Something for everyone: Using digital methods to make physical goods*

by

ginger coons

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Information University of

Copyright by ginger coons 2016 Distributed under a Creative Commons Attribution-ShareAlike 4.0 International license Something for everyone: Using digital methods to make physical goods* ginger coons Doctor of Philosophy 2016 Faculty of Information, University of Toronto

Abstract

In this dissertation I draw a link between the purported changes being wrought on society by the adoption of digital production technologies and previous waves of technological change in the production of goods. I use two case studies to provide detailed accounts of methods of production and development which use digital fabrication technologies to negotiate relationships between individuals and standards. The first case study, a collaborative Free/Libre

Open Source Software development project, represents a kind of digital production which creates digital goods. The second case study looks at the digitally-aided fabrication of a physical good: 3D-printed sockets for prosthetic legs. I further argue that we need new frameworks for studying the intersection of the digital and the physical, or the inextricability of the two concepts. Scrutinizing the way mass methods of production almost always account for the needs of some kind of normative user, resulting in mis-fit for non-standard users, I then question arguments about mass-customization as a solution to mis-fit. One of the contentions I advance in this dissertation is that positioning mass production as necessarily harmful and marginalizing, while seeing mass customization as a solution, creates a counter-productive dichotomy. In service of that argument, I draw on historical work about a pre-digital custom industry: dressmakers in the pre- and early-Industrial period. Finally, I trouble the role of the user in production, and contend that, increasingly, the distinction is not between who is a user or a producer, but in which circumstances, and when, one is acting in the role of the user or producer.

ii Acknowledgements

I don't want to write a long list of acknowledgements, but I don't want to miss anyone who deserves thanks. A significant number of people have helped this dissertation into the world, including the usual suspects like my parents (for the moral support), my committee (for the comments, insight and direction), and my friends and colleagues. By name, I need to thank, of course, my supervisor, Matt Ratto, and my committee members, Patrick Keilty and Brett Caraway. But if this were a book, with just a little dedication page, far different from the wall of text expected of the dissertation acknowledgements page, I'd write simply “To Rhonda, who taught me how to schedule.”

iii Contents

1. Introduction 1 2. Theoretical framework 16 3. Methodology 35 4. Custom and mass production 51 5. Interfaces between bodies and standards 97 6. The body in digital production 145 7. The users of custom goods 167 8. Conclusion 190 References 204 Appendices 211

iv List of illustrations

Illustration 1 – p. 55

A drawing of a shirtwaist, printed in the 31 March, 1904 edition of the Tacoma Times.

Illustration 2 – p. 56

Emiline, a lithograph by James S. Baillie of New York, NY (1845), showing the fitted style of bodice that was common at the time and for much of the 19th century.

Illustration 3 – p. 59

The Fitting by Mary Cassatt (1891) depicts a woman being fitted for a dress, a task almost uniformly carried out by the mistress dressmaker herself. It is unclear whether the fitting takes place in the woman's home or in the dressmaker's shop. It is also unclear whether the painting is intended to reflect life in Cassatt's native Philadelphia, or Paris, where she lived at the time it was painted (Weinberg, n.d.).

Illustration 4 – p. 77

LGRU sticker with the "Practice Shapes Tools Shape Practice" slogan. (Photograph by the author.)

Illustration 5 – p. 100

The standard form of a trans-tibial prosthesis, with the socket highlighted. The cosmetic fairing, which serves to make the prosthesis look more leg-like, is not shown. (Illustration by the author.)

Illustration 6 – p. 103

Dependency grid, circa summer/fall 2013. (By the author.)

v List of appendices

Appendix 1 – Glossary p. 211 Appendix 2 – Interview guides p. 212 Appendix 3 – Production colophon p. 213

vi 1 Introduction

*Though the title of this dissertation implies a clear distinction between the digital and the physical in the phrase “Using digital methods to make physical goods,” it is impossible to find hard and fast differences between the digital and the physical, even if that were a desirable outcome. Rather than reify a distinction between the digital and the physical, this dissertation is instead about processes which blur the lines between what we might historically or popularly have thought of as the physical and the digital.

Introduction

This dissertation is about mass-customization, 3D printing, Makerism, the idea of “the digital,” human bodies and embodiment, and the promises often offered by those who argue that the future will be full of customized, digitally-built objects made to individual specifications—or something for everyone. I argue in the next eight chapters that, in order to understand the promises being offered up by advocates of digital fabrication technologies (especially in the service of mass customization), it is necessary to interrogate how such technologies and their uses differ from existing modes of production, and what kinds of interventions they are meant to make in the negotiations individuals face on a daily basis between their selves and the world. Crucially though, before I begin those interrogations, I must first provide some context on a paradigm shift or revolution that is supposed to be taking place: the next industrial revolution, a revolution which leverages digital fabrication tools and a can-do attitude to redefine the way we produce and consume.

1 On June 18th 2014, an event called the White House Maker Faire took place. In conjunction with the event, the President of the United States issued a proclamation, announcing the National Day of Making. The proclamation reads, in part:

Today, more and more Americans are gaining access to 21st century tools, from 3D printers and scanners to design software and laser cutters. Thanks to the democratization of technology, it is easier than ever for inventors to create just about anything. Across our Nation, entrepreneurs, students, and families are getting involved in the Maker Movement. My Administration is increasing their access to advanced design and research tools while organizations, businesses, public servants, and academic institutions are doing their part by investing in makerspaces and mentoring aspiring inventors. [...]

As we observe this day, I am proud to host the first-ever White House Maker Faire. This event celebrates every maker -- from students learning STEM [Science, Technology, Engineering and Math] skills to entrepreneurs launching new businesses to innovators powering the renaissance in American manufacturing. I am calling on people across the country to join us in sparking creativity and encouraging invention in their communities. (Obama, 2014)

Those two paragraphs are an excellent example of how new technologies in general, and desktop fabrication technologies in particular, are often talked about and promoted by politicians, industry, and journalists. Packed into under 150 words, this portion of the proclamation invokes economic growth, democratization of access to new technology and opportunities, personal fulfillment, learning outcomes currently considered valuable, and the idea of a "renaissance" in manufacturing, a field in which the United States formerly considered itself dominant. In large part, it is a story about the ways in which the Maker movement can help the United States to be a successful, globally-competitive country. The idea that the adoption of a new technology can so radically change the economic and social landscape of a country and its people is one which has been frequently trotted out in service

2 of the Maker movement and its attendant rhetoric. As the headline of an op-ed by Sadowski and Manson (2014) cheekily puts it, statements like those from President Obama are very much selling the idea that it is possible to "3-D print your way to freedom and prosperity."

A few months before the presidential proclamation, in a February 2014 post on the official blog of the White House, Kalil and Miller (2014) recounted the story of “Joey Marshmallow,” a young boy who displayed a marshmallow cannon at the 2012 White House Science Fair. They used the story of Joey's marshmallow cannon—and the business card he handed the president, which had the slogan “Don't be bored, make something” printed on it— to contextualize the idea of being a Maker, rather than a consumer (with the two positioned as opposites), and to introduce the White House Maker Faire. They described Joey and his Maker compatriots as “the millions of citizen-makers driving the next era of American innovation” (Kalil & Miller, 2014). This emphasis on a move away from passive consumption, towards active, individualized, entrepreneurial production as something which has economic and social benefits for a nation is one thread in the rather complex weave of public adoption of digital fabrication technologies.

***

PROFIT, a Canadian business magazine, has been covering mass customization for some time. In 2011, an article published in PROFIT proclaimed

But now a new miracle is being performed by progressive firms across the business spectrum: mass personalization. Also known as mass customization, it's the process in which companies allow clients to tailor products—from jewelry to pet food—to their personal needs without paying "custom-made" prices. And it gives companies new opportunities to differentiate their products and claim customers that won't settle for "one size fits all." [...]

Meanwhile, the rise of user-generated content coupled with the proliferation of on-demand, iPod-style purchasing—which allows

3 customers to buy exactly what they want; nothing more, nothing less— has fuelled both the demand and the expectation for customized products. (Beaton, 2011)

Three years later, anotherPROFIT writer drew a connection between mass customization and the newest hot topic in business: 3D printing.

The evolution of 3-D printing will give mass customization another big push. Over the next decade, as the technology becomes cheaper and more sophisticated, 3-D printers will replace offshore suppliers for some product categories, encourage local manufacturing and simplify supply chains. (Castaldo, 2014)

In effect, the message being peddled by this well-read publication1 for small and medium enterprises (SMEs) was that not only will mass customization change how customers choose products, but 3D printing will be increasingly used to produce customized, on-demand products. Castaldo raised some of the points shown in the arguments supporting the White House Maker Faire: the use of digital desktop fabrication technologies will allow small enterprises to locally produce innovative products themselves, rather than dealing with large, off-shore production houses and the onerous overheads of mass production processes.

Though the accuracy of the above statements—and the verdict on how much rapid prototyping is capable of reinvigorating local production—remain to be determined, the articles published in PROFIT and the Presidential declaration are evidence of a growing interest in how post-Industrial nations and their enterprises can use technologies like 3D printing. Providing further evidence, Lindtner and Guimarin (2014) describe the major argument behind Chris Anderson's 2012 boook, Makers, intended as a manifesto for the Maker movement. They say “Making gets people excited (again). It is the story of adventure and of conquering unfamiliar territory to reinvent how technological futures are made today — at its heart it is a vision of technological and social progress. Journalists, scholars, and makers alike have been busy telling this story, joining in on the promotion of making as the harbinger of an

1 Its publisher claims a combined online and print readership of 216,000 (Rogers Publishing, n.d.), a not inconsiderable amount in Canada's relatively small market.

4 industrial revolution” (2014). Such grandiose ideas and motives, pushing the agenda of the Maker as innovator, of the new technology as paradigm shifter, is not unusual for Anderson, a former editor of Wired magazine. However, its adoption as potential policy, or at the very least as feel-good governmental encouragement of can-do, entrepreneurial behaviour, combined with connections to ideas such as mass customization, learning, agency and active citizenship add to the potency of what might otherwise be seen simply as boosterism by tech journalists.

In this dissertation I tackle some of the topics associated with the public discussion and adoption of technologies like 3D printing. In particular, I examine ideas and practices around mass customization, as aided by both software- and hardware-based rapid prototyping and desktop fabrication tools. In so doing, I take the opportunity to ask how the increasing popularity and viability of digital methods impacts the production (and, by necessity, consumption) of physical goods, as well as the interfaces we build between our idiosyncratic selves and our standardized production systems and institutions. In addition to this major question, some other issues arise, such as:

▪ What kinds of constraints do digital methods impose on the production of physical goods? ▪ What ideas and ideals from digital production methods are kept in the application of those methods to physical goods? ▪ How formulaic (or not) are our engagements with digitally-produced physical goods? Are we developing deeper, more engaged relationships? ▪ How do we negotiate the relationships between our bodies and the standardized world? Are digital production methods changing these relationships?

One of my goals in addressing these questions is to critique the ways in which both individuals and institutions are heralding a new industrial revolution while imagining a world in which we can achieve almost perfect consumption, literally something for everyone, a one- size-fits-one model in which new technologies allow us to assuage our alienation from the products in our lives. In part, the tension between the culture of innovation championed in the

5 Presidential declaration and the more effective consumption envisioned in the articles from PROFIT, provides grist to the mill of this dissertation. As something of a counterpoint to such conflicts, another purpose of this dissertation is to engage in a hands-on way with issues of custom production through desktop fabrication, and to provide some much-needed, first-hand descriptive accounts of such activities, absent the can-do rhetoric of most popular accounts on the subject.

In this dissertation I will explore a variety of practices which can be encompassed by terms like “digital fabrication,” specifically in the service of custom production and customization. While a business-oriented view of custom production might position it as an activity undertaken by one party on behalf of another, for pay, I choose to draw a wider boundary around the concept. Instead, I take custom production, and specifically custom production achieved through digital means, as something which can be done either by an individual who is undertaking the task on their own behalf, or something which can be done for others, whether for pay or not. I draw a distinction between custom production—an activity in which something is newly-produced to a custom specification—and after-market customization undertaken by individuals who wish to modify mass-produced goods. However, despite the distinction between custom production and after-market customization, I do occasionally use the term “customization” in referring to the process of making an object custom, in relation to the gathering of specifications for instances of custom production. In the examples set forth in chapters four and five, custom production is undertaken by artists, software developers, medical practitioners and end users. The important similarity in all of these cases is that an effort is being made, through means of digital production practices, to resolve some kind of mis-fit between an individual and a good.

6 A brief note on technological utopianism

In exploring the social impacts of new technologies, wading through a mire of extreme enthusiasm, fired-up futurists, profiteers and technical prophets is unavoidable. The dynamic between excited entrepreneurs and nay-saying academics is a well-clichéd Punch and Judy show. One of the purposes of my dissertation is to bring new empirical, descriptive data on current practices of digitally-aided mass-customization into the conversation. Salvos like the following, from well-known 3D printing booster Hod Lipson (an academic, but with an agenda which prizes technical innovation), show a degree of aggravation on the side of the utopians:

"3D printing technologies will close the gulf that divides the virtual and physical worlds. Of course, a skeptic would quickly point out that the digital and physical worlds already intersect at several points. After all, design and manufacturing processes have been driven by computers for decades. Mass production these days is nearly fully automated (except for the last step—the human-intensive assembly line)." (Lipson & Kurman, 2013, p. 14)

In the war of words which pits people who study the infrastructure of digital systems against people who want to make claims about the potential of new technologies, statements like Lipson's are a calculated way to re-inflate the importance of claims about the world-changing qualities of 3D printing.

On the other side of the spectrum, Sadowski and Manson take great pains to satirically position 3D printing and its attendant boosterism as a sort of false consciousness:

“Are you feeling alienated by the economic structures of society? Don’t you worry about silly things like collective politics, socioeconomic transformation and philosophical reflection. Just fire up your 3-D printer, come on down to the hackerspace, and engage yourself in an “authentic” craft. Or even turn it into a lifestyle – that is, if you have the privilege to access the necessary expensive equipment, space, skills and time.” (Sadowski & Manson, 2014)

7 This reciprocal effort, with promoters and skeptics taking jabs at one another from opposite sides of an issue is a popular theme in the discussion of any new technology. Futurists are always unnecessarily optimistic, nay-sayers and critics are always overly-pessimistic. It is my aim in this dissertation to tread a relatively careful line between blue-sky imaginings about a 3D printer in every home and doom and gloom prophecies about robots taking our jobs. Rather, my interest (and one of my driving forces) in this dissertation is in constructing an argument which questions received knowledge from both sides, while proposing both theoretical and methodological tools for moving forward.

Purpose of the dissertation

In this dissertation I draw a link between the purported changes being wrought on society by the adoption of digital production technologies and previous waves of technological change in the production of goods. I also use case studies to provide detailed accounts of methods of production and development which use digital fabrication technologies to negotiate relationships between individuals and standards. Further, I argue that we need new frameworks for studying the intersection of the digital and the physical, or the inextricability of the two concepts. As what we have tended to call “digital” becomes more a part of life, increasingly without recourse to screens or items that look like computers, it becomes even more important to develop appropriate ways of describing the roles and functions of computational goods. A further theme running through the dissertation is the very modern idea of statistically-determined normativity. Mass methods of production (in which one or few products are made to suit the needs of many or all users) almost always account for the needs of some kind of normative user, whether that user is real, or, more likely, based on an abstract notion of the average. The construction of an average or normative user of a product, baked into the product itself through its design, can have a detrimental impact on marginalized and non-normative users. This problem is a potential starting point for the positive things mass- customization can do. However, one of the contentions I advance in this dissertation is that positioning mass production as necessarily harmful and marginalizing, while seeing mass customization as a solution, creates a counter-productive dichotomy. In service of that

8 argument, I draw on historical work about a pre-digital custom industry: dressmakers in the pre- and early-Industrial period.

The literature with which this dissertation engages comes from a broad spectrum, informed predominantly by science and technology studies, critical information studies, and the digital humanities. However, given the interdisciplinarity of these fields, the work I draw upon could also be characterized as digital studies, communication, surveillance studies, gender studies, among other interconnected areas of the social sciences and humanities. I also make use, as appropriate, of materials from marketing, business history and consumer behaviour. What matters most to me in aligning with particular bodies of scholarship is their utility in tracing the stakes and impacts of new technologies and their attendant systems, as those impacts are felt in the context of the place of the individual in society. Methodologically, I make use of participant observation methods for data collection and grounded theory for analysis. In effect, the empirical components of my dissertation come from observation of projects which involve themselves in the production of custom goods through digital fabrication technologies. This observation is conducted with varying degrees of participation. In one project, I have been a somewhat privileged onlooker, with access to materials and people which would not be publicly available. In the other project covered, I have been an active participant, helping to steer the progress of the project and providing labour in its service. I approach the data from both cases using a framework informed by grounded theory, in which I analyze both primary materials and my own field notes. More specific details on this methodology will be provided in chapter two.

Contribution

The public discussion and adoption of digital fabrication technologies like 3D printing, 3D scanning, and computer-controlled cutting is often polarized, either wildly utopian or staunchly negative. Critical scholarly analysis is emergent and often focuses on studies of the Maker movement (for example, the October 2014 issue of the Journal of Peer Production addresses the theme of “Shared Machine Shops” by publishing a series of articles which offer critiques and recountings of practices in hackerspaces, Makerspaces and Fab Labs). My work

9 fits into this emergent area by examining ideas and practices around mass customization, as aided by both software- and hardware-based rapid prototyping and desktop fabrication tools. In particular, I ask how the increasing popularity and viability of digital methods impacts the production (and, by necessity, consumption) of physical goods, as well as the interfaces we build between our idiosyncratic selves and our standardized production systems and institutions. I achieve this through diverse methods: in two empirical case studies, I use participant observation, semi-structured interviewing and user testing to drive an analysis of two custom fabrication projects, one which studies a Free/Libre Open Source Software project called the Libre Graphics Research Unit, a group aiming to build software tailored to the tastes of artists; the other using 3D printing and 3D scanning to produce custom components for prosthetics legs, in partnership with a rehabilitation hospital in Uganda, an associated NGO, and the research arm of a multi-national software company. These case studies are contrasted against existing ideas about mechanization, the atomization of labour, and mass production, as well as the role of human bodies and minds in the labour process, and against a historical case, looking at the practices of 18th through early-20th century dressmakers.

The increasing use of digital fabrication technologies is important because it addresses technologies which are being adopted more widely (an issue of public concern), and have the potential to illuminate the tense relationship between our conceptions of the digital and the physical (a pressing scholarly concern). Existing, highly instrumental work on mass customization often sees it as merely a benison with few, if any, drawbacks. Using my position at the intersection of design, information studies, and science and technology studies, I address that shortfall by bringing concepts such as the potential violence of categorization, the necessary but often unrecognized embedding of values in designed objects, and the use of technologies to construct users. Because digital fabrication technologies are arguably at the intersection of the digital (immaterial or intangible) and physical (material or tangible) worlds (positing for the moment that those are two different worlds), this dissertation also gives me an opportunity to speak to the similarities between the digital and the physical, and to trouble existing definitions of those two terms. Rather, one of the contributions of my dissertation is that the physical and the digital aren't inherently different (which is, in itself, not a new idea), but that we need to look at the ways in which the modes of engagement within what we typically call “digital” and what we typically call “physical” differ from one another, and how

10 the characteristics ascribed to the two concepts shape the way we apply them. I contend that, in looking at the modes of engagement present across projects which fall into various camps along the traditional physicality-digitality spectrum, there are valuable insights to be found about the materials (whether they are called digital or physical) themselves. A further contribution of this dissertation is a framework for doing the kind of looking I suggest above. I argue that three relatively common concepts from existing literature and theory—the social organization of labour, the material conditions of production, and the role of the user in production—should be taken together as a framework through which to examine projects, processes, and the groups of people carrying them out.

My final contribution, based on my own engagements with the framework I mention above (and detail further in later chapters of this dissertation), is that there are crucial differences between custom production as practiced by the artisans described in my case studies and what is sometimes called “parametric customization,” or the process of allowing a user/purchaser to make choices about the design of an object, based on a set of options given within constrained parameters. I argue that parametric customization does not provide for significant agency on the part of the user, operates within a constrained, unchanging industrial system, and mimics mass production in most ways. This contention is important at present because of the rise of such parametric customization systems, under the heading “mass customization” and the positive traits ascribed to such mass customization. In addition to suggesting that parametric customization is closely tied to digital fabrication, I also argue that many of the benefits to the customer of other forms of custom production (benefits which include attention to fit and appropriateness, as well as the ability to have a meaningful dialog with the producer) are implied to be present in parametrically customized objects, but may well not be. This situates arguments about parametric customization very much in the same area as those about the emancipatory power of 3D printing. Both are often given a veneer of user power through the application of idealistic or romantic ideas about the value of pre- Industrial or pre-mass production. However, I also highlight some of potential of customization through digital fabrication through one of my case studies, which uses 3D printing as an aid to the work of craft practitioners.

11 Chapter outline

This dissertation takes the following structure:

In chapter two, I provide grounding in the literature and theory relevant to my dissertation. In part, that grounding introduces concepts and authors I return to and use in later chapters, but it also situates my work in ongoing scholarly discussions. Those discussions include the nature of “the digital” and whether or how it might be separate from the physical (as I allude to in the asterixed portion of my dissertation's title); the interconnected issues of standards, categories and residuality; the role of the body in discussions about materiality; and the fractionalization of labour under scientific management. Taken together, these themes form the foundation on which the remainder of my dissertation is built.

In chapter three, I outline the methodology used in conducting my major case studies. Drawing on a variety of methods and lenses, including grounded theory and participant observation, chapter three details the nuts and bolts of the methods used. Chapter three also offers a brief introduction to the two case studies. At the end of chapter three, I make sense of how the cases were handled, and how the carrying-out and interpretation of those cases has been influenced by a fairly broad cross-section of both scholarly and popular literature, as well as my own past experiences in F/LOSS projects. I discuss issues like the decision to include (or not) new data in the LGRU corpus of documents. In this chapter I also briefly introduce some of the substantive issues encountered in creating sociotechnical interventions in sensitive contexts. Chapter three also provides a rationale for what were, at times, difficult decisions, and decisions which have ultimately and necessarily altered the form the final dissertation has taken. In considering methodology in a nuanced way, the chapter offer both a rationale for my own work and, potentially, a path forward for others conducting multi-sited research which touches upon both new technologies and marginalized groups.

Chapter four addresses the differences between custom and mass production processes, while contrasting relevant historical material from the time of the Industrial Revolution against the study of a recent software development project. Using data collected from an initiative called the Libre Graphics Research Unit, and from selected other Free/Libre Open

12 Source software and hardware projects, this chapter draws parallels between historical changes to the production of goods, exemplified by the work of custom dressmakers in the 18th through early-20th centuries, and to current initiatives in the production of software and electronic hardware. This chapter contrasts a historical perspective of production against a recent initiative in which bespoke software was built. By tracing the trajectory of custom-to- mass production, I build an argument about the changing place of the end-user in production, which I then contrast against and apply to the LGRU case. Because the LGRU case is one in which I was largely not a participant, it offers the opportunity to examine the practices of others, in a contemporary setting, and with a highly accessible set of documents and stakeholders, which serves as preparation for the case detailed in chapter five. As such, the LGRU study and the dressmakers serve as the basis of a framework which I then apply to the prosthetics project detailed in chapter five.

Chapter five, under the heading “Interfaces Between Bodies and Standards,” details a project which uses 3D scanning, 3D modelling and 3D printing to develop a new method for the production of sockets for prosthetic legs. Grounded in my own participant observation in the project, this chapter both describes the project, its development and outcomes, and also engages in discussions around normativity, bodies, and the problems of implementing technological systems for marginalized groups. The second of the two present-day empirical cases in my dissertation, the prosthetics project serves as a site for the close examination of decision-making, organization-building and stakeholder relationships in a project which seeks to use desktop fabrication technologies for custom production. Because this chapter documents a project in which I was a participant and, at times, a decision-maker, it tackles a huge corpus of data on how a desktop-fabrication-for-customization project was carried out. This chapter uses that wealth of data to ground further arguments about the ways in which desktop digital fabrication tools are starting to be used, and how such implementations line up with or go against the promises and arguments surrounding the emancipatory potential of desktop fabrication and mass customization.

13 Chapter six takes the discussion of the body's relationship to digital production further. In exploring the presence of the body in digital production, as well as the relationships between human bodies and digital production practices, I contrast the idea that bodies are typically removed from automated digital labour against the notion that in manual automated labour, the mind is removed. Based largely on existing literature, this chapter also takes up some of the data from my case studies to question the place of human labour in digital production practices. This chapter carries on some of the themes first advanced in the empirical work of previous chapters, but gives further weight to such ideas by drawing on existing conceptions of bodies and labour. In engaging with literature on labour in general, as well as with what is specifically called “digital labour” this chapter asks questions about the nature of the body in labour, the nature of the body in digital labour and now, the nature of the body in desktop digital fabrication. Engaging with the trajectory that others have previously indicated (eg: Zuboff, 1988) in which the atomization of labour takes both embodied knowledge and exertion away from the labourer, this chapter poses the question “do desktop digital fabrication tools really de-alienate us from the products of our labour? And under what circumstances?”

While chapter six focuses mainly on the body in labour, chapter seven picks up a different thread first presented in chapter two and expanded somewhat in the empirical chapters: the relationship between new production technologies and the end-users of customized products. This chapter draws both on some of the historical and F/LOSS vignettes from chapter four, and on emerging uses of mass-customization through digital fabrication technologies. The major question driving the chapter is something of a companion to chapter six's question. In this case, I ask “does mass customization introduce user agency and choice into the production of goods, or does parameterization of goods merely represent a broadening of possible consumption, without new chances of intervention?” To that end, I further elaborate on the idea of parametric customization, suggesting that it is the form taken by much of what is currently sold as mass customization.

14 Chapter eight offers some conclusions about the subjects at hand. Linking findings, stories, and descriptions from chapters four and five, while bearing in mind the more substantial theory-building present in chapters six and seven, in this chapter I synthesize the arguments constructed elsewhere in the dissertation. I also make a case for the significance of the dissertation, both as a contribution to ongoing scholarly arguments, and as a way of thinking about a set of technologies and practices which are gaining new prominence. Finally, in this chapter I tie up a variety of loose ends, bringing back some of the contentions from the introduction and explaining their relevance in relation to both the empirical and theoretical chapters. With these issues in mind, the chapter closes by identifying areas which might be examined further and on which I might base future research.

15 2 Theoretical framework

Introduction

In this chapter, I situate my dissertation in ongoing discussions—both scholarly and popular—about the revolutionary and emancipatory potential of digital fabrication technologies; the nature of “the digital;” the place of the body in the digital world; the limitations of categories; the fractionalization and alienation of labour through scientific management practices and attempts at embedding knowledge in machines; and the current state of mass customization. All of these areas colour the arguments I advance later in this dissertation. As such, I use this chapter to outline the topics and concerns with which I wish to engage, and to provide some background useful to understanding both what I contend in later chapters and how those contentions fit into existing scholarly and popular debates.

In order to provide an adequate introduction to the areas with which my dissertation is concerned, and to give a sense of how I will approach those areas, I first treat the popular idea that we are currently experiencing another industrial revolution, which—in addition to providing a financial boon to businesses—promises to de-alienate and emancipate individuals; following on from the rhetoric about the emancipatory potential and revolutionary nature of digital fabrication technologies, I then provide a small overview of major issues in the fractionalization of labour and alienation from the means of production. Because fractionalization relies on routinization and standardization, I then situate my work in discussions of standards, categories, and problems of mis-fit. I then take a detour into the concept of “the digital,” which is a hugely thorny problem on its own, but which forms an

16 important cornerstone for discussions of digital fabrication. Because many of the case studies in this dissertation are related to the way we negotiate the interfaces between individual bodies and mass standards, I then move on to an overview of the body as it relates to both standards and digitality. To an extent, the tension in the relationship between the body and the digital is about the pain of conforming to the world, and the science fiction promise of bodilessness. I end the chapter with an overview of current ideas about custom production, specifically digitally-aided custom production. I recognize that each of these concepts is large enough and complex enough to be its own dissertation. My goal in this chapter is to provide a brief overview of some of the more substantive issues as they relate to my dissertation, and to place my dissertation in ongoing discussions within those concepts and areas. Further sections of this dissertation will engage with and elaborate on the concepts introduced here.

The nth industrial revolution

One of the biggest common threads in discussions of desktop fabrication and the maker movement is the idea that an industrial revolution is currently underway. As an article in The Economist puts it,

“As manufacturing goes digital, a third great change is now gathering pace. It will allow things to be made economically in much smaller numbers, more flexibly and with a much lower input of labour, thanks to new materials, completely new processes such as 3D printing, easy- to-use robots and new collaborative manufacturing services available online. The wheel is almost coming full circle, turning away from mass manufacturing and towards much more individualised production. And that in turn could bring some of the jobs back to rich countries that long ago lost them to the emerging world.” (Markillie, 2012)

The idea that desktop manufacturing represents another industrial revolution has gained such traction that articles are being written refuting that suggestion (for example, Chalmers, 2013). Whether or not desktop fabrication represents an industrial revolution, a legion of writers and organizers are determined for it to be at least some kind of revolution. The potential revolution

17 is represented by the Maker movement (helped along by Make magazine and global Maker Faire events), in which not only production methods apparently change, but individuals are empowered to do for themselves things which previously seemed impossible, or at least prohibitively expensive or difficult. As Sadowski and Manson (2014) somewhat sarcastically put it, “What’s not to like with a revolution that — according to tech gurus, media and politicians alike — is seemingly so democratizing, empowering and profitable?” That idea, that the rise of desktop fabrication is being billed as a revolution, whether industrial or social or both, is one of the foundations of this dissertation, and one of the sites at which I engage with existing discussions about labour, alienation and fractionalization.

Of course, it is not new to think of technology as inherently empowering. Barbrook and Cameron remind us that Marshall McLuhan also exhibited such technologically deterministic ideas, and "preached the radical message that the power of big business and big government would be imminently overthrown by the intrinsically empowering effects of new technology on individuals" (1996, p. 48). They suggest that such ideas—including the idea that decentralization of speech would lead to decentralization of power—gave rise to what they refer to as “the Californian Ideology,” which “promiscuously combines the freewheeling spirit of the hippies and the entrepreneurial zeal of the yuppies" (p. 45). Langdon Winner (1986) makes a similar argument to McLuhan about the relationship between technologies and social forms, suggesting that nuclear reactors require centralization and control, while solar power is, by its nature, more conducive to decentralization. Arguments about digital fabrication take a similar direction: the decentralization of production (through individual ownership of tools like 3D printers) will decrease individual reliance on centralized modes of production, allowing individuals a greater degree of self-reliance. This move is seen by many as emancipatory, putting the tools of production into the hands of individuals.2 Because debates about the emancipatory potential of technologies are founded in existing concerns about centralization, alienation and lack of control of individual labour, I now move to a brief review

2 It might be instructive at this point to look at a concept from Ruth Schwartz Cowan (1983), who argues that housework is a decentralization of tasks which, if given over to industry, would be centralized. She offers the examples of laundry, meal preparation and child care. Industrial versions of these services take advantage of economies of scale in a way that individuals cannot, and might even be capable of providing a better service than a single housewife (or increasingly, a cooperative, task-sharing family), given the ability of industry to specialize. We could view the home 3D printer as an equivalent, but opposite move: turning more of the work of the manufacturing industry into housework.

