Digital Media Ecologies

Entanglements of Content, Code and Hardware

Sy Taffel

• Publisher: Bloomsbury Academic
• Year: 2019
• ISBN:
  9781501349256
• Edition: 1
• Topic: Ciencias sociales
• Subtopic: Sociología

Our digital world is often described using terms such as immateriality and virtuality. The discourse of cloud computing is the latest in a long line of nebulous, dematerialising tropes which have come to dominate how we think about information and communication technologies. Digital Media Ecologies argues that such rhetoric is highly misleading, and that engaging with the key cultural, agential, ethical and political impacts of contemporary media requires that we do not just engage with the surface level of content encountered by the end users of digital media, but that we must additionally consider the affordances of software and hardware. Whilst numerous existing approaches explore content, software and hardware individually, Digital Media Ecologies provides a critical intervention by insisting that addressing contemporary technoculture requires a synthetic approach that traverses these three registers. Digital Media Ecologies re-envisions the methodological approach of media ecology to go beyond the metaphor of a symbolic information environment that exists alongside a material world of tantalum, turtles and tornados. It illustrates the social, cultural, political and environmental impacts of contemporary media assemblages through examples that include mining conflict-sustaining minerals, climate change blogging, iOS jailbreaking, and the ecological footprint of contemporary computing infrastructures. Alongside foregrounding the deleterious social and environmental impacts of digital technologies, the book considers numerous ways that these issues are being tackled by a heterogeneous array of activists, academics, hackers, scientists and citizens using the same technological assemblages that ostensibly cause these problems.

PART ONE

Theorizing digital ecologies

1

Technology, complexity and agency

On 6 May 2010, the US equity market experienced an extreme and novel form of turbulence. Commonly known as the ‘Flash Crash’, around US$800 billion was wiped off the value of stocks between 2:32 and 2:45 p.m., with the vast majority of the fall taking place between 2:41 and 2:45. After trading in Standard and Poor (S&P) E-Mini futures was paused for 5 seconds, the market began to recover, with over two-thirds of losses being recovered by the close of the day. In the wake of this event, media coverage focused on speculation surrounding the role that computer-based trading played in the Flash Crash, how this new class of digital actor had demonstrated the ability to impact financial systems in unpredictable ways that could lead to catastrophic economic consequences.

In this chapter, I will use the events of the Flash Crash as a way of examining a range of questions surrounding technology, agency and complexity. The concept of agency addresses who and what can and does act, how these actions impact upon other entities, and how these agencies are distributed between individuals and collectives, humans and nonhumans. While agency has traditionally been invoked as a purely human and individual affair, it is approached here as a distributed capacity that can never solely be attributed to a single, isolated entity. Instead, agencies are understood as unstable, relational and multiple rather than as the expression of an individual’s will. In order to formulate an ecological model of mediation and political action, addressing questions of how agency is distributed between human, organic and technical systems is a necessary pre-requisite; conceptualizing how to mobilize and enact change in response to ecological crises requires us to first consider how agencies coalesce, cascade and erupt within complex, nonlinear systems.

The chapter begins by outlining the key actors involved in the Flash Crash, and how their collective interactions shaped this event. This then leads to a discussion of several concepts and genealogies that can help us to create a map of how agency functions within complex, dynamical technocultural systems. This begins by thinking through the multiple legacies of cybernetics, which includes the tropes of information theory, feedback and nonlinearity. I then move on to discuss several interrelated system theories that have been influenced by cybernetics: autopoiesis, complexity theory, and Deleuze and Guattari’s concepts of assemblages and abstract machines, all of which assist in refining an ecological model of agency. Finally, I conclude the chapter by employing discourses of cyborgs and posthumanism to consider the inhuman agencies and politics of computer-based trading.

Algorithmic trading

To grasp what happened in the Flash Crash of 2010, we must first become familiar with the main entities that were involved in the event. This means exploring several types of computer-based trading systems, alongside structural changes that have occurred within financial markets which enable computer-based trading. In particular, this section will explore algorithmic trading (AT) and high-frequency trading (HFT) systems as two different types of digital agents whose actions were central to both the onset and the subsequent trajectory of the Flash Crash. Electronic trading is a relatively new phenomenon, which in certain contexts has largely replaced human-to-human transactions, to the point where now over half of all trades on financial markets are executed by algorithms. Whereas financial trading may still conjure up images resembling the human-led pit trading depicted in films such as The Wolf of Wall Street, by 2010 this was no longer the major form of exchange within digitized and globalized financial systems.

Pathways towards electronic trading have not been smooth and continuous. The evolution of automated trading systems is episodic and discontinuous, with technological changes surrounding software and hardware being accompanied by structural pressures arising from both transnational financial markets and the specific local politics that surround futures exchanges. While there clearly would be no automated trading systems without computers, connectivity and code, a range of human agencies also played key roles in these events. For example, the formation of the Globex trading system – which in 1992 became the first fully electronic trading system, when it was introduced as part of the Chicago Mercantile Exchange (commonly known as the Merc) – was largely a response to the perceived threat of futures markets in London and Hong Kong that were open while trading pits were closed in the United States. Donald Mackenzie (2015: 660) documents how ‘electronic trading shifted from being an unimportant adjunct to the pit to becoming a replacement for it’, with the introduction of the S&P E-Mini stock market index futures contract in 1997.¹ Shortly after its introduction, the new E-Mini began to outsell the pit-traded ‘big’ S&P contract, in part because the near-instantaneous electronic Globex system meant that the inherent delays of embodied open outcry pit trading could be avoided.

