Digital Organizing

Revisiting Themes in Organization Studies

Ursula Plesner, Emil Husted

• Publisher: Bloomsbury Academic
• Year: 2019
• ISBN:
  9781350305328
• Edition: 1
• Topic: Business
• Subtopic: Information Management
• Index: Business

This important new textbook offers a lively and topical discussion of how digital technologies impact various aspects of organizations, such as structure, knowledge, collaboration, communication, identity, legitimacy and power. Taking a critical and nuanced approach, this engaging textbook introduces readers to central themes in organization studies and reflects on how changes brought about by digitalization have important implications for private, public and voluntary organizations, and on practical disciples such as strategy, management, innovation and entrepreneurship. Contemporary case studies drawn from a wide range of international organizations demonstrate the real-world relationship between digital technologies and organizing. This is an essential textbook for final year undergraduates, postgraduates and MBA students taking a module in technology and organization. It is also suitable for any student of organizational studies wanting to understand more about the role that the digital plays in contemporary organizing.

PART

I

DIGITAL ORGANIZING

1

THE DIGITAL AND THE ORGANIZATIONAL

This book is titled Digital Organizing, and the present chapter zooms in on the implications of these two words. We begin our account of the digital in the 1850s, when we saw the first attempts to develop computational machinery and the binary thinking that forms the basis of present-day digital technologies and the automation of work. We then jump to the 1940s, when advances in engineering resulted in the transistor, which allowed computational power to double every 18 months since then. This exponential growth in computing capacity has resulted in an enormous amount of innovation within, and diffusion of, digital technologies and their applications in organizational contexts. We end our account of the digital with a discussion of how a new epoch of connectivity between individuals and organizations began in the 1960s, when the first steps toward the invention of the internet were taken. The first part of this chapter is therefore dedicated to fleshing out what is implied by the terms ‘algorithmic thinking’, ‘exponentiality’, and ‘connectivity’. These terms are important in understanding the technical foundations of digitalization as we experience it today. Based on this very basic understanding of what the term ‘digital’ refers to, we can then discuss what is implied when we talk about digitalizing products, processes, and entire organizations or industries.

Following the description of ‘the digital’, the second part of the chapter presents our approach to thinking organizationally about digitalization. We explain how organizational thinking differs from more technology-oriented perspectives on digitalization. Our goal here is to show that although new organizational dynamics emerge with digitalization, the established concepts and approaches from organization studies remain useful tools for understanding the digitalization of organizations. The chapter ends with a presentation of the structure of the book, describing seven major organizational themes: structure, production, collaboration, knowledge, communication, legitimacy, and power. We will show how each of these classical concepts can be linked up with a more contemporary concept, which together can elucidate various dimensions of digital organizing not necessarily captured by the classical concepts in their original form.

Algorithmic thinking

The idea of producing machines that could compute arithmetical tables dates back to the early nineteenth century. Sadie Plant tells the poetic story of how, in 1833, an amazed audience was invited to inspect a futuristic device “which seemed to have dropped into [the] world at least a century before its time” (Plant, 2016, p. 5). The device in question was the Difference Engine, invented by the engineer Charles Babbage. Babbage’s project was to use machines to compute arithmetical tables, and his Difference Engine was able to raise “several Nos. to the 2nd and 3rd powers, and extracted the root of a quadratic Equation”, as one of the onlookers observed (Plant, 2016, p. 5). That same year, George Boole (1854) developed his ‘symbolic logic’, later published in his An Investigation of The Laws of Thought. Based on formal logic and algebra, the book presents the idea that logical relations can be expressed in symbolic form: thus, the logical dichotomy ‘true or false’ can be translated into the binary digits ‘0’ or ‘1’. As Boole himself writes, his objective was “to investigate the fundamental laws of those operations of the mind by which reasoning is performed; to give expression to them in the symbolical language of a Calculus” (Boole, 1854, p. 1).

Today, we talk about Boolean algebra as that branch of algebra where the values of the variables are ‘true’ and ‘false’ and the main operations are ‘and’, ‘or’, and ‘not’. Consider how Boolean algebra differs from elementary algebra, where the values of the variables are numbers and the main operations are addition and multiplication. Boole’s work laid the foundation for the design of modern computers. Now that logical propositions could be expressed in algebraic form, logical deductions could be drawn by algebraic calculations. But it would take decades before operational computers could be designed and built. In the 1930s, Alan Turing developed a theoretical model of the computer: “With a tape drive and a computation unit, this hypothetical, abstract machine was capable of reading, erasing, and writing digits on a single line of type. It processed zeroes and ones on a tape of infinite length which passed through the drive, and followed a series of basic commands” (Plant, 2016, p. 82).