18 of Taylorization, atomization, and the division of labour as a means of alienating individuals from the products and processes of their labour and, to a lesser extent, the goods they consume.

Alienation and fractionalization of labour

Though proponents of digital fabrication technologies embody—to paraphrase Barbrook and Cameron—both hippies and yuppies with regard to their attitudes about capital, I am concerned less with the applications of 3D printing and its kin to the improvement of the bottom lines of businesses, and more concerned with its role as a potential de-alienating force. While this dissertation does admit the experience of both the user and the producer, the user is admitted in her role as a contributor to production, rather than as an entirely passive consumer of a good produced at a distance from her. As such, I will focus here more on the idea of alienation as it relates to the alienation of the worker from the object of their work. Beginning with that alienation, I will then move on to the fractionalization of labour, and end this section with a brief sojourn into the idea of knowledge as embodied in workers versus knowledge embodied in objects or documents.

Marx refers to “the product of labor” as “labor which has been embodied in an object, which has become material: it is the objectificationof labor” (1844). Because the object is an embodiment of the worker's labour, Marx argues that “the more the worker spends himself, the more powerful becomes the alien world of objects which he creates over and against himself” (ibid). As the worker expends his labour in the production of objects and in the service of an employer (who ultimately owns the objects produced through the worker's labour), he loses the resources he does possess and can profit from: his time and energy (ibid). Marx further defines the alienation of labour:

“What, then, constitutes the alienation of labor?

First, the fact that labor is external to the worker, i.e., it does not belong to his intrinsic nature; that in his work, therefore, he does not affirm himself but denies himself, does not feel content but unhappy, does not

19 develop freely his physical and mental energy but mortifies his body and ruins his mind. The worker therefore only feels himself outside his work, and in his work feels outside himself. He feels at home when he is not working, and when he is working he does not feel at home. His labor is therefore not voluntary, but coerced; it is forced labor. It is therefore not the satisfaction of a need; it is merely a means to satisfy needs external to it. Its alien character emerges clearly in the fact that as soon as no physical or other compulsion exists, labor is shunned like the plague. External labor, labor in which man alienates himself, is a labor of self-sacrifice, of mortification. Lastly, the external character of labor for the worker appears in the fact that it is not his own, but someone else’s, that it does not belong to him, that in it he belongs, not to himself, but to another. Just as in religion the spontaneous activity of the human imagination, of the human brain and the human heart, operates on the individual independently of him – that is, operates as an alien, divine or diabolical activity – so is the worker’s activity not his spontaneous activity. It belongs to another; it is the loss of his self.” (Marx, 1844)

The alienation of labour, characterized by lack of control and expenditure of one's only salable resource in order to subsist, has moved—since Marx's time—to encompass labour which does not simply exploit the physical capacities of workers, but also their mental capacities. Braverman (1974) provides the example of clerical workers, an occupational category which transformed from—in its early days—a small group of craft workers engaged closely with the owners of businesses in the management of such businesses, to an army of functionaries perceived by the managerial class as in need of rational management and control. It is this kind of control that writers like Matthew Crawford decry when they somewhat glibly refer to “affable complaisance” as a key feature of the modern knowledge worker (2009, p. 9).

20 Those who write about the history of scientific management (eg: Haydu, 1988; Braverman, 1974; Bahnisch, 2000) frequently describe it as a process of removing planning tasks from formerly autonomous workers and placing those tasks into the hands of managers, which was at its peak3 at the turn of the 20th century. In removing planning tasks from the hands of workers, scientific management both reduces the need for skill in the worker (thus reducing the cost of the worker as well) and reduces the potential for the worker to exercise autonomy in ways which might interfere with the goals of the employer. Through methods which centralize planning and break formerly complex tasks down into smaller, more routinized tasks, scientific management (as espoused by people like Frederick W. Taylor) extracted expertise from craft workers and placed it into the hands of management. Other tactics, like increased supervision and a system of payment contingent on piecework, assisted in fitting workers into a rationalized shop (Haydu, 1988). When a skilled labourer is difficult to replace, has knowledge not possessed by the factory owner, and is capable of colluding or combining with other workers, the ability of the employer to extract value from labour is somewhat contingent on the willingness of labourers to cooperate in that goal. Dyer- Witheford lets us in on a lesser-known facet of Charles Babbage's work (beyond his status as the inventor of the programmable computer), namely an interest in using mechanisms to “reduce and eventually eliminate from production a human factor whose presence could appear to the new industrialists only as a source of constant indiscipline, error, and menace” (1999, p. 3). The use of machines to supplant human labour in industrial production ultimately leads to a related change which took place roughly concurrently: the extraction of expertise from skilled workers, and its embedding in machines and documents.4

3 Or at least its peak of popular recognition. Braverman (1974) argues that we no longer talk about scientific management not because it no longer exists, but simply because the dominant paradigm does not need to have a name. 4 Scientific management didn't just stop in factories and offices, but also made its way onto bodies, and especially those of women: "Recasting corset fitting as a science in the 1920s relied on the widespread knowledge and faith in the practices of scientific management. The transformation of industrial work in the early twentieth century through implementation of the concepts of efficiency and rationalization, as well as the turn to technology for problem-solving promoted de-skilling of workers, and thus loss of an important basis of their power in the workplace. Utilizing the ideologies of scientific management, corset manufacturers transformed the consumption experiences of saleswomen and their customers when they bought, sold and wore corsets. While this strategy sought to keep women customers bound in corsets, it did, at least temporarily, give corset saleswomen a measure of new status and prestige. However, women's bodies were literally the vehicle for the successful shifting of scientific management ideologies from the workplace to the marketplace and the home" (Fields, 1999, p. 370).

21 Dyer-Witheford tells us that, though the deskilling at the heart of scientific management was “at first attempted primarily through organizational innovation” it was “subsequently mechanically embedded in the Fordist assembly line” (1999, p. 73). Though perhaps that is true of scientific management specifically, concerns over where knowledge resided existed some time before the advent of Fordism. In 18th century Britain, the export of tools used in the textile manufacturing and metalworking industries was prohibited by law, roughly concurrent with two Acts of Parliament intended to “prevent the emigration of skilled artisans” (Musson, 1972, p. 21). Keeping artisans in Britain kept their labour and expertise from being exported to competing industrial interests on the Continent, while keeping the tools of their trade from export prevented both the use and the reverse-engineering of mechanisms which were considered to give Britain a competitive advantage (Musson, 1972). The idea that knowledge can be embedded in objects has since come into its own. The immutable mobile, for example, gives us a way of approaching objects which carry knowledge and expertise. Law describes immutable mobiles as including “instruments, devices and technologies which also hold their structure when shipped from one location to another” as well as textual vehicles, like scholarly or technical papers (2009, p. 8). Although it is worth noting that the immutable mobile does not exclude humans from its ranks (ibid). Radder (2003) similarly offers the idea of “thing knowledge”, or knowledge which inheres in objects and suggests that “[i]nstruments, not beliefs, are the carriers of thing knowledge” (p. 43) and that such knowledge can be transferred to objects while still residing in the individuals who produce it. This take on the embedding of knowledge in objects concurs with British Industrial era policy on the export of expertise. However, both the immutable mobile and thing knowledge fail us somewhat in drawing a distinction between the (tacit) knowledge embodied in an object and the (explicit) knowledge represented by systems of scientific management. The task card of the scientific manager can be an immutable mobile just as much as a skilled metalworker or the parts of a loom.

22 Standards, categories, and things that don't fit

The process of embedding knowledge and expertise in objects and documents (and divorcing it from skilled workers) is one which carries through in the development of formalized standards. Rationalization, classification and standards go hand-in-hand. Gandy reminds us that, on a basic level, “classification involves the assignment of individuals to conceptual groups on the basis of identifying information” (1993, p. 16) and that “classification always includes an assessment” (1993, p. 17). Such assessments, he argues, when carried out in aggregate, are used “to determine norms” (ibid). Busch concurs, suggesting that “[s]tandards are the rules by which we are told we should live, and the range of possibilities presented to us when we make choices” (2011, p. 24). Standards, in a social sense, are the rules individuals in a society are expected to live by, determined by assessing what is normal or common.

Standards in another sense, of course, are documents which set forth technical norms for compliance or interoperability, agreed-upon by groups of concerned stakeholders. In this sense of the word, O'Connell (1993) offers us the example of the legal volt, a standard which inheres in a set of batteries, carried in a case and certified to be a reference against which others can ensure they have achieved a standard unit of measurement. Though embodiments of standards (the movement of which O'Connell refers to as the “creation of universality by the circulation of particulars” [1993, p. 129]) like the legal volt or the International Prototype Kilogram are determined through centralized authorities, they have many of the same characteristics as less formal social standards, acting as norms to which others (whether they be instruments of measurement or individuals in a social structure) can be compared. Standards of the technical variety can encompass those that are widely adopted and used in day-to-day life around the world, like the metric system, and can also be more esoteric, used to gain credibility within specialized groups, like ISO's process standards. Regardless, technical standards exist as a metric by which other attempts can be measured.

But classification is imperfect. It is impossible to entirely account for the diversity of the world in a constrained set of standards or categories. This, Star (2010a) tells us, is the purpose of categories like “other,” which she names “residual categories.” She argues that “simplifications reduce people to one thing, and residual categories are often a way of forming

23 the discarded complex” (2010a, p. 152). Anything that does not fit into the explicit categories offered is encompassed by such a residual category, intended to account for those who do not fit into the common norms.5 In cases like the classification of race, gender or sexual orientation, such categories can wield huge amounts of power: if a census or other formal information-gathering tool does not represent you, relegating you to the “other” category, what hope is there of having your interests accounted for in policy decisions which make use of the data collected? Standards also hold other marginalizing powers. Pargman and Palme relate the story of the town of Hörby in Sweden, which suffers from the lack of an “ö” in the ASCII character set used for domain names on the internet. The substitution of an “ö” for an “o” causes embarrassment, as “part of the name, -by, means village or hamlet in Swedish. Unfortunately, the meaning of the Swedish word hor is adultery or fornication” (2009, p. 177). In this case, the marginalizing power of a standard developed in the English-speaking world is in its ability to alter meaning and prevent those outside of its world from expressing themselves accurately.

Standards, whether they be social norms or technical documents, both articulate a view of the world and have enforce such views. Social norms and standards can perhaps be resisted, though those which formally structure societies have more enforcing power. Selecting “other” on a tick-box-based form marks someone as not belonging to an accepted categories and, because the boxes are intended to reflect reality, most likely relegates them to a statistical minority. On a self-administered paper form, such decisions can be circumvented: leave the field blank, write in an unauthorized answer. The analog nature of the paper form leaves room to—if not exert power—circumvent or disagree with the question or its formulation. A key distinction in this dissertation, which I ground next, is how the digital might or might not be conceived as different. In the case of categories, one might argue that circumvention is more difficult, that dissent has a more binary embodiment. However, that characterization reifies the idea that the digital is distinctly different from the analog or the physical in a straightforward way. The next section lays out the groundwork on which trouble that assumption.

5 Conversely, one might try to fit into norms and standards through tactics like—in the case of fitting into the norms of fashion—the use of constrictive undergarments like the corset. As Fields (1999) argues, through the use of charts and classifications, women shopping for corsets in the early 20th century were encouraged to find fault with their bodies, as “[m]anufacturers and retailers colluded in subjecting women to the scrutiny and discipline of scientific rationalization” (p. 373).

24 “the digital”

The concept of “the digital” is popularly taken as somehow separate or discrete from the world we knew prior to its introduction. Rather than being simply another medium, as physically-embedded as any other, it is treated as both a profound game-changer and as something apart from the physical world. This distinction is reified not only in popular writing (for example, design website PSFK offers an article in which it conducts some trend spotting in relation to what it calls bringing “digital interactions into the physical world” which roughly amounts to enacting online tropes in “real life” [Gonzales, 2014]), but also in the implementation of policy and responses to problems which arise in the apparently separate digital world (for example, the Government of Canada has published a “Cyber Security Strategy” [Public Safety Canada, 2014] and the Department of Justice has convened a cybercrime working group which has, among other things, prepared reports and recommendations on the issue of cyberbullying [Department of Justice, 2014]). I take a position which has become fairly common in a number of intersecting academic fields (the digital humanities, media archaeology, some quarters of information studies and human- computer interaction, to name a few), that the digital is not another world, discrete from everything else. Instead, my view of the digital is coloured by work from scholars like Kirschenbaum, Hayles, and Dourish, which takes for granted the physicality of the digital and treats the activities which take place in so-called digital environments as part of a complex system of infrastructure, objects and practices which are entirely part of the “real world.” However, I use the popular dichotomy between the digital and the physical as a useful point of tension and leverage and as an area for investigation.

The digital has some corollaries which need to be defined somewhat before going further. Chiefly, the concepts of “digital goods,” “digital fabrication” and “born-digital” are at issue.

A digital good, popularly, is something which is made up of bits and bytes, or ones and zeroes, or off and on. Digital, against analog, is something made up of a set of discrete states, against something which exists on a continuum. The common distinction is between bits and atoms, per Negroponte (1995). Of course, as Kirschenbaum, Blanchette, Hayles, and others

25 show us, every bit is an atom. For practical purposes, we might first suggest that the digital must be something relating to computers, made by computers, or displayed by computers. But of course, computers have existed in a huge variety of forms in the last century and a half, being anything from thought experiments to specialized office clerks. We might instead, then, suggest that the difference between analog and digital is the difference between a machine which works mechanically and one which does not. But that distinction doesn't hold true either, as something like a mechanical punched-card reader might be considered a digital mechanical computer. Perhaps the difference between an analog computer and a digital computer really has more to do with how changeable the two are, rather than what they're made of. Perhaps a general-purpose computer is the foundation on which the idea of the digital can rest. Such a definition does not, however, exclude the clerks or the thought experiments, both of which are fully capable of running new programs. The ability to be reprogrammed, then, is not an exclusive trait of digital computers. Is a punched card digital, as I suggested above? It seems on the face that it should be, as its programming relies on the distinction between a punched-out hole and un-punched portion of card. But surely it can't be entirely digital, because there exists the possibility that a hole might be imperfectly punched, which would make it neither a one nor a zero. A child filling in a bubble wrong on a Scantron sheet is an example of an imperfect paper-based digital computing medium. We can view these imperfections as bugs, or we can view them as possible departures from systems which could otherwise be considered digital. On or off, one or zero.

Blanchette (2011) describes what he calls the "purported independence from matter" of the digital by saying that "[a]s a mere collection of 0s and 1s, digital information is independent of the particular media on which it is stored—hard drive, optical disk, etc.—and the particular signal carrier[s] which encode bits, whether magnetic polarities, voltage intensities, or pulses of light" (p. 1042). Anyone who has ever made a web page or written a program will be familiar with how dependent digital goods are on their carrier medium. It is not for nothing that the programming language Java, developed in the early 1990s, was billed as a universal programming language, something one could “write once, run anywhere” often abbreviated as “WORA” (Langley, 2002). It is also not for nothing that developers of websites and applications keep whole stables of devices in order to test the compatibility of their work

26 across platforms. And, as Kirschenbaum argues, the substrate on which a digital program is housed is actually quite unique, with two different disk images of the same game containing different information (2008). In short, the idea of the digital is thorny. But it would be somewhat tautological to define the digital as that which is routinely called “digital.” Instead, broadly and for the sake of convenience, I will take the digital to be forms of media facilitated by contemporary computers. In this sense, a website can be thought of as digital, but a ream of fabric from a 19th century Jacquard loom is not. However, a ream of fabric produced by a computer-controlled loom might be considered the output of a digital production process.

Given the extreme complexity of defining a digital good, the idea of a good being “born-digital” could present a morass of definitional problems. For the moment, I'll advance a tautological explanation. The characterization “born-digital” applies to a vast collection of processes and tools, which can broadly be thought of as digital goods produced originally using digital tools. This category goes deep, in that we can consider such massively complicated products as computer operating systems to be digital systems concerned with the production and management of digital goods. Other examples, such as certain 3D modelling software, the mailing lists of Free/Libre and Open Source Software projects6, and version control systems, are somewhat smaller in scope, but rich in application. In libraries and archives, the term “born-digital” is applied to holdings which are created using what we currently think of as digital means. From this kind of pragmatic approach, a born-digital good 6 F/LOSS, Free/Libre Open Source Software, is an inclusive term for a whole family of software development and distribution practices and ideologies. The Free/Libre Software and Open Source Software movements/communities share some practical similarities and have a number of ideological differences. Crucial to the narrative of this dissertation is that all F/LOSS prizes the idea that source code, the human-readable code in which software is written, should be available to all users, who should also be free to modify and re-distribute it. This holds true of both Free/Libre Software and Open Source Software. The primary differences between software termed “Open” and software termed “Free” or “Libre” run along ideological lines. As Richard Stallman, one of the founders of the Free Software movement puts it, “[o]pen source is a development methodology; free software is a social movement. For the free software movement, free software is an ethical imperative” (2009, p. 31). All of this is to say that the concept of Open Source is often used pragmatically, while the use of the term Free Software, and the adjacent term Libre, implies a fundamentally political viewpoint which puts the idea of user freedom at the centre of its thinking. In practice, possibly because of this distinction, the idea of Free/Libre software is often associated with activism and grassroots applications for software and its development, while Open Source is considered a business-ready mode of software development and distribution. For the purposes of this dissertation, however, all of these factors are collapsed into one concept. To me, there is something inextricable in the relationship between the Free/Libre Software movements and the Open Source software paradigm. What is crucial to the understanding of the F/LOSS project covered in this dissertation is the idea of collaboration, agency and free distribution of software. In this sense, the important ideas are that a variety of users and developers (and user-developers) can get together to work on software together, to mutual benefit, and that the software they produced can then be released into the wild, in a way which permits others to make use of the codebase that has been developed. That is the functional and ideological definition of F/LOSS taken in this dissertation.

27 differentiates itself from any other digital good by not being a reproduction of, we might assume, a born-physical good. An e-book typeset digitally might be born-digital, while a scan of a Gutenberg Bible or of Blake's Poetical Sketches would not be. Born-digital goods are digital goods produced, apparently exclusively, through digital means, whatever “digital” does mean.

Digital fabrication, then, is a process of using a computationally-controlled machine to produce an artefact. Crucially, such computationally-controlled machines take their guidance from files generated using specialized software on what we currently think of as computers. Returning to the example of the Jacquard loom, we might say that a 19th century loom controlled by punched cards or some kind of mechanical assemblage is computationally controlled, because it is controlled by a mechanical computer, but it is not performing digital fabrication. Instead, it is performing some kind of computational (but not digital) fabrication. In this sense, we can also return to specialized clerks. A team of female clerks calculating star charts could be said to be producing those charts computationally, but for our purposes here, they are not producing them digitally. To be clear, this distinction is totally arbitrary and based on the idea that, when we talk about new and supposedly world-changing technologies, it isn't rhetorically helpful to mention that such technologies have not only past precedents, but also previous implementations and definitions. The line between what we might call “computationally-produced” artefacts and “digitally-produced” ones is vanishingly small and entirely predicated on an idea that the computers we use today somehow represent a sharp break from the computers, both mechanical and organic, that performed our computational tasks over a century ago. However, for the purposes of this dissertation, the rhetorical context is important. When a business publication writes an article about how digital fabrication will change everything, the discussion is normally about current developments in computationally-aided production. At the risk of sounding glib, I'd hazard to say that few of the readers of publications like Fast Company or PROFIT are interested in the Jacquard loom, except perhaps as a piece of history. Because of this rhetorical distinction, I use the term “digital fabrication” to describe a set of production processes which make use of contemporary computers, very much in the spirit of my above working definition of the digital.

28 The body in the digital

The digital, when posited as something separate from the physical, becomes a playground for those who wish to deny the body. In fiction, we see, over and over, the trope of the elimination of the body, of the body replaced by disembodied minds, living in digital realms. Whether it is the singularity (per Vernor Vinge, among others) or the Matrix (Silver, Wachowski & Wachowski, 2001), fictional accounts of the development of computation often end in human minds becoming one with the digital, and discarding or subordinating bodies. This subordination of bodies comes, perhaps, from the idea that information can beat matter. Dyer-Witheford (1999) describes knowledge's triumph against matter as one of the key information age conceits. Hayles, too, suggests that “[t]he great dream and promise of information is that it can be free from the material constraints that govern the mortal world" (1999, p. 13). Of course, as we see in debates about the materiality of the digital, information is nothing without the infrastructures on which it lives. Even the Matrix cannot be populated without the vats. Why, then, is the conception of the bodiless mind so popular?

It has frequently been suggested that the unruliness of the human body poses a problem to rational views of the world. Foucault (1978) argues that capitalism “would not have been possible without the controlled insertion of bodies into the machinery of production and the adjustment of the phenomena of population to the economic processes” (p. 141). The body (and the reproduction of working bodies) is a part of production and needs to be subordinated to the process. Maddalena and Packer argue further, suggesting that the human body has often been treated as “an imperfect medium – an intermediate medium, a way-station on the teleological track to perfect transmission, to the disappearance of the human” (2015, p. 97). The human as an intermediary in a technological system, a disciplined part of a machine, is also a common feature in scientific management. Indeed, as we saw above, Babbage imagined machines as tools for removing the unpredictable human from the production process. But Babbage's take is profoundly Industrial. Hayles (2006) offers us a description of the posthuman: “the posthuman in its more nefarious forms is construed as an informational pattern that happens to be instantiated in a biological substrate” (p. 160). This conception of the posthuman as information trapped in a body resonates with both its antecedent, the Cartesian mind-body divide—in which the mind rides along in and controls a physical vessel,

29 but is somehow separate from it—and modern equivalents, in which the needs of the body are treated as inconveniences to be overcome. This seems to be increasingly the case as knowledge work overtakes physical labour, and parts of the body which were previously essential to work now become surplus to the needs of capital. A computer programmer working for equity at a Silicon Valley start-up might well drink Soylent7 to avoid the time loss and cognitive load entailed in the process of acquiring and eating a meal. This choice positions the work of the mind (and the eyes, and the fingers tapping the keyboard) as something which is unprofitably interrupted by the exigencies of the body. Once again, the body is an intermediary, but this time, it is the intermediary between the ideas in the mind and their execution on a digital substrate.

In the sense that the body is treated as unruly, with its needs representing a hindrance to the work of the mind, the primacy of the screen-mouse-keyboard configuration in knowledge work sharply curtails the parts and motions of the body which are called upon in doing work. Because the important aspects of the work are supposed to inhere in the mind of the worker, the body ceases to be something which brings value to the work, and instead might be seen as a hindrance. Where the work of the Industrial era made use of the bodies and energy of workers, the intellectual labour of the post-Industrial era wants only the contribution of the worker's mind and the parts of the body necessary for interfacing the mind with a computer. The ideal worker in such a case might well be the vat-person wired into the Matrix, or the mind uploaded into the cloud, reducing the friction represented by the needs of the body. But Hayles (among others) rebuts the suggestion that the body can be seen as separate from the mind. She suggests that the concept of “the body” has often been erased or rendered manageable and predictable through its description in idealized or essentialist representations. However, she further suggests that the concept of embodiment is capable of re-inserting lived experience (which necessarily involves individual bodies) into our conceptions of human activity. She invokes the idea of habits situated in our bodies, an “inscribed practice,” which is “an action that is encoded into bodily memory by repeated performances until it becomes habitual” (1999, p. 199). She invokes the work of Hubert

7 Soylent is a comparatively new entrant to the meal-replacement market, purporting to have all of the micro- and macronutrients necessary to sustain a human. The promise of Soylent is that it takes away the concern for balancing nutrition and leaves its users able to enjoy food as they wish and choose, but to not depend on food for nutrition.

30 Dreyfus, who argued that such embodied practices could not be divorced from the human body and thus could not be replicated by computers. The body at issue in Hayles's conception of embodiment is one which carries with it individual specifications, experiences, and embodied knowledge and habit.

But the body has a special place in custom production. The body, as an individual body rather than an idealized one, is seldom standard in any respect, and never standard in all. With requirements, proportions, measurements and uses varying from person to person, objects of the body are often those which seem to break out of standard solutions and require reckoning or attention: the garment that doesn't fit quite right, the golf club that isn't the optimal length for one's swing (Zipkin, 2001). The intersection of the body and the standard is, first, where the idea of custom production comes in, and, second, where some tension then arises between the capabilities of digitally-aided custom production practices and the non-standard individuals they are meant to serve.

What does it mean to be custom?

One of the assumptions of custom production is that a mis-fit between a user and a good is a problem. Regardless of the exact nature of the mis-fit, someone who does not feel that a good is appropriately meeting their use case is subject to a mismatch between their expectations or needs and what they have been provided or found available to them. This is a problem which impacts not just the person suffering from the mis-fit, but also the systems through which they have navigated to get the good with which they are feeling a mis-fit. From a purely instrumental perspective, someone who does not feel that a good has met their needs might abandon the good, sell it on to someone else, or simply not re-purchase it when it comes to the end of its useful life. For actors like producers and marketers of consumer goods, mis-fit in any great measure thus becomes problematic when it negatively impacts sales or customer retention. As van Oost suggests, such mis-fits often occur as a result of mis- configuration8 of the user: “Configuring the user as 'everybody' in practice often leads to a product that is biased toward young, white, well-educated male users, reflecting the

8 Channelling Woolgar's idea that an object can “configure” a user.

31 composition of the designer's own group” (2003, p. 196). Categories which are underrepresented in the designing class are then underrepresented in the products of their work.

Perhaps more significantly for the user and their relationship with the world, mis-fit is a problem of mis-categorization. Because any categorization scheme necessarily fails to account for the entire diversity of whatever set it is intended to categorize, there will always be cases which do not fit (as in the above description of Star's work on residual categories). Mis-categorization then leaves room for marginalization, with those who do not fit relegated to the sidelines through lack of representation and recognition. Or, from a consumer perspective, falling outside of accepted categories leads to a lack of consumer choice and availability of goods capable of meeting one's needs. Crucial to the problem of mis- categorization is one of the popular assumptions underlying mass customization: adding more categories resolves mis-categorization and mis-fit. If every individual can become a category unto themself, the problem of mis-categorization must surely be resolved. As Borges has so memorably envisioned this scenario, the only map capable of fully and accurately representing a terrain is a map which is a one-to-one representation, effectively becoming and overshadowing the terrain (Borges, 1998).

In von Hippel's (2005) take on lead users, we see one proposed solution to the problem of mis-fit. Because individuals have varied needs, it takes a large number of products to satisfy all users. With industrial methods, mass production is more cost-effective than small-run or individualized production. Thus, in von Hippel's logic, when a user finds herself un-served by the results of mass production, she is forced to develop her own solution. This is what von Hippel refers to as “user-innovation.” However, companies are becoming increasingly willing to take the task of fulfilling individual needs away from users, offering a broader range of products under the banner of mass customization, custom production which makes use of the tools and processes of mass production, but achieves a better fit for an individual user. Zipkin outlines three traits in mass-customization, from the perspective of a business: “elicitation (a mechanism for interacting with the customer and obtaining specific information);process flexibility (production technology that fabricates the product according to the information);

32 and logistics (subsequent processing stages and distribution that are able to maintain the identity of each item and to deliver the right one to the right customer)” (2001, p. 82). Zipkin goes on to provide the example (which was current in 2001) of custom jeans offered by Levi's, which made use of manual measuring of customers in store, followed by a production process leveraging most parts of the existing mass production infrastructure, but with CNC cutting machines used to cut pattern pieces without introducing prohibitive set-up costs to the process.

A key development in mass-customization in the last decade has been the easy-to-use, web-based “configurator” offering a preview of the object being customized and an interface through which to customize it. When Zipkin wrote his critique of the value of mass customization in 2001, the state of computationally-generated previews of objects was a far cry from what it is now. Aichner and Coletti (2013) argue that web-based tools—including configurators “are the ideal tool for a dialogue with the potential customer, providing necessary information about the product and collecting his/her preferences in a totally automatic way” (p. 22). Such tools offer an opportunity to enforce constraints on customization, keeping the user's options in line with the capabilities of the production process being used. For the kind of process embodied by the configurator (and the production it then leads to), I will, throughout this dissertation, use the term “parametric customization,” which is used casually in much customization literature, although generally without the accompaniment of a clear definition. For my purposes, parametric customization is roughly a synonym for many kinds of configurator-aided mass-customization: it is a process which allows a user to make choices within a constrained set of parameters, with those parameters being directly tied to the capabilities of the production process on which the customization relies.

The case studies I engage with in the remainder of this dissertation are all, quite emphatically, not examples of parametric customization. I use parametric customization instead as a foil against which to position my three case studies. I make that comparison because of the popularity and, indeed, increasing dominance of parametric customization in current implementations of mass customization. I find it instructive to contrast what I would

33 think of as customization against the kinds of parametrically-defined customization which appear to make the most sense for those wishing to truly put the “mass” in mass customization. I further incorporate the other issues raised in this chapter, including how digital fabrication projects might differ from manual fabrication methods; the relation of the body to “the digital” and to digital labour; the fractionalization of labour in customization and digital production; and how all of those issues are inflected by our uneasy relationship with rationalization and standardization. I make use of those concepts in building a better understanding of mass customization and digital fabrication as they relate to individuals, embodiment, and labour. I frame my case studies in relation to three themes: the social organization of labour, the material conditions of production, and the role of the user in production. I seek to illuminate, through my case studies and framework, concerns about the body as a point where concerns about labour, the idea of “the digital,” and questions about the nature and value of custom production intersect. What is crucial about the trifecta of the social organization of labour, the material conditions of production, and the role of the user in production is that they provide a frame within which cases which produce goods in different mediums, at different times, and with differing ways of organizing labour can be compared to one another. In the next chapter, I outline my case studies and the methods by which I studied them.

34 3 Methodology

This chapter covers the methodology used in my dissertation, in two parts. The first half of the chapter is very much a traditional methods chapter, outlining the nuts and bolts of my research and describing the intentions I had at the outset of my work. The second half of the chapter (under the heading “Observations after the fact”) does a more reflexive exploration of the ways in which my methodology was changed by the act of doing the work. In the first half of the chapter, I lay out facets like framework and sample, as well as brief descriptions of the two case studies. In the second half of the chapter, I address the somewhat more complex issues that arose during the two case studies, and how some of my methodological considerations and concerns changed over time.

Participatory methods

Before introducing my case studies and the specific methods used to collect and analyze the data emerging from those case studies, I feel that I need to provide a kind of framework or justification for the active role I have taken on in one of the cases. Though it is not unusual for researchers to embed themselves in the communities they study, and tactics like participant observation are well-grounded in both past work and strong current development, the idea of being not just an active participant, but a stakeholder, seems like a somewhat less common path. While I do take on many of the tactics of participant observation, I feel that it is worth elaborating on my additional position as a stakeholder in my cases.