Electronic trading does not have an unproblematic history. A prominent early example of the destabilizing impact of electronic trading surrounds the severity of the 1987 market crash commonly known as ‘Black Monday’. This event was partially attributed to portfolio insurance, an early form of electronic trading that was ‘designed to protect individual investors from losses, but when used by many investors simultaneously … helped make the fall in prices a systemic event with a feedback loop’ (Carlson 2007: 15). While this high-profile, high-impact case where electronic trading played a prominent role in a stock market crash dampened the enthusiasm for computer-based trading in the late twentieth century, subsequent increases in computational processing power accompanied by the introduction of more algorithmically complex forms of electronic trading subsequently led to computer-based trading performing the majority of trades on global markets.

The final report of the Foresight project, which was commissioned by the UK government following the Flash Crash to explore the future of computer-based trading in financial markets, argues that there are two fundamental classes of computer-based trading system in operation today. The first is AT systems that perform trades that would have been undertaken by humans in the past, and the second is HFT systems ‘doing jobs that no human could ever hope to attempt’ (Beddington et al. 2012: 33). HFT systems execute trades at speeds grossly exceeding those which human traders are capable of. They execute huge volumes of these trades, with each individual transaction only designed to produce a fraction of a cent in profit. Over time, however, the massive quantities of miniscule amounts add up to a significant source of revenue, while vastly increasing the overall volume of exchanges in financial markets. This expedited pace of exchange can be grasped through the fact that whereas in 1945 US stocks were held for an average of four years, by 2011 this had decreased to a mere 22 seconds (Toscano 2013). HFTs are designed not to build any significant portfolios of stocks, so most assets are traded moments after they are acquired, and portfolios are not held overnight. While proponents claim that under most circumstances HFTs add liquidity to markets due to the increased volume of transactions, as we will see, there are also situations in which they have contributed to sudden shortages of liquidity and thus to the formation of Flash Crashes.

While HFT has been a central figure in press coverage and popular debates surrounding the Flash Crash, the Securities and Exchange Commission (SEC) report into the Flash Crash identifies an AT program as the instigating factor for the event (SEC 2010). At 2:32 p.m., a large trader initiated an automated execution algorithm designed to sell a total of 75,000 E-Mini contracts, one of the three largest single-day sell programs executed on the E-Mini within a twelve-month period. AT programs can be instructed to take price, time and volume as variables into account when completing the order, but in this specific case, the AT program was set to sell orders at a rate of 9 per cent of the volume of trades occurring over the previous minute without specifying price or time as additional variables.

As the AT program began to sell large numbers of E-Mini contracts, many were initially bought by HFTs; however, as HFTs are designed not to hold large numbers of contracts at any time, as they began to accumulate contracts in a declining market, at 2:41 HFTs began aggressively selling contracts. The large AT sell program’s response to this increased volume of trades was to increase the rate at which it sold contracts, as the only variable it was using to govern its activity was the volume of transactions, thereby adding further pressure to the market which had seen orders on the buy-side fall to less than 1 per cent of that morning’s level (SEC 2010: 3). This caused a liquidity crisis, and consequently, AT systems that had been instructed to buy or sell particular stocks without price variables in some cases executed trades at irrational prices of either 1 cent or $100,000, ‘stub quotes’, or placeholders that are never intended to actually be traded. At 2:45:28 p.m., E-Mini trading was paused for 5 seconds by the Merc to prevent a further cascade of declining prices and irrational trades. This brief break in trading allowed the sell-side pressure to relieve, and when trading resumed prices stabilized and then recovered.

What does this event tell us about the relationships between technology, agency and complexity? On the one hand, the Flash Crash confronts us with the scope of particular forms of nonhuman agency within the ecology of partially automated, digital financial trading. The AT and HFT systems provide definite advantages in terms of speed when contrasted to human traders and that temporal advantage entails that in a competitive marketplace there is a strong rationale for replacing human traders with automated systems. Indeed, when it comes to HFT, we see strategies that would be impossible for humans to execute being highly profitable. However, these nonhuman decision-making entities also have the potential to behave in unpredictable ways that can amplify the impacts of crises and crashes, generating systemic instabilities which are highly undesirable. Furthermore, some of these behaviours, such as buying stub quotes at $100,000 each, the highest price that can be listed, are forms of irrationality that would almost certainly not occur with human traders.

