex 3/95 Tina Smith Legal tøxpert Systems: Discussion of Theoretical Assumptions NOK 278
CompLex 4/95 Ole Tokvam Personvern og straffeansvar - Straffelovens § 390 NOK 198
CompLex 5/95 Johs. Andenæs Juridisk utredning om filmen «To mistenkelige personer» NOK 138 COMPLEX Bestillingskupong
Ja takk, vi bestiller herved følgende eks. av COM PLEX:
.....eks. av nr...... å k r ......
...... eks. av nr...... å k r ......
...... eks. av nr...... å k r ......
...... eks. av nr...... å k r ......
Av fremtidige COMPLEX utgivelser (se omtale foregående sider) bestilles herved:
eks. av nr...... å k r ......
eks. av nr...... å k r ......
N av n :......
F irm a:......
A d resse:......
P o stn r./ste d :......
Kupongen sendes til:
flkademika■ inIKNMIONI i P«GIO>manD((
Karl Johansgt. 47 0162 Oslo 1
The book contains a collection of articles discussing various legal pro blems connected with the application of information technology in public administration. Important keywords are: • Data protection • Rule of law • Regulatory management
Many of the contributions take, as an explicit point of departure, exam ples from the Norwegian public sector. However the problems discus sed are of a general nature. Several of the questions the book pin points are also of topical interest for the private sector.
ISBN 82-518-3416-3
TANO 9 788251 8341 62