35 In engaging in practices which foreground personal involvement in research, I fit into a tradition as old as the social sciences. August Comte, in his treatise of 1848, in which he laid out the idea of social science, also presented another idea. Accompanying social science, which was to be a discipline designed to learn about the social practices of humans, Comte also described what he called “social engineering.” Social engineering was intended to be the practical application of findings gleaned from social science. Unfortunately, the concept and its attendant term were appropriated to serve some truly heinous purposes in the early-to-mid 20th century, which may well have led to what seems to be collective amnesia about the long- lost sibling of social science. Though I'm certainly not advocating for a return to the term, I do feel that some of the more useful aspects of the originally-posited social engineering have been taken up in concepts like Action Research, where stakeholders are involved in research projects, and the idea of gaining insight is not divorced from providing help and value to the people being studied. One of the grounding ideas behind Action Research is, indeed, to consider and include the agency of marginalized groups when working with them, and to help them solve problems (Masters, 1995). In many ways, one of the case studies in my dissertation (PrintAbility, which I describe below) takes up this banner, seeing research and community outcomes as inextricable. I hope that the investment I feel towards the parties involved in my case studies, and my commitment to assisting and doing right by them, will show through in both the more descriptive portions of my dissertation, and the accompanying analysis. I feel that this sensitivity and investment is key in doing research with communities, whether they be marginalized or comparatively privileged.

36 An introduction to the two case studies

The Libre Graphics Research Unit (LGRU) The LGRU is a multi party partnership project funded over two years by the European Commission Culture Programme. The LGRU's mandate is to explore the future of art and design software through collaborations between artists, software developers, arts organizations and standards bodies. Its primary organizational membership is comprised of four media arts organizations: Constant (Belgium), WORM (Netherlands), Medialab Prado (Spain), and Piksel (Norway), alongside a number of smaller participants in other European nations. It carries out its goals through a collection of conferences, meetings, events, hack fests and artistic commissions with the goal of re-making software around its future use cases.

The collaboration carried out by the members of LGRU is an attractive case study in the larger cadre of this dissertation. Pushing back against instantiated practices, developing new standards, and engagement on a collaborative user-developer level are all activities carried out by the LGRU. In particular, the reciprocal relationship between tools used in artistic practice and the shape of the practice produced by those tools—a foundational interest in the LGRU initiative, is an ideal example of one of the tenets of mass customization: that individuals seek out products which fit their needs. As the text of the LGRU project description puts it, “[b]ecause of the availability of source code, F/LOSS tools are virtually polymorphic: they welcome divergence, alteration and exchange” (LGRU Wiki, 2011). The polymorphous nature of F/LOSS tools, combined with the blurring of lines between artists and developers means that software has the potential to be a tool which is informed by the needs of its users, though that potential is seldom realized.

PrintAbility PrintAbility is another multi-stakeholder project. It brings together a non- governmental organization with a focus on providing assistive devices in developing countries (CBM Canada, the Canadian division of CBM, an NGO with global reach), a rehabilitation hospital in Uganda (CoRSU), the research arm of a multi-national software company (Autodesk Research), and a university-based experimental research group (the University of Toronto's Critical Making Lab). The purpose of the project is to develop, test, implement and

37 scale a method for using 3D scanning, 3D modelling and 3D printing to produce prosthetic sockets more quickly. The project springs from a confluence of factors: the lack of trained prosthetists and prosthetic technicians in many developing countries; the slowness of current methods of socket production; the low capacity for production of prostheses caused by the first two factors; and a growing interest in the use of 3D printers and scanners in medical applications.

The project was initiated by CBM Canada and CoRSU, who made contact with the University of Toronto to find a collaborating researcher. The scope of the development phase of the project extended from July 2013 to January 2015. Though the project continued after that point, in both a testing phase and in a scaling-up of the process into a social enterprise, the case study presented in this dissertation covers only the year and a half of development. As such, activity at CoRSU is somewhat more peripheral to the case study than it might otherwise have been. Rather, the chapter concerned with PrintAbility recounts and analyzes the development of the process, which has largely taken place at the Critical Making Lab, with the assistance of Autodesk Research and CBM Canada, and with resources and input from CoRSU.

Methodological framework

A suite of related methods and frameworks informs my handling of both case studies. In varying amounts, the two studies involve a combination of participant observation, analysis of pre-existing data, and collection of new data. The data varies in the two cases, with the LGRU case relying on documents describing the group, its events and its output, and the PrintAbility case making heavy use of field notes from participant observation, as well as semi-structured interviews and user testing. The analysis of both studies draws on grounded theory, reflecting a desire to let analysis be led by data. As such, coding is done, memos are written, and categories are saturated in order to arrive at findings which, ideally, are reflected in both cases. In addition to these two empirical or first-hand cases, I include, alongside the analysis of the LGRU, a historical case based on secondary sources describing the practices of 18th to early-20th century dressmakers.

38 In my analysis of both case studies, I make use of grounded theory. As such, it's worth taking a moment, at this point, to elaborate on my interpretation of grounded theory. I draw heavily on Kathy Charmaz's work on the subject. In particular, Charmaz (2009) describes grounded theory as a method which seeks to find theory emergent in data, with an emphasis on being open to new theoretical findings, rather than fitting new data into existing frameworks. As Bryant (2002) tells it, this idea of deriving new theory from evidence originally came as a response to established sociological practices in the 1960s, which saw empirical, quantitative work done, more often than not, as an effort to create proofs for the grand sociological theories of the late 19th and early 20th centuries. The grounded theory method, according to Bryant, was an effort to use rigorous, empirical, but qualitative methods to develop new theory, rather than simply verifying existing theories. Grounded theory has, as any such methodology must do, evolved since its original development in the 1960s. Contrary to some descriptions of grounded theory, such as the commonly-repeated (and vilified) exhortation from Glaser that researchers constructing grounded theories should attempt to approach their research with few or no pre-existing theories or hypotheses (1978), Charmaz does not advise that researchers avoid contact with existing literature prior to their own analysis. Instead, Charmaz gives a mellower instruction, that researchers should be open to finding new ideas, and should allow themselves to be guided by their own findings, rather than slotting findings into a priori assumptions.

Of particular use in my dissertation is the concept of axial coding. In axial coding, the process is something like the following: as data is coded and themes emerge, one should make note of those themes, write memos making sense of the themes, continue coding data, being aware of the themes that have already emerged, and try to saturate the thematic codes (Charmaz, 2009). Saturation in this case refers to a state of finding no new conceptual attributes to assign to a code. In an axial coding process, that saturation might be achieved by conducting further research, using new data to confirm that no new codes are emergent or that existing codes are fully articulated (ibid). We might also think of this process as being somewhat iterative, refining categories and attempting to find their boundaries before presenting a finished analysis.

39 LGRU My study of the LGRU relies on the analysis of existing LGRU documentation, in the form of meeting minutes, mailing list discussions, wiki pages, websites, programs for events, catalogues and publications, and other textual and visual output. I analyze these documents through the twin lenses of grounded theory (in the vein of Charmaz, described above) and actor network theory (per Law, 19929). In particular, my work makes reference to Law's assertion that actor-network theory “tells empirical stories about processes of translation” (1992, p. 387). While the above description of grounded theory may make it seem paradoxical to apply an external theory to a grounded theory analysis, I feel that the open coding methods of grounded theory will, in this case, be augmented and complemented by an added focus on the relationships between actors. However, rather than fitting my findings into an existing theoretical framework, I use actor network theory simply as a way of focusing my activities. In a sense, actor network theory serves as a theoretical underpinning, but not as a container for my findings.

In total, the corpus of existing texts to be analyzed is made up of 29 pages of data from the LGRU wiki and 115 pages from the LGRU website, which includes documentation from LGRU meetings, some of which has also been published in other formats, such as catalogs. In analyzing this corpus, I read all of the listed documents, conducting open coding as I read, with codes from early readings informing my analysis of later portions. Because earlier codes necessarily informed the construction and organization of later codes, I started my analysis with the foundational description and purpose of the LGRU, which are articulated on the LGRU Wiki. Throughout much of the LGRU's duration, I also subscribed to its mailing list (the archives of which are also publicly available). Though I did not formally code mailing list conversations, those conversations did inform my understanding of the LGRU's wider context.

The second contributor to the study was my own participation in an LGRU event. Acting as a facilitator of an LGRU-presented workshop/hack week, I engaged personally in the

9 “This, then, is the core of the actor-network approach: a concern with how actors and organizations mobilize, juxtapose, and hold together the bits and pieces out of which they are composed; how they are sometimes able to prevent those bits and pieces from following their own inclinations and making off; and how they managed, as a result, to conceal for a time the process of translation itself and so turn a network from a heterogeneous set of bits and pieces each with its own inclinations, into something that passes as a punctualized actor.” (Law 1992, p. 386)

40 user-development process championed by the LGRU. In this phase, I collected my own reflections on the development process through, predominantly, field notes. The field notes cover my experiences in the two-week long Interactivos? workshop organized by the LGRU in in 2012. I consider this portion not as a site in and of itself, because of the incompleteness of the field notes, but think of it rather as a first-hand window into the kind of interaction facilitated by the LGRU project. All told, the materials from the LGRU case offer a coherent look at the period of two years over which the LGRU was active, and will provide a set of novel findings, supported by the large corpus of primary data.

PrintAbility The PrintAbility case study covers a shorter period of time than the LGRU, but is one in which I have been actively involved throughout. Asaparticipant in the project, the variety of methods I was able to use is somewhat more diverse, with participation and study blending into one another, methodologically and with regard to desirable outcomes. Because the PrintAbility project has a concrete outcome (the testing and eventual implementation of a new method for the production of prosthetic sockets in hospitals in the developing world), my work has involved tasks which contribute both to this dissertation and to the successful outcome of the project. One such task or method (depending on which context it is taken in) is an environmental scan. Throughout the project, one of my tasks has been to track and monitor existing and new developments in the use of computer-aided design and manufacturing (CAD/CAM) in the production of prostheses. This has involved the use of article aggregators like Web of Science, to get a view of the historical state of CAD/CAM prostheses as represented by the academic literature. It has also involved keeping abreast of new developments, largely through websites which chronicle 3D printing news. Though not all information collected in the course of these activities is rigorously coded, it forms the background against which my data collection and analysis take place, much as reading the LGRU mailing list formed the background for my analysis of the LGRU documentation. Some particularly interesting or relevant sources make an appearance, informally, either bolstering observations, or acting as the impetus for particular activities. Sometimes, these sources provide an oppositional motivation, acting as a foil against which I compare the PrintAbility project.

41 In addition to the ongoing environmental scan, one of the important foundations of my work on the PrintAbility project has been a series of interviews with prosthetists, prosthetic technicians, orthotists and users of prostheses. These interviews, conducted over a period of eight months, have included five medical practitioners (prosthetists, orthotists, prosthetic technicians) and three users of prostheses. Many of the user-participants have attended multiple interviews, with the normal course being a preliminary interview, followed by a user testing session and demonstration of the PrintAbility process at a later date. The preliminary interviews ask a series of open-ended questions about the user's history with their prosthesis, and provide an explanation of our goals with the PrintAbility project. The later sessions are structured as part demonstration, part user testing session. These sessions take place in front of a computer which has scanning and modelling software installed on it. The medical practitioners interviewed participated in user testing/demo sessions similar to the users, but with somewhat more ability to construct an accurate socket during the course of the session. In some of these cases, the sockets designed were retained, with the permission of the practitioner, and later printed.

Beyond the environmental scan, interviews, and user testing sessions, a major portion of the research in the PrintAbility case is the study of the collaborative efforts involved in the development of the process. These efforts are chronicled through my own field notes, through emails collected over the course of the study, and through materials produced during the study. A corpus of field notes, collected over a year and a half, is augmented by emails and sketches, process work and photographs documenting the project. The analysis of these materials is particularly beholden to the idea of iteration. Because the project took place over a relatively long period of time, and because I was a participant throughout, I produced memos and conceptual insights earlier on in the process than in the LGRU study. Often, these memos or insights were either incorporated into my thinking on the project, or were more formally taken up. In one notable instance, an idea about the use of grease pencils became a short position paper, on which I was a co-author, which then moved in multiple directions, resulting in a longer paper for a special issue of Design Studies; and a talk and a paper with which I was not involved, but which had a particular concern for the use of craft practices in digital

42 production; and an increased concern on my part (manifested in my ongoing interviews) with the invisible or ignored work carried out in the process of socket production.

All of the sources and methods described above play into the analyses which appear in the following two chapters. The two cases also reflect on each other, with the analyses often occurring concurrently. Though, in reality, my involvement with the LGRU case began before the PrintAbility case started, active engagement, preliminary coding and some memo production took place earlier for the PrintAbility case than for the LGRU case. However, because I had already been an observer and, in small part, a participant in the LGRU, some of its themes were already present when I embarked on the PrintAbility case. Concerns about the interplay of the two cases, as well as the ways in which each case was ultimately handled, are addressed more fully in the next section.

Observations after the fact

“The world is very complex. There are no simple explanations for things. Rather, events are the result of multiple factors coming together and interacting in complex and often unanticipated ways. Therefore any methodology that attempts to understand experience and explain situations will have to be complex. We believe that it is important to capture as much of this complexity in our research as possible, at the same time knowing that capturing it all is virtually impossible. We try to obtain multiple perspectives on events and build variation into our analytic schemes. We realize that, to understand experience, that experience must be located within and can't be divorced from the larger events in a social, political, cultural, racial, gender-related, informational, and technological framework and therefore these are essential aspects of our analyses.” (Corbin and Strauss, 2008, Ch.1)

43 This second portion of the chapter has as its goal a kind of radical transparency. As Corbin and Strauss argue above, contextualizing data and capturing complexity are essential in representing as well as possible the world being described in a given piece of research. As the chapters covering my case studies will show, this is an issue dear to my heart, and something which I find to be of both methodological and theoretical use. For me, because I have been complicit, in a variety of ways, in much of the work described in the two case study chapters of this dissertation, the contextualization and capturing of complexity described by Corbin and Strauss is absolutely essential in describing the ways in which I handled my research. Because much of what happened and the events and decisions that caused me to make certain decisions is not necessarily relevant to my work, at least as a finding, I use this portion of the chapter to convey the things I find important about how the work was carried out, and what that meant in the context of my own research outcomes and my ability to write this dissertation. The following sections describe the circumstances and decisions which took place over the course of the two case studies. I hope that they will aid in effectively contextualizing the work presented in this dissertation, and exposing some of the reasoning involved in my decisions about what information should “count” in the case studies, what should not, and how that changed over the course of the work, given the extreme richness of the total data available.

Coding

Though I mention in the first half of this chapter that my initial plan was to make use of axial coding, it is in the context hindsight that I can provide a more detailed account of how I coded my data. I coded some of my data on an ongoing basis, with analysis of the PrintAbility project taking place as materials became available. Experiences, emails, diagrams and user testing videos all formed the basis of small memos, as they happened. Many of the findings from interviews became the basis for discussions about the progression of the PrintAbility software and were thus coded in a different way: they were literally coded into the software. It was through reflection immediately after my user testing interviews that those decisions were made. That process is emblematic of a large proportion of my PrintAbility work. However, my fieldnotes were a different matter. I analyzed my fieldnotes some time

44 after they were taken, waiting until I had several notes before going back and reading over them. The coding of fieldnotes was somewhat informal: I would read through a series of notes and look for emergent themes As themes emerged, I copied relevant passages from the original notes into new documents representing codes. As those codes solidified, I then wrote memos, using the snippets of text as fodder for more analytic engagements with the codes. Those memos form the basis for many of the insights I gathered from the PrintAbility project. However, I gained other insights by re-reading the PrintAbility memos through the lens of the LGRU. I describe the process of coding the LGRU materials in the following paragraph.

Because the vast majority of the LGRU data is pre-existing, publicly-available text which was produced and put online before I began my analysis, my coding was more systematic and less based on the availability of materials than was the case for PrintAbility. I coded the LGRU documents similarly to the PrintAbility fieldnotes. As I read through the documents and saw themes emerging, I collected relevant snippets into text documents which represented codes. Those codes emerged at times from themes I saw recurring in the documents (such as the importance of the material conditions of production), and at other times, from the terminology used by the authors of the documents (like “practice shapes tools shape practice”). As the codes became saturated with snippets, and as the number of new codes emerging from each document decreased, I began to write memos. I took these memos into account when looking for higher-level codes capable of encompassing a number of other codes. It was this process which led me to the three high-level themes which ultimately formed the framework I use to organize my dissertation: the social organization of labour, the material conditions of production, and the role of the user in production. I then supplemented the high-level codes derived from analysis of the LGRU documents with other examples from the LGRU, which I derived both from other documents available online, and from my own first-hand experiences of collaborating with LGRU member organizations. Later, when writing portions of this dissertation, I used the three high-level codes to re-read some of my PrintAbility memos. The codes also formed the basis for my targeted reading of secondary sources on dressmakers.

45 Changes in the LGRU corpus

The corpus of LGRU materials changed during my analysis. Because I was more closely acquainted with some parts of the project than others, during coding of the original corpus, I uncovered themes which were heavily represented in some of the artefacts and sources I hadn't originally intended to analyze. In this respect, I found myself drawing on a document produced during the “Tools for a Read-Write World” Interactivos? hacksession which took place at Medialab Prado in spring 2013. The document, a zine describing all of the projects produced during the week, offered descriptions of projects which closely met the theme of distributed collaboration, which emerged strongly from my analysis of other LGRU documents. I participated in an editorial capacity in the production of the zine, collaborating with Femke from Constant (who conducted interviews with the Interactivos? participants) and Ana and Ricardo from Manufactura Independente.10 Though I did not take field notes documenting the production of the zine, the procedure we used closely followed a workflow that Manufactura and I have developed over our years of working together on another F/LOSS-based publishing project. Some insights from the production of the zine are incorporated into my analysis of the LGRU case. My participation in Interactivos? made it a richer source than some of the textual analysis, as I had access to information about the material conditions of production that was not as present in the texts analyzed.

Also significant in the LGRU case was the decision about whether or not to use git repositories as sources of data. One emergent theme, the conflict between producing new tools and modifying old ones led me to seek out more information on how the commissioned projects were carried out. Because git repositories for some of the commissions were available, I found myself needing to decide whether or not source code, readme files and other documentation available in the repositories should be admitted to my corpus. Ultimately, rather than treating these repositories formally as data, I found myself using them to confirm my understandings about the division of labour in the commissions. In this respect, the commit histories were far more important to me than the actual content of the commits, serving to confirm, for example, that contributions to particular pieces of software were

10 In treating much of my data, I do not anonymize, as anonymization would require me to also change the names of the projects in question. Instead, if a name is a matter which can be easily discovered by anyone doing a cursory look at the materials being described (as in the case of the work done on the zine mentioned above), I use it.

46 largely taken on by people who are more explicitly positioned as software developers. In this respect, I need to admit to some less systematic methods, which inhere in my long-standing association with many of the people involved in the LGRU. My assessment of who to count as a software developer was based on the histories of the activities of these participants, and an awareness of their activity-level in other software projects. Were I to make a judgment like that outside of a community with which I have a longstanding relationship, I would need to use other methods, such as looking at a user's activity history on Github or other code sharing websites, to make assessments about whether or not someone might be considered to be primarily a developer. Still, this is one area I feel some uncertainty over, given the equivocal feelings that many participants in the Libre Graphics community hold about being classified as either a developer or an artist. Some align strongly with both identities, and some individuals who might be considered as developers—especially if viewed outside the context of software-oriented communities—consider themselves artists.

In a similar vein, in substantiating some of the claims I wanted to make about the LGRU, I found myself needing to bring very peripheral sources into my corpus. One such example came up in discussing the role of developers in the production of documentation. In order to make a claim about developer involvement in booksprints, I needed to come up with a way to determine who can be said to be “involved” in the development activities of a F/LOSS project. To do this, with relation to the team of authors listed in the credits of a book published by FLOSS Manuals (one of the smaller member organizations in the LGRU), I made recourse to the publicly-archived development mailing list of the software project about which the book was written. An analysis of who participated in the discussions on that mailing list served as a proxy for determining which of the participants in the booksprint could be said to be actively involved in the development of the software project. This, I use as a comparatively objective method of gauging commitment and contribution to the life of a project.

Taken as a whole, the addition of data to the LGRU corpus was generally in service of enriching my understanding of both the social organization of labour (as in the case of the git repositories) and the material conditions of production (as with my experience in the Interactivos? process). These additions were made either because certain findings didn't feel

47 complete, or because questions were raised for me by the analysis of the LGRU texts. These additions also elucidated questions around the relationships between developers and artists, and, as such, between the imagined or intended users and the people who would ultimately build the software, with the input of those possible users. Enriching the corpus through the integration of git repositories gave me access to a kind of fine-grained data about the LGRU's development activities.

Changes in the PrintAbility corpus

Throughout the year and a half of progress on the PrintAbility project, it was my intention to not include our trip to Uganda in my corpus. This had been my intention for a number of reasons, both scholarly and organizational. I intended to study the development of the process, rather than its implementation. This would line up more accurately with the LGRU case, which similarly followed development but not use of software. I was interested in the relationships and working methods of the organizations and individuals building the PrintAbility process. The vast majority of those, and the ones that made the largest contributions to development and structuring of the project, were in Canada, even if the idea had originally come from CoRSU. I was also concerned about the ramifications of doing a study in a developing world context. I had the ethical scholar's concern about not wanting to parachute into a comparatively under-resourced milieu, get my data, and leave. I maintained my resolve to not admit data from CoRSU into my corpus through the flight to Uganda, and during the weekend we spent there prior to the beginning of our time at the hospital.

Things changed when we got to CoRSU. For one, in my interactions with both the hospital workers and administrators, I did not feel that I was in a position of comparative power. I do recognize that, objectively, with regard to affluence and power in a global context, access to resources and social goods, I am of course at a huge advantage. Uganda does not even have a public ambulance service, while Canada has socialized health care. There is certainly a gap in access to resources. My purchasing power is such that I can buy a plane ticket without worrying too much about the expenditure. The workers in the Orthopaedic Workshop at CoRSU cannot say the same, though they represent a comparatively

48 professionalized class of worker. Regardless of those structural issues, in the context of the Orthopaedic Workshop itself, they were on their home ground and had far more applied expertise. They were familiar with the structure and functioning of the hospital, while I was an outsider. In their work, they were experts. In their hospital, they were similarly expert, even if they were not always capable of exerting power on a larger institutional scale (which is also true of many medical practitioners at comparable levels in developed-world contexts). This led me to feel more comfortable about admitting data from CoRSU into my corpus. In addition, I had been worried about what it would take to get the appropriate ethics clearance for that context. This was made easier by the work of a CBM Canada employee, Emily, who made great efforts to go through the appropriate channels for all of our ethical clearances, whether clinical or otherwise, in a country where such procedures are often ignored.

As I wrote my chapters, I found myself admitting more and more data from CoRSU. Though it was made easier by the considerations listed above, what made it feel necessary was the relationship between the development of the project and its implementation. Though I followed the development of the PrintAbility project, and was thoroughly implicated in large parts of its data collection in Canada, the trip to CoRSU felt, in a way, like an exhalation of breath. In seeing the process in action, so much of what we had done during the year and a half of development finally felt both more meaningful and more appropriately contextualized. In this sense, the project differs somewhat from other research practices which engage critically with values and ideals about design and production. In particular, though one might perhaps align the work with critical design, which creates interventions that reject current systems and norms (Bardzell & Bardzell, 2013), I suggest that the work we have done with PrintAbility is very sensitive to existing infrastructure, and attempts to make use of it, rather than reject it. In relation to critical making, a concept from which the work does originate, even if only because the project comes from the Critical Making Lab, I see a somewhat uneasy connection, in that in past, critical making has not sought to build interventions for deployment, but has instead aimed to build understanding of systems through production. Though the PrintAbility project certainly does that for its participants (and might well be, in that sense, critical making on a very grand scale), its outward focus and more instrumental goals distance it somewhat from work which solely has critical or research outcomes in mind.

49 The following two chapters elaborate further on both the LGRU and PrintAbility, first providing more detail on the projects and then making use of the collected data to illuminate traits shared by the projects, as well as differences, in order to construct an argument about the conditions under which digital fabrication and customization practices are carried out. I use the framework outlined above, concerning the social organization of labour, the material conditions of production, and the role of the user in production to frame the cases and make them clearly comparable.

50 4 Custom and mass production

Introduction

In this chapter, I use two cases to denaturalize mass production. First, through a historical case study of custom dressmakers in the in the 18th to early-20th centuries, I provide an account of a historical production process which managed to persist well into the Industrial Revolution and apparently to resist the effects of that revolution for some time. In addition, by providing some context around the move from custom to mass production in dressmaking, I highlight the slow, uneven and contextually-dependent trajectory towards standardization. Through the dressmakers, I build an analytic scaffolding—described further in the next paragraph—which I apply, in this chapter, to both the dressmakers and my second case. In the next chapter, I take the scaffolding forward and apply it to my third and final case. The second case at issue in this chapter highlights the activities of a software development initiative called the Libre Graphics Research Unit. This case gives me an opportunity to study how an interdisciplinary group of artists, designers, organizers and software developers collaborate together to do what I argue amounts to a kind of computational custom production, building new software which is informed by discussions between artists, designers, and developers. Through the two cases, I use this chapter to look at the processes by which goods are produced and to draw parallels between a pre-digital customization practice and a very resolutely digital one. In tracing a set of arguments about the transition from custom production to mass production during the Industrial Revolution and in contrasting those arguments against the LGRU case study, I argue for the importance of considering the blurring of the distinctions between user and producer; the difference between custom production

51 when one does it for themself and custom production when the end-user is not implicated in the production process; and the difference between custom production of born-digital goods and custom production of other, non-digital goods.

In addition to contentions about the history of custom production, in this chapter I build an analytic scaffolding which highlight three key areas of concern in considering customization and custom production practices: the social organization of labour; the material conditions of production; and the role of the user in production. These three key areas, though they happen to line up with existing theory, are also emergent from the case studies I analyze in this chapter. Though current arguments about free labour might argue that the third area, the role of the user in production, is a subset of the social organization of labour thanks to the use of un-paid user work to create value, I make a point of differentiating the two. I do this because of the need to recognize the unique stake of the person who intends to own or make use of a custom good, even if they are implicated in its production process. I return to this issue in more depth in chapter seven, where I look at the distinction between use value and exchange value in custom goods. All of this is in service of positioning the move from custom to mass production, an artefact of the Industrial Revolution, against the current, much-touted move from mass to digitally-aided custom production (as exemplified by the passages from the popular business press found in chapter one of this dissertation). By contrasting the custom- to-mass move of the Industrial Revolution against the mass-to-custom move apparently arising from the growth of digital fabrication technologies (themselves also often being positioned as yet another industrial revolution), my goal here is to illuminate some of the contrasts and similarities between the custom production that dominated the 18th and 19th centuries and persisted into the beginning of the 20th, and the emerging custom production of the 21st century. In essence, the purpose of this chapter is to indicate that there are a number of strong similarities and shared organizing principles between what is often seen as an outmoded form of production (the handwork and small-businesses and craftspeople embodying pre-Industrial Revolution custom production) and what is increasingly touted as an exciting new paradigm in modern industry (namely, mass customization and digital fabrication). I use the case of a software development initiative because many assumptions about digital fabrication find their original home in contentions about strictly digital goods.

52 The abstractions, replicability and transferability of digital goods would, at first glance, appear to be at odds with the extremely local and contextual nature of customization practices like those of 18th through early-20th century dressmakers.

The professional dressmakers of the 18th through early-20th centuries were engaged in detailed practices of customization and, by dint of the types of products they made, were not only obliged to take the needs of their customers into account, but were also among the later groups of fabricators to have their industry fully industrialized (though certainly some industries feeding into dressmaking, such as textile manufacture, industrialized significantly earlier11). In order to look at the work of 18th to early-20tth century dressmakers, I draw on existing analyses from social historians specializing in business, labour and women's history. These sources cover western dressmakers, focusing on those in the eastern United States, England (and occasionally Scotland), and France, with an additional, passing example from Australia. My second case, which comes from the very near past, is the Libre Graphics Research Unit, a European consortium of software developers, media arts practitioners, and media arts organizations. The LGRU was a limited term initiative which took place from 2011 to 2013 and engaged practitioners from across Europe (and occasionally beyond) in co- production projects. In involving artists and designers in the development of new kinds of design software, I contend in this chapter that the LGRU represents a kind of customization which is, in fact, very similar to the work done by professional dressmakers, but which creates digital artefacts, rather than garments.

In presenting the cases of the dressmakers and the LGRU, I first address the custom production practices of 18th to early-20th century dressmakers, as well as their place in society and their relationships with their customers. I then briefly describe some of the factors involved in the contraction of the custom clothing industry, and, at that point, admit some detail about tailors, whose work was, in many places, industrialized some 50 years before the work of the dressmakers. Throughout my description of the work of custom dressmakers, I codify their practices into three organizing categories: the social organization of labour, the material conditions of production, and the role of the user in production. I use these categories 11 Musson (1972) tells us that the British cotton industry of the early 19th century “was in many ways exceptional in its use of power-driven machinery and factory production” (p. 18).

53 both as a framework for comparing the custom dressmakers against the other case studies present in my dissertation, and as a way of connecting the case to the existing literature. Following the historical sketch of the custom-to-mass move in clothing, I introduce and describe the LGRU project. Biographical information about the project is followed by findings (based on a mode of analysis outlined in chapter two) which are enriched through comparisons to other Free/Libre and Open Source Software and cultural production projects as necessary, as well as to the framework described above and to relevant literature. Finally, the conclusions I have drawn about custom production of digital goods are related back to the Industrial Revolution era custom-to-mass move in the garment industry and the practices displaced by it. In drawing that comparison, I aim to highlight the similarities between the manual custom production of the dressmakers and the digital custom production of the LGRU participants, while also beginning to demonstrate the interconnections between the three elements of my framework of analysis. By demonstrating those interconnections and showing the applicability of the three elements of the framework across time, distance and industry, I aim to lay the groundwork for its application to the case study detailed in the following chapter, which melds a number of the elements present in this chapter, but goes further in detailing the implementation of a digital fabrication process.