As Plant notes, all “subsequent computers are implementations of this most general of general purpose machines” (Plant, 2016, p. 82). Early computer technology (Image 1.1) was developed by engineers working with electronic circuits. An electronic circuit can be in an ‘on’ mode and an ‘off’ mode, corresponding to a ‘1’ when the current passes through the circuit and a ‘0’ when the current does not pass through it. This again corresponds to the Boolean ‘true’ and ‘false’. The basic operation of a computer starts with these ‘on’ and ‘off’ circuits, which are interpreted as ‘1’ or ‘0’. The ones and the zeroes are ordered into patterns, and the patterns are what gives them meaning. To take a random example, the pattern 1110001100 corresponds to the number 908. However, it is not only numbers that can be represented by zeroes and ones. Any object or experience that is produced or processed by digital technologies (graphics, sound, photo) consists of combinations of zeroes and ones. Computer operations are also just numbers, encoded to make different things happen. Even the most advanced software is nothing more than a very long set of on and off signals, which – combined – tells a computer what to do (Image 1.2). The sets of rules defining the operations that a computer must perform are called algorithms. Since we are interested in the organizational dimensions of digitization, we will not delve further into a discussion of binary logics or computation as such. We will simply underscore here that binary logics are the foundation for the development of both algorithms and computers and argue that they are also becoming an important force in structuring organizational life.

Image 1.1 Computer scientist Grace Hopper presenting a computer in 1952. Hopper popularized the idea of machine-independent programming languages, and she is credited with popularizing the term ‘debugging’ for fixing computer glitches – inspired by an actual moth removed from the computer.

Source: Science Source, Ritzau/Scanpix.

Binary logics in organizations: The case of digitization-ready legislation

The binary logics discussed here, besides being foundational to computer programming, have also become a factor in redesigning organizations, ostensibly to make them more efficient. The goal is that certain types of routine administrative tasks can be carried out by algorithms that can process information and even ‘make decisions’. Let us take an example from the organization of public service delivery. In countries where the digital infrastructure is advanced, public organizations that deal with certain kinds of standard administrative case processing can eliminate a number of meetings between citizens and public servants, and all the interaction previously required in connection with preliminary registration, application forms, information storage, processing, and decisions that needed to be made in relation to a case. Some cases may be quite simple: in a country that pays child subsidies, it is enough that the ‘system’ is informed that a woman has given birth to a child. Based on the mother’s national ID number, the system then accesses the parents’ income, determines the amount of payment, and sets up a monthly transfer of child subsidy to the mother’s bank account. This happens in Denmark, where the public sector is among the most digitized in the world, and where all persons have a national ID number and a linked bank account. In the Danish public sector, much of the routine administrative work has become automated in this way (Justesen & Plesner, 2018). Data about the citizens are already available in digital platforms shared across the public sector, so when citizens apply for various services online by entering specific information, their application can in some cases be processed by an algorithm, untouched by human hands, or unread by a human case officer. An algorithm is programmed to produce a ‘decision’ and then inform the citizen immediately that they are accepted/rejected for a subsidy, or that they are being refunded X amount of income tax.

The efficient use of algorithmic decision-making requires that the law and the procedures be simple. If this is the case, an algorithm can quickly decide whether a citizen is entitled to a particular social benefit on the basis of objective factors such as age or income level. To make the most of this technological possibility – and to achieve the potential benefits in efficiency – all new legislation in Denmark needs to be ‘digitization-ready’. This means that the law – to the extent possible – must be written in simple, unambiguous terms that can be translated into binary codes for use in the administration of the law. The automation of work processes presupposes binary options and a consistent vocabulary. This is because even the most sophisticated algorithms cannot handle ambiguous categories or problems. Complex administrative laws call for professional judgment and can end up preventing the use of digital solutions. Only a human employee can consider special cases or interpret ambiguous language (Justesen & Plesner, 2018). In the Danish public sector, the automation of administrative work based on binary thinking has led to extensive analyses of work processes in order to determine what kind of work can be automated and what needs to be handled by people.

