Social Analyses of Computing: Theoretical Perspectives in Recent Empirical Research
Citation: Rob Kling (1980) Social Analyses of Computing: Theoretical Perspectives in Recent Empirical Research. ACM Computing Surveys (RSS)
DOI (original publisher): 10.1145/356802.356806
Semantic Scholar (metadata): 10.1145/356802.356806
Sci-Hub (fulltext): 10.1145/356802.356806
Internet Archive Scholar (search for fulltext): Social Analyses of Computing: Theoretical Perspectives in Recent Empirical Research
Tagged: Computer Science (RSS) Social Computing (RSS), Computer supported cooperative work (RSS), Social Computing (RSS), Human Computer Interaction (RSS)
In his introduction, Kling argues that speculative analysis of the roles computing will play in the social world is very important for informing decision making and planning for the potential consequences of technological development. However, to think that such speculations are objective and not based in prior beliefs about society is naive technological determinism. Social analysis of computing requires choosing assumptions about the nature of the social world. Configurations of assumptions are clustered into theoretical perspectives.
He focuses on two theoretical perspectives: systems rationalism and segmented institutionalism. Systems rationalists are optimistic about the roles technology will play in social life, they assume that there is consensus about social goals. They tend to be holistic, emphasize efficiency, and make computer users the center of their analysis.
Systems rationalists are concerned with "legitimate" aspects of social life (e.g. economic efficiency, the workplace as officially represented by managers). Segmented institutionalists are also concerned with "illegitimate" aspects like class (.e.g conflict, ambition). They value individual agency and social equity above economic efficiency. They consider broader implications of technology and how it may impact parties outside the workplace or firm.
Subtypes of systems rationalism include management scientists, who often place the interests of managers above subordinates; structural analysis are interested in the broader social world, but still focus on optimal decision making in the interests of the firm; and human-relations analysts, who see worker satisfaction as important for efficiency.
Segmented institutionalized include political analysts who focus on conflict between groups and power, interactions who focus on social construction of meaning through interaction, and Marxists who focus on how technology furthers labor exploitation or potentially liberating.
While Kling has lumped all these fields into two high level categories, he emphasizes that they actually differ in important ways about social assumptions. There is an amazing table on page 4 that summarizes the two theoretical perspectives and 6 sub-perspectives. He concludes the first section of the article with an example about the introduction of a new kind of organizational control system into a manufacturing company. The goal of this paper is to draw from prior empirical studies to learn the relative usefulness of the two perspectives.
Early studies of computing (50s and 60s)
During this period the systems rationalists mainly wrote speculative minded articles about how technological development might alter the work place and very few empirical studies of the "actual impacts of specific developments" on social life in or out of the workplace. On the other hand, the segmented institutionalists (Mann, Blum, Hoos, Mumford), found empirical evidence of the political roles of computer use, what happens to jobs that get automated (in the early days people were moved into different roles, but from the 70s on they got laid off), and how clerical jobs changed with the introduction of new technology.
Contemporary studies (70s)
In the 70s the systems rationalists finally got their act together. Whisler (1970) found that automation increased centralization of decision making in insurance firms. Gotlieb and Borodin (1973) "avoided largely speculative analyses" in their book reviewing empirical studies of computing in the workplace that refrains from the naive optimistic utopian accounts of the 60s.
Computing in organizations (70s)
Systems rationalists often assume that technological innovations diffuse in order to meet a "need of individuals or organizations." However defining "need" is a problem. Needs are not just economic conceptions, but also by "social features of an organization." Organizational efficiency is often subordinate to the private rationales of administrators. Computer systems can be introduced in order to "fit the political contours of existing organizations" and are selected as "instruments of bureaucratic politics" for their application to "ongoing conflicts and coalitions." For example computers can be used to impress auditors (who can't examine the code) in ways that and calculations can't. This fact is incongruous with the accounts provided by systems rationalists. Firms are much more than engines of economic production.
Although human relations people often suggest that work will be radically improved by technology, it turns out that computer use "did not profoundly alter the character" of jobs in the real world. Some jobs are made much more efficient through technology, but for others it has little effect. It seems most helpful for white collar workers and managers.
One common claim by structural analysis is that computing affords better monitoring of employees, but Kling didn't find strong evidence that this was common in settings like police detectives or accountants. Computers make records, but people were making manual records before. "Computing is selectively exploited as one strategy among many for organizing work and information." Political analysts have a better take because they consider the existing degrees of prior autonomy.
Computers cause problems! While management analysts often think that computer problems have systematic solutions, this doesn't stop problems from happening. Kling suggests that we should think of computers not as tools but as packages of skills, organizational units, physical devises, and understandings of what computing is and means. The interactionist approach sees these as sociallly embedded. Kling and Kraemer did an empirical study to decide whether the tool or package concept is a better explination for why copmuter problems happen. They find that policies are retroactively not proactively and so cannot prevent new kinds of problems from arising and that new kinds of problems are more likely to arise in more technically and socially complex settings. They interpret this as a favorable to the segmented institutionalist perspective and in favor of the 'package' conception. Computing is not simply a 'tool'.