54 Dressmakers

Custom work and made-to-measure In North America and Europe, the cusp of the late-19th century and early-20th century was a watershed time in clothing production on a large scale. According to Wendy Gamber (1992), the advent of the shirtwaist (a term referring to what we would now call a blouse, see Illustration 1) in the 1890s represented the first real inroads made in ready-to- wear women's garments. Prior to the shirtwaist, fashions requiring a high degree of customization—largely due to their close-fitted bodices and jackets, visible in Illustration 2 —were de rigeur and necessitated the work of skilled tradespeople if they were to be constructed properly. A woman looking to get a new dress had essentially two options: she could fabricate something at home with the Illustration 1: A drawing of a shirtwaist, printed in the 31 March, assistance of her female family members or, if she was 1904 edition of the Tacoma Times. affluent enough, could have a dressmaker or seamstress make a dress for her. For poor or rural women in the United States at the beginning of the 19th century, dresses were made at home, often constructed from the recycled parts of existing dresses—indeed, in these types of households, fabric was an expensive and precious commodity (Fernandez, 1994). As styles changed, dresses might be made over to reflect small changes such as a different style of sleeve or bodice, but having an all-new dress was a rarity (ibid). Clothing for men and children in such families was fashioned in much the same way, without the intervention of professional sewers or cutters. For affluent women and men, dressmakers and tailors were available on a contracting basis, producing garments for individuals, based on their measurements and tastes (ibid). Though mass produced clothing did exist at the time, it did not enjoy a positive reputation (an issue about which I go into further detail later in this chapter).

55 Though dressmakers almost uniformly operated very small businesses and, as Gamber (1997) notes, such small businesses have often been ignored in the study of business history, a small slew of historians have taken up the task. Crowston (2001) offers an insight into the various classes of women-garment makers in Old Regime France, delineating the high-skill, high prestige dressmakers who counted nobles among their clients, from the seamstresses who catered for working women and the wives of less-monied citizens (all of whom were eclipsed in prestige by the male tailors who specialized in ladies' two-piece court Illustration 2: Emiline, a lithograph by James S. dresses during much of the 17th and 18th Baillie of New York, NY (1845), showing the fitted style of bodice that was common at the time and centuries). Parisian seamstresses gained th for much of the 19 century. their corporate charter in 1675, with a guild charter defining their scope of activities, encompassing the sewing of clothing for women and children (ibid). Dressmakers worked in home-workshops, and performed mending and alterations in addition to their dressmaking activities. These mending activities included the making-over of dresses to meet new fashions. Though many of the activities of the different classes of dressmakers were similar, Crowston suggests that the more marginal dressmakers might have produced more aprons and skirts than dresses, given the income of their clientele (ibid). This general structure holds true in 19th century England, with different classes of bespoke dressmakers catering to the aristocracy, the middle classes and the working classes, respectively (Rogers, 1997). Rogers indicates that 19th century England had added an additional category to this taxonomy: the “show shops” that offered ready-made garments to those who could not afford custom-made. In the Old Regime France described by Crowston, this sort of activity might have been taken up by the fripiers who sold used garments. This was also a niche occupied by American merchants of used

56 clothing, turning the secondhand clothing trade in the United States into a business concern on a national level (Fernandez, 1994).

Mistress seamstresses and their employees (some of whom—especially those involved in the apprenticeship system—would likely have lived with their mistress, while others—day labourers generally employed on short term contracts—lived out) worked from sun-up to late evening, roughly eight or nine o'clock (Crowston, 2001). Over the course of her apprenticeship, a dressmaker might start out by plain stitching, as well as fetching, carrying and household chores, move up through more complicated sewing tasks (such as fashioning the bodice) and ultimately become responsible for the cutting (Gamber, 1992). In the Parisian dressmaking system described by Crowston, five years of training and work of this sort would be capped by the trainee seamstress producing her masterpiece and becoming fully qualified to be a mistress. After finishing her training (whether in the Parisian system requiring the formal completion of a masterpiece, or the less formal plethora of American systems), she might eventually save up enough money from her wages, or by borrowing from family if her family had such means, to open her own dressmaking shop (Gamber, 1992). In mid-to-late 19th century Boston, dressmakers going into business for themselves were generally in their 30s or 40s, owned little or no property, and operated as sole proprietors with a small number of employees. The vast majority did not own enough property or earn enough income to be suitable risks for loans and are, as a result, badly-documented in surviving records (Gamber, 1992).

The curtailment of women's business activities was symptomatic of social factors originating beyond the walls of their shops. Though a large proportion of the businesses run by women in both Britain and France during the 18th and 19th centuries were millinery and dressmaking establishments (Honeyman, 2007; Craig, 2001), what Walsh (2005) refers to as the “so-called female business sector of millinery and dressmaking” (p. 183), women's businesses in general were “less capitalized” (Craig, 2001, p. 200) than those of men. Craig suggests, in fact, that dressmaking and millinery were among the only areas in which businesses run by women performed favourably and had lower failure rates. According to Hoeneyman, in Britain, until the 1870s, women were subject to the “law of coverture, [...] which gave

57 husbands control of their wives' property during marriage, restricted married women's access to independent activity and contained harsher terms than the continental equivalents” (2007, p. 479). Though Honeyman among others troubles the idea that the doctrine of separate spheres of activity significantly prevented women from entering into commerce and argues that women were more active in business than is often suggested, the prominence of women in dressmaking and millinery does reinforce the idea that reputable trades for women to engage in were those which catered to other women and which placed them in comparatively gentile, domesticated settings. Given that a large number of women who owned businesses were either never married or were widowed (Honeyman, 2007), those of middling means might well have looked upon dressmaking or millinery as one of the few options available to them for earning a stable living in a socially acceptable line of work.

Inside the workshop Crowston (2001) provides an insight into the working conditions of French seamstresses during the Old Regime. In period depictions of seamstresses' shops, workforces are shown as ranging from two (including the mistress) to six (also including the mistress), with workers sitting on chairs and hand-sewing fabric which is either rested on trestle tables or on their laps. The mistress is depicted as either providing direction to her workers, or as measuring a client (see Illustration 3, next page). Crowston notes that this is a similar division of labour as in images of tailor's shops from the period, with the difference that workers in tailoring shops are pictured as sitting, cross-legged, on tables to do their work, a pose which was apparently typical and iconic for French tailors at the time (and which is consistent with Ginsburg's [1972] description of English tailors). Production and sales were co-located, with seamstresses of varying levels of affluence making use of their apartments in different ways. A less affluent seamstress with a small, one or two room residence might well conduct her business in the same room that she slept and ate in, while a more affluent one might have an entire room of her apartment given over to the conduct of business. She might also have made use of light, windows and mirrors to make the room appear larger and brighter. Crowston further draws on probate records to detail the tools and furnishings present in the dressmaker's workshop, which included scissors, thread, mirrors, screens, workbenches, chairs, storage furniture, and irons. This relatively spare set of tools was apparently all that

58 was required at the time to practice the trade of dressmaking, or at least all that was considered of value and thus, all that appeared in accountings of the goods in the typical seamstress' shop.

This characterization of the Parisian dressmaker's shop reveals the dressmaker as a tradeswoman on a fairly small scale with direct oversight of her employees (and indeed, responsibility for their training), a constrained set of tools, and little or no stock kept on her premises. In a business which trades in the production of image, she enhances her own through the use of relatively simple items like mirrors. The mistress dressmaker, as the owner and most senior member of the shop, engages with her customers while less senior members of the shop are relegated to the production of the garment. The mistress dressmaker's investment of capital and risk in the business (and its location in her home) Illustration 3: The Fitting by Mary Cassatt affords her a place of seniority. (1891) depicts a woman being fitted for a dress, a task almost uniformly carried out by the mistress dressmaker herself. It is unclear whether Rogers (1997) draws attention to a the fitting takes place in the woman's home or in th the dressmaker's shop. It is also unclear whether distinction in the mid-19 century the painting is intended to reflect life in Cassatt's needle trades (a term used at the time to native Philadelphia, or Paris, where she lived at the time it was painted (Weinberg, n.d.). encompass everything from glovemakers to sweatshop workers to mistress dressmakers), between respectable workers in “'honourable'” work milieux (custom dressmakers catering to private trade, as in the Parisian description above) and workers in “'dishonourable'”

59 portions of the trade, including those doing “slop”12 work, or piecework contracting in the home, the factory and the workhouse (pp. 591-592). Even in the so-called honourable13 sector, the exigencies of the market (with busy periods requiring extremely long work hours, and slow periods resulting in temporary un- or under-employment) were such that, according to the period reports cited by Rogers, the health and wellbeing—and the morals—of women in the needle trades were seen as being in some danger. According to a series of reports published in the popular press during the 1850s, downward mobility was a distinct possibility in the needle trades, with “honourable” seamstresses finding employment in slop and sweated work during periods of slow business in their normal trade (Rogers, 1997). Rogers argues that much of the moral panic surrounding these assertions may well have been manufactured by reformers and government commissioners intent on demonizing women in the needle trades, and cites the testimony of mistress dressmakers who argued that their employees were capable self- regulators who would not stand for unseemly behaviour in their workplace. Crowston (2001) offers a similar assertion, that the ability of a trainee dressmaker to take on the trappings of gentility was a determinant in her success in the industry.

Constructing a garment The actual activity of high end dressmaking was thoroughly custom. In constructing a garment, the dressmaker was responsible not only for an accurate fit (accomplished by taking a series of measurements on a long strip of paper, which was kept for future reference) (Ginsburg, 1972; Kidwell, 1979), but for the diminishment of any flaws the client's body might be perceived to have (Crowston, 2001). Pre-made patterns, though we may now think them to be standard, took some time to develop, originating in mathematical systems of extrapolating proportional body measurements from a limited set of measures (Kidwell & Christman, 1974). As Crowston tells it, “[p]rior to the eighteenth century, most seamstresses and tailors made life-sized patterns for garments on their clients out of thin cloth, using the pattern as the lining of the finished garment” (2001, p. 151). Fernandez (1994) confirms that this practice

12 The term “slop” was used from at least the 14th century to refer to a loose garment, a “jacket, tunic, cassock, mantle, gown, or smock-frock” (Slop, n.d.), but had expanded by the 15th century to also include baggy trousers, including those of sailors (ibid). By the 17th century, the term encompassed “[r]eady-made clothing and other furnishings supplied to seamen from the ship's stores; hence, ready-made, cheap, or inferior garments generally” (ibid), which is the sense I employ here, and the sense apparently used in the term “slop shop.” 13 The work of custom dressmakers might be considered to be “honourable” because it kept women in the company of other women, in small shops which could be seen to have a domestic flavour to them, and out of factories and the company of men.

60 persisted in 19th century America, with thin cloth used as a medium for taking measurements. In determining the style or fit of the dress, dressmakers (and tailors) took their cues from clients.14 As Miller (2006) describes it, the movement, in the 18th and 19th centuries of affluent women both around the United States, and on trans-Atlantic journeys, facilitated the movement of fashion trends in an era before popular depictions of clothing in the press were normal. Clients, upon seeing a new style worn by a prominent woman, would ask their dressmakers to replicate the style. In this way, changes in fashion were driven by the social circles of clients, rather than by the edicts of dressmakers, and made the clients very much partners in the devising of their dresses. In this sense, we might well see a division in the activities of dressmakers based on their clientele, with more affluent women with better access to fashion-forward circles bringing demands or guidance which differed from the requests of women who travelled in other social circles. Gamber (1992) further suggests that clients in the eastern United States often brought in their own fabric, which was then made up by the dressmaker. Indeed, because it was normal for customers to provide their own fabric and trimmings, dressmakers, unlike other tradespeople (including those in another largely female industry, millinery) did not need to carry much by way of stock (ibid). In these ways, not only was the taste (and by connection, social status) and body of the client represented in the way her dress was made, but a degree of power over its construction was held by the customer and her own interpretations of current fashion, as well as the means available to her in purchasing cloth and trimmings.

The importance of professional skill among dressmakers is highlighted in the practice of having a seamstress in to improve or finish a homemade dress. Fernandez (1994) recounts that, towards the end of the 18th century and during the 19th, American families would contract with dressmakers to do cutting and fitting (the most skilled activities in making up a dress), while the women of the family would do the plain-stitching, saving money on tasks which they were capable of carrying out themselves, while still enlisting the specialized expertise of the dressmaker for more complex or specialized tasks. This activity took place in the home of the client, rather than in the workshop of the dressmaker. Gordon (2004) notes that, in the early-20th century American midwest, this habit persisted and helped more affluent rural

14 Although Stoneley (2006) conveys an account from Thoreau about his inability to convince his tailoress to fashion him a waistcoat in what she considered an out-of-date style.

61 women to avoid the appearance of wearing home-sewn garments, without incurring the prohibitive expense of having a dress cut and sewn entirely by a professional. It could be argued that the participation of the rural woman in the sewing of her own dress is an early form of free labour. Terranova (2000) refers to free labour as the “moment” in which “knowledgeable consumption of culture is translated into productive activities that are pleasurably embraced and at the same time often shamelessly exploited” (p. 37). The plain- stitching carried out by the less affluent clients of dressmakers certainly shares some characteristics with Terranova's definition of free labour. A woman who carries out her own plain-stitching, possibly with the assistance of her family members, is indeed engaged in a productive activity, and one which it is not her professional onus to perform. She may well take it on with pleasure, or at least with good humour, if she sees it as an opportunity for togetherness and enjoyment with her family or friends. However, I quibble with the idea that the client in such a case is being shamelessly exploited. Rather, she is making a decision which has both economic and social implications. The dressmaker would most likely just as happily sell the complete dress at a higher price, rather than invite the participation of the customer— and the lower fee. The customer makes a choice to use her own labour to turn a dress into a more accessible good. The dress itself is another reason to trouble the idea of free labour: though the client is investing her labour in productive activity, she does gain something tangible from it (aside from any enjoyment she may get from the communal activity). She gets a dress, and the benefits that come with it.

A side-note in dressmaking, but one in line with debates about free labour, is the appearance, in the early 20th century, of “ready-to-make” garments offered by Australian catalogue retailers (Miller and Merrilees, 2004). The ready-to-make process involved a client choosing a style of dress, sending in her measurements, and receiving a pre-cut, pre-pinned set of fabric pieces which needed only to be sewn together in the appropriate configuration (ibid). With sewing machines more commonplace by this time, the effort involved in putting together such a garment would have been significantly reduced, if one owned a sewing machine. It is, however, interesting to note that, even in the ready-to-make system, the cutting of fabric was still a task carried out by a professional. Ginsburg (1972) echoes the importance of cutting, suggesting that it was the dearest skill in the arsenal of the mistress dressmaker,

62 and the last to be taught to a dressmaker-in-training. The skills and status distinction between cutter and sewer persisted into the 20th century American garment industry. When garment production became an industrialized trade, carried out in mixed-gender factories, the task of cutting was given to men, who were paid higher wages than their women co-workers, with women habitually restricted to sewing tasks (Von Drehle, 2003).

In short, the world of the bespoke dressmaker, at least in many western countries, during the 18th to early-20th centuries, was one which offered a degree of potential for upward mobility, provided that a dressmaker-in-training was able to acquire the tools, skills and manners of her trade. In favourable circumstances, a successful professional dressmaker might well make a career and a life for herself, and might employ a series of trainee labourers. The mistress dressmaker owned her tools, rented her premises, and controlled the labour of her employees. In some cases, she may even have been considered responsible for the moral welfare of her workers. The class of clientele she was capable of attracting would form one of the bases for her success, with high class clients attracting other high class clients, and with even one or two particularly good clients serving as an attraction for other clients of an aspirational mindset (Crowston, 2001; Gamber, 1992). The dressmaker succeeded or failed by her skill, her thrift, her relationships and her management of her employees, contractors and apprentices. She acted much as Marx (1861) describes the prototypical craftsperson, with almost all of the hallmarks of a capitalist (her ownership of the tools, her purchase of the labour of her employees), with the one crucial exception that she, too, knew her trade and was skilled in her craft.

The introduction of graded sizing The trajectories of clothing for men and women varied in the 19th century. In part, because the fashion in women's clothing was for tight-fitting bodices, women's garments were custom-produced into the end of the 19th century and, depending on locale or social class, well into the 20th century. Gordon notes that, as late as "1890 women's styles accounted for only 25 percent of factory-made clothing" (2004, p. 70). By comparison, show shops and slop shops catering to men began to proliferate in England at the end of the 17th century, apparently thanks to changes in fashion which made clothing for men significantly more forgiving

63 (Ginsburg, 1972). Crowston notes that “the production of men's suits involved a more structured division of labor than did women's dresses” (2001, p. 90). This allowed tailors of men's garments to take on more workers, and to leverage the efforts of home workers and sweaters15 more effectively.16 Trautman (1979) posits that the move to mass production of men's garments was helped along by the relative homogeneity of styles for men, compared against those offered to women. In addition, when fit was not a concern, mass production proliferated, as with garments for prisoners and soldiers, as evidenced by the existence of the British slop trade (Rogers, 1997).

For men in Britain, the tailoring trade was changing at the end of the 18th century. Linebaugh (1992) argues that "the material conditions of production that had formerly permitted the tailor his 'cabbage'17 were removed. By the 1790s cutting by pattern had ceased to be the exclusive practice of secretive masters,18 and had become widespread with the publication of printed patterns that provided exact directions as to how to cut each piece, and illustrations showing the most economical way to lay out the patterns of the cloth" (p. 439). For women, prior to the advent of mass production, a brisk trade in patterns and drafting systems started to take off in the 1800s, serving women who could not afford to pay dressmakers. By 1854, unsized paper patterns for undergarments were available in the United States from Mme. Demorest, a pioneer of paper pattern making and selling (Kidwell, 1979). By 1864, Demorest's was also offering a custom cutting service, in which customers could send in a set of measurements and get a pattern for a jacket or waist, cut to the appropriate size (ibid). By 1867, E. Butterick & Co. was offering paper patterns in graded sizes, allowing women to choose a pre-defined size and modify if desired (ibid). These moves represent a kind of “thing knowledge” (per Radder, 2003), embodying the expertise of the dressmaker, at least in her capacity as a fitter or pattern designer, into a paper pattern. Though Radder describes thing knowledge as inhering in scientific instruments, the thing knowledge of the pattern or pre-cut dress pieces is knowledge of how to draft and size a dress pattern, and indeed how to perform

15 A “sweater” in this case being not a garment, but a capitalist controlling a stable of sweated workers. 16 Marx (1861) argues that the condition of the sweated worker or outworker is one of having “sunk down below the average level of the normal workers,” allowing sweaters to act as middlemen, taking a standard wage from a factory owner and putting work out at a reduced rate, deriving profit from the differential. 17 “Cabbage” being a tailor's perk, bundles of scrap fabric picked up off of the floor of the tailoring establishment and sold to scrap merchants, with the profit going to the individual tailors who collected and sold the scrap (Linebaugh, 1992). 18 The implication here being that tailors had adopted patterns long before dressmakers did.

64 what was historically seen as the most skilled portion of the process of custom dressmaking. In this move, we can see the division of labour not necessarily changing (for women who assisted in the sewing of their dresses, based on the expertise of a hired dressmaker), but the milieux of labour becoming more distributed. Certainly, the interaction between the dress pattern drafter and the purchaser/user changes in this scenario. Rather than the dress being the result of a number of discussions and interactions between the customer and the mistress dressmaker, the mail-order dress cutting or pattern purchasing systems turn the interaction between the customer and the pattern drafter into one in which the customer simply sends off her measurements (or, in the case of graded pattern sizes, simply asks for the desired size) and receives the makings of a dress (or a pattern) in return. She has no direct interaction with a professional dressmaker, and is doing her sewing in isolation from whoever drafted the pattern and cut the pieces. In this sense, we see, in patterns and pre-cut dresses, a move away from the craftsperson as “repository of the technical knowledge of the production process (Braverman, 1974, p. 133). Rather, the expertise of the dressmaker becomes embedded in the pattern or the pre-cut pieces of fabric, and the interaction between customer and dressmaker lessens.

In France, Green (1994) argues, the wars of the 19th century assisted the development of ready-to-wear garments for men by both “stimulat[ing] demand” and “encourag[ing] a reorganization of the production process that spread from military to civilian workshops” (p. 728). Green also suggests that the “mass measurement of soldiers” aided in the development of standard clothing sizes and that, paired with an understanding of proportion and an interest in anthropometry19, these new data points helped to bring graded clothing sizes for men into the mainstream (ibid), civilizing mass produced clothing and moving it out of the slop shop and into a more reputable sphere. Women had no such war to fight and were thus exempted from being the beneficiaries of what is one of the first instances of large-scale mass production of clothing. Green further suggests that similar changes in women's garments coincided with their move in more significant numbers into the workforce towards the end of the 19th century, and to a “masculinization” of women's garments, which concurs with Gamber's suggestion that the shirtwaist was a watershed moment in the mass manufacturing of

19 Anthropometry being the science of human measurement and proportion. See coons, 2014 for a more thorough history and critique.

65 women's garments. It appears that the confluence of greater numbers of women entering the workplace, the popularity of the shirtwaist and the newfound respectability of mass produced garments in menswear provided an opportunity for existing methods of mass production to move into the women's garment trade. That move heralded changes in the conditions under which women's garments were produced (moving from small workshops to factories and the outworking system), and indeed, in which they were used.

In the segment of the clothing trade devoted to slop shops and military uniform contractors, we see the most extreme kind of intermediation between the end-user of the garment and the person or persons producing the garment. In stark contrast against the individualized, reciprocal relationship between a lady (the term is used here advisedly) and her dressmaker, the soldiers fighting France's 19th century wars had no relationship whatsoever to the factory workers producing their garments. Because of this intermediation, and because of an effort to speed up and make affordable the process of producing thousands upon thousands of uniforms, the sizing and proportions of garments were necessarily rationalized, a departure from the high-quality garments of tailors, made to fit and flatter a particular individual. This highlights a radical change in the production of garments: the change in the role of the user in the process of production. In short, the place of the user became simply that: a user of a product, with no say in its production and no influence on its shape or construction. Contrary to the contribution of the affluent th19 century woman who even sourced her own fabric, France's soldiers were given a uniform standardized in every way. Green (1994) characterizes this is one of the major changes in garment manufacture: “the production of garments in the absence of known, measurable customers” (p. 729). This is the quintessential move from custom to mass production, from a customer having some degree of control or input in the good being produced, to one simple choice: buy or don't buy the product, as is. In the case of the soldiers being given uniforms, not even that choice would have been offered.

As the garment trade gave way to divided, fragmented labour, the efficiencies of the sewing machine, and the adoption of graded sizes, there was some push-back, particularly from tailors. As Green puts it, “highly skilled and articulate mid-nineteenth-century tailors defended their craft by identifying their mode of production with art and relegating ready-to-

66 wear to an artless, industrial production” and “castigated putting out and homework as fragmenting the work, demeaning the worker's labor, and lowering taste” (p. 730). Schorman (1996) suggests that the tailors were joined by customers in resisting the move to mass production of men's garments, stating that “[a]t the end of the [19th] century, clothing manufacturers still defensively advertised their wares as being 'equal to custom made' and took pains to distance themselves from the industry's past” (p. 111). Regardless, by the turn of the 20th century, at least the lower rungs of both men's and women's fashion had become occupied by readymade, with other segments soon to follow.

In the account of dressmakers in the 18th through early-20th centuries, we see a complex system of production and trade which had, in many locales, a clear hierarchy of work, defined by skill and experience, and sometimes instantiated in government-granted charters. In addition, the place of women in society, and the intimate nature of the work involved in dressmaking caused it to be an industry almost entirely dominated by women, and one which afforded women an opportunity at a livelihood and a chance to run a business. Crucially, it was a process which also involved the purchaser in the process of production, through the sourcing of her own fabric, the provision of guidance on style, and the attendance at multiple fitting sessions for each garment. In the move to mass production, many of these factors changed. Not least of all, with the popularization of sweated work and putting out in the needle trade, the production of ladies' garments ceased to be an opportunity for advancement. Outside of the small shop employing workers of varying levels of skill, a formal path towards advancement became less clear. The direct involvement of the customer in production ceased, and, in the absence of a definite customer, skills like fitting were removed from the arsenal of the newly formed needlewoman.

Introducing the LGRU: Looking back to look forward As we see above, the move from custom to mass is not nearly so tidy as one might imagine it to be. Even in a single industry, such as the garment trade, the move from custom to mass takes place at different times for different portions of the population. It is certainly not a story of change directly caused by technical innovation. Instead, as with so many changes, it is a story of parallel evolution, of regulatory changes, of conflicts between actors, and of

67 persistent co-existence of old and new methods. In many ways, this coexistence of old and new methods is present in the case study I present next.

In the description of the Libre Graphics Research Unit project, the need to examine old tools and methods is emphasized as crucial in developing new tools. Describing one particular theme area, the LGRU wiki states "We will analyse the history of lay-out (from moveable type to compositing engines) in order to better understand how relations between workflow, material and media have been coded into our tools" (LGRU, n.d.). In the context of the LGRU project, which seeks to create design software capable of breaking out of current established paradigms (most of those paradigms being based on the Adobe Creative Suite), the appeal to look back at historical methods shows a desire to see how things might have been done before the current modes became dominant, and to use that perspective to consider what assumptions are built into existing tools. The exhortation to examine old methods in devising new ones applies not just to the LGRU, but to all production practices. Some production practices, such as the use of Taylorism as a means of organizing work, are so ingrained that they almost entirely cease to be visible, or they come to be taken for granted. As Braverman (1974) points out, we no longer think of Taylorism as present in the factory, but only because it has so imbued every action of the modern production line that it simply is the dominant form, no longer needing to be called by a specific name (a point which I will return to later). In providing an example of how to trouble dominant forms of production, in this case through the creation of bespoke software, the LGRU offers an interesting insight into how we might de-naturalize our modes of organizing and doing production. In the remainder of this chapter, I contrast the LGRU case against the early-Industrial dressmakers in order to highlight the differences and similarities between two practices of custom production. Though both cases are compelling examples of custom production processes that integrate the labour and input of the user, they exist over a hundred year apart, and with hugely different tools and outputs. By comparing the two, I aim to both show the similarities between them, and to demonstrate the utility of considering the social organization of labour, the material conditions of production, and the role of the user, as those three considerations apply to the study of custom production practices, both present and historical.

68 The LGRU

Scratching one's own itch It's a truism of F/LOSS that many people build software to scratch their own itches. Scratching your own itch works like this: you have a problem and can't conveniently find something that solves it. You write yourself a piece of software to solve the problem. Your itch is now scratched. You might well post your software somewhere later, maybe on a repository- hosting site like Github, or just for download from your blog or website, because someone else might have the same problem as you, and there's no reason not to make your solution available. But even if your program does ultimately get used by someone else at some point in the future, the thing that caused it to exist was your need, your itch. That rationale is a kind of one-to-one, entirely custom production. One man, one donkey, in a Waldenesque rhetoric of self-reliance. One man, one software compiler. You have a problem, you fix it for yourself. It's the opposite of mass production, where someone other than you makes a product that solves an average problem, moderately well, for a large number of people in order to profit from the sale of that solution, and you buy it in an effort to meet your own instance of that supposed collective need.

What, then, about a project suggesting that, rather than simply scratching our own itches, people should get together in multidisciplinary groups and scratch each other's itches, or even some specialized collective itches? I now turn to some of the process and output of the Libre Graphics Research Unit (LGRU), a two-year, European Commission-funded initiative to bring software developers, designers and other stakeholders in the F/LOSS graphics space together to develop new methods, tools, insights and paradigms for graphic design. Based on the premise that much existing Libre Graphics (a broad term used to describe all Free/Libre and Open Source Software used in computer graphics applications) software mimics already- existing proprietary graphics software, the LGRU had as its goal the production of new software which does not aim to replicate established proprietary tools. Running from 2011 to 2013, the project involved four major partner organizations and a group of collaborating partners. The four media arts organizations forming the core of the project were Constant, based in , Belgium; WORM, based in Rotterdam, Netherlands; Piksel, based in Bergen, Norway; and Medialab Prado, in Madrid, Spain. A common trait of the four partner

69 organizations was their existing stake in F/LOSS methods and tools. The project was structured around a number of research meetings, each encompassing a specific theme, and each hosted by one of the major collaborators (with the exception of the final meeting, which was hosted in Porto, Portugal by one of the smaller partners, with the assistance of Constant). In addition to the meetings, there was also a series of commissions, a collection of texts (referred to as the Reader) intended to disseminate theoretical outcomes, and a number of public, show-and-tell-style activities.

In major part, the LGRU was intended as a way to "[p]rovide [a] platform to artists and designers to engage in the construction of tools and standards that matter to them" (LGRU Wiki, 2011). This is a theme that carries through the outputs produced and the discussions carried out. In service of this goal, the LGRU meetings in Brussels, Bergen, Rotterdam, Madrid and Porto were attended by artists, designers, software developers and engineers, publishers of crowd-written software manuals, and a number of other stakeholders involved in the production and use of Libre Graphics software. The meetings and outputs (as well as the contents of the Reader) were organized around four themes:

• Networked Graphics, focusing on distributed collaboration, networked collaboration and changes to the relationships between users and developers, as well as the lines drawn between those categories; • Co-position, addressing layout and the ways in which existing and historical layout tools determine what it is possible to produce, with the ultimate goal being to imagine new ways of doing layout; • Piksels and Lines, which explored existing ways of producing images, with both the somewhat eponymous pixels and lines represented by raster and vector graphics;20 • Abstracting Craft, which interpreted the manual as an interface between users and developers.

20 Raster-based graphics are comprised of a number of pixels set out in a grid. Each pixel contains colour information, allowing the grid of pixels to make up an image. In vector-based graphics, images are generated by programmatic or mathematical definitions of points, lines and curves. This means that vector graphics can, in effect, be written the same way one would write other software code. In open formats like SVG (Scalable Vector Graphics), the easy parsability of the markup language defining the graphics makes programming or auto-generating graphics especially accessible.

70 Research meetings organized around the four themes were carried out between 2011 and 2012, hosted by three of the four partners. Two meetings (Networked Graphics and Abstracting Craft) were held in Rotterdam and hosted by WORM; Co-position was held in Brussels, hosted by Constant; and Piksels and Lines was hosted in Bergen and hosted by Piksel. During this time, commissions were also carried out around each of the themes. 2013 saw more public dissemination of LGRU works: the Libre Graphics Workstation was installed in Madrid at Medialab Prado; a performance of the Piksels and Lines Orchestra was given at Libre Graphics Meeting 2013, hosted by Medialab Prado, which coincided with a week-long hack session—given in the cadre of the LGRU—called “Tools for a Read-Write World;” and a final meeting, focusing on collecting documentation of the LGRU, was held in Porto, organized collaboratively by Constant and Manufactura Independente, one of the smaller partners in the project.

This chapter makes use of the corpus of documents and outputs produced during the LGRU project to uncover insights into the ways in which the practices of the LGRU represent a kind of custom tool production led by user-developer teams. The corpus of data used here is comprised of the public-facing documentation produced over the course of the LGRU, which includes websites, wikis, meeting schedules, project descriptions, catalogues, software outputs, and git repositories. The following sections highlight some of the themes which have emerged in my analysis of these public-facing LGRU texts and artefacts, starting with an idea which places the LGRU work squarely in the world and practices of Free/Libre and Open Source Software development.

