Visualization in the Age of Computerization

Annamaria Carusi, Aud Sissel Hoel, Timothy Webmoor, Steve Woolgar, Annamaria Carusi, Aud Sissel Hoel, Timothy Webmoor, Steve Woolgar

• Publisher: Routledge
• Year: 2014
• ISBN:
  9781135077365
• Edition: 1
• Topic: Ciencias sociales
• Subtopic: Estudios de medios

Digitalization and computerization are now pervasive in science. This has deep consequences for our understanding of scientific knowledge and of the scientific process, and challenges longstanding assumptions and traditional frameworks of thinking of scientific knowledge. Digital media and computational processes challenge our conception of the way in which perception and cognition work in science, of the objectivity of science, and the nature of scientific objects. They bring about new relationships between science, art and other visual media, and new ways of practicing science and organizing scientific work, especially as new visual media are being adopted by science studies scholars in their own practice. This volume reflects on how scientists use images in the computerization age, and how digital technologies are affecting the study of science.

Part IVisualization in the Age of Computerization

1 Algorithmic Alchemy, or the Work of Code in the Age of Computerized Visualization

Timothy Webmoor

DOI: 10.4324/9780203066973-2

Introduction

“I’m doing something very dangerous right now” (Informant 1, July 8, 2010).

“Yah, now is not a good time for me!” (Informant 2, July 8, 2010).

Ethnographic silence can speak volumes. Despite prompts from the anthropologist, the dialogue dried up. Falling back on observation, the two informants were rapidly, if calmly, moving between their multiple program windows on their multiple computer displays. I had been observing this customary activity of coding visualizations for nearly a month now—a visual multitasking that is so characteristic of the post–Microsoft Windows age (Friedberg 2006). Looking around the open plan office setting, everyone was huddled in front of a workstation. Unlike ethnographic work in “wet labs,” where the setting and activities at hand differ from the anthropologist’s own cubicled site of production, this dry lab (Merz 2006) seemed so mundane and familiar, as a fellow office worker and computer user, that admittedly I had little idea of how to gain any analytic purchase as their resident anthropologist. What was interesting about what these researchers and programmers were doing?

It wasn’t until I had moved on to questioning some of the PhD researchers in the cramped backroom that the importance of what the two informants were doing was overheard: “Well done! So it’s live?” (Director, July 8, 2010). The two programmers had launched a web-based survey program, where visitors to the site could create structured questionnaires for other visitors to answer. The responses would then be compiled for each visitor and integrated with geo-locational information obtained from browser-based statistics to display results spatially on a map. It was part of what this laboratory was well known for: mashing up and visualizing crowd-sourced and other “open” data. While focused upon the UK, within a few months the platform had received over 25,000 visitors from around the world. The interface looked deceptively simple, even comical given a cartoon giraffe graced the splash page as a mascot of sorts. However, the reticence the day of its launch was due to the sheer labor of coding to dissimulate the complicated operations allowing such an “open” visualization. “If you’re going to allow the world to ask the rest of the world anything, it is actually quite complicated” (Director, June 11, 2010). As one of the programmers later explained his shifting of attention between paper notes, notes left in the code, and the code itself, “I have to keep abreast of what I’ve done because an error … (pause) … altering the back channel infrastructure goes live on a website” (Informant 1, July 21, 2010).

Being familiar with basic HTML, a type of code or, more precisely, a markup text often used to render websites, I knew the two programmers were writing and debugging code that day. Alphanumeric lines, full of symbols and incongruous capitalizations and spacing, were recognizable enough; at the lab this lingua franca was everywhere apparent and their multiple screens were full of lines of code. Indeed, there were “1,000 lines just for the web page view itself [of the web-based survey platform]” (Informant 2, June 6, 2011)—that is, for the window on the screen to correctly size, place and frame the visualized data. Yet when looking at what they were doing there was little to see, per se. There was no large image on their screens to anchor my visual attention—a link between the programming they were engrossed in and what it was displaying, or, more accurately, the dispersed data the code was locating, compiling and rendering in the visual register.