Automation can improve efficiency but it can also be uesd to produce an air of innovation and sophistication.
Decision making is one of the most theoretically interesting applications of computing, but at the time a small share of computing resources were employed in decision making compared to the automation of mundane business operations.
Systems rationalists were interesed in technical problems with databases and when decisions could have used the data, but didn't. Interactionists and organizational-politics folks consider how computing systems interact with the external relations involved in decisions. "Computing was used because it convinced important parties that decisions were being carefully made."
When it comes to automating mundane tasks, there can be a surprising efficiency tradeoffs. People might spend more time doing less important work that can be partly automated. E.g. increasing police efficiency at minor offenses "further clogged the already jammed courts."
Systems rationalists usually focused on how well procedures worked. When a system is completely automated this is a Good approach, however when skill and cooperation are involved control systems can be "gamed" (see Campbell's law).
Control systems can exist for bureaucratic or political purposes and despite claims of importance may be basically useless. How 'efficacy' is constructed can be just as important as the capabilities of the technological system when it comes to whether models will be used or not. Computers can be sites of conflict between different groups, each seeking to enforce their desired forms of measurement by encoding them in the system.
Models are often only used when they justify already made decisions. They are often used rhetorically by advocates. They can be too difficult and expensive to develop and deploy quickly. Modelers rather than models become political tools because the modelers have an air of objective authority. "Computer-based analyses are a social resource used by political actors in the same manner as any other social resource."
Introducing automation systems can alter power structures in organizations. Influence can shift and people's jobs cans change. These changes are not intended in the system design, or by managers. In politicized settings (e.g. municipal governments) computer systems can profoundly alter power relations. In municipal governments power shifted from the council to top administrators and data custodians.
Lots of people, remembering McCarthyism are concerned about how computer use may lead to considerable loss of privacy because computers can collect a lot of information that can be shared. But according to studies by systems rationalist, "major institutional barriers ... prevent organizations from routinely pooling information about their data subjects." (Boy has this changed!!!)
Rule, a segmented institutionalist, was concerned with mass surveillance and social control by credit agencies. He studied TRW which was a highly centralized credit agency with 20 million records that collect and provide "derogatory information." While Rule anticipated the political abuse of such systems, he didn't find evidence of it being currently abused,
While the rationalists proposed "hygienic reform of record keeping systems," Rule thought that such mechanisms would not be able to limit the capacity of such information to "profoundly constrain" through organizations acting legitimately. "Rule is concerned that only economic costs serve as a barrier to collecting and using similarly delicate information." Kling argues that this is a much more serious consideration and that the rationalists misguidedly attempt to make their analyses "value free." The segmented institutionalizes do better by making their value commitments explicit.
How are computer systems going to be made socially accountable? The public seems to have difficulties correcting errors encoded in computer systems. Systems rationalists tend to say that efficiency, authority and professionalism will be enough. Alternatives involve the state regulation, law, and citizen action. Because computer systems are often used for bureaucratic and political purposes not efficiency the systems rationalists are hopelessly naive. In settings like accounting where law already enforces accountability systems involving computers are likely to be accountable, but in other settings like police work they may be easily abused.
In conclusion, Kling reiterates that computers have little causal agency themselves, rather computer use is "purposive and varies between social settings." Impacts of computing cannot be decoupled from the social situations. Computer scientists need to learn that researchers is always informed by the researchers' prior theoretical commitments.
He ties up his argument that the two perspectives are complimentary but that as the focus becomes broader the segmented institutionalist perspective becomes more powerful.
With the development of the chip-based microprocessor (emerging at the time) the scope of computing is likely to become broader and more profound in the near future as email, ETF, and "wired societies" take off.
It is important that our theoretical assumptions correspond to reality not to some idealized utopian notions that can be easily analyzed.
This article shifted units of analysis (firm, municipality, broad social world, industry) in order to make a broad point about the kinds of theoretical assumptions on which analyses are based, but individual studies should be careful and explicit about levels of analyses.
Theoretical and Practical Relevance
Rob Kling was an early researcher into studies of the role of computing in organizations. This seminal article argues for a theoretical approach that incorporates symbolic interactionism and political theory instead of assuming that organizations are rational and that systems can be uncritically designed to solve clearly defined problems. He reviews and synthesizes a lot of prior literature so this is a very good source for early perspectives into the role of computers in organizations.
This work sketches two categories of often-oppositional inquiry which have persisted from the time of its writing to today. Although naming conventions have shifted, of particular use are the summary table on page 64, which outlines both views and their sub-disciplines, and their general perspectives on technology, settings, organizing concepts, dynamics of technical diffusion, their notion of what constitutes good technology, and their affiliation with different workplace ideologies.