A “Classic set of tools”(+git) In the wiki page describing the LGRU, the combination of a mailing list, a wiki, and an IRC channel is referred to as a “Classic set of tools for communicating news and materials between project participants” (LGRU Wiki, 2011). These three tools are stalwarts of F/LOSS development and organizational practices. To describe them in brief, a mailing list is a channel of communication via email, in which subscribers can (normally) both write to the list and receive anything anyone else writes to the list. It is used for both broadcast communication

71 and discussion amongst list subscribers. Most mailing lists also have online archives of their activity, automatically generated by the software running the list. Depending on the settings chosen by the list administrator, the archive might be either public or private. Mailing lists in the sense used here are distinct from uni-directional mailing lists used solely as a broadcast medium. A wiki, possibly the most well-known tool of the set, follows a somewhat similar principle, but on a different substrate. A wiki is a website which can be both read and edited by all of its members (and sometimes, by the general public, without membership), and which shows the history of its edits. Many software projects use wikis as a more permanent, public- facing organizational tool, using them to display things like project roadmaps and planned features. Finally, IRC, Internet Relay Chat, is a protocol for both group and individual synchronous discussion. Many F/LOSS projects maintain their own IRC channel as a place to have meetings, offer support to users, or simply for those involved in the project to hang around with each other and be available to chat when necessary. The three tools are often used to complement each other, with, for example, the log of a meeting held in a project's IRC channel being subsequently sent to the project mailing list or posted for posterity to a project wiki.

Along with the so-called classic set of tools, the LGRU also makes use of git, a popular version control system. In effect, git allows developers to both upload code into a centralized repository (while also keeping a synchronized copy on each contributor's own computer), and to create a history of the development of the code, with contributions attributed to their contributors. Version control systems allow changes to be reverted (which is also a feature in many popular wiki platforms) and for a process called “branching” or “forking” which allows a project repository to be duplicated and modified, independent of the original repository. Branches can be incorporated back into the main project through a process called a pull request in which the owner of a branch requests that someone with administrative permission on the main trunk accept a request to incorporate particular changes back into the primary project. This is one of the major ways that F/LOSS development is currently done, with commit privileges being a badge of strong inclusion and commitment in many F/LOSS projects.

72 It is significant that this set of tools is used for the LGRU project. Though it may seem obvious that a project which aims to produce software should employ the tools of software development, the specific “classic set” is one which speaks strongly to the heritage of F/LOSS. In chapter two, I suggested that a definition of F/LOSS which serves both functional and ideological purposes involves the sharing and re-use of codebases and other artefacts, and the tools and licenses which make that sharing possible. In practice, this is achieved with tools like mailing lists, wikis and IRC channels, which allow discussion between developers (and users, and any other category of participant we might choose to delineate), and with version control systems, which allow code and assets to be both shared across distance and have their histories recorded. While the arts organizations involved in the LGRU project are already at home with some or all of the tools described above, the use of those tools is still something of a case of meeting developers in their own space, or taking on the trappings of development. Rather than change the toolchain being used in development, the LGRU rationale specifically recognizes the functional value of those tools, not only for software development, but for the organization and implementation of other kinds of projects.

This tactic—incorporating the tools of F/LOSS software development into projects which have outcomes other than software—is in evidence elsewhere. Many projects producing cultural works or doing organizing and advocacy work in the F/LOSS space use some or all of the classic set, as well as git. Mailing lists and wikis may be the most prevalent, with groups engaged in F/LOSS-related organizing and Free Cultural21 work commonly using mailman, a popular mailing list software, to organize amongst themselves. Wikis are also commonly used for everything from collaborative writing to organizing conferences. A particularly strong example of a non-software project which makes use of the classic set is LibrePlanet, an umbrella group (or “planet”22) sponsored by the Free Software Foundation, which serves as a

21 The term “Free Culture” is derived from the work of Lawrence Lessig, who wrote Free Culture, a book which has become something of a bible for individuals and groups who want to emphasize sharing and enrichment of the public domain in their works. This desire is most often instantiated in the use of either permissive (licenses which place minimal restrictions on the distribution or modification of the work) or copyleft licenses (licenses which require re-users of a work to license their derivative works under a similar license), allowing others to re-distribute and modify free cultural works, with varying stipulations about the conditions of re- use. 22 The term “planet” is commonly used in F/LOSS circles to indicate something like an aggregator, a venue (whether it be a blog, an RSS feed, or something more complex, like the infrastructure of LibrePlanet) for representing a set of related interests. The major hallmark of a planet is that it hosts material produced by a number of different people or groups engaged in one particular subject or interest area.

73 hub for organizing around Free Culture and Free Software advocacy. Its website, libreplanet.org, is wiki-based. Smaller interest groups within the planet have mailing lists, and an IRC channel (#libreplanet) exists for synchronous communication. The LibrePlanet wiki also serves as a venue for facilitating the organization of an annual conference.

Among the groups represented in the LGRU, many are existing users of the classic set+git. Open Source Publishing, an offshoot of Constant, uses the set+git in its design practice, with particularly interesting implementations of git. CREATE, the global community for Libre Graphics, bases its activities around a mailing list and a wiki. Manufactura Independente, one of the smaller partners in the LGRU, uses git to great effect, for both its traditional use, and in hosting branchable cultural works other than software. This is one of the more interesting trends in the use of the classic set+git. A number of groups in the LGRU, like Open Source Publishing and Manufactura Independente, blend more traditional forms of cultural production—like graphic design and publishing—with software development. These groups use their git repositories to host small, purpose-built scripts alongside other elements implicated in design and publishing. This blurring of design and software development leads to a second important thematic in the LGRU work, the role of the user in production or, perhaps more accurately, the blurring of the line between users and producers in the LGRU.

Interfaces between users and developers One of the stronger themes present in the LGRU work is a blurring of the distinction between users and developers. This is particularly interesting in the context of this dissertation because it offers the possibility that custom digital goods might be produced by the individuals intending to use them, creating a one-to-one relation between user and product, or, as it is termed above, scratching one's own itch. In the context of the LGRU, this blurring takes a number of forms, breaking down the distinctions between amateur and professional, user and developer, and artist/designer and developer. These kinds of blurring are present in both the work performed under the aegis of the LGRU and its four strands of enquiry, and in the other work of its members and member organizations in both their own practices and in their involvement with the LGRU.

74 As in the examples of Open Source Publishing and Manufactura Independente, the blurring of user and developer can be a move from a more designerly background towards greater use of development methodologies. In their use of the classic set+git, these designers take on practices and tools originally used in software development environments. This is one sense of the blurring between users and developers, with even those who are not necessarily contributing to the codebases of software projects adopting practices which help them understand and empathize with the ways of working favoured by their developer compatriots and, beyond empathizing, to make use of practices they find valuable. Beyond the use of tools traditionally found in development, there is also the practice of scripting small programs or add-ons. Manufactura is a useful example of this, having created a suite of scripts they refer to as “Tiny type tools,” which assist in type design tasks traditionally seen as time-consuming or tedious. The in-house development of small, specialized tools by OSP does similar work, with members identifying problems (itches, even) and developing small tools or add-ons for other tools in order to solve those problems (scratching) and then making them publicly-available through their website or git repositories for secondary use by others. These examples of user- developers identifying areas where they find their work tedious, inefficient, or un-optimal offer a glimpse into an idea I will elaborate further in chapter five: skilled practitioners automating (or accepting automation of) tasks that they don't find enjoyable or essential to control manually.

An interface of a different sort, performing bridging between users and developers is the manual or documentation for a piece of software. Manuals, as with those produced during the booksprints championed by FLOSS Manuals, one of the smaller partners in the LGRU, offer an introduction to a piece of software, or instructions on how to carry out a task or class of tasks. In the FLOSS Manuals booksprint format, a group of stakeholders gets together over a period of time (something in the neighbourhood of a week) to collaboratively write a manual. In one example of a manual produced by a booksprint, upon comparison of the list of book contributors to the currently active members of the development mailing list for the software being described in the book, I found that fully half of the authors in the booksprint (3/6 co-authors) have been active on the software project's development mailing list over the last calendar year. This speaks to that manual as being a comparatively direct communication

75 between those intimately acquainted with the project—through its development process—and those wishing to learn about that piece of software. In this particular case, the manual truly is an interface between developers and users, providing a bridge between the development community and the software's potential users.

This speaks to something of a difference in kinds of communication. One could argue that the medium of mailing list, or at least the development mailing list, is a way for developers to communicate with each other, or for particularly motivated users to engage with development. Git repositories serve a similar purpose, collecting additions and suggestions, and providing a venue from which to download the most current version of a piece of software. Manuals, wikis and other forms of more didactic, static documentation serve as ways of communicating to the outside world, and to those who are interested in learning about the software, but not necessarily in participating in its development. In many ways, these public- facing documents exist to educate and perhaps even indoctrinate potential users. The documentation, whether developed by people directly involved in creating the software or not, gives users and potential users an insight into how the software works, what it is capable of doing, and, in some cases, who does what or what recourse there is when things go wrong. These channels encode knowledge about projects; provide space for discussion, interaction and rapport building; and, in a small way, sell the project to potential new users or developers.

76 Illustration 4: LGRU sticker with the "Practice Shapes Tools Shape Practice" slogan. (photo by the author)

"practice shapes tools shape practice" The LGRU project has a unique and compelling way of articulating something that has become a kind of received knowledge among a certain subset of academics and designers: that, to paraphrase Langdon Winner, artefacts have politics. Winner (1980) uses the story of the parkways leading out of New York City—spearheaded by Robert Moses, who was the chief planner for the city for much of the mid-20th century—which had overpasses too short to admit busses, thus preventing the poor (poor in this case being defined as non-car owning) from visiting the beaches and parklands to which those roads led. Though Moses' overpasses are something of an apocryphal story (Joerges, 1999), the observation has spurred much reflection, and has been adopted as a value in urban planning and design discourses. Such observations are even key to a small but growing field, Values in Design, which studies the

77 embedding of social and moral values into designed objects, and seeks to make thoughtful interventions. In the case of the LGRU, the concern with ideas and capabilities being embedded into tools is articulated in the phrase “practice shapes tools shape practice” which appears on a number of LGRU materials, including a rather ingenious sticker in which the words are laid out in a circle, forming an infinite loop (see Illustration 4).

In the LGRU articulation of the idea that values can be embedded in designed objects, a number of themes come up: the relationship between content and form; the way the characteristics of a tool can shape the content it produces; “overdetermined functionality;” relations between tools and practice; and the idea of tools that “think with (or against)” their users (LGRU, 2013). Though these themes are articulated throughout the LGRU project, they are perhaps most closely examined in the Co-position strand, which seeks to understand the constraints of existing layout tools in order to imagine new tools, and in which the idea of looking back to look forward is highlighted. The research meeting and worksessions structuring the Co-position strand took place in Brussels in February 2012 and were hosted by Constant. Like other strands of the LGRU, Co-position included a set of deliverables, including a commissioned piece of software. The Co-position project starts from the conceit that page layout is based on 600 years of history, with very little change in the way its process is imagined, even as it moves from something executed in lead type to something done on a modern computer. In “16 Case Stories,” the catalogue produced as a deliverable of the Co- position strand, the authors state that

“Each tool-of-a-trade is tightly linked to professional conventions and ways of doing. While using them side by side, we understand how much they are 'path dependent', meaning that software, like any technology, is often the result of more or less arbitrary conflations of people and situations, sometimes leading to unreflected fossilisations (the text-box model) or other times unnecessarily overlooking relevant practices (chevronnage)” (from “16 case-stories”)

The authors/compilers of “16 case-stories” go on to suggest that, in the realm of page layout, “[w]hat exists, seems to have been directly exported from Gutenberg and shows surprisingly little traces of it's [sic] encounter with digital systems” (ibid). The purpose of the strand, then, was to challenge such received knowledge, and to conceive of new ways of imagining text

78 layout, while considering the constraints and histories of the tools employed. One of the ways the authors/compilers of “16 Case Stories” illuminate embedded values in software is with the inclusion of prompts, like the following, in which they ask readers to think about the decisions underlying tools:

“Exercise: Comparing different ways code reveals itself in design tools (GIMP-scripting, Blender, , Scribus etc.)” (ibid, italics mine)

As with the example above, many of the experiments and discussions arising from the Co-position work look at the decisions taken in programmatic contexts. In one vignette provided in the catalogue, cartographic depictions of a contested region in Spain are compared, with the rich data provided by OpenStreetmap in stark contrast to the blankness and erasure of Google Maps. In a less obviously politically charged example, the different methods vector graphics programs use to interpret paths are discussed, with questions arising about the impact of such interpretations on applications like milling and laser cutting. In both examples, differing decisions taken in the development of supposedly analogous tools are compared, highlighting the ways in which a supposed equivalency between the two tools might in fact be false. This is a critique of values embedded in design in the clearest sense: side-by-side comparisons of tools, with their differences outlined, and the impacts of those differences made clear, whether those differences be overtly political, or constrain the range of functions available to the user. Carrying on from the identification of such issues, in the next point of analysis, I explore the interventions produced and proposed in some of the strands of the LGRU project.

New design through new tools v. re-imagining existing tools Two ideas from the LGRU which emerged in my analysis seem at first glance to be in direct conflict: facilitating new ways of doing design by creating new tools, contrasted against re-imagining, re-writing and remixing existing tools. At first glance, the two themes appear to be represented in two discrete portions of the LGRU work, and do not seem to be in dialogue with each other. In the initial description of the Co-position theme, the idea of producing new tools in order to do new work is well-represented, with work in this area aiming to think

79 about the decisions embedded in existing tools as a jumping-off point for considering new tools with different embedded assumptions. In the description of the Piksels and Lines project, the idea of re-imagining existing tools is heavily present, with a "particular focus on improvements, interoperability between and fringe use of F/LOSS graphic bitmap and vector software, as well as generative software used in performative contexts." This theme, embodied by the Piksels and Lines Orchestra commission, builds a new practice on top of existing software.

In practice, the two ideas seem to merge and co-exist. In the early phases of my engagement with the LGRU documentation, it appeared that the two ideas existed independent of each other, and possibly were only able to co-exist because they were instantiated in different research meetings and commissions. However, upon reviewing the catalogue produced to document the Co-position theme, my interpretation changed. Here, we see the early work of the Piksels and Lines Orchestra being represented as an exploration of the layout concerns articulated in Co-position. Though the work took place in Bergen, at the Piksels and Lines themed worksession, it was then collected up in the discussion of Co- position ideas, under the heading “Process Aware,” along with other ideas about how to make design processes legible. A recounting of the early work done on the Piksels and Lines Orchestra articulates it as a way of adding a new layer to existing design software, sonifying the work of layout, providing a new layer of awareness of what is happening, as it happens. Later, the Orchestra became a commission, undertaken under the Piksels and Lines mandate, and ultimately presented at the Madrid meeting.

Conversely, the Co-position strand, under its own aegis, produced a new application, which takes the strand's thinking and exemplifies the idea of creating new tools to do new design. Where the Piksels and Lines Orchestra offers a way of being more conscious, reflexive, or even entertained by existing design practices, the Graphical Shell commission builds a new way of making images. Combining an online canvas which can be concurrently accessed by a number of different users at once, with command line tools for manipulating objects on the canvas, Graphical Shell seems to jump off from the questions about scripting and legibility of actions presented in the Co-position catalogue. It very explicitly enacts ways for users to

80 understand what their tools are doing for them at a given moment, and, through that transparency, to counteract some of the overdetermination found in the way other tools handle issues around image creation and layout.

Ways of working together One of the key takeaways from the LGRU project and its attendant events and commissions is a diversity of ways in which heterogeneous groups can work together to achieve collective outcomes. An interesting aspect of this is how each of the four major partners used their own methods of organizing, their own formats for events, and their own assets to produce meetings and work sessions. Though the particulars of the meetings were often different, with specific emphases or terms and organizing tactics used by individual partners, a number of elements stayed roughly the same. One such element is the importance placed on informal, un-structured work and discussion time. An example of this is found in the program of the first meeting hosted by WORM in Rotterdam:

"20.00 an optional, ad-hoc evening programme in which we can use WORMs space for further discussion, extended meet&greet and project presentations. Projectors, sound systems and workspaces available."

This approach makes a point of providing the necessary resources to carry out discussions (space to work, projection capabilities) while leaving the time un-structured, and is equally present in the common concept of the “unconference.” An unconference is an event during which the program is determined by the interests of the participants, leaving the schedule to emerge, rather than being a prescriptive structure put in place before everyone has had a chance to meet and talk. The unconference approach was used in the Piksel-hosted LGRU meeting, with room left open for participants to work on projects which had not been integrated into the schedule ahead of time. This approach is common in a variety of venues, and is heavily present in F/LOSS events. PyCon, one of the larger and more prominent events in the F/LOSS event calendar, offers what it calls “Open Spaces.” These spaces give attendees of the conference a chance to organize themselves on an ad hoc basis and invite other interested parties to join in discussions which are scheduled based on mutual agreement during the

81 conference (Python Software Foundation). Though, of course, it would be possible for attendees of events like PyCon to decide for themselves that they wish to discuss something not included in the event schedule, the official sanctioning and provision of resources (space, technology, etc.) is a notable feature of both the LGRU implementations of un-scheduled time and its analogues in larger conferences. For the LGRU, though, it is especially crucial, as the size of the event and its status as a work meeting makes it less convenient or acceptable for attendees to simply drop out of a scheduled event in order to have an off-topic discussion.

Beyond the un-structured time that is, in effect, a standard format in F/LOSS events, several of the practices of working together evidenced in the LGRU meetings offer interesting, localized formats. In the Constant-hosted meeting, what are termed "Kitchentable presentations" appear in the schedule. These presentations, which involve discussion accompanied by a meal, offer a less hierarchical way of discussing issues and projects by placing everyone equally around a table, and effectively changing the presentation dynamic. On the other hand, in the Interactivos? format championed by Medialab Prado, we see something more akin to a hack week, with participants given space and resources to work on a project. Like other structured hacking events, the Interactivos? sessions start with presentations from individuals championing ideas, who then become the nucleus for larger working groups made up of previously-unattached participants. The groups, thus formed, then discuss their ideas with advisors, who attend the event throughout, acting as sounding boards for the projects under development. In the LGRU-sponsored Interactivos? event which took place in spring 2013, groups approached the structure in different ways. Some chose to work off-site, convening in areas they found to be quieter or more conducive to productivity. Some participants drifted from project to project. Other groups stayed fiercely cohesive. All of this was supplemented by periodic, scheduled check ins with the group as a whole, and with the advisors. A notice board was also installed, asking groups to check in on a daily basis, indicating their progress, questions or needs that had arisen, and feedback or assistance they wanted from the rest of the group. Sometimes, those requests were triaged efficiently and resolved, other times, they remained unfulfilled. Regardless, the implementation of an asynchronous way of providing project updates allowed the individual groups more freedom, in the same way that having scheduled meetings with advisors allowed the groups to work on

82 their own terms and in their own spaces, knowing when their next formal opportunity for feedback would be.

While the methods used in the LGRU meetings and work sessions are, in and of themselves, an interesting case (and somewhat typical of broader themes in face-to-face F/LOSS collaboration), perhaps the more relevant discussions about working together are rooted in the explicit discussions around collaboration highlighted in the LGRU projects. There is a clear interest, among a number of the LGRU themes and projects, in the issue of distributed collaboration. In the Networked Graphics thread, one of the guiding premises is the following: "The web has changed our perspective on creative production; work has become networked and distributed, challenging conventional ideas about specialism and professialism [sic]." The Co-position theme takes "decentralized collaboration" as a concern and asks the specific question "How can we understand what workflows, distributions of work and media are coded into our tools?" All of these questions and statements show a clear interest in querying the nature of collaboration, often at a distance, and how our tools shape the ways in which such collaborations happen. This is evidenced further by the collaborative nature of the Graphical Shell project outlined above, and several of the projects which arose from Interactivos?, many of which had collaboration, discussion and version control at their core (displayed in the publication which resulted from the event). Whether the strong interest in collaboration, and particularly distributed collaboration, springs from the LGRU's relationship to F/LOSS, which foregrounds distributed collaboration, or whether the interest comes from a more general interest in current changes to design methodology is an open question.

83 How are software developers like dressmakers?

Because this chapter is about the parallels between pre/early-Industrial Revolution dressmakers and the participants in the LGRU and because I introduced it with the question “how are software developers like dressmakers?”, I will now draw some direct comparisons between the two. These comparisons fall into three high-level categories visible in and emergent from both cases: the social organization of labour; the material conditions of production; and the role of the user in production. Taken together, these three categories offer an insight into some of the similarities between women working in the 18th through early-20th century custom garment trades and a distributed group of pan-European artists and software developers working together to create new design tools. They also form the framework through which I will explore the case study presented in chapter five, and around which I will organize the further, more speculative and theoretic chapters of this dissertation.

The social organization of labour Material to both the LGRU and the dressmakers are issues of social organization, social hierarchy, and the division of labour by skill, experience or volition. In the case of the dressmakers, class is also a determinant in the organization of labour. In both cases, the skills (both level and type) of participants vary, with expertise in different areas distributed amongst members of the group. For the dressmakers, such division is hierarchical and based on job requirements and, in some cases, corporate definitions of status. This is, to put it mildly, less the case in the LGRU, which largely eschews formal hierarchies in the treatment of its members. Regardless of those differences, there are still some productive parallels to be drawn between the ways the two groups organize themselves (or are organized by others).

Tasks in the dressmaker's shop were divided by skill level, with more junior employees taking on more formulaic, less skilled tasks. The hierarchy of progression, in some cases, led from apprenticeship and menial labour, to the acquisition of specialized trade skills, up to the top-level tasks of business management and dealing with customers, taken on almost uniformly by the mistress herself. Beyond the role of the employees, some tasks were apportioned (or, if not apportioned, then more genteelly given over by custom or request) to the client, giving her a creative stake in the outcome of production, whether those tasks be the

84 purchasing of fabric or, among poorer clients, doing plain-stitching in order to reduce the cost of the dressmaker's services.

In addition to the formal hierarchies of work amongst the members of the dressmaker's workshop, some attention was given to the social circumstances of workers. Regardless of the attitude of the individual mistress dressmaker towards the stewardship of her employees and apprentices, in 19th century England, there was an assumption that such a thing should be considered and that it was a large enough social issue to necessitate government commissions of enquiry and florid newspaper reports (Rogers, 1997). This attention towards the moral welfare and working conditions of women in the needle trades was also directed towards men in similar trades, and towards women in other industries and coincided, according to Rogers, with a public discourse which emphasized the harm of the city and the wholesomeness of rural and family life for women. Efforts to regulate the labour of women in the needle trades were spurred on by such attitudes (ibid). In a different country and time, in th17 century Paris, the obtaining of a corporate charter allowed female dressmakers the opportunity to work at all, giving them a sense of legitimacy and a set of guild rules under which to trade. Those rules also happened to lay out the tasks they needed to do and the skills they had to acquire in order to become fully-fledged members of their trade. In both cases, a combination of advocacy and government intervention and legislation structured the ways in which dressmakers were allowed to work, or the ways in which employers were meant to treat their employees.

Though the LGRU, being situated in the 21st century, does not explicitly concern itself with the moral welfare of its participants and does not operate under a government-granted guild charter with clear, hierarchical steps towards mastery (or mistressry, as the case may be), suitably modern social structures and modes of organizing are still visible and are, at times, not so different from some of the structures adopted by (or imposed on) the dressmakers. As a project funded by the European Commission, the LGRU does, in many ways, rely on the buy- in of a government authority in order to function. Though documents of incorporation do not exist (because it is a fixed term project, rather than a formal organization), a proposal document outlining the roles and duties of the various LGRU members does lay the groundwork for collaboration and participation. The strands of the LGRU work are laid out in

85 its initial proposal documents, as are the particular roles of each group (eg: the responsibility of Constant to serve as the administrative organization). Though the membership of the LGRU is significantly smaller than the number of Parisian seamstresses privy to guild membership at any given point in the 17th and 18th centuries (by 1691, for example, there were on the order of 1,000 mistresses [Crowston 2001]), the function of the organizing document is similar: to lay out ground rules and responsibilities, subject to individual implementation, by a group of members and the affiliated organizations to which they belong. Like a guild, the LGRU is an organization of organizations. Unlike the dressmakers, labour in the LGRU is not explicitly gendered. Where the charter of the Parisian dressmakers enforces a system in which women are in the business of catering to women, the LGRU, unsurprisingly, does not have a mandate which restricts the gender of its participants. More notable, however, is the composition of genders represented in the LRGU compared to other F/LOSS activities. In the group photo taken during the Brussels Co-position meeting in 2012, of 36 participants pictured, nine of them are visibly identifiable as women, which amounts to 25%. In F/LOSS circles, that proportion is an achievement. By contrast, PyCon (mentioned above as one of the larger F/LOSS community events) achieved a proportion of 20% women speakers and attendees in 2013 (Skalski, 2013), which was considered an achievement and an example to other F/LOSS communities. That the LGRU included such a comparatively high proportion of female participants is perhaps a testament to its organizers and their position as employees and members of arts organizations, which have a greater heritage of inclusivity than software development communities.

Back at the shop level, there are comparisons to be drawn from the relation between individual workers and the structure within which they work as factors in both the example of the dressmakers and the LGRU. Within the whole of the LGRU project, and within the individual working strands, there is evidence of a nuanced interplay between individual participants and the projects on which they work. Both the LGRU and the dressmakers are, in ways and in varying degrees, loose or temporary affiliations of individuals, with an investment in the current, collective work (whether it be a dress or a piece of software), but with a further, longer term investment in the benefits derived by the individual. Whether those are immediate, tangible benefits (like the payment for a commission within the

86 framework of the LGRU, or the lodging, board and wages of a live-in worker in a dressmaker's shop; in other words, the tools with which to reproduce her labour), or less tangible benefits which manifest themselves over time (networks built, items added to CVs, new skills learned; a trade acquired, progress towards a masterpiece, the eventual potential for independence and self-employment), the individuals implicated in both the LGRU and the dressmaker's shop have an individual investment in their own future, as well as a present investment in the work being done. Though some of these traits (like an individual investment in one's own future) are common to a number of ways of organizing work, they differ from more formal modes of work in that a strict division of life and labour is not enforced. In the case of the dressmakers, this is because of the somewhat maternalistic structure of the shop. In the case of the LGRU, this is because of the concentrated nature of the activity, taking place in short, intense periods of time during which participants from different geographic locales come together to collaborate.

Similarly, though the concept of an apprentice's indenture somewhat belies it, the dressmaker's employees and the participants in the LGRU are all, ultimately, only temporarily bound up into an organization. Granted, the levels of investment and obligation vary hugely, and there is really no comparing the life of an indenture-paying apprentice in the 18th or 19th century against the work of a present-day artist/developer executing a commission, but the overarching social structure has similar traits. Individuals within the groups are loosely or temporarily affiliated, but ultimately able to leave or seek out other work and affiliations (unless they're an indentured apprentice, of course). Though the commitment and time scale differ, the apprentice ultimately serves her indenture in order to one day become a comparatively self-determining mistress (comparatively, that is, against the capacity for self- determination of a woman living during the same historical period but working at something other than a skilled trade, such as domestic service). Similarly, the members of the LGRU participate for a time in order to achieve a set of goals, before ultimately dissolving and moving on to new things (which could be compared against formal employment, an arrangement that assumes a longer term of investment). Similar arguments have previously been made about the role of participants in distributed work, in general (as with Castells' concept of “the space of flows” in, eg. Castells, 1999), and software projects in particular (as in

87 D'Mello & Sahay, 2007), but the relationship between the individual and the assemblage has often been portrayed as atypical (for example, Weber [2004], which argues that Open Source software development represents a novel kind of political economy which might be exported to other areas; or anything by Eric Raymond [esp. Raymond, 2001], who makes similar arguments). Instead, I suggest that there is a corporateness (in the historical sense, but certainly not in the modern sense of an obligation to generate financial profit) in projects like the LGRU which mirrors similar practices of temporary association familiar to the structures present in the skilled trades of the pre/early-Industrial era. Though I do not mean to romanticize the pre/early-Industrial guild system, which certainly had problems of its own, there is an instructive comparison to be made between what are, in effect, two different kinds of organizations which encourage temporary association/employment, but which, in theory, at least, seek to foster a level of skill and self-development in their members which, when members become fully trained, can be used as a pre-requisite to starting their own shops or furthering their own practices. Crowston (2001) suggests that as many as half of all apprentice seamstresses in Paris during the 17th and 18th centuries never finished their training and, thus, were never allowed the opportunity to become either journeywomen or mistresses themselves, while both Crowston (2001) and Nenadic (1998) make the compelling assertion that the early stages of indenture were less about learning a trade and more about making coffee (something that bears a resemblance to the role of present day interns, and which bears similar marks in terms of its role in using free or even fee-paying variable capital to produce value), sweeping, and providing a paying labour force capable of doing menial tasks and offsetting costs. Indeed, the comparatively high number of women trying to make a living in the dressmaking field was cited by social reformers as a factor in their ill-use and exploitation (Rogers, 1997). Without taking the position of seeing trainee dressmakers as purely economic actors, I would like to contend that the model of limited term, loose association claimed by the apprenticeship system (whether or not it was actually enacted in its ideal form) does provide a structure similar to that of the LGRU in its apparent commitment to helping grow skill and experience among a corps of participants who then take those skills to other venues.

88 More generally, there is an interesting parallel between arguments about software development (beyond the scope of the LGRU and even beyond F/LOSS) and arguments about manual skilled trades. Ensmenger (2010) tells a portion of the story of the development of IBM's OS/360 operating system, and the book spawned by it, The Mythical Man Month (now considered something of an integral work in thinking about software engineering), written by Frederick Brooks, one of the managers of the project. Ensmenger argues that in The Mythical Man Month, "Brooks' larger point was that computer programming, as an inherently artistic activity, was resistant to most forms of industrial production" (p. 47). The solution, then, was to make the programmer the focal point of the process. Ensmenger goes on to suggest that though "[t]he development methodology that Brooks outlined was never widely adopted in industry, [...] his larger argument about the inherently creative nature of programming was" (p. 47). Taking for granted the creative value of the programmer and his apparent inability to be mechanized or routinized makes an interesting departure from the trajectories of other craft trades. In the history of programming, it is seen as something of a discovery: that the people programming the machines who, in early days, were seen as clerical workers, might actually be exercising control and creative agency and, as such, should be treated as skilled23 (Ensmenger, 2010). This is arguably the reverse of the fate of the dressmaker, whose trade was ultimately routinized, compartmentalized and de-skilled down to a level where sewers in mass-production today do the work of 19th century slop workers, exercising no creative control over their labour and very little by way of approved or authorized self-determination (and with limited scope for even un-authorized self-determination). This classic Marxist argument has been made elsewhere, in essentially any analysis of an industry undergoing Taylorization (eg: Braverman, 1974; Crawford, 2009; Bahnisch, 2000), but is particularly interesting when contrasted against a trade like computer programming which appears, at least in part and for the moment, to have followed the opposite course. In the case of the garment trade, what Braverman refers to as “the separation of conception from execution” (1974, p. 114) is out in full force. Managers or planners wield the power to define and allocate tasks, while sewing operatives carry out allotted tasks repeatedly—what Braverman calls “the division of labor in detail” (1974, p. 72). Though this kind of division of labour is present in the

23 Ensmenger also notes that this discovery was accompanied by a notable shift in the gender of computer programmers. When programming was considered clerical work, it was dominated by women. When it became skilled, creative, analytical work, it masculinized, to the detriment of the women who had pioneered the field at a time when software was considered secondary to hardware.

