Organization and Information Processing: Elements of a History of Early Digitization in the Swiss Federal Administration

The Government as Machine

In his book The Government Machine about the history of the computer in the British Civil Service Agar (2003) picked up on a metaphor that Charles Babbage used in the 19th century. Babbage spoke of the government as an “analytical engine.” The metaphor can also be found later in the scientific literature. Peter Collin and Klaus-Gert Lutterbeck used the image of an “intelligent machine” (Collin & Lutterbeck, 2009) in 2009. Recently, the metaphor has also been used by Nick Schwery in his master thesis at the Swiss Federal Institute of Technology (Eidgenössische Technische Hochschule, ETH) in Zurich (Schwery, 2017), entitled “Die Maschine regieren. Computer und eidgenössische Bundesverwaltung” (“Governing the machine. Computers and the Federal Administration”). Agar connects such metaphors to various forms of government. He describes the development in the late 18th century as a “liberal balance” of power and trade between different actors, as “checks and balances” that change to a “machine” or “engine” in the 19th century (Agar, 2003, pp. 21-44). His reference is Charles Babbage, who defined progress as mechanization and control of the governing forces. The best symbol for this is the punch card that materializes thoughts: “In fact, there is no reason why mental as well as bodily labor should not be economized by the aid of machinery” (Agar, 2003, p. 43).

One of the first measures to mechanize the British Civil Service was the introduction of the typewriter. It was accompanied by a division of labor between “rule-giving generalists” and “rule-following mechanicals” in the administration (Agar, 2003, pp. 45-63). We will discuss this division of labor in the Federal Administration using the example of the ZAR. Gradually, it became clear that the distribution of functions within the governmental and administrative apparatus was central. Richard Burdon Haldane saw the government not as a machine but as a whole that should be functionally recorded for efficient operation. He identified three principles for efficiency: specialization, organization of intelligence and command, and financial control (Agar, 2003, p. 69).

Agar describes a next big step with the invention of the universal machine by Alan Turing, who identified the term “mechanical” with a routine process, as did Max Newman. Soon, mechanical came to mean “done by a machine” (Agar, 2003, pp. 63-74). Turing combined it with the “decidability problem” and the “generalist-mechanical split.” Agar concludes that “Turing’s [universal] machine and Babbage’s engine are examples of how political history—the metaphor of government as a machine, mobilized for quite different reasons during the modern period—intermeshed with the history of technology” (Agar, 2003, p. 74). In support of this statement, he cites the case of the General Register Office and ideas about “one universal register” for Great Britain. Modernization of the economy and society led to a growing need for coordination and control by the European states. World War I reinforced this trend. The states needed information to produce reliable statistics for planning (censuses) and controlling citizens (registers; Agar, 2003, p. 102ff.). I will return to the functions of registers in the section on the ZAR. Before doing so, I will present a short overview of the development of the Swiss Federal Administration and describe its path toward automatic data processing.

The Development of the Swiss Federal Administration—A Short Overview

To follow Jos C.N. Raadschelders, a history of the administration has to be embedded in an analysis of the internal operations of the administration but must also consider aspects of external interdependencies with society, the economy, science, and so on (Raadschelders, 2009, p. xiii). The administration assists the government in executing decisions made by the parliament. In political science, there is an ongoing discussion on the separation and boundaries of government and administration. In the United States, Woodrow Wilson spoke of a “politics-administration dichotomy” (Wilson, 1887). His concept distinguishes fundamentally between elected representatives of the people in a parliament or government and officials of a public administration. In Switzerland, Fritz Fleiner, professor of Public and State Law, referred to the administration as the “acting state power” (Fleiner, 1911, p. 7). As part of assisting the executive, the administration’s tasks include the preparation, implementation, and execution of laws, regulations, and directives. In such a process, the administration produces “binding decisions” (Niklas Luhmann) that not only set a framework for society but also increasingly entail a controlling function (Luhmann, 1970, p. 182).