Within the internal logic of financial markets though, these unwanted impacts are insufficient for investment banks, hedge funds and other trading entities to consider jettisoning computer-based trading; this would leave these actors at a competitive disadvantage. The broader system therefore has a structural role in determining the agencies of individual trading entities, and while periodic instability may be unwelcome, the advantage of trading at the speeds of networked computational systems rather than those at which human bodies and communication acts function is perversely understood to outweigh these systemic risks. As Foresight conclude, a consequence of exchanges being conducted at speeds which outcompete human traders and prevent human oversight in real time from removing structural risk is that ‘computer based (and therefore mechanical) trading is almost obligatory, with all of the system-wide uncertainties that this gives rise to’ (Zigrand, Cliff and Hendershott 2011: 9).

Moving beyond this general understanding that computer-based trading has some form of agential capacity within contemporary financial markets, and that this both increases profitability and systemic risk, requires exploring how nonlinear systems function in some depth. We have seen that during the Flash Crash, HFTs created a feedback loop which amplified the risks and issues that emanated from the large AT sell program. In order to map these issues surrounding nonlinearity, technology and agency, I next turn to the history of understanding processes of control and feedback whose genealogy can be traced to the formation of the interdisciplinary field of cybernetics in the mid-twentieth century, before examining how processes of feedback and homeostasis that emerged in cybernetics have subsequently been reformulated within complexity theory, systems biology and ecology. Engaging with these theoretical and historical accounts allows us a more detailed and nuanced way of grasping how relational and distributed agencies flow through open systems, which not only are key to comprehending the specific case of computer-based trading but are pervasive within digital media ecologies.

Nonlinear agencies

Cybernetics emerged as a field of academic inquiry during the 1940s and 1950s from the collaboration of a transdisciplinary group of academics including Norbert Wiener, Warren Weaver, Gregory Bateson, John von Neumann and Margaret Mead. The term ‘cybernetics’ was coined by Wiener (1948) in Cybernetics: Or Control and Communication in the Animal and the Machine, and as the title denotes, cybernetics aimed to explore mechanisms of control and communication alongside organizational and configurative patterns common to living and nonliving systems. From its inception then, cybernetics muddies the distinction between living and nonliving systems (George 1977: 2). In addition to examining biological and technical entities, cyberneticists recognized that ‘it is certainly true that the social system is an organization like the individual, that it is bound in a system of communication, and that it has dynamics in which circular processes of feedback play an important part’ (Wiener 1948: 24), additionally blurring the boundaries between individuals and collectives.

The histories and legacies of cybernetics are not only relevant in terms of their demarcation of feedback and nonlinear dynamics though; the branch of cybernetics associated with Claude Shannon and Warren Weaver’s information theory, and Jon von Neumann’s work around digital computers are pivotal to the technological genealogies that manifest today as pervasive networks of digital devices. Equally, as we will see later, the paradigm of control and communication has been advocated as the fundamental logic or diagram that defines contemporary societies, for example, in Deleuze’s work surrounding societies of control. Conversely, later strands of systems theories that pay a genealogical debt to cybernetics include systems biology, complexity theory and Earth Systems theory, which are key fields for science-led comprehensions of the Anthropocene. As a consequence of cybernetics’ influences upon contemporary discourses of technology, control and ecology – the central themes of this book – it is useful to recount various strands of cybernetic praxis in order to elucidate how they came to be so influential, in addition to contrasting the models of agency that arise from these differentiated and often contradictory models.

Cybernetics effectively formed as a discipline from a series of conferences held in the United States between 1946 and 1953, commonly referred to as the Macy conferences, that were formally titled ‘Feedback Mechanisms and Circular Causal Systems in Biological and Social Systems’. Feedback loops occur when elements are causally connected so that an initial causal factor circulates around the system, so that effects feed back to the start of the loop. Whereas in a linear chain of causality A effects B which effects C which effects D, in a system with circular causality (feedback) A effects B which effects C which effects A. Wiener uses the example of a man steering a boat as an example of a feedback-based system; the steersman’s job is to visually assess any deviation from the desired course and compensate by moving the ship’s rudder to counter-steer. This may even overcompensate, in which case the steersman reassesses the situation and alters direction. As such, the steersman navigates through a process of continuous feedback. Indeed, the term cybernetics originates from the Greek word kybernetes, meaning steersman, as cybernetics studies processes of control or steersmanship.

Early cybernetics research explored a diverse array of feedback-based systems: biological systems, such as human coordination in walking or picking up cigarettes; mechanical systems, such as the thermostat and the governor of a steam engine; and systems which link living and nonliving components such as the steersman. In all of these examples, ‘the feedback tends to oppose what the system is already doing, and thus is negative’ (Wiener 1948: 111). Negative feedback is self-corrective or homeostatic; feedback counteracts systemic perturbations. Positive feedback, by contrast, involves feedback which reinforces change, leading to vast alterations given only minute changes to inputs, as difference becomes iteratively magnified. Although cyberneticists discovered the equations governing positive feedbacks, they were largely conceptualized as undesirable noise which led to systems rapidly becoming unpredictable. Consequently, positive feedbacks were neglected by cybernetics research, which was characterized by minimizing noise while explicating processes of homeostatic balan...

Table of contents

• Cover
• Half-Title
• Title
• Contents
• Acknowledgements
• Introduction: Re-thinking media ecology
• PART ONE Theorizing digital ecologies
• PART TWO Ecologies of content, code and hardware
• Bibliography
• Index
• Copyright