89 custom dressmaker's workshop, it is on the assumption that the employees participating in it have some chance of learning the skills required to eventually undertake all parts of the task.

The material conditions of production Among custom dressmakers, it was the norm to employ live-in apprentices and workers, and to keep minimal stock. Dressmakers generally lived and worked in the same place, and, along with their employees, worked long hours. The divisions in labour were based on divisions in skill and the use of materials. It is not for nothing that the term tailor derives from tailler, to cut (Crowston, 2001), a skilled task, while seamstresses, sewers by appellation, were seen as less skilled.24 The prestige of cutting inhered in its comparative risk, with a bad cut ruining cloth, wasting money and ultimately raising the possibility of an upbraiding by the customer. As Crowston argues, stitches can be undone, while a bad cut cannot. In this sense, the hierarchy of labour in the dressmaking shop is very specifically housed in the material tasks of the workers. A cutter is, by dint of her relationship with the fabric, given more authority and more trust than a sewer. As such, it is immediately visible, by looking at the tasks of individual workers, to see their place in the hierarchy of the workshop. The mistress measures the client, the skilled workers cut and do fancy sewing, the apprentices fetch and carry and do repetitive, un-skilled work. Crowston puts it plainly when she writes that "[m]aking dresses required many hours of relatively simple stitching that could be relegated to a capable apprentice" (p. 124). The menial labour of the apprentice was essential to the final product, but certainly not desirable as a task for skilled workers.

The above does not hold true for the LGRU. Where the seamstresses show their place in a formal hierarchy through the particular tasks they take on, with those tasks being closely tied to their relationship with the fabric, the lack of a formal hierarchy in the organization of the LGRU somewhat precludes such signifiers of rank from manifesting themselves in the material tasks of the members. Indeed, even if we were to draw some kind of arbitrary hierarchy based on, for example, skill at computer programming, fitting the highly heterogeneous work of the LGRU into it would be exceptionally difficult. Instead, the material conditions of production in the LGRU manifest themselves in different ways. Those ways have

24 It is also possibly for this reason that skilled needlewomen did not call themselves seamstresses, but instead went by such appellations as dressmaker or mantuamaker (Crowston, 2001; Miller, 2006).

90 less to do with the maintenance of hierarchy, and more to do with the facilitation of collaborative work. As I suggest above, it is problematic to compare the working conditions of 18th to early-20th century seamstresses against the conditions in which a largely voluntary25 group of artists and developers elect to produce software together. Though the LGRU does, unlike many other projects, use money as a way of facilitating work, the status of F/LOSS developers as generally affluent (or at least not entirely precarious and affluent in comparison to the conditions experienced by pre- and early-Industrial tradeswomen) cannot be ignored. It would be irresponsible for me to draw a direct parallel between the conditions of ill-paid and often ill-treated needleworkers and present day professionals of any kind. However, with that caveat in mind, I wish to make a comparison between how the material conditions of the dressmakers' labour structure and are structured by their object of production, as well as their social structure, and how similar work is done in the LGRU setting. Within the LGRU, those organizing the events and commissions are in the role of enabler more than anything else. Rather than directing work, they facilitate the self-guided or collectively-guided work of others. Rather than serving as overseers or even the most skilled people in the room (both of which were ideally the case for the mistress in the dressmaking shop), those organizing LGRU events ensure structure in areas where it helps, diminish it or change it when it becomes counterproductive, and seek to provide the resources required to work productively. In part, this difference may be one of control: the LGRU is a loose partnership of organizations and individuals, without a monolithic hierarchy, and without, for example, a direct chain of command or a formal structure which requires that participants meet the expectations of their employer. Inasmuch as some participants were paid to do things like execute commissions, and others were employees of the member organizations, relations between individuals involved in the LGRU still did not involve the same kind of direct control that a mistress could exert over her employees and apprentices.

In the LGRU project, the importance of the material conditions of production is made far clearer than in many other F/LOSS enterprises. Whether in the collective meals written into the schedule for the meeting hosted by Constant, or in the list of audiovisual materials available during informal work sessions at WORM, the parties involved in organizing LGRU

25 I say “largely voluntary” because some members of the group were paid for their labour, either in the form of commissions, or as employees of the sponsoring organizations.

91 events make it clear that a set of material infrastructures are necessary for the execution of their work. In some ways, this is unsurprising, given their staunchly material habits. Compared against software projects which base themselves in ad hoc organizations formed around digital spaces26 and only occasionally meeting up in person, the arts organizations comprising the core of the LGRU all have a large stake in space and infrastructure. They have offices, galleries, meeting spaces, and employees, they hold exhibitions and meetings, and forge relationships with the arts communities and granting organizations in the cities, regions, and countries in which they are situated. They must spend money on overheads including those involved in the maintenance of offices and staff. In many ways, they centralize ownership of basic tools and resources which other, less infrastructure-oriented groups might assemble on an ad hoc basis, out of the possessions or means of their individual members. A painfully simple example of this is the suitcase full of extension cords and power bars owned by Constant, which enables the computer-toting participants in its activities to fulfill a basic need. At Medialab Prado, an organization with a somewhat more permanent public-facing mandate, the infrastructure includes a re-configurable lecture hall and technicians who are employed to make it work. All of these assets and infrastructures, whether as trivial as a power cord or as major as a lecture hall, are distinguishing factors in the work of the LGRU's member organizations, which somewhat buck the F/LOSS habit of outsourcing physical infrastructure.

In the material conditions of production, there is also the matter of the artefact that is being produced. The dressmakers are in the business of producing a dress for a specific client, most likely using material she has brought in. Because the dressmakers in question were operating before the widespread adoption of paper patterns, each dress was unique, based on the mock-up of thin fabric built around the client, which eventually became the lining of the dress. The dress, in this sense, is a rivalrous good, one which embodies the knowledge and skill of the dressmaker. The dress could, potentially, have its stitches picked out in order to derive component pieces from which a similar shape could be traced, but that is the only physical instantiation of the dress's plan. The measurements of the client, kept on a strip of paper, represent her to the dressmaker, but do not tell the whole story of the dress. By 26 Although this is not to say that the digital spaces in which they meet are not also reliant on physical infrastructure. Indeed, as I have stated elsewhere in this dissertation, it is impossible to untie the digital from its physical substrates, to such an extent that the term “digital” as a differentiator seems almost meaningless.

92 contrast, the work of the LGRU is embodied in software and documents which are infinitely re-producible and distributable—non-rival. In part, this distinction means that multiple members of the LGRU, with the aid of version control systems, can work on a given piece of software or documentation at the same time. It also means that barriers to participation are slightly diminished compared to the dressmakers. When the fabric of a dress is not only valuable, but not the property of the dressmaker, there is a huge impetus to prevent a less skilled member of the workshop from damaging it. By contrast, with a piece of software housed in a version control system, not only can changes be reverted, but multiple versions of the software can be kept, allowing parallel development. This is one key distinction between the material conditions of production experienced by the dressmakers and the LGRU participants: the LGRU, thanks to its focus on non-rival goods, has the potential to take greater risks in development and to allow members to follow their own paths, should they so desire, rather than carrying out a hierarchically defined master plan.

The role of the user in production The introduction to this chapter highlights a contrast between custom when one does it for themself and custom when the end-user is not the producer. In the case of the dressmaker, the identity of the end user is fairly easy to determine, being the client for whom the dress is made. Quite possibly, someone else will later become a secondary user, with a servant being gifted a garment by an affluent employer, or in a less affluent scenario, a sister or other female relative repurposing a dress given over by its first owner. Unlike software, which, within the constraints of access to infrastructure, can be infinitely shared and copied, the movements and re-uses of the dress are relatively straightforward. As such, it is far easier to think about the place of the dress's end user in its production than it is to think about the involvement of any number of possible, eventual end users of a piece of software. With that in mind, I consider here the place of the most immediate end user of LGRU software who, in many cases, will have at very least been one of the people conceiving of it in the first place.

Both the custom dressmakers, and the activities carried out in the course of the LGRU project speak to an interest in creating individual solutions to problems, and in end-users being both stakeholders and partners in the production of solutions, although I suspect that

93 the motivation behind this decision differs in the two contexts. As with the families who brought in dressmakers to execute the more complex, skill-intensive work on dresses, while the women of the family sewed skirts and did other plain-stitching, we can look upon the relationships between designers and developers (although I cringe a little at reifying that distinction) in the LGRU project similarly. In that case, designers and developers are engaged in collaborative work, with developers bringing more specialized programming skills, while designers, artists and organizers help to set the direction of the work and do the metaphorical plain-stitching, at least with regard to the production of the software. Though in the case of the LGRU work, there isn't a business relationship present, the idea that many artists and designers, by doing the scripting and small modifications of software that many of those engaged in the LGRU certainly do, are engaging in something akin to home sewing. If left entirely to their own devices, the products of their work might prove to be extremely time- consuming to make, and most likely of lower quality or complexity than similar goods made by those with greater knowledge of software development. Instead, through the collaborative process, they are creating goods that they want or need, which would be ill-fitting or inappropriate if bought mass-produced, but prohibitively expensive if contracted for in a completely bespoke way. The willingness for the two groups to collaborate opens up a middle ground, where the developer gets a good idea of the needs being met and a knowledgeable collaborator, while the designer or artist, with a practical stake in the project, can assist where possible, or at the very least, feel enough of an investment and commitment to support the developer in the execution of the project. In the two pieces of software documented above, such a division of labour is apparent. The conception of the project is a collective enterprise, with even some prototyping done collaboratively, but the work of execution is given over to a skilled programmer.

This kind of behaviour is carried out in other F/LOSS projects. People who do not consider themselves to be “developers” as such, but who want to be involved with or contribute to a project, might find ways of contributing that match their skillsets. Common ways of making non-code contributions include writing documentation, creating graphics or other visual assets, doing translation work, and so on. In making these kinds of efforts, which can contribute to both the life of the project and the software itself, self-professed non-

94 developers once again take up the role of metaphorical plain-stitching, doing tasks within their abilities, while advancing the progress of the larger project. Though the direct benefit to the plain-stitcher might not be as apparent in software contributions as it is in dressmaking, there is arguably a degree of self-interest. As in many characterizations of communal work, like the barn raising or the more general Norwegian concept of dugnad,27 many members of a community contribute to the execution of a project, often for communal gain, and sometimes with no directly visible positive impact for certain individual members. We could liken this to family members trading labour, pitching in on the plain-stitching on each other's hemlines. The community, whether it be a family or a software project, obtains a collective benefit from the pooled labour of the individuals, and each individual gains something from the gains of the community, whether it be a dress (or the promise of a hand-me-down in the future) or an improvement to a piece of software.

The two cases explored in this chapter present what appear to be opposite sides on a spectrum of production. The historical case looks at a process at home in the early-Industrial period, when some tasks were mechanized already (like the production of cloth), while others were still accomplished entirely by hand (like the cutting and stitching of a dress). Regardless, it is a case which, for a time, defied wholesale mechanization. By contrast, the LGRU, though coding may be seen as something of a craft practice, is firmly situated in a time when production of a great deal of things has become both mechanized and digitized, with computer control of industrial production now commonplace. Indeed, even software development is more automated than it was fifty years ago (Ensmenger, 2010). The LGRU sought to produce software, screen-based goods which can be distributed through the internet, but which were based on the interests and provocations of the LGRU participants, while custom dressmakers created goods which were paid for and, in some senses, defined by their clients.

The next chapter looks at a case which falls somewhere between the pure software of the LGRU (which, regardless of the purity of its output, still requires a diverse array of physical inputs) and the early-Industrial craft labour of the dressmakers. In short, the case represented in the next chapter takes the lessons learned here and apply them to a context in

27 Dugnad being a form of collective labour, much like a barnraising, but with the distinction that, unlike in a barnraising, the ultimate benefit is collective rather than individual.

95 which digital fabrication meets the still very traditional and craftsmanlike work of the prosthetist, while still engaging with the tensions between digital production and custom goods.

96 5 Interfaces between bodies and standards

In this chapter, I draw on field notes taken during the project described herein. Field notes, when used verbatim, are indicated by being set in italics and indented.

PrintAbility

This chapter maps roughly to the structure used in chapter four. I start by providing some background and context for the case study under examination here, then follow with a series of significant vignettes which arise from my documentation and analysis of the project. In recounting and analyzing the vignettes, I provide some initial connections to relevant scholarship, as well as to the high-level themes identified in the previous chapter. Following the vignettes, I go into more detail in connecting the current case to the cases presented in the previous chapter, with a particular eye to mapping the vignettes and other findings to the social organization of labour, the material conditions of production, and the role of the user in production—concepts which structured the previous chapter, structure this chapter, and will provide the organizing principle for the remaining chapters of this dissertation.

The case study I present in this chapter concerns a project called PrintAbility (sometimes written as printability, printAbility, or even 3D PrintAbility)—officially instantiated

97 in summer 2013—which seeks to use commodity 3D scanning, 3D modelling and 3D printing tools to speed up the production process for prosthetic legs in a hospital in Uganda. One of the hopes driving the project is that, with a successful first implementation in Uganda, the process might then be brought to other milieux. Though the implementation of the system as it currently stands specifically aims to speed up the process of producing a socket for a child receiving a trans-tibial prosthesis at one particular hospital, a second goal of the project, and one which is taking place in the development stages of the process, rather than the implementation, is to explore the ways in which people might become lead users through access to expertise and equipment. This element of the work took place in Canada, with participants drawn from the local population of prosthesis users. The findings from the portion of the work which concerns lead users were then used as a form of feedback in developing the system, in the form in which it was implemented at the participating Ugandan hospital, CoRSU (Comprehensive Rehabilitation Services in Uganda). The model of consulting with lead users has also been carried over to the CoRSU context, with one patient so far taking on the role of lead user/expert patient.

The project involves a number of different actors. In the Critical Making Lab (which is part of the Semaphore Research Cluster on Mobile and Pervasive Computing, or Semaphore for short) at the University of Toronto, Dr. Matt Ratto, the Principal Investigator (PI) of the lab, and myself, form the core of the group, which is to say we are the two participants from UofT who have stayed consistent throughout the duration of the project covered here. However, in addition, a rotating cast of research assistants and interns is present, with a summer intern completing a bachelor's degree in architecture; a member who started out as a research assistant in his final semester of an Electrical and Computer Engineering degree (also at the bachelor's level), before graduating and continuing to volunteer on the project; and a Master of Engineering student with a background in aerospace engineering making up the rest of the group. After the end of my formal period of involvement with the project, other students also joined the team, coming to the project with different responsibilities and expertise. Beyond the group at the Critical Making Lab, the major partners in the project are CBM Canada, a non- governmental organization focusing on providing assistive devices and medical treatment in the developing world; CoRSU, a rehabilitation hospital located in Entebbe, Uganda which

98 works closely with (and receives funding from) CBM (both CBM Canada and other branches of its global parent organization); and Autodesk Research, the research arm of a large software company specializing in 3D modelling and computer-assisted design (CAD). Those sets of actors can largely be considered as three organizations: the university, CBM Canada, and Autodesk. In the early portions of this chapter, CoRSU is considered as part of CBM Canada28 because all communication with the staff members at CoRSU was mediated by the CBM Canada employees with whom the university team had direct contact. At the end of the chapter, CoRSU comes into its own, with a brief account of some of the activities which took place on the Canadian team's trip to visit CoRSU. Autodesk's involvement is embodied by the participation of one person: Ryan, who leads the Design and Fabrication team at Autodesk Research, and who is also the primary developer of Meshmixer, a piece of software which plays a major role in the PrintAbility project. As such, Ryan is, in practice, constituted as a member of the research team, though he is not currently affiliated with the university.

The trans-tibial prosthesis Some of the specifics of a trans-tibial prosthesis bear mentioning, if only to provide context and terminology necessary for the rest of this chapter. In very broad strokes, a trans- tibial prosthesis is a prosthetic leg worn by someone who has a functioning knee and upper- leg, but who is either missing much of the leg below the knee, or has some kind of condition below the knee which cannot be remedied with a brace or orthotic. A trans-tibial prosthesis (Illustration 5) is made up of a socket, a pylon, some kind of ankle joint, a foot, and a cosmetic covering known as a fairing (the fairing is not pictured in Illustration 5). The residual limb (also called a residuum, or a stump) fits into the socket. The socket is the one component of the prosthesis which is, in all cases, customized to the particular user. In varying degrees, depending on the context, available time, and available money to spend, other components of the prosthesis can be customized. In the case of the established method used in many developing world hospitals—comprising tools, procedures and a supply chain developed and maintained by the International Committee of the Red Cross—the socket and its liner are the only fully custom portions of the prosthetic leg, while other components are supplied by the ICRC or an affiliated distributor.

28 Though it may seem slightly cumbersome to consistently refer to CBM Canada by its full name, I do so in order to prevent confusion in the moments when the global organization enters the picture.

99 Because components like the pylon, foot, and ankle assembly are mass-produced and available in a very small range of variations (three different colourways, adult and child sizes of the alignment parts, and a slightly broader range of foot sizes), the socket in the ICRC process is quite literally an interface between an individual and a standard. Between the highly-personal residual limb, which can be even more varied than one might expect from a piece of human anatomy, and the standardized parts of the ICRC trans- tibial prosthesis kit, the socket acts as a mediator. However, because of its custom nature, the socket is the portion of the prosthesis which is the most time-consuming and skill- intensive to make. It is also an important element to get right. If a Illustration 5: The standard form of a trans-tibial prosthesis, socket is ill-fitting, there is a good with the socket highlighted. The cosmetic fairing, which serves to make the prosthesis look more leg-like, is not chance that its owner will stop shown. (illustration by the author) wearing it, either because it is too uncomfortable, or because it fails to make the wearer's life easier. As such, the socket is the point of intervention in the PrintAbility project, and the one element which we aim to replace with a 3D-printed component.

100 Background

During the summer of 2013, CBM Canada approached the Innovations and Partnerships Office (IPO) at the University of Toronto, looking for a researcher to partner with on a project involving 3D printing for medical applications. The IPO approached Dr. Matt Ratto, who began talks with CBM Canada. Their discussions at that point revolved around scoping out what kind of project would be achievable and impactful. A number of options were floated, including the possibility of using 3D printing to produce orthotic devices. The conclusion of the discussions was, however, that the 3D-printing of prosthetic sockets for children requiring trans-tibial prostheses would be useful, doable and most beneficial.

Around the time the initial discussions between Matt and CBM Canada were taking place, I was trying to solidify my doctoral thesis proposal. While discussing with Matt what the major case study in my dissertation might be, he suggested that the generation of prosthetic sockets might be a good subject. I had previously been thinking about convening a workshop or hackathon in which a group of people with a similar interest and problem (I was considering urban cyclists, as this is a group I belong to and have connections with) would gather and use digital fabrication tools to come up with individualized solutions. The idea was to examine how different individuals with an ostensibly similar need could find differing ways to serve that need, based on their own preferences or requirements. Matt, in typical style, read a precis I had written about this idea and said “What if you changed 'cyclists' to 'prosthetics users?'” It seemed a reasonable suggestion. Users of prostheses exemplify the contrast I was hoping to study: a common problem, but with different individual needs, which might be addressed through a digital production process. I joined the project and became involved in much, although not all, of the planning work. This chapter addresses most all of the work and discussion to which I had access during the period from the inception of the project up to (and very partially including) our trip to Uganda in January 2015 to implement the project's pilot phase. After the trip to Uganda, though the project has continued, my involvement became drastically reduced, as I focused on this dissertation.

In July 2013, we completed an application for a competitive grant intended to support early-stage research on innovative health interventions in the developing world. At the time,

101 the PrintAbility process had not even remotely been solidified and was largely speculative. We knew that the three basic stages of the process would be “Capture, Design and Print,” but had not yet settled on much of the technology to be used. The only known portion of the toolchain at that point was that Meshmixer—a (monetarily) free piece of software from Autodesk with a strong capability for sculpting and good handling of scan data—would be used for modelling. Though we did not win a grant during the first funding round we applied in, we subsequently applied again, in a later round in spring 2014, and succeeded. However, in the interim, CBM Canada drew on its existing pool of donors to raise money to fund the project. They successfully raised $100,000 through these avenues, helped along in their efforts by a visit to the Critical Making Lab by a number of potential donors who took the opportunity to see the process as it was at the time, and to learn more about both the short- and long-term goals of the project.

Vignettes

The dependency grid In the early days of the project, Matt defined three stages in the process we would be developing: “Capture, Design and Print”, which have since morphed into “Scanning, Modelling and Printing.” Under the three headings, we set priorities, tasks, and budget items, which were used both for internal planning and in grant applications, and even appear codified in Gantt charts developed by an employee of CBM Canada who was responsible for creating the project's initial business plan. Within this framework, based on discussions we had at the time, and based on our literature reviews around issues in the production of prostheses using Computer-Assisted Design and Computer-Assisted Manufacturing (known, especially in the literature, as “CAD/CAM”), I devised what I termed a “dependency grid” (Illustration 6), which broke down the possibilities for each step, and listed what the dependencies or problems with each method might be. This grid was distributed among the group and served, at least in part, as a framework for thinking about which specific technologies to use for the three steps. Being able to make effective decisions about technologies has been a key issue throughout, as both 3D scanning and 3D printing are moving targets, with speedy development and commercialization happening in both areas.

102

Illustration 6: Dependency grid, circa summer/fall 2013

Ultimately, the dependency grid played very little role in the actual technical29 choices made for the project. This is largely because the grid served to educate members of the group who did not know very much about the technologies in use, rather than to organize the thoughts or criteria employed by the stakeholders who actually made technical decisions. Those of us who

29 Rachel and Woolgar (1995) make the very instructive point that what counts as a “technical” problem is a moving target and varies from group to group and moment to moment. They describe the delineation of a task as “technical” as representing a “break between those outside and those within” (p. 261), with the technical acting as a kind of black box, accessible only to those inside its space. They couch this observation in the apparently obvious, but often ignored relationship between the technical and the social, suggesting that "[t]echnical work is construed and displayed in such a way that it acquires a robustness which defies deconstruction. This is not just a result of the amount of conspicuous resources expended on technical work. The point is that such expenditure commits people to a course of action whereby they display and reaffirm the separateness and distinctiveness of the technical from the human world which necessarily creates, maintains, and uses it in all aspects of its existence" (p. 270). This characteristic of “the technical” is of particular interest in the way the barrier between technical and organizational work is represented in the PrintAbility case.

103 were responsible for making technical decisions did not need a didactic framework to make decisions, as the grid merely made explicit the things we already knew about 3D scanning, 3D modelling and 3D printing. What is worth noting about the disuse of the dependency grid is the particular role it played, and for whom. Though we originally imagined the grid as a way of structuring our technical decisions (and, indeed, did use the headings as the basis for other technical planning and task setting), because the grid effectively made explicit the implicit knowledge of the members of the research team, it was not necessary for our work. But as an educating tool when approaching other members of the larger project team, it served the purpose of creating a clearer understanding about the technical requirements of the project. In this regard, I'm somewhat tempted to think of it as a boundary object (per Star, 2010b), in that it aided in the facilitation of collaboration between somewhat divergent groups, not necessarily in agreement with one another, but working towards the same overall goal. However, I hesitate to think of the dependency grid as a boundary object, as any actual evidence of its efficacy in doing boundary work is slim. Instead, I think of the dependency grid as one place where a discussion of the social organization of labour becomes particularly relevant. In its capacity as a communicative tool, the grid offers a way to bridge between different groups within the larger project team, a subject about which I'll go into greater length later in this chapter.

Making decisions about scanners One of the major technical tasks in the initial period of the project (spanning roughly July-November 2013) was the evaluation of different 3D scanning technologies. We experimented with photogrammetry software, which makes use of a set of photographs encompassing the whole object to come up with a 3D model of the object pictured. In this vein, we tested both proprietary (Autodesk's 123D Catch) and F/LOSS (Insight 3D) tools. Some limitations stopped us from going further with these options. We disliked the slowness involved in taking a large set of photos (on the order of 20-40) and the lack of detail in the models produced. Attempting to use photogrammetry, as it currently stands, in capturing a body part also introduced more error than we wanted into the scan data, as keeping even an adult sufficiently still for long enough proved difficult. We were also concerned, in the case of 123D Catch, that using a cloud-based service, rather than a piece of software capable of doing

104 all of the processing on a local computer, might hinder the implementation of the scanning technology. We didn't want a steady internet connection to be a requirement for the process. The major appeal of photogrammetry had been that it required no specialized hardware, given that images could be captured with a standalone digital camera, phone camera, or tablet camera. However, in hindsight, the ready availability of such devices in the Ugandan context, especially those of sufficient quality, could be questioned. Indeed, subsequent observations at CoRSU show that feature phones are in more evidence than smartphones.

In November 2013, we began working with a hardware-based scanning solution. During a hackathon in which we participated, thanks to an acquaintance from a medical imaging laboratory, we tested out a PrimeSense Carmine, which is an RGBD (Red Green Blue Depth) camera. The Carmine is very similar to the sensor, also made by PrimeSense, which formed the basis for the first generation Microsoft Kinect sensor device, which is used to detect movement and gestures in order to control video games on the Microsoft Xbox 360 video game console. The Carmine we tested produced a high quality model in a short period of time when paired with a piece of scanning software (at the time, we used the demo version of a product called ReconstructMe), while also providing real-time feedback on the efficacy of the scan. After a successful test, we decided to purchase some Carmines to integrate into our system. Unfortunately, our decision roughly coincided with the purchase of PrimeSense by Apple, which caused the Carmine to become abruptly unavailable, just before we tried to purchase some. We opted to switch to the Kinect, which for the time being at least, had a reasonable certainty of continuing to be available, despite its use of a PrimeSense sensor as integral hardware.30

Later on, in early 2014, we changed our scanning software. Based on a lack of response from the company which makes ReconstructMe (we tried to procure a number of licenses, which is a task initiated by filling out a form on the company's website and then waiting to hear back), we switched to a piece of software called Skanect, which was easier to source. In fact, we ended up purchasing five licenses of Skanect, bundled with Kinect devices, from a website which specializes in selling equipment to students and educational institutions. This

30 Since then, Microsoft has released a new generation of Kinect sensor which is based on entirely different technology. Our early tests suggest that it is not currently usable as a 3D scanning device.

105 kind of purchasing based on either convenience or ability of vendors to comply with the rules of the academic institution was a common facet of the early stages of the project, as all procurement at that stage was being done through Semaphore, which is bound by university procurement policies. Our plan was to get these Kinects, order a couple of laptops capable of running the software, and send the whole package to Uganda for initial testing and use. At the same time, however, we also ordered two Sense scanners, which are another 3D scanner based on a PrimeSense sensor, but organized in a different form factor. Where the Kinect is designed to sit on a surface and have an object move around in front of it, the Sense is hand-held (although it still needs to be tethered by a USB cable to a computer) and designed to be moved around the object. In addition, the Sense is powered by a USB cable, while the Kinect requires an additional power cable. This is due to the form factor and functionality of the Kinect, which incorporates a small motor to rotate the device on its stand. The Sense (which is produced and sold by 3D Systems, a major manufacturer and distributor of 3D printers) was not available when we were initially making choices about scanning hardware and had thus had not been included in our original decision-making. When we initially received our order, we did some testing with the Sense, but not much. One frustration I had with the Sense was the aggressive way it installed its driver software. Due to an incompatibility with camera drivers, the simple act of plugging a Sense into a computer rendered the Kinect/Skanect combination unusable and then required manual deletion of software packages in order to make the Kinect functional again. I spent an entire work day troubleshooting this problem the first time we experienced it. This was one of the factors involved in my early dislike of the Sense. We later switched from the Kinect to the Sense. The decision was made by Matt and our summer intern, during July 2014, while the intern was tasked with the development of a protocol for carrying out the 3D scanning process. However, as with all technical decisions made over the course of the project, it was taken with a degree of impermanence in mind. Knowing that the Carmine is no longer available, the stability of the Sense as a solution is short-term at best, and the supply will eventually dry up. As with our decisions about computers and 3D printers, one of the ongoing tasks of the project, going forward, will be to consistently find replacements for the Sense. In this respect, the use of commodity or consumer-targeted technology makes PrintAbility a slightly different proposition than other medical technologies. Where a large piece of equipment specifically designed for a medical application

106 might have some assumption of longevity (and accompanying long-term support), we fully expect that our technologies will quickly be superseded by others. This is both an opportunity and a worry, as it allows us to consistently improve our process, but also causes it to become obsolete more quickly than most medical technologies.

The material conditions of production Decisions around scanning hardware and software reflect the importance of a concern for the material conditions of production. In highlighting which factors led to making a particular choice, we see far more clearly how a different choice might instead have been made, in a way denaturalizing what might otherwise appear to be a simple decision, or even an opaque fait accompli. In this regard, the tools of actor-network-theory might be of some use. As Latour (2005a) puts it, “ANT claims to be able to find order much betterafter having let the actors deploy the full range of controversies in which they are immersed” (p. 23). He further suggests that order comes not from attempting to settle controversies, but from “trac[ing] connections between the controversies themselves” (p. 23). I take this as a kind of call to action, and an exhortation to treat the conditions which make up an eventual decision, rather than simply describing the outcome of those decisions. As Latour does in Aramis, or, The love of technology (1996), I use my embeddedness in the PrintAbility project as a way of watching such controversies in action, as they form the eventual outcome. This commitment to looking at actors and relationships also provides a good reason to examine work as it happens, rather than the outcome produced, and to make a concerted effort to denaturalize the development of systems and technologies, through attention to the conditions under which those systems and their eventual outcomes are produced. The denaturalization of development is also a concern addressed by the study of the conjunction of the social organization of labour, the material conditions of production, and the role of the user in production. Those three frames for studying production processes offer a useful lever with which to dig into systems which might otherwise appear opaque. I pick up on this theme further in my comparison of the PrintAbility project against the LGRU and dressmaker cases, at the end of this chapter, highlighting, in particular the different kinds of ad hoc organization and allocation of work present in the PrintAbility and LGRU projects.

107 When is a 3D printer like a sewing machine? One of the constant frustrations in following public discourse around 3D printing is how easy everything is made to sound. Every 3D printer is a seemingly perfect, limitless factory for producing whatever its operator's heart desires. 3D printers can be dropped into any situation and are then able to flawlessly provide goods. In this sense, the 3D printer is treated as a thoroughly Modernist machine, transferrable to any venue with little change or differentiation.31 Stories like the 3D printing of a wrench on the International Space Station (NASA, 2014) reinforce the idea that a 3D printer is a black box with few or no externalities or potential problems. In a similar vein, the unproblematic black box is also presented as a danger and an opportunity for hysteria, as evidenced by the popular fear that 3D printers, combined with files for printing hand guns, will cause a sudden uptick in the availability of dangerous weapons. People who spend large amounts of time working with 3D printers know otherwise, and spend large amounts of time coping with the externalities and technical issues implicated in 3D printing (see, for example, Record, coons, Southwick & Ratto, 2015 on externalities in 3D-printed guns). One of the issues we have had to concern ourselves with in the development of the PrintAbility project has been the breakage, tweaking and maintenance of 3D printers. A primary worry has been that, in shipping 3D printers to Uganda, there's a huge possibility that parts will break or become misaligned. The field note below exemplifies such problems, and details a representative moment of printer troubleshooting.