In the 20th century, in Europe, governments and administrations increasingly devised, planned, and realized measures to steer various political fields. The substance, direction, and extent of these coordinating and controlling functions take different national and supranational forms. This article considers the example of the Swiss federal government, the “Swiss path” toward administrative automation and digitization (Koller & Schüpbach, 2016).

The modern Swiss Confederation was founded in 1848, after a short civil war in which progressive liberal cantons (states) prevailed over Catholic conservative cantons. As the decentralized administrative structures largely dating back to the Ancien Régime no longer met the requirements of a modern 19th-century political system (Bäumlin, 1961, pp. 69ff.), a new central administration—a “classical state apparatus” (Germann, 1998, Chapter 2)—was needed.

The groundwork was laid by the Federal Act on the Organization and the Course of Business of the Federal Council (Bundesrat) of July 7, 1849. ² The Federal Council is the government of the Swiss federal state. It was particularly in charge of designing the framework for a modern economy and society, supported by the Federal Administration (Bundesverwaltung). These measures—which were based on laws and resolutions of the Federal Assembly (Bundesversammlung), the parliament of the Swiss federal state—aimed at unifying a culturally, linguistically, religiously, and economically highly fragmented territory (Bäumlin, 1961, p. 69ff.).

The growing tasks of the Administration led to its repeated enlargement and reorganization. The Federal Council would be divided into seven departments, which still exist to this day, with the clearly defined tasks shared out among them. ³ This delegation of tasks was set out in concrete terms in the Federal Council’s resolution concerning the competences of the departments and their heads of April 9, 1897 (Altermatt, 1991, pp. 291-305). That civilian-like nature, with the ad hoc involvement of private individuals for certain actions such as negotiations and so on characterized the political system in Switzerland for a long time, as the German sociologist Max Weber also remarked. Weber’s chapter on the “Honoratiorenverwaltung” in “Wirtschaft und Gesellschaft” from 1922 compares U.S. townships with the “directly democratic Swiss cantons” (Weber, 1922). Decisive advances in professionalization were made during World War I ⁴ and World War II, especially, for the purposes of steering a war economy.

The federal government extended its administration massively during the 1930s and 1940s. Because of the development and expansion of the welfare state after World War II and a corresponding population increase, a significant part of the government’s savings efforts after 1945 were unsuccessful. Therefore, the Federal Council set up the ZOB. ⁵ Among other things, it was responsible for conceptualizing and planning data processing in the different government offices. This professionalization of the administrative service was comparable to those in other countries, for instance the British Civil Service, as Agar described.

The Fuel for the Machine: Data and Information

The most important task of the public administration is the production of “binding decisions” (Niklas Luhmann). The modern state is required to make decisions for other functional areas of society. To this end, the administration collects and analyzes data and information in the political field for which it is responsible to the government (or politicians). This knowledge is the basis on which the government takes decisions and the administration executes them.

In times of rapidly growing information volumes, workflows had to be standardized and the processing of data automated. George W. Brown and Louis N. Ridenour discussed this in respect of the United States as far back as 1953 (Brown & Ridenour, 1953). Luhmann analyzed legal aspects of automation in public administrations in 1966 (Luhmann, 1966). In the case of the Swiss Federal Administration, we can distinguish three different paths toward automation and digitization: “operations research” in the military field, “automatic data processing” in the fields of registers and statistics, and “digital applications” in various political fields that overlap with scientific areas such as agriculture, meteorology, materials testing, and others. The need for research and innovation as part of the production of knowledge in public administrations is sometimes underestimated and therefore deserves to be studied seriously. However, it is not possible do that here. In the next section, I will concentrate on the development of registers and databases, with the ZAR as the focus of the investigation.

As a consequence of the growth of the Federal Administration after World War II, ever more information had to be processed in less and less time. Two important examples of this are the old-age and survivors’ insurance (OASI; Alters- und Hinterlassenenversicherung, AHV), which provides each person with an identification and insurance number, and the ZAR, which was set up to establish reliable statistics and developed into the main instrument to control the residence of foreigners in Switzerland. A technological change began that would have a lasting effect on the Administration. During the 1960s, the collecting of data was standardized and the procedures increasingly computerized. Automatic data processing appeared to fulfill the promise of greater efficiency. To test this, we should first clarify our terminology. We have to differentiate terms used in the 1960s from concepts we use nowadays.