The web-based survey and visualization platform was just one of several visualizing platforms that the laboratory was working on. For other types of completed visualizations rendering large amounts of data—for example, a London transport model based upon publicly available data from the Transport for London (TfL) authority—you might have much more coding: “This is about 6,000 lines of code, for this visualization [of London traffic]” (Informant 3, June 10, 2011) (Figure 1.1).

Over the course of roughly a year during which I regularly visited the visualization laboratory in London, I witnessed the process of launching many such web-based visualizations, some from the initial designing sessions around the whiteboards to the critical launches. A core orientation of the research lab was a desire to make the increasingly large amounts of data in digital form accessible to scholars and the interested public. Much information has become widely available from government authorities (such as the TfL) through mandates to make collected digital data publicly available, for instance, through the 2000 Freedom of Information Act in the UK, or the 1996 Electronic Freedom of Information Act Amendments in the US. Massive quantities are also being generated through our everyday engagements with the Internet. Of course, the tracking of our clicks, “likes,” visited pages, search keywords, browsing patterns, and even email content has been exploited by Internet marketing and service companies since the commercialization of the Internet in the late 1990s (see Neuhaus and Webmoor 2012 on research with social media). Yet embedded within an academic institution, this laboratory was at the “bleeding edge” of harvesting and visualizing such open databases and other traces left on the Internet for scholarly purposes. Operating in a radically new arena for potential research, there is a growing discussion in the social sciences and digital humanities over how to adequately and ethically data mine our “digital heritage” (Webmoor 2008; Bredl, Hünniger and Jensen 2012; Giglietto, Rossi and Bennato 2012). Irrespective of what sources of data were being rendered into the visual register by programmers and researchers at this lab, I constantly found myself asking for a visual counterpart to their incessant coding: “Can you show me what the code is doing?” I needed to see what they were up to.

Figure 1.1 Lines of code in the programming language C++ (on right) rendering the visualization (on left) of a London transport model (Informant 3, June 10, 2011).

Codework

Code, or specifically working with code as a software programmer, has often been portrayed as a complicated and arcane activity. Broadly defined, code is “[a]ny system of symbols and rules for expressing information or instructions in a form usable by a computer or other machine for processing or transmitting information” (OED 2013). Of course, like language, there are many forms of code: C++, JavaScript, PHP, Python—to name a few more common ones discussed later. The ability to “speak computer” confers on programmers a perceived image of possessing inscrutable and potent abilities to get computers to comply. ¹ Part nerd, part hero, programmers and specifically hacker culture have been celebrated in cyberpunk literature and cinema for being mysterious and libertarian. ² The gothic sensibility informing such portrayals reinforces a darkened and distanced view of working with code.

The academic study of code, particularly from the science and technology studies (STS) perspective of exhibiting what innervates the quintessential “black boxes” that are our computing devices, has only recently been pursued ethnographically (e.g., Coleman and Golub 2008; Demaziere, Horn and Zune 2007; Kelty 2008). Oftentimes, however, such studies scale out from code, from a consideration of its performative role in generating computerized outputs, to discuss the identity and social practices of code workers. More closely examining working with code has received less attention (though see Brooker, Greiffenhagen and Sharrock 2011; Rooksby, Martin and Rouncefield’s 2006 ethnomethodological study). Sterne (2003) addresses the absent presence of code and software more generally in academic work and suggests it is due to the analytic challenge that code presents. The reasons for this relate to my own ethnographic encounter. It is boring. It is also nonindexical of visual outputs (as least to the untrained eye unfamiliar with “reading” code; see Rooksby, Martin and Rouncefield 2006). In other words, code, like Thrift’s (2004) “technological unconscious,” tends to recede from immediate attention into infrastructural systems sustaining and enabling topics and practices of concern. Adrian Mackenzie, in his excellent study Cutting Code (2006, 2), describes how software is felt to be intangible and immaterial, and for this reason it is often on the fringe of academic and commercial analyses of digital media. He is surely right to bemoan not taking code seriously, downplayed as it is in favor of supposed higher-order gestalt shifts in culture (“convergence”), political economy (“digital democracy” and “radical sharing”) and globalization (“network society”). No doubt the “technical practices of programming interlace with cultural practices” (ibid., 4), with the shaping and reshaping of sociality, forms of collectivity and ideas of selfhood; what Manovich (2001, 45) termed “trans-coding” (e.g., Ghosh 2005; Himanen, Torvalds and Castells 2002; Lessig 2004; Weber 2004; for academic impacts see Bartscherer and Coover 2011). However, these larger order processes have dominated analyses of the significance involving the ubiquity of computer code.