13/September/2014 Unpacking the second Z18, afternoon of Friday, 12 September

It's around quarter to 11 on a Saturday morning and I've just remembered that I should take down some notes on what happened yesterday afternoon. I'm reminded of the incident because I've just been reading Zuboff, and she mentions that the Singer Sewing Machine Company had to employ skilled tradesmen to assemble its machines, mainly because they couldn't produce perfectly uniform parts.

Yesterday, Matt and Dan tried to get the second Z18 working. This is the one we're planning to ship to Uganda. The plan with this has been, for a couple months at least, to get it up and running here, test it out, run some prints, and

31 Harvey notably refers to Modernism as aiming to “speak to the eternal” (1990, p. 21)

108 then re-pack it, more securely, and ship it off to CoRSU. So yesterday afternoon, Matt and Dan tried to do that. I wasn't involved, other than as a spectator, mostly because I was practicing my digital rectification skills in advance of a bunch of workshops and demonstrations next week.

I wasn't there at the start of the process, because I'd gone off for lunch. When I got back, I found that they'd just finished taking the box off of the machine. It was still sitting on its palette, and they'd happily discovered that, rather than having to lift the printer out of the box, or destroy the box by cutting it down, the box was designed to be lifted off of the printer. So it was sitting out in the main room, on its palette. Just after I started casually observing what they were up to, they discovered that a part was loose. It was sort of metal plate, which is, I believe, supposed to be on the bottom of the printer. It wasn't affixed and, what's more, there seemed to be some screws missing. Dan determined that he would have to get in touch with the Makerbot people [...]. This is the second printer we've received in bad condition. So Matt and Dan spent their afternoon moving in and out of the fishbowl32 (where they ultimately moved the printer). At some point, I think after Dan left, Matt found a screw on the floor and realized that it was one of the missing screws from the plate. He wondered out loud if that meant that there would be other screws, hiding in the box.

When I next checked in on Matt's progress with the printer, he was expressing frustration with it. It had problems recognizing that it had a print-head attached and was alternating error messages. It also, at that moment, had filament stuck in the print head. [...]

Anyway. It was around 5:30 and Matt was grumbling at the printer. He was also bothered by the problems on another level. One of the worries about the prosthetics project is that printers might arrive at their destinations in bad shape. There's only so much we can do to insulate against that. One of the ways of doing that is what we're doing now: unpacking the printer in our lab, making sure it's working, and then re-packing it more securely. But there's still a risk.

32 In the Semaphore space, the “fishbowl” is a glassed-in, ventilated area which centralizes the activities of the Critical Making Lab and its members. The fishbowl is where 3D printers, laser cutters, power tools and other potentially disruptive, loud or smelly items are kept.

109 The activity described above was part of one of many days of work needed to prepare the Z18 for shipping to CoRSU. Beyond ensuring that all of the parts of the printer were there, its ability to complete a print accurately needed to be tested. Both of these tasks used the existing skill-base available in the lab. In this way, one of the biggest issues in the public understanding of 3D printing (as the technology currently stands) is the erasing of the role of the technical specialist. The status of the 3D printer as something that requires a great deal of time, skill and expertise to use and maintain is often ignored in narratives about its capabilities. In the case of the Z18 that was shipped to CoRSU, the labour of, ultimately, three people who use and maintain 3D printers on a daily basis was a necessary precondition for the printer's eventual functioning (and even then, it needed a little more work done once we got to CoRSU, an issue which is covered in a later portion of this chapter). Ignoring or collapsing this labour and expertise in order to argue that the 3D printer is somehow all-powerful makes the machine look like a far more potent actor and world-changing influence than it is currently capable of being.

So when is a 3D printer like a sewing machine? Arguably, the 3D printer is like any technology early in its development and popularization. Zuboff provides a very brief anecdote from industrial history. She says that “the Singer Sewing Machine Company was not able to produce perfectly interchangeable parts. As a result, they relied on skilled fitters to assemble each product” (1988, p. 39). The comparison to sewing machine parts that are imprecisely stamped or cut and need to be adjusted by skilled tradesmen in order to form part of the final product could just as easily be a comparison to the home computer components which require a level of knowledge and skill to be formed into a functioning assemblage. Sanding a cog so that it fits into a mechanism could just as easily be checking the compatibility of a sound card with a given motherboard. Both of these have parallels in the work done to render the newly- arrived 3D printer functional. None of these processes “just work” or function immediately out of the box. O'Connell gives us a useful way to think about how the “just working” or not of a collection of socio-technical parts can be interpreted: "When a bomb made in Massachusetts, a bomber made in California, and a bomber pilot trained in Colorado are brought together for the first time in Panama and expected to fight a war, they must fit together as if they had not been formed in separate local contexts" (1993, p. 163). The sewing

110 machine, the 3D printer, the constituent parts of a military action: all are examples of the need to attend to both the social organization of labour and to the material conditions of production. The two issues are inextricably linked in these cases because the success of the components relies on both the way the individuals using them are organized (for example, the part-stamper being a different person from the machine assembler) and the way the goods are produced (whether that be the imprecision of the part, the tools used by the assembler, or the simple act of noticing that the assembler is a skilled worker).

Connecting the socket One of the consistent topics of discussion over the course of the project has been about what hardware should attach the socket to the pylon, and how. In a typical prosthetic leg made in Canada, a metal fastener called a pyramid joint is attached to the distal end of the socket, and used as an adjustable connector which joins a metal pylon to the assembly. In the method currently being used at CoRSU33, which is provided by the International Committee of the Red Cross, a plastic cup is built into the bottom of the socket, which is attached to a polyurethane pylon by an adjustable, plastic alignment assembly. In previous work on prostheses for the developing world, novel, locally-sensitive solutions have been devised for building either components of the prosthesis, or the whole thing. For example, one process developed in India makes use of plastic drain pipes, which are molded to fit in much the same way as the sheet plastic used in the ICRC method (Andrysek, 2010). The decision to use the ICRC assembly in the first implementation of our project is based on the idea that the time it takes to produce a socket, and the precision of the socket's fit is currently the biggest problem, and that, though the ICRC pylon and alignment assembly might be more cumbersome than the pieces used in Western prostheses, they are an established solution to which the hospital already has access and is experienced with. Because PrintAbility is, first and foremost, designed to be used at CoRSU, the kind of local sensitivity we needed to exercise in our thinking about materials is somewhat different from the kinds of local sensitivity which see the implementation of systems which try to do without specialized tools or materials. Our local sensitivity needed to come in the form of recognizing the existing base of expertise and

33 In reality, two different methods are used at CoRSU. The ICRC method, described above, is the one with which the PrintAbility project is in dialogue. However, for affluent patients able to afford more expensive medical care, CoRSU also offers components which would be entirely at home in a Canadian prosthetist's office, including the pyramid joint.

111 equipment already present, rather than assuming a tabula rasa, or, worse, denying the expertise and skill of the practitioners in the Orthopaedic Workshop at CoRSU and the capacity of the hospital. As Fouché (2011) puts it, many western technological interventions carried out in the name of aid are a kind of “technological missionary work” (p. 62) and “sadly, though not purposefully, construct the receivers of these technological tools as empty vessels into which Western technological knowledge must be poured” (p. 73). Recognizing the pride of CoRSU staff in the capabilities of their institution was a key sensitivity in making decisions about what is and is not a reasonable opportunity for change in the development of at least the CoRSU implementation of the PrintAbility project and was crucial in our interactions with the hospital staff both before and during our trip to Uganda. Later in this chapter, I detail a specific example—the case of a patient named Rosaline and a choice about feet—in which local sensitivity was especially important.

Despite our conviction to work with the ICRC alignment assembly and pylon (and despite the installed base at CoRSU which supports its use), the idea of using different elements has been in circulation throughout the development of the project, as has the idea of adopting some ICRC parts but not others. One motivation for this has been the concern that the ICRC alignment hardware does not provide as much scope for re-adjustment as other solutions might, compromising the ability of the prosthesis to have its alignment adjusted as needed. At various times, in the lab, we've discussed the possibility of producing something like a pyramid joint, to be attached to a piece of metal pipe. This solution has been explored as far as the sketching phase, with a number of metaphorical napkin-back sketches of new alignment assemblies circulating in the lab. The field note extract below details one of the moments in which we grappled with the capabilities of the ICRC alignment hardware.

112 8/Sept/2014 [...] We wound up having a bit of a chat about how the socket will be attached to the rest of the ICRC leg. Matt had realized, over the weekend, I guess, that what we'd talked about last week (eliminating the washer and the alignment cup from the set-up and replacing them with a screw shaft built into the socket) wouldn't work. Basically, his realization amounts to: the people who designed the ICRC parts had more nuance in mind than we'd given them credit for, because the big hole in the alignment cup allows for some pretty decent alignment work to take place. So the big hole, the washer, the screw, the little round alignment thingy, and the top of the post/pylon all work together to allow for robust alignment of the socket to the pylon. The hole and washer setup make it so that the screw has a larger range in which it can be placed in relation to the pylon. The hole needs to be big, so our idea with a screw shaft built into the socket doesn't work. Smaller placement range. So the three of us [Matt, a research assistant, and myself] discussed alternatives for a bit. A few of the thoughts: can we drop the washer into the socket as it prints (my thought, apparently shared by Mitch [a high-ranking employee of CBM Canada]); a bunch of other stuff that I've now forgotten, because we just had a whole thing about the travel.

In cases like the one above, the negotiation has been between the existing ICRC hardware and the design of the socket, trying to use the capabilities of the ICRC materials, while also ensuring a strong connection to the socket. However, another avenue has been pushed by outside actors, involving the use of non-ICRC hardware. A sample of one such solution, produced by an organization called Limbs International, came with us when we went to CoRSU, and is comprised of a milled metal alignment disk and a metal pylon which can be cut with a hacksaw to achieve an appropriate length for the patient. These pieces are paired with a style of foot known as a Niagara foot, which has superior ankle action to the SACH (Solid Ankle Cushion Heel) foot used in the ICRC kit, but does not look like a foot, which can be problematic in terms of patient uptake and satisfaction.

113 The role of the user in the selection of components and the desire to look human With regard to prosthetic components like the Niagara foot, it's worth noting that many wearers of prostheses do not wish to look like cyborgs, and are willing to trade some functionality for the ability to integrate more closely into their communities. This is a concern both in venues like CoRSU, and among the Canadian wearers of prostheses I interviewed over the course of the project. Sobchack (2004) sheds some light on this issue by making a comparison between two different conceptions of the prosthetic. Herself a wearer of a trans- tibial prosthesis, Sobchack contrasts her attitude to her own prosthesis against Barthes' concept of the jet-man and other ideas of prosthetically augmented humans. She argues: "Not only do I see myself as fully human (if hardly singular or glamorous), but I also know intimately my prosthetic leg's essential inertia and lack of motivating volition" (p. 205). She goes on to suggest that “the experience and agency of those who, like myself, actually use prostheses without feeling 'posthuman'” has been largely erased from current conceptions of the prosthesis in academic discussion (p. 208). This popular, somewhat metaphorical, conception of the wearer of the prosthesis as somehow superhuman is only reinforced by public personalities like athlete, model and actor Aimee Mullins, a double trans-tibial amputee, whose collection of prosthetic legs allows her to perform a gamut of roles from paralympic runner to towering model to Barbie-legged ideal of feminine beauty (ibid). Mullins owns her identity as a cyborg, her performance in various areas of her life augmented by her choice of prostheses. By contrast, we see the distaste many wearers of prostheses feel towards the Niagara foot, or the case of Rosaline, a patient at CoRSU, whose brother, in the face of a lack of right-feet in the appropriate size, chose for her to have a left-foot fitted to her prosthesis, rather than a custom-built but not very-footlike foot made for her by Abdullah, a senior member of the Orthopaedic Workshop. These decisions are not about function, in a biomechanical sense, but about function in integrating into local communities. Many wearers of prostheses would simply rather map to what is viewed as “normal” or “human” than to augment or take on some kind of role as cyborg or superhuman.

In aiming to be sensitive towards local conditions, the issue of the socket connector and the attendant problems with component choice provides a first example of how the three overarching themes highlighted in this dissertation must work together. In considering the

114 social organization of labour in the example of Rosaline's foot, we see the patient (or, in Rosaline's case, her older brother) both engaging with the construction of the prosthesis, and determining its final shape. In considering the environment into which the prosthesis will be put, some apparently functional concerns are subordinated to social concerns. Though the foot built by Abdullah might well be biomechanically superior, the sanded down left-foot Rosaline was ultimately given looked less artificial, an important factor in the eyes of her brother. The iterative process involved in the production of the prosthesis, with a number of fittings taking place over the course of two days, provides opportunities for the patient or her guardian to exercise agency and have a say in the construction of the prosthesis. In this sense, the social and the material are inextricably linked, with the layout of the workshop and CoRSU's decision to provide on-site lodging to patients combining with the relationship between the workshop technicians and the patient to influence the outcome of the production process.

The Niagara foot (or even the foot made for Rosaline by Abdullah) can be seen as an artefact of thoroughly modern scientific rationalism. There is a focus on function, ignoring affect and social circumstances. Harding (2011) refers to the interrelation between scientific rationalism and post-World War II development activity, particularly noting the role of non- governmental organizations (NGOs) and international governmental organizations (IGOs, like the United Nations) in applying scientific rationalism as a cure for poverty, and as bringers of science to developing nations. Escobar picks up the same thread, emphasizing development as a process of turning “traditional” societies into “modern” ones (2011, p. 280). An easily portable modernity may be a popular goal among development agencies, but Harding stresses that, even when new technologies are dropped into developing countries, “the recipient society always changes what it borrows so that the new ideas, processes, or goods fit into the existing social order with minimum disruption" (2011, p. 9) rather than upending or changing the character of the society. To this end, it is especially key to highlight the importance of attending to the social organization of labour and material conditions of production in the workshop and the institution within which it operates. In the example of CoRSU, this might mean recognizing that patient scheduling takes place differently than it would at a Canadian hospital, and accepting the need to adapt the technical system to that mode of scheduling, if necessary. In so doing, we remember to admit local conditions and the agency of the people

115 working in those conditions. At the risk of sounding neoliberal or under-representing structural concerns, I believe that allowing room for local choices in the construction of descriptions and analysis of a project or workshop helps to somewhat upend the idea of direct technological transfer.

Choices about the values we embed in our software A major point of discussion and, occasionally, contention, over the course of the project has been what role our custom software, Socketmixer, should play in the learning and professional development of the clinicians and technicians using it. Is it a learning tool, in the vein of a manual, or is it a scaffold that its users will always need to employ in order to produce a complete socket? Over the course of the project, we've gone back and forth about how much the software should structure the actions of its users, and how easy it should be to break out of Socketmixer and only use Meshmixer. One of the pivots on which this discussion has turned is Socketmixer's ability to automate some tasks which would involve a number of complex steps if done directly in Meshmixer. If we look at Socketmixer as something that supports a clinician's learning of Meshmixer, then having automated steps that cannot be easily replicated manually is a problem. If, on the other hand, the goal is to have the clinicians always using Socketmixer alongside Meshmixer, with the ability to choose which steps to accomplish with the assistance of Socketmixer and which to do manually, then having some heavily-automated steps is more a benefit than a problem. Making decisions about Socketmixer's role has involved, more than we expected, interrogating the user interface decisions we've made as the project has progressed, as we see in the following field note.

7/October/2014 Tuesday morning, while I'm going through user testing videos and making notes about the configuration of the wizard34

Matt has a meeting with Vincent [the member of the project responsible for the development of Socketmixer] later this morning to talk about the wizard and to check in on progress. Because of that, and also because I need to, I'm going

34 Throughout the majority of the project's duration recounted in this chapter, the piece of software now known as Socketmixer did not have a name. Instead, we referred to it simply as “the wizard” because we considered it analogous to the software installation wizards, common in the early days of home computing, which walked users, step-wise, through the process of installing a piece of software.

116 through the videos from my user testing sessions. One thing I realized, in watching the most recent video, is that there's no way to skip one of the small steps that makes up a big step. So, for example, in the cleanup section of the wizard, there's a set of steps you get walked through in order to clean up your model. One of those steps, right now, is a smoothing step. But we don't want the smoothing step to stay in there. Instead, we want to replace it with a remesh. So, in my user testing, when we hit the point where the wizard was going to do something I didn't want to do, I had the prosthetist just jump over into Meshmixer and start doing things manually. Basically, because I didn't want to perform a remesh, and because I had the skill to not need to rely on the wizard, when the wizard started enforcing things, I abandoned it.

So, just now, I went in and mentioned this to Matt. I said something like "I've just realized that there's no way to skip a step in the wizard." And he said something along the lines of "Yes there is, you just need to go to the next page." At which point, I clarified that I meant the small steps within the larger steps, and explained that the arrows function as "accept" keys. [...] [I]n the current system, if you want to reject one step, you basically wind up having to reject all of the subsequent steps in that assemblage of steps. If I don't want to invert, my options are to either jump over to Meshmixer and thus lose all further steps in the cleanup portion of the wizard; invert anyway and then, if I know about the "I" key, re-invert back, or just put up with doing a step I don't actually want to do. Not a very satisfactory set of options. On the one hand, it really forcibly scaffolds people in the actions they take. On the other hand, it might prove really frustrating and routinizing.

Anyway, [...] one of the funny things this is making me think about is that, though we study the ways people are configured by the technologies they use, it's kind of amazing me how the instrumental goal of putting out a system that works is, in many ways, making us be a little blinder to some of the things we'd notice and care about if we weren't the ones engaged in making the software.

Over the course of Socketmixer's development, we made a number of observations like the one above. Some of those observations resulted in changes to the software, while others didn't. Perhaps an amusing side effect of coming from a context informed by science and technology studies, the development of Socketmixer was often a process of weighing up our desire to make and implement decisions, against questions about user agency and what kinds

117 of work should devolve to the software, rather than to its user. Some of these questions have remained unresolved, with the version of Socketmixer implemented at CoRSU being a somewhat uneasy compromise between permanent scaffolding and temporary education. In the implemented version of Socketmixer, actions which can be taken by the user are, to an extent, explained. Automated steps, on the other hand, are somewhat more opaque.

Our decisions about how Socketmixer should work and how much it should do automatically were at least partially influenced by testing sessions with Canadian prosthetists and prosthetic technicians (as the field note above suggests).35 In these sessions, the prosthetists were informed both by the manual processes of rectification and socket building that they are familiar with and, in some cases, knowledge of other specialized CAD-based socket design software. Based on the testing sessions with prosthetists, our ultimate decision was to automate parts of the process which seemed to be situated less in the realm of professional expertise and more in the realm of tedious, manual work. The desire in making this decision was to respect the skill of the prosthetists, while attempting to make their socket design process faster and less tedious. That manifests in things like the alignment process, which automatically flips and rotates the 3D model of the limb into a mostly appropriate position (the right way up, in relation to the ground plane), while leaving to the prosthetist the task of doing more minute alignments (what degree of incline the model has, for example). In particular, we included a very manual, granular rectification process, given that the rectification is one of the areas in which the professional expertise and discretion of the prosthetist manifests itself. Each of these decisions, whether it be a question about the functioning of a button, or what kind of assistance to provide in the rectification process, was informed by both our own concerns about the role of the software, and by the preferences and practices of the consulting prosthetists.

Many of the issues emerging from our software development process are similar to the concerns present in the LGRU work. In particular, the status of the software as a venue for communication between its developer and its user. The issue of embedding values into designed objects, which arose in chapter three, comes out even more clearly here. Nissenbaum

35 I address the Canadian testing sessions further in a white paper on the subject of how prosthetists engage with new-to-them 3D modelling software (coons, 2015).

118 (2001) provides a way of looking at the values embodied in computer systems. She suggests, first, that “[v]alues affect the shape of technologies” and that, as such, “[w]e must also study the complex interplay between the system or device, those who built it, what they had in mind, its conditions of use, and the natural, cultural, social, and political context in which it is embedded” (p. 120). In the case of Socketmixer, there was a tension in our development work between values of ease-of-use and professional development. Decisions about which tasks to automate and which should have room for user intervention were coloured by our discussions with prosthetists, and by concerns about providing opportunities for skill development. Attention to the conditions of production, the assumptions of the developers, and the context of use is an important lynchpin in connecting issues around the material conditions of production—with regard to communication between developers, access to potential users, and the capacity for discussion and interplay—to the role of the user in production. This theme will emerge both later on in this chapter, and in chapter seven, which will explicitly address issues around the role of the user in production, especially with regard to the agency of the user in the production process.

Two notes from 24 August, 2014 on the initial scans from CoRSU By August of 2014, many of the technical decisions in the project had been made (though we still recognized that technical choices would have to be periodically reviewed in future). In June/July, we had finally settled on a scanning solution and had possession of the scanners and capture software, as well as two laptops capable of running the software. Our intern, Jeff, had also mostly completed the how-to guide for 3D scanning that was to accompany the scanning equipment. That's how, when Mitch from CBM Canada took a trip to Uganda, we were able to send a scanning kit along with him. The kit was comprised of a laptop with Skanect installed on it, a Sense scanner, a USB key with a number of assets included on it (not least of all, a backup image of the computer's software, in case something went wrong) and a printed-out version of the instructions detailing how the scanning process should work (supplemented by a digital version, included on the USB drive). All of this set the scene for the events described in the following field notes.

119 It's one o'clock in the morning on the 24th of August 2014—a Sunday. I wasn't sleeping very well, so I opened my computer and checked my email. Earlier in the day, Matt had sent out an email to Emily and Mitch, which he had included me on. He was asking, essentially, if there was any news from the training Mitch has been doing in Uganda. (Mitch is in Uganda right now, and will be back some time next week. We sent him off with a 3D scanner and one of the Alienware computers with Skanect installed, as well as spending a day before he left training him on the use of the scanner.) So Mitch wrote back. The basic gist of his email is that he's trained five people on the scanner, with varying degrees of success. They've been practicing, mostly arms, he says (which is a bit of a frustration, since scanning arms isn't going to result in any sockets under the process we're currently working on). He mentioned that the only person who really took to it right away was a visiting student from the local university. The health care practitioners have been having a tougher time of it.

Crucially, he also mentioned that he's bringing us back three complete ICRC socket sets, which he was surprised to find have a total value of 1300 USD. So I'm a little gleeful, and far more awake than I should be at this time of night. This is one of those moments when it's really hitting home for me that we may actually change the world with this project. The old design joke that you can have a project good, fast, or cheap; pick two, doesn't apply here. Not only are we improving the consistency of the quality of the prostheses, as well as the speed with which they're made, but we're also actually going to make them significantly cheaper. It's pretty damn remarkable. So I'm having one of those moments where I can see that, if we do this right, we can actually change things.

///////////////////////////////////////////////////////////////////////////////////////////////////

A little reflection, later on in the morning of August 24, 2014, round about 10AM.

I've been thinking a little bit about the training of medical practitioners that Mitch has been doing. I've wondered a little how much of the difficulty the practitioners are having is because of the ill-training of their trainer. Remembering back to the day we trained Mitch, he tried scanning maybe once, and then declared himself prepared. We asked if he was sure, if he wanted to practice some more, but he said no. He said that he'd maybe practice once or twice more at home, but that he didn't need to be expert on the process in order

120 to effectively train others on it. I questioned that, to myself, at the time, and I question it now. Now, I question whether or not the lack of success the clinicians in Uganda are experiencing is because Mitch isn't modelling successful use of the process to them. I question whether or not he can accurately convey the process to them if he isn't fluid and confident with it himself. [...] But this leads me to something bigger: What would an effective training program for clinicians look like?

So now we're into things about pedagogy and training and practice and how to help people build useful expertise. How much training does a trainer need in order to teach the process? I mean, in universities, we demand expertise of our teachers. In university programs that have practical, job-oriented outcomes, we want instructors who have practices outside of the university, or at least substantial prior experience in the field. When we got certified on our crazy expensive 3D scanner, buddy from Trois-Rivieres36 put us through a two-day bootcamp and an exam, and refused to certify those of us who missed bits of it. He, himself, was an employee of the company selling the scanner and made an entire job out of going around, training people on the scanners. Now, I don't think that a concentrated training program and an exam necessarily proves any kind of expertise going forward. The value of the training is in giving people the tools they need to get started, to become expert enough, and to then build their expertise as they work with the process or equipment more.

Shortly after this set of field notes was written, we started getting back scans from CoRSU. And they were very good. The clinicians at CoRSU had been practicing with the scanner and were starting to produce very high quality scans. Using a shared networked folder, they were sending those scans back to us. We started to make use of the scans in our own activities, designing a number of sockets based on one of the earlier scans from CoRSU. We used that scan not only to prototype by ourselves, within the lab, but also when we had prosthetists, orthotists and wearers of prostheses visiting to test out our method. Prior to that point, we had been using either a scan of a plaster positive (a positive mould of the residual limb, based on the cast taken of it in the conventional fitting process), or a digitally-amputated version of first, a scan I found on Thingiverse37, and then later, a scan I took of Matt's leg and,

36 This is a glib reference to a certification program offered to us with the purchase of a professional-grade 3D scanner, which involved a trainer travelling to Toronto from the company's office in Quebec. 37 Thingiverse is an online repository of models for digital fabrication, run by Makerbot Industries. Though it is largely used to host and share models designed to be 3D printed, there are also users who distribute control

121 as with the previous model, digitally amputated it. Having access to scans of real residual limbs gave us more confidence in the prototypes we were producing, and allowed us to finally confirm that our process was able to produce a socket that mapped well to the residual limb of a real patient (or at least, as much confirmation as was possible without fitting the socket). The remaining confirmation of the process would only be available to us once we had been to CoRSU and had the chance to fit real patients with the digitally-produced sockets.

The task of translating an activity across distance brings to mind what Latour (2005b) calls an immutable mobile, and what O'Connell (it should be noted, drawing on the immutable mobile) (1993) refers to as “the circulation of particulars.” In both concepts, objects circulate through space (and sometimes time), bringing with them information which would not otherwise easily make the transition from one place to another without experiencing significant change. O'Connell uses the example of the legal volt, an important piece of reference information contained in a few batteries. Manuals and instruction sheets, I would argue, serve a similar purpose. They provide a baseline for discussion, a shared set of understandings which, if used in conjunction with an appropriate context, serve to collapse the distance between the manual writer and the manual reader, or between all of the different, dispersed readers the manual might have. For all my talk above of resisting the modern impulse to impose sameness through technology transfer, the manual and training work to homogenize the experience of using the scanner. In this case, with the research team in Toronto acting as developers, and with the technicians and technologists in the Orthopaedic Workshop acting as early users of the system, the negotiation is between providing a level of consistency (that sameness and collapsing of space, so that troubleshooting can happen and, eventually, user peer review and training communities can be built) and building in opportunities for agency and active engagement. I return to this thread later in discussing the blurring boundaries between user and producer.

files for milling machines and laser cutters.

122 Getting ready to go to CoRSU The portion of the PrintAbility project which comprises this case study effectively ends the morning we took off for Uganda. At the time the scanner was taken to CoRSU, we thought we would be following it shortly, travelling to Uganda in October 2014. In preparation, we worked full-tilt to prepare for the trip. Socketmixer, still called “the wizard” at that point, became a key priority, with development speeding up in August and September. Vincent, who started working on the wizard while still an undergraduate student at UofT, continued working on it in the evenings, having graduated and taken a computer programming job at a large software company. I devoted my efforts to developing text for the wizard, as well as helping to tweak the user interface, a process which involved repeatedly walking through the process with both prosthetists and wearers of prostheses, getting them to work on sockets, scaffolded by the instructions and controls provided in the wizard. During that process, I recorded videos of their testing, noting areas where they stumbled, and areas where their expertise came into play. Based on these observations, we aimed to both make the wizard more legible at first use, and to take advantage of the existing expertise and professional capacities of its users. We were aided in this process by having access to a former president of ISPO, the International Society for Prosthetics and Orthotics, which is the organization responsible for the development and standardization of training guidelines for prosthetists, orthopaedic technologists and orthopaedic technicians worldwide. Having intimate knowledge of best-practices in orthopaedic training proved useful in confirming that the practices we were embedding in our software were appropriate and mapped relatively well to existing training and competencies.

Concurrent with fine-tuning the wizard, we shipped a 3D printer to Uganda. We had previously taken delivery of two Makerbot Z18s, the first affordable, consumer-grade 3D printer with a sufficiently tall print area for our purposes. The availability of these printers was a cornerstone of the project's feasibility, as even a very small child would likely require a socket longer than six inches (the height of the Makerbot Replicator 2, the previous flagship Makerbot printer, and still one of the most reliable 3D printers owned by the Critical Making Lab). One Z18 went to CoRSU, packed in a crate, while the other stayed in the Critical Making Lab, in theory to be used to print sockets, but in practice it became a test bed for extensive

123 troubleshooting. The troubleshooting continued until we left, and even beyond. On a previous project, members of the Critical Making Lab had developed a process we humorously likened to the functioning of mission control for the Apollo 17 mission38, with those members of the lab remaining in Toronto serving as mission control, continuing to tweak the tweakable aspects of a process, while other members of the lab travelled. Having sixteen extra hours to troubleshoot, while other members of the team were stuck on a plane, proved extremely useful. This was ultimately to be the case on our trip to CoRSU. However, an unexpected public health situation bought us a few more months of development time before the trip.

In late September 2014, shortly before we were scheduled to visit CoRSU, a case of Marburg Virus was diagnosed in Uganda. The victim was a radiologist at a hospital in the east of the country, who had gone home to Kampala when he became ill. What would have been a concern at the best of times was rendered even more worrying by the ongoing Ebola outbreak in west Africa. Our trip was ultimately postponed until January, during which time Ugandan public health authorities cleared all of the patient's contacts and ensured that no one else had contracted Marburg Virus. As such, the portion of the PrintAbility project chronicled in this chapter ended in mid-January 2015, when the Canadian team travelled to Entebbe, Uganda, to implement the pilot phase of the project and to train the clinicians in CoRSU's Orthopaedic Workshop. Some work continued while we were in Uganda, making it difficult to draw a clean line under the development phase of the project, but for the purposes of this dissertation, the case needs to end somewhere. The following sections diverge somewhat from the descriptive style adopted above and instead link up the findings from the development of PrintAbility with the three high-level themes identified in the previous chapter.

38 In 1972, the fender of an Apollo 17 mission moonbuggy broke in half and was repaired, ad hoc with duct tape, by one of the astronauts on the mission. His fix didn't last long, but allowed, with repeated reapplications of tape, a day of exploration. But the problem needed a better fix. As a story on NASA's website tells it, “[b]ack in Houston, NASA engineers understood the seriousness of the situation. If they couldn't come up with a solution while Cernan and Schmitt slept, the next day's exploration could be severely curtailed. The astronauts might even be limited to walking distance” (NASA, 2008). And indeed, the engineers did spend the time overnight devising a solution, which the mission astronauts were able to implement the next day with the materials at hand: duct tape and laminated maps (ibid).