In the 1960s, data processing meant the collection and management of information objects. When data, numbers, texts, and other items are processed largely automatically by a machine on a given program, we speak of automatic data processing. In the 1970s, the term often used was “electronic data processing.” The term “Information and Communication Technology (ICT)” was not used at the time. Its use today stresses the role of unified communication and the integration of hardware and software to enable users to access, store, transmit, and manipulate information.

At the beginning of the 1960s, the Federal Administration was running three data centers and had around 10 big calculating machines: one at the Federal Statistical Office (Bundesamt für Statistik) that functioned as an electronic data center for the whole Administration, three at the ETH in Zurich, two at the Military Department, and three others for social security, customs, and taxes. At that time, the ZOB developed a policy paper on data processing in the Administration. ⁶ This article tells us much about the perception of the problems in processing information automatically and the first concepts for dealing with such problems. It begins by stressing the advantages of data processing systems: great speed in performing the operation (once it is prepared; see below), more comprehensive analysis with little additional effort, print-ready tables, and easy, space-saving archiving. But there are also disadvantages: Data processing is a rigid system and has no flexibility, so that a lot of work went into preparing an operation. Finally, computers were very expensive and created staff problems because experts were rare and costly (Gugerli, 2002).

The premises for operating such a system set out by the ZOB are also very interesting. Data technology is always in advance, the ZOB said; not all applications were predictable, and electronic processing was only suitable if computational tasks fulfilled the following conditions: The task must be described exactly and logically; it must have a minimum volume and should have a certain routine character. But the main condition for operating such a system, according to the ZOB, was the adjustment of management methods and accurate and full planning. The experts on organization said there could not be any short-term changes of rules and workflows.

Measured by numbers, the ZOB’s policy was quite successful. By 1978, there were already 47 mainframe computers running for the federal government: The Interior Department had 16 machines, the Post Office 13, the Military 7, Finance and Economics 5 each, and the Foreign Department 1. This machinery was worth around 195 million Swiss francs and was run for more than 71,000 hr by 660 persons, most of them experts. As a result, the costs were very high: 116 million Swiss francs. The Statistical Office still functioned as the data center for 33 government offices. It was running three mainframe computers now with a staff of 128 and costs of 32 million Swiss francs. ⁷ In 1980, the ZOB’s responsibility for data processing was institutionalized by law (see below).

If we try to assess the ZOB’s policy paper of the early 1960s, we may say that it did not correspond to a digital vision but rather aimed at solving problems in the management of ever more information. As Thomas Haig remarked, corporate “systems men” tried to establish the computer as a managerial information system rather than a simple technical extension of punch-card data processing and sought to define computing as the primary agent of managerial control. The computer supported a new conception of information (Haig, 2001). Haig’s findings are also true for the early digitization of the Swiss Federal Administration. Data processing developed incrementally on a path that would ultimately lead to the digitization of important parts of the Federal Administration and the operationalization and standardization of administrative workflows. It is therefore necessary to look more closely at the position of the ZOB to better understand its role in the digitization of the Federal Administration.

In the Heart of the Machine: The ZOB and the Automation of the Federal Administration ⁸

As already mentioned, the ZOB was responsible for the organization of the Federal Administration. The office was set up by the Federal Act (Bundesgesetz) of October 7, 1954. It had a special legal status: Although it was organizationally part of the Federal Department of Finance, it was directly responsible to the Federal Council. The ZOB was therefore regarded as a privileged staff unit of the government. It constantly examined utility and potential savings in the Administration. It assessed the establishment, extension, restriction, or abolition of offices and services of the Confederation. It initiated and carried out investigations to this effect or delegated them to third parties. ⁹ The offices and services of the Confederation gave the ZOB unlimited access to their business and provided the necessary documentation (Hongler, 1978).