Image 1.2 Old computer punch card. Punch cards are pieces of stiff paper with holes. They have been used throughout much of the twentieth century to contain digital data (with the punched/unpunched holes representing ‘on’ and ‘off’ or ‘1’ and ‘0’). Similar cards are still used in voting machines.

Source: EyeJoy, iStock.

The digitalization of case-handling raises numerous questions: How can we ensure that automation enhances the quality of services? What does it mean to be treated ‘fairly’? What role should programmers play in the legislative process? And, most importantly, are there limits to automation? Are there certain processes that are so complex or sensitive that they should require face-to-face interactions and human judgment? Given the speed of technological advances and the constant quest to improve efficiency, digitalization is becoming more a question of how we follow binary logics to automate organizational processes rather than whether we should automate certain processes.

The phenomenon we have described here is also known as computational thinking. Computational thinking attempts to re-define a problem so that it can be solved by a computer. This process usually entails breaking down a complex problem into ever smaller, more manageable parts (Google for Education, 2017). This way of thinking is both essential to creating computer applications and creates the basis for the digitalization of organizations (Table 1.1). Across many types of organizations, computational thinking is applied in analyzing work processes, simplifying decision-making, and automating case-handling.

Table 1.1 The concepts of digitization and digitalization.

Beyond binary thinking: The challenge of complicated mathematics

‘Digitization-ready legislation’ and the resulting reconfiguration of public administration are examples of digitalization of organizations based on algorithmic thinking. The examples emphasize the practices of extracting the human element from the equation so that work processes can be automated. But this simplification process is not the full story of algorithmic thinking. Although algorithms are based on simple binary logics, they can perform very complex tasks. They can be designed to organize decision-making, services, trade, and entire industries in ways that are impenetrable for anyone who is not a skilled mathematician. This is the point made by Cathy O’Neil (2016) in her book Weapons of Math Destruction. A mathematician herself, with a PhD in algebraic number theory, O’Neil worked for a hedge fund, joined a risk analysis company, moved to an internet start-up, and finally decided to use her insights about mathematical models to enrich the public debate. Based on her experiences in the field, O’Neil has grown increasingly skeptical of how we allow algorithms to make quick, but ultimately opaque, decisions:

This was the Big Data economy, and it promised spectacular gains. A computer program could speed through thousands of résumés or loan applications in a second or two and sort them into neat lists, with the most promising candidates on top. This not only saved time but also was marketed as fair and objective. After all, it didn’t involve prejudiced humans digging through reams of paper, just machines processing cold numbers. By 2010 or so, mathematics was asserting itself as never before in human affairs, and the public largely welcomed it. Yet I saw trouble. The math-powered applications powering the data economy were based on choices made by fallible human beings. Some of these choices were no doubt made with the best intentions. Nevertheless, many of these models encoded human prejudice, misunderstandings, and bias into the software systems that increasingly managed our lives. Like gods, these mathematical models were opaque, their workings invisible to all but the highest priests in their domain: mathematicians and computer scientists. (O’Neil, 2016, p. 3)

In her book, O’Neil discusses a number of areas where organizations rely on advanced algorithms to process data and to make decisions that may end up having grave consequences. A spectacular example is that of a very popular schoolteacher who was fired because of a score generated by a ‘value-added modeling system’. The school district had hired a consulting firm to develop an evaluation system that could rate teachers in order to weed out the worst of them (those in the bottom 2 percent). The algorithm used was supposed to eliminate human bias and be objective. At the same time, it was bound to be very complex because it needed to integrate a myriad of factors such as students’ socioeconomic background, learning disabilities, various measures of educational progress, teacher evaluations, and so on. In the end, this meant that, although the teacher got excellent reviews from her principal and from her pupils’ parents, and herself believed that she was a good teacher, she got laid off due to the low scores. Moreover, she could never obtain any explanation for how these scores were derived. In O’Neil’s analysis, the first problem with the algorithmic basis for firing the teacher was that nobody could explain how the score had come about. “It’s complicated,” as the managers said. Some of the more fundamental problems with the algorithm were that it could not judge whether the data entered into the system were valid, and that it was designed to make inferences on the basis of a very low number of cases – which is statistically unsound.

Despite such weaknesses, it is clear that algorithmic thinking has become important in the reorganization of work and the relations between employees, managers, and citizens or customers. It is also worth noting how much the implementation of algorithms ultimately depends on the mathematicians who develop them. Normally, mathematicians and statisticians live a quiet life a...