Boring and analytically slippery, code is also highly ambiguous. The Oxford Dictionary of Computing (1996) offers no less than 113 technical terms that use the word “code” in the domain of computer science and information technology. So despite acknowledging the question “Why is it hard to pin down what software is?” (2006, 19), Mackenzie, a sociologist of science, admirably takes up the summons in his work. For Mackenzie, code confounds normative concepts in the humanities and social sciences. It simply does not sit still long enough to be easily assigned to conventional explanatory categories, to be labeled as object or practice, representation or signified, agent or effect, process or event. He calls this “the shifting status of code” (ibid., 18). Mackenzie’s useful approach is to stitch together code with agency. Based upon Alfred Gell’s (1998) innovative anthropological analysis of art and agency, Mackenzie (2006, 2005) pursues an understanding of software and code in terms of its performative capacity. “Code itself is structured as a distribution of agency” (2006, 19). To string together what he sees as distributed events involving code’s agency, he takes up another anthropologist’s methodological injunction to pursue “multisited ethnography” (Marcus 1995). In terms of how code is made to travel, distributed globally across information and communication technologies (ICTs) and networked servers as a mutable mobile (cf. Latour 1986), this approach permits Mackenzie to follow (the action of) code and offer one of the first non-technical considerations of its importance in the blood flow of contemporary science, commerce and society.

Given its mobility, mutability, its slippery states, code can usefully be studied through such network approaches. Yet I am sympathetic with recent moves within ethnography to reassess the importance of locality and resist the tendency (post-globalization) to scale out (see Candea 2009; Falzon 2009; in STS see Lynch 1993). While there is of course a vast infrastructural network that supports the work code performs in terms of the “final” visualizations, which will be discussed with respect to “middle-ware”, most of the work involving code happens in definite local settings—in this case, in a mid-sized visualization and research laboratory in central London.

Describing how code works and what it does for the “hackers” of computerized visualizations will help ground the larger order studies of cultural impacts of computerization, as well as complement the more detailed research into the effects of computerization on scientific practices. I am, therefore, going to pass over the much studied effects of software in the workplace (e.g., Flowers 1996; Hughes and Cotterell 2002) and focus upon when technology is the work (Grint and Woolgar 1997; Hine 2006). Staying close to code entails unpacking what occurs at the multiple screens on programmers’ computers. Like a summer holiday spent at home, it is mundane and a little boring to “stay local,” but like the launch of the new web-based open survey visualizer that tense day, there are all the same quite complex operations taking place with code.

With the computerization of data and visualizations, the work with code weaves together many formerly distinct roles. This workflow wraps together the practices of: sourcing data to be visualized; programming to transform and render data visually; visualizing as a supposed final stage. I term these activities “codework.” Merging often sequential stages involved with the generation of visual outputs, I highlight how proliferating web-based visualizations challenge analytic models oriented by paper-based media. Computerized visualizations, such as those in this case study, are openended. They require constant care in the form of coding in order to be sustained on the Internet. Moreover, they are open in terms of their continuing influence in a feedback cycle that plays into both the sourcing of data and the programming involved to render the data visually.

Codework, as an emergent and distinct form of practice in scientific research involving visualization, also blends several sets of binary categories often deployed in visual studies: private/public, visible/invisible, material/immaterial. While these are of interest, I focus upon the manner in which code confounds the binary of creativity/containment and discuss the implications for the political economy of similar visualization labs and the accountability o...

Table of contents

• Cover Page
• Half Title Page
• Other Title
• Title Page
• Copyright Page
• Table of Contents
• List of Figures
• Introduction
• Part I Visualization in the Age of Computerization
• Part II Doing Visual Work in Science Studies
• Contributors
• Index