In 1960, the Federal Council entrusted the ZOB with the planning and coordination of automatic data processing in the Federal Administration. To this end, it granted the ZOB technical directive authority, although the ZOB did not run the computers. This was done by the Federal Statistical Office—as already mentioned, the first government office to process data automatically. In the 1970s, there was a growing demand for a more rational organization and better workflows in the Federal Administration. For this reason, it was felt that the position of the ZOB within the Federal Administration should be strengthened and more computer specialists should be employed in the ZOB and selected federal agencies. ¹⁰ The Federal Council therefore asked parliament to upgrade the Central Office to a Federal Office for Organization in 1979. ¹¹ Parliament passed the Act of December 19, 1980 on the Federal Office for Organization (Bundesamt für Organisation, BfO).

The 1980 Act explicitly mentioned the promotion, coordination, and monitoring of automation and data processing in the Federal Administration as one of the main tasks of the BfO (Article 5). The BfO was responsible for issuing new directives and technical instructions on appropriate and economical workflows and tools (Article 6). The first technical instruction on data processing in the Federal Administration dated from January 6, 1961. ¹² Subsequently, the BfO continued to expand the conceptualization, planning, and coordination of automatic data processing in the Federal Administration. Its key tool for implementing the concepts was HERMES—the acronym stands for Handbuch der Elektronischen Rechenzentren des Bundes, eine Methode zur Entwicklung von Systemen (Manual of the federal electronic data centers, a method for the development of systems)—a management method for projects in the area of IT, service and product development, and business organization adjustment. ¹³ HERMES is the continuation and elaboration of the technical instruction on data processing in the Federal Administration dated January 6, 1961 (see above) and is the main IT project method used today.

Although today we speak of a digital revolution, the technical and organizational change toward computers and digitization was, on the whole, incremental, as James Cortada observed for instance in his fascinating book “All the Facts” about the history of information in the United States since 1870 (Cortada, 2016). In the administrative sector, data processing developed into ICT during the 1990s. This change was accompanied by a shift of perception. The focus on organization was replaced by a focus on technology: In 1990, the Federal Council dissolved the BfO and transferred its tasks to the Federal Office of Information Technology (Bundesamt für Informatik, BfI, now: Federal Office of Information Technology, Systems and Telecommunication). ¹⁴

The Central Register of Foreigners: A Model for Digitization in the Administration ¹⁵

The ZAR, the Central Register of Foreigners of the Swiss Foreigners Police (Eidgenössische Fremdenpolizei), is an early example of data processing and automation in the Swiss Federal Administration. Its development took place in three phases. Its operational objective was to process more information in a shorter time. This was due to a growing flow of information on foreigners living in Switzerland: Between 1960 and 1965, the number of foreign residents increased from 496,000 to 810,000. They were mostly migrant workers from Italy. Right-wing politicians were warning of the country being swamped and calling for action to reduce immigration. The Federal Council adopted measures to “restrict the influx” in 1963. To assess the effectiveness of these measures, reliable statistics were necessary. For the Foreigners Police, the problem was “essentially a question of numbers and statistics.” Therefore, it proposed to use federal statistics as a way to chart and monitor foreigners in the Swiss cantons. Based on a report from a group of experts, the Federal Council decided in 1970 to develop a central register. In 1971, the project was pioneered in four cantons, the first phase in the development of this tool. 2 years later, statistics for all Swiss cantons were available for the first time. In a second phase after 1975, the application was expanded with more comprehensive table tools and a network of stations and screens. Incrementally, this statistical tool evolved into a permanent inventory—a database (the third phase).

Implementing the ZAR project was quite complex: In 1973, during the first phase, Swiss communes collected information about 1,493,573 individuals using paper forms. They transmitted the data on paper to the ZAR where it was copied, mostly by women, to punch cards that were then transferred to the BfS, the data center of the Federal Administration. After a plausibility check, the data were linked to the AHV (OASI) number that was used as the official Swiss social security number at the time (Brügger, 2018). With this link, a data set was considered official.

Programming the ZAR took 17 months, and subsequently the data were updated continuously. Some 23 employees, mostly women, made about 8,000 changes a day. Initially, there were some challenges with the input and output of the ZAR: Because around 30% of all foreign individuals left Switzerland without going through the necessary formalities, their departure was not registered. Other problems arose because births were registered as new entries, and some professions were wrongly coded. Furthermore, some of the tables were too complicated to be used.

Despite these problems, statistics were now more reliable than before. Some 18,000 duplicates were eliminated. However, the operation of the ZAR was very costly, and the automated data exchange required intensive cooperation with the other parties involved. In the beginning of the second phase, in 1975, the application was extended with more comprehensive tools for the production of tables. Afterwards, the ZAR was equipped with terminals for working with the permanent inventory. Gradually, a statistical tool evolved into a database for real-time use. This database was cross-linked with the registers of other authorities at the federal and cantonal level. In 1977, 36 different terminals were integrated into a network between the ZAR and the data center in the BfS. Afterwards, in the third phase, the ZAR became the most important tool for the Swiss Foreigners Police. It enabled them to control how long people had been in the country and whether they left on time and was used to find addresses (to collect taxes or alimonies), to monitor entry bans, and as a basis for the calculation of entry quotas. The Bundesanwaltschaft (Office of the Attorney General of Switzerland) even took data from the ZAR to politically monitor foreigners. This led in 1989 to the so-called Fichenskandal, which was inquired by a parliamentary commission (Kreis, 1993).

The Swiss Foreigners Police believed in the concept of a permanent inventory that “has brought a much improved control mechanism” to Switzerland. The ZAR became a model application that was shown all over the Federal Administration and was visited by experts from abroad. However, establishing a network caused a new problem: securing and protecting the data. In 1983, the Federal Council published an ordinance for the ZAR that for the first time contained regulations on the registration of persons by the Swiss Foreigners Police. In addition, the ZAR was responsible for a tacit centralization. Essentially, alien’s policy was a cantonal responsibility; but the development of the ZAR infrastructure led the cantons gradually to hand over the relevant operations to the Federal Administration.

Discussion: Elements of a History of the Early Digitization of Public Administrations

The general characteristics of the ZAR project and the development of information processing in the Swiss Federal Administration under the auspices of the ZOB can serve to describe a historical path for the early stages of digitization in public administrations: The need to process more information in less time led to a search for economic solutions and investment in data processing. The result is an operationalization of administrative processes for the production of binding decisions that is perhaps reminiscent of the image of the government as a machine. In the 1960s, consultants, journalists, and some scientists promoted the “totally integrated management information system,” as Thomas Haig remarked, the computer supported a new conception of information. Haig’s findings are also true for the early digitization of the Swiss Federal Administration. The state’s need for coordination and control entailed a growing demand for information. The technology and instruments to collect and manage this data are well known: forms, the interfaces between citizens and the state, and registers, the systems to store and manage these data. They were mechanized and processed with the aid of punch cards on a machine. When they were updated constantly, they incrementally developed into permanent inventories, into databases. What we have here is a model for the early automation of workflows accompanied by a process of standardization of information.

The history of the Federal Administration in the 1970s and 1980s shows that early digitization is driven not only by technology but also by the organizational problems of large administrations, as, for instance, James Cortada described. The first calculation machines that the ZOB helped to plan and coordinate were a response to this. They helped to solve the constantly increasing information problems of the Swiss federal state. The example of the ZAR showed the importance of the organizational aspect in this development. It also showed the hybridity of the process, the interplay between analog and digital processes.

Data processing is developing incrementally on a path that ultimately leads to the digitization of ever larger parts of public administrations. As a consequence, the question of trust between states and citizens is an integral part of the history of digitization. For a comprehensive history, it will be important to understand that automation is an interaction, a constellation among human being, machine, and program. We have to see administrations as hybrid processes with numerous passages from digital to analog procedures (Machlup, 1983, p. 660). New forms of administrative organization will be based on technologies that form part of the digital infrastructure. Digital technologies and administrative action will be inextricably linked (Stalder, 2018).