124 Connecting PrintAbility to custom dressmakers and the LGRU

In this section, I aim to take the lessons learned from the custom dressmaker and LGRU cases presented in the previous chapter and apply them to my analysis of the PrintAbility project, both in addressing the vignettes presented above, and in admitting some new examples from the PrintAbility project. As an organizing principle, I take the three meta- codes derived from the comparison of the LGRU against the custom dressmakers and apply them to the data collected over the course of the year and a half of PrintAbility development. Some detail from the trip to Uganda is also provided here, as issues like the social organization of labour and the material conditions of production cannot be accurately or reasonably presented without, at least in part, addressing the context of the project's implementation. As such, examples from CoRSU are interspersed with discussions of the previous year and a half of development, all with a focus on how the project was shaped by its social organization and material conditions, and what the role of the user looked like. In crafting the comparison, one particularly interesting aspect of the PrintAbility case is that it integrates the production of software, the use of software, and the development of robust procedures for building objects which ultimately need to map accurately, comfortably and reliably to human bodies. In these respects, PrintAbility shares a number of concerns with both the LGRU and the dressmakers.

The social organization of labour

In the first part of this chapter, I presented the dependency grid and its (dis)use, highlighting that its role was indicative of the way labour within the project was organized and conceptualized. With the grid behaving as something of a boundary object, it is an artefact which—perhaps more obviously than many of the other artefacts produced during the project—highlights the negotiations involved in dividing the project work, and in making determinations about who a given class of task should belong to. This highlights an issue of scale present in the project (and in both the LGRU and dressmaker cases). In all cases, there are local ways of organizing work (for example: who repairs the 3D printer in the lab; who takes on the development of a piece of software; who is allowed to take the measurements of a client), as well as broader forms of organization (the need for a memorandum of understanding; the partnership between four organizations; the entrenchment of a guild

125 through the granting of a charter). The analysis that follows takes up those two different scales in comparing the social organization of labour in the PrintAbility project to the relevant examples presented in chapter four.

The role of institutions as organizing agents in multi-party projects is strongly highlighted in the PrintAbility project. From the outset, the project was motivated by relationships between institutions. An existing collaboration between CoRSU and a Belgian university planted the seed of CoRSU's interest in 3D printing. The strong relationship between CoRSU and CBM (CBM keeps a staff member embedded at CoRSU, for example) meant that CoRSU's administrators were comfortable proposing a new project to CBM. The existence of the Innovations and Partnerships Office at the University of Toronto meant that a clearinghouse was available through which CBM Canada could seek a principal investigator to spearhead and define the project, with regard to the capabilities of low-cost 3D printing. The existence of the Critical Making Lab and Semaphore as entities meant that funding and infrastructure were in place to do early research, before dedicated funds had been raised to support the project. Existing procedures within the university for both employing students and for incorporating lab-based research into coursework or major student research projects helped to provide a labour pool. In short, the existence of a set of institutions made it possible for the project to move from idea to research project to pilot implementation.

Conversely, the activities of institutions placed restrictions on the ways in which the project could be carried out. For example, the need for a memorandum of understanding (MOU) between CBM Canada and the University of Toronto caused a great deal of institutional wrangling and negotiation, which included the efforts of a large number of staff members and other stakeholders on both sides. The development of an MOU relied on the varied understandings of value and novelty, as held by the two different organizations. From a university perspective, value inheres in intellectual property, rendering projects which do not produce unique IP relatively valueless, or at least of little interest, in the eyes of the Innovations and Partnerships Office, which made getting institutional support in the drafting of an MOU difficult. Another structuring factor was the application for funds from an external granting organization, which spurred the development of a set of materials describing the

126 project, and also laid the groundwork for a business plan, in compliance with the subsequent “Transitions to Scale” grant also available (by application, after the receipt of the phase one grant) from the same organization. The ability to apply for further funding after the pilot project also meant bearing later application requirements in mind while developing the initial pilot implementation.

On a smaller scale, the social and institutional organization of labour on the Canadian side of the PrintAbility project involved the allocation of specialized tasks to the individuals most able to carry them out. This structuring of work happened on a number of levels, and in a number of ways. One example is my own involvement: given my role as a doctoral level researcher situated in a faculty with a social science bent, I was involved in drafting ethics documents, devising research protocols, and interviewing local informants, both prosthetists (and other orthopaedic professionals) and users of prostheses. In addition to these research activities which contributed to this case study, I took on a paid role for four months, working with an intern to develop procedures for the scanning of residual limbs and documentation explaining the procedures we developed. In part, that work was motivated by my existing expertise (in writing documentation), but in another way, it also helped me to develop a new research interest and area of specialization, namely 3D scanning and the use of 3D scan data in digital 3D modelling. As such, after the completion of that task, I moved on to assisting with the development of Socketmixer—which, in its most major capacity, provides scaffolding for the 3D modelling process—writing instructions, carrying out user testing, and ultimately training the CoRSU clinicians on it.

The above speaks to a kind of ad hoc organization of labour which emerged over the course of the project. Based on individual interest and expertise, as well as factors emerging from the project, different individuals involved with the project took on different roles. Some roles were predictable, like Ryan making changes to Meshmixer, which he, as its original developer and the only Autodesk employee engaged with the project, was uniquely qualified and permitted to do (qualified because it was originally something he developed himself, permitted because its codebase is not public). It also follows that he would be the one to provide support in using the more opaque functions of the Meshmixer API.39 Similarly, the

39 An API (Application Programming Interface) is a set of specifications which allow third-party developers to

127 work on materials testing done by our master's-level research assistant was entirely predictable and in-line with the interests he expressed when joining the project. On the other hand, as in the case of my involvement with the 3D scanning portion of the project, some elements devolved to particular people not based on any pre-determined organization or decision about roles, but based on individual interest, expertise and availability. Matt, unsurprisingly, spent more time working with the 3D printers than I did. Though his expertise in 3D printing is the major reason he was approached to be the Principal Investigator for the project, it did not need to follow that he would do much of the hands-on work with the 3D printers. Similarly, the involvement of two PhD students from the Critical Making Lab (Dan and Gabby) who were not formally implicated in the project arose more from their presence in the lab and willingness to troubleshoot 3D printers than from anything else. The efforts of one of those students, Dan, at modelling the component that would connect the socket to the ICRC alignment pieces was based entirely on his facility with Solidworks, a piece of CAD software well-suited to the design of mechanical parts.

With particular reference to the work done by Dan and Gabby, there is an interesting case to be made for the similarities between PrintAbility and the intersections of the custom dressmakers with the LGRU. Like apprentice dressmakers and F/LOSS developers, doctoral students often do work which has no direct monetary reward, but which provides them with expertise, experience, or an ability to otherwise grow in areas they find relevant. A lab model, which provides graduate student members with enough employment opportunities to feel economically comfortable, while also providing them with opportunities to work on projects which they might not otherwise have come into contact with, seems to operate on a similar (but slightly less economically exploitative and more directly skill-building) model to an apprenticeship. By taking on some of the trappings of an apprenticeship (skill-building, indoctrination into a trade, some kind of certification of expertise, the completion of a masterpiece), while also, at least in the case of the Critical Making Lab, encouraging a kind of robust self-interest and self-direction, the lab model seems to combine traits present in both the guild training system and the ad hoc and voluntary ways of organizing labour present in F/LOSS projects. build applications compatible with an existing software platform. For example, the API for Twitter allows developers to build clients which make use of the Twitter infrastructure, providing access to the functions required for operations like publishing a tweet or sending a direct message.

128 Outside of the context of the small research team, the differing kinds of labour involved in the Canadian side of the project, writ large, cannot be discounted. Participants had different stakes in the development of the process, with some doing it because it became a function of their existing employment (as with the CBM Canada employee who developed the project's business plan), through the buy-in of their organization, some doing it out of interest, but with the approval of their employer, some doing it because one of their job responsibilities is to produce novel research, and some as contract workers specifically employed in the cadre of the project. For some participants, these roles changed over time. The constraints that these commitments placed on the different parties involved in the project are hard to determine and even harder to articulate, but contributed to the level of activity each participant was capable of, and the ways in which they negotiated their activities. For example, for the employees of CBM Canada who worked on the project, travel is part of their job description and, as such, the trip to Uganda, though it required planning, was not only authorized by their workplace, but organized by it. On the other hand, participants in the project who had day jobs at private companies needed to negotiate the expectations of their employers and, in some cases, colleagues, in committing to the trip. Though it is difficult to say “no” to an employee who wants to take time off in order to pilot a new kind of prosthesis in a developing-world hospital, especially when the patients are children, giving that time off is still less of a foregone conclusion than is authorizing a work trip in the cadre of an ongoing project that forms part of one's work duties.

The social organization of labour when we got to CoRSU Though our work at CoRSU was not initially intended to be part of this case study, what happened during our week there relates strongly to the theme of the social organization of labour. A major concern in our preparation for the trip was that we should be there as trainers, but not as owners of the system we had developed. Recognizing the importance of localization and modification by the Orthopaedic Workshop workers was essential in trying to avoid the Modernist tropes of aid and development work, such as the idea that development is a process of turning a “traditional” society into a “modern” one (Escobar, 2011, p. 280) or that, within the frame of Western exceptionalism, “[t]here is one world, and it has a single internal order” and that the only society capable of developing the tools to recognize that true order

129 springs from the West (Harding, 2011, p. 6). From our perspective, especially as non- prosthetists with no practical experience in building prostheses for children, it was important to foreground the expertise of the clinicians at CoRSU, even if that expertise is developed in the framework of international (Western) standards and methods for the production of prostheses. Enacting respect for the expertise of the clinicians, while providing them with useful training and a toolset for further training and development, was one of our primary concerns, and one which could have come into conflict with our other major goal, the printing of at least one functioning socket during our week there. In our ten-strong group, that balance was negotiated in different ways, by different people, at different times. A complicating factor was that each member of the group had a somewhat unique set of responsibilities and goals for the trip. Some of those goals were distinctly big-picture (and were generally embodied by the work done by Matt and Mitch) while others were practical (as with the software tweaks carried out by Vincent, or the 3D printer repair done by Ryan). This is not to say that parties with big-picture tasks did not also carry out smaller or more practical tasks. Likewise, in our capacity as a visiting group, we were all treated with a high degree of care and attention, as epitomized by the half-day tour of the hospital given to us by its CEO. In this respect, at least at the beginning of the trip, there was a distinct discomfort among some members of the group. While the intention of the UofT team (a loose agglomeration which includes everyone who worked on the technical development of the project) was to, in effect, muck in and get as much work done as possible during the trip, the institutional work of being toured around the facility and shown its capacities was necessary.

For me, coming from a large, North American university, and from a social science- oriented faculty populated by professors and doctoral students embedded in a generally critical school of thought, a key worry was inadvertently enacting a colonial mode of interaction, with the CoRSU clinicians as supplicants to the new technological knowledge we were bringing. A few factors mitigated my concerns slightly before the trip40: the fact that the project had originally been initiated by CoRSU; the expertise of the clinicians; and CoRSU's status as an important regional centre of expertise in rehabilitation medicine. These three

40 In realizing these mitigating factors, I owe a debt of gratitude to Professor Cara Krmpotich, whose generous sharing of experience from her background and interest in both post-colonial and anthropological issues helped me solidify an understanding about the capacity and agency of CoRSU's employees and administrators.

130 factors were reinforced by some of the pre-trip interactions we had with CoRSU staff. As described in a field note above, on a trip to CoRSU in August 2014, Mitch from CBM Canada brought over a computer and a 3D scanner. A syncing system was set up so that scans done at CoRSU could be sent to a linked computer in Toronto. Through these scans, the clinicians at CoRSU showed that they were taking ownership of the scanning process, producing high- quality scans, with little intervention from the Canadian team. We tried to consciously continue that trend when we arrived at CoRSU, with a degree of success. In part, we were aided by the strong professional knowledge on display at CoRSU. Because the practitioners in the Orthopaedic Workshop effectively had one over on us with regard to clinical and technical knowledge of traditional methods for producing prostheses, we began on a more equal footing than might otherwise have been the case. Rather than bringing in expertise that was fully- formed, our understanding of 3D printing, 3D scanning and 3D modelling was strong, while the members of the Orthopaedic Workshop had a far stronger knowledge—both practical and theoretical—of the process of producing a prosthetic or orthotic device.

With regard to the goings-on during our time in Uganda, the facet of the work during the week at CoRSU that I am most familiar with is the 3D modelling training. Over the course of four days, I conducted training and evaluation sessions with five different clinicians from CoRSU's Orthopaedic Workshop. Other clinicians spent time producing sockets with members of the Canadian team. Two different models were used: in the training model, the goal was to develop skil with both Socketmixer and Meshmixer, helping clinicians gain an understanding of the basic principles of 3D modelling. To that end, the training sessions involved one or two clinicians, sitting in front of a computer, with me sitting in a chair next to them. They were tasked with walking through the instructions provided in Socketmixer, and executing rectifications in Meshmixer. In the co-production model, someone from the Canadian team, already familiar with 3D modelling, occupied the proverbial driver's seat (which is to say, had control of the computer and mouse), but took instruction from a member of the Orthopaedic Workshop on what rectifications should be done to the residual limb, and what dimensions the socket should adhere to. This process produced the majority of the successful sockets printed during our week. It was the method by which the socket for Rosaline, the almost four-year-old girl who received the most successful socket of the week, was made.

131 In the training-focused modelling, two goals were key: getting clinicians to go through the entire modelling process on their own, without any direct intervention from me; and to get clinicians who had been trained to feel capable of training their colleagues. In this respect —helping the clinicians go from no knowledge of 3D modelling to a level at which they felt able to train someone else—there was some potential to upend existing organizational structures in the workshop, at least temporarily. As with some of the activities during the Canadian development phase of the project, there was, as far as we knew, no existing consensus on who would become responsible for given tasks in the new process during their implementation at CoRSU. To our knowledge, everyone in the workshop had participated in the scanning training that had taken place before our arrival. For the modelling training, the first session took place with one of the most highly-ranked members of the workshop staff, but due to time constraints and technical issues, he did not successfully make a full socket. Offering the training to the highest-ranking member of the Orthopaedic Workshop team first was a conscious choice, made out of deference to his position in the workshop. The next day, another more senior member of the team, an orthopaedic technologist, went through the training and produced a socket. However, thanks to an illness later in the week, she was unable to become a peer trainer during our time at CoRSU. Because of these external factors, the first peer training session paired two orthopaedic technicians (a ranking which is a level below a technologist in the ISPO taxonomy of practitioner training levels). The two technicians went through the process together, making the trainer one of the most immediately experienced staff members in the use of both Socketmixer and Meshmixer. Towards the end of the week, he was also involved in the training of another technologist.

On both the Canadian and Ugandan fronts, the organization of work was at times based on pre-defined roles, and at times based on the availability of individuals at particular times. Tasks were taken on either because of existing role and rank, or based on who happened to show a willingness or immediate aptitude. In the Canadian context, different members of the project team joined for different periods of time, and based on varying assumptions about things like compensation and long-term involvement. Hierarchies in the two teams seemed to enjoy comparable levels of flatness, but varying degrees of formality. On the Canadian team, most conversations about the toolchain involved all of the member of the

132 research team (encompassing members of the Critical Making Lab, an intern, participants who started as students and then graduated, and Ryan, the participant from Autodesk), and always eventually devolved to Matt in his role as Principal Investigator. Decisions which could not be siloed under the heading of research would involve, either from the outset, or subsequently, both of the involved staff from CBM Canada and, as needed, other relevant CBM Canada employees. On the CoRSU team, day-to-day decisions devolved to Moses, one of the more senior technologists, or to Abdullah, the head of the Orthopaedic Workshop. Almost uniformly, when telepresence or email communication took place before our visit to CoRSU, Moses was the representative with whom we worked. During our time at CoRSU, other tasks were taken on by administrative members of staff, each of whom brought their own authority and ability to their assigned tasks.

I would be remiss to not mention the role of the patients at CoRSU and the Canadian volunteers in discussing the organization of labour in the project. Our relations with Canadian prosthetists and orthopaedic technicians were negotiated in a variety of ways. At the outset of the project, gaining access to prosthetists was a difficult task. Perhaps understandably, prosthetists, prosthetic technicians, and orthotists were wary of claims that digital fabrication could be applied to their field. Direct approaches to prosthetists and the hospitals at which they work often failed. Though some CAD/CAM processes have made their way into the practices of prosthetists, a those processes have also been accompanied by projects which have more hype than execution. Some such projects have turned out to be vaporware41, while others have been one-off academic projects with little future or practical application. Such outcomes have led to healthy skepticism among both prosthetists and wearers of prostheses. We ultimately gained the attention of prosthetists by being able to show both knowledge of the past CAD/CAM precedents in prosthetics, and by actually presenting the preliminary outputs of our work. Media attention also helped lend credibility to the project, and to cause practitioners to come to us, rather than us to them. A similar trend was evident in our relationships with wearers of prostheses, many of whom were excited about the project, but also had stories to tell about past experiences with short-lived, experimental sockets. In developing the process, one of our efforts has been to include the input of both the prosthetists 41 The term “vaporware” implies a project which is promised but never delivered. Vaporware, in contrast to other kinds of “ware” (hardware, software, freeware), has no characteristics other than its absolute etherealness.

133 and the wearers in what we have built. In this way, the wearers of prostheses can be seen as taking on the role of expert patients, providing feedback based on their lived experiences with their prostheses and the systems through which those devices have been acquired.

One particularly poignant example of the importance of users was in our development of a soft liner for the sockets. In talks with prosthetists and orthotists, we had been told that tube socks made adequate liners for sockets. In the North American context, however, it is common to use an expensive silicon liner. These silicon liners wear out quickly and are difficult to clean, compared to a regular sock. Tube socks, in short, seem like a rational decision when looking for a practical, cost-effective socket liner. However, in discussions with wearers of prostheses, we heard over and over again a disdain for the idea of wearing a tube sock (or multiple layers of tube socks) as a liner. Whether the concern was bunching, comfort, or sweat retention, our wearer-informants strongly negated the idea of using tube socks as liners. This sort of interplay allowed us to take into account the concerns of different stakeholders, and to compare professional expertise against lived experience.

At CoRSU, some relations with patients took on a similar model, while others did not. A major difference between the two contexts is the actual need for a new socket. In the interviews with Canadian wearers of prostheses, none of our informants were expecting to get a new, functional, durable, comfortable socket from us. This put them in a position where they were fully capable of assessing parts of the system, and of offering insights into traditional methods of producing prostheses, but without needing to bring their concrete, immediate requirements into the discussion. Conversely, at CoRSU, all of the patients we interacted with were there because they needed a new socket or at least an adjustment to their existing prosthesis. This meant a more serious kind of give-and-take, with our intellectual interests necessarily being subordinated to their immediate needs. An interesting contrast in this regard, though, is between Rosaline and Ruth. Rosaline, being too young to be very articulate about her own feelings or needs, was far less capable than Ruth, a young adult, of critiquing her new socket. For the purposes of a fitting, Rosaline's responses (given through her older brother) were enough. Ruth, however, falls closer to the profile of our Canadian informants. She is an adult, educated, and has strong opinions about her prosthesis. Because of these traits,

134 and because of her interest in the project, it was agreed that she would take on the role of expert patient, providing further insight throughout the project, and offering up candid assessments of the new sockets.

The material conditions of production

Of the vignettes detailed in the first section of this chapter, the importance of the material conditions of production is most evident in the decisions taken about 3D scanners. In deciding which scanners to use, and in sourcing them, issues like compatibility with different computer systems, the availability of certain technologies, and university procurement policies come into play. These factors may seem trivial or secondary, but paying them at least some attention is key in understanding how the decision to use the Sense came about. It is far too easy to simply naturalize a decision, taking it as a fait accompli or erasing the process by which it was reached. In looking at the case of the 3D scanners, and several other examples from both this chapter and the previous one, it becomes more obvious that attending to the conditions under which work is done and goods/systems are produced is essential to a full understanding of their final form.

The development of the process at the heart of PrintAbility was based on the infrastructure of the Critical Making Lab and Semaphore. The presence of a group of researchers and practitioners in one place, whether formally involved with the project or not, facilitated the development and troubleshooting of the process that was ultimately implemented as a pilot project at CoRSU. As I highlighted in my analysis of the LGRU, though it is easy to ignore the importance of the material and the specific in crafting analyses of the digital and the broad, we do so at our peril. Whether it is the stack of cases of Club Mate42 fuelling a Free Software conference, or the bored graduate students fiddling with the extruder heads of 3D printers, the incidental labour and resources consumed in the process of creating novel digital tools are unavoidable.

42 Club-Mate is a caffeinated soft drink popular with members of European hacker communities.

135 Proximity and artefacts In the development of the PrintAbility project, the key material conditions impacting the outcomes of the project included things like the quality and reliability of a given 3D printer, the weight of a spool of plastic filament, the availability of a piece of software (consider the incident described above in which a software decision was, in part, based on the ability of a vendor to reply to an email), or the proximity of different team members to each other. It is not for nothing that hackathons exist to bring groups of software developers together in a confined space. In our case, many thorny technical and social questions were addressed on days when many of the members of the team were together in the lab. Examples given above speak to the importance of proximity: the sketches of possible new alignment pieces for the sockets, which were passed around the lab, left on tables, and ultimately lost, without ever being digitized. Discussions about the placement of equipment at CoRSU also involved the use of sketches. At a time when, of the members of the PrintAbility team, only CBM Canada employees had visited CoRSU, their ability to produce sketches of the hospital layout informed the rest of the team's understanding of the spaces available at CoRSU, as well as their proximity to each other.

Another material issue is the availability of equipment for comparison. It would be difficult (and pointless) to deny that a major part of the Critical Making Lab's institutional expertise in 3D printing comes from the capacity to use and compare a variety of different 3D printing technologies, both on the commercial end of the spectrum and the consumer end. Though, in theory, it might potentially be possible to do effective evaluations of different 3D printers without actually using them (comparisons of, perhaps, manufacturer claims about speed, resolution and other apparently clear technical issues), one of the strengths of the Critical Making Lab in recent years has been an applied knowledge of the more idiosyncratic aspects of consumer-level 3D printing. Whether this is in finding out what it really means to print ABS without a heated print bed,43 knowing what thickness of card is actually best for manually levelling a print bed, or discovering just how difficult it is to perform basic maintenance on a black-boxed extruder assembly, these are facets of the printers that could only otherwise be gleaned by reading the mailing lists and forums of other 3D printer 43 ABS (Acetyl-Butyl-Styrene) is a plastic used commonly in 3D printers. However, if the print bed of the 3D printer is not heated, ABS has a tendency to not stick to the bed properly, causing the edges of the print's lower (first) layers to lift up, a phenomenon which often causes the deformation or failure of the print.

136 operators. Such apparently small concerns, often glossed-over in discourses promoting the potential of 3D printing, are extremely important to the actual implementation of projects which rely on the technology and are, thus, highly material (in both sense of the word) to the successful implementation of such projects. More broadly, being able to understand the capabilities of different exemplars of a technology allows for a more nuanced, less monolithic critique of, in this case, 3D printing as a whole. This aligns well with the actor-network-theory idea of watching issues and controversies-in-progress with the goal not of resolving them, but of understanding their interconnections and issues. In understanding the issues germane to implementing robust 3D printing projects, the problems and decisions implicit in the technology need to be denaturalized in order to even be accounted for, never mind understood and parsed in a larger socio-technical context.

Mission control The “mission control” metaphor mentioned above also speaks to the importance of the material conditions of production. The role of mission control is best exemplified with a short recounting of an incident which took place the week before the trip to CoRSU. One of the flaws with the Makerbot Z18 is its irritating habit of losing extrusion during print jobs, thanks to the force it takes to pull filament into the print head, especially when the spool of filament is particularly heavy, or positioned below the print head, or both. The loss of extrusion effectively ruins the print, either causing the adhesion of the layers of plastic to be weak, or causing the print to simply abort altogether. For our purposes, this problem was temporarily resolved through the intervention of a Critical Making Lab PhD student, Dan, who designed a rack which could be fitted onto the side of the Z18 in order to position the spool of filament above the print head, thus taking advantage of gravity in improving the consistency of the feed. The rack was designed to take a standard size of metal tubing, attached together with 3D- printed pieces, and with appropriate fit achieved through the wrapping of masking tape on the ends of the metal tubes. All of this was done so that the rack could be produced locally, using a combination of standard materials and 3D-printed parts (and which also spurred questions about the availability of masking tape in Uganda). However, while we were in Uganda, Dan and another PhD student, Gabby, continued to experiment with the Z18 that was left behind in Toronto, and subsequently discovered a simpler solution to the extrusion problem, involving

137 mineral oil. Because mineral oil is a relatively established part of the 3D printing toolkit, we had brought some along on the trip. An email from Dan and Gabby allowed us to move to a slightly less complex remedy to the extrusion problem. Thanks to an awareness of the conditions under which we were working, and the knowledge of our available toolset, a solution could be developed at a distance.

One of the issues in popular discussions about 3D printing and other kinds of digital fabrication is the assumption that material issues will be easy. We see this over and over again. In coverage of the Liberator, a 3D-printable firearm, the material conditions required for its production are not accounted for (Record, coons, Southwick & Ratto, 2015), and the production of a gun is instead positioned as something as easy and tidy as the downloading of a file. As historians of production (for example, Crowston, in the case of the working conditions of Parisian dressmakers) sometimes exhort, considering output is not enough. In order to more completely understand the place of a product in society, and how it gets there, the conditions of its production must be taken into account. If it is so easy to erase the process of production from goods produced before the advent of digital production, there is the very real danger that we will move further towards an ignorance of production and its conditions as digital fabrication technologies—promoted, at best, as a mode of production with fewer strings attached and at worst as a kind of magic—become yet more mainstream.

The role of the user in production

In the two field notes detailing the early days of 3D scanning at CoRSU, there is an interesting blurring of the line between user and producer. The blurring, as the prosthetic technicians and technologists become users of the 3D scanner and its attendant software, comes in the eventual use case for the scans, and the way training will eventually be handled. Though the technicians are acting as trainee users during the early days of their scanner use, their experiences and opinions form an important source of experience and expertise in devising and iterating the PrintAbility toolchain as a whole. While they are using consumer- grade tools to do their work, they have the power and opportunity to make changes to both how they use the tools (Eglash, 2004), and how the tools will be developed (or swapped out for

138 other tools) in future. While all of this is going on, they are also producers in quite a straightforward sense: they are using these tools, alongside their expertise, to build prosthetic sockets for their patients. All of this speaks to a multi-layered relationship between the user and the producer: it is exceptionally difficult, in the case of the technicians, to determine where their status as user ends and where their role as producer begins. This is echoed both elsewhere in the PrintAbility project, an in the cases from chapter four.

3D printing and other forms of digital fabrication open up new questions about the division between the user and the producer. In the example of the seamstresses in the previous chapter, the end user is relatively clear: the woman who buys and wears the dress is the user. In the LGRU, the user is the real or hypothetical designer or artist whose itch is being scratched. In the PrintAbility project, the user is the child who ultimately walks away on a 3D- printed socket. However, as in both the dressmaking example and the LGRU case, the question of who is truly the producer or developer in the PrintAbility case is in question. From the perspective of tools and materials, a whole ecosystem of producers is implicated in any production activity. Whether the factory operative who oversees the melting of pellets and their extrusion into filament is more the producer than the farmer whose corn becomes PLA44 is up for question. Whether the mill-worker who weaves the fabric is more the producer than the seamstress is a moot point. They both have a hand in the ultimate production of the dress. It's the proverbial turtles all the way down, making the identity of the “true” producer difficult and, really, unnecessary to define. As we make use of the tools and materials of others who come before us in the supply chain, we're users all the way down. Or, in a different, more active view, because we are all adding on and acting on the object of production, we're producers all the way down. We might also take Marx's point that production requires the “consumption of the means of production” and that, as such, an “act of production is therefore in all its moments also an act of consumption” (1973 [1857-61]). In the example of PrintAbility, we might choose to say that the clinicians decide the form of the socket, within the constraints of their materials, and with the scaffolding provided by their existing professional best practices and the shortcuts offered by Socketmixer. They then make use of the materials available to them, whether those are the filament spools of the prinatAbility project or the

44 PLA: Polylactic Acid, a type of plastic popular in consumer-grade 3D printing. Made from vegetable sugar, most often derived from corn.

139 pre-formed feet of the ICRC system. In this sense, the socket is co-produced by the clinician, the constraints built into Meshmixer and Socketmixer (some of which, themselves, are informed by existing guidelines devised by ISPO and the ICRC), the mechanical capabilities of the 3D printer and the material it uses, the requirements of the parts with which the socket must interconnect, and the form and availability of materials and best practices.

But none of those concerns are particularly new or unique. The currents of global production and trade are already well-documented (eg: Cowen, 2014). Rather, what is interesting about both PrintAbility and the examples provided in the previous chapter is the way in which the end user can be incorporated into the process of production. This is, after all, one of the promises of customization: that the needs of the end user will be taken into account in the production of their custom goods. From a more traditional customization perspective, PrintAbility offers end users a socket which conforms to the shape of their residual limb and, hopefully, to the requirements and exigencies of their daily life. But this is also the goal of a well-fitted prosthetic socket made by conventional means, and is a value already built into the practices of the clinicians who fit and make the sockets. What is exciting and novel about PrintAbility is the potential for co-creation by the wearer and the clinician. Equally, tensions arising between the expertise of the clinician and the lived experience of the wearer (itself a different kind of expertise) are a fascinating site of development. That potential for co- creation, tension and differing stakes is not new either, being in evidence in the work of the custom dressmakers and even in the work of prosthetists using existing methods. What is of concern, more broadly, is the possibility that other implementations of mass-customization, by dint of their massness, will ignore or parameterize the issues of need and individuality that arise from the use cases of individuals. Rather than the kind of co-production we see in PrintAbility, we might instead see an offloading of responsibility and choice entirely onto the end user, with the fiction of choice represented by a parameterized mass-customization framework standing in for an actual concern about function and fit.

In the process of fitting a prosthesis (a task that, by appearances, could take up as much time as can be given over to it), there is a clear interplay between skilled patients, especially those who have been wearing a prosthesis for a period of years, and their

140 prosthetists or other clinicians. Skilled users come to the process with opinions about fit, comfort, appearance and function. Examples provided in my work with Canadian users of prostheses include a desire to colour-match fairings to the wearer's exact skin tone, the need for prostheses which can be worn while doing sports, and the tension between appearance and ruggedness. Wearers devise their own methods for making the most of situations which can be difficult or uncomfortable, experimenting with ways of caring for the skin on their residual limbs, keeping old prostheses for scenarios like showering, or experimenting with new liners for their sockets. Similarly, the expert user recruited by CoRSU to consult on the PrintAbility project brings her own experience, offering commentary on which areas of the prosthesis pinch, which areas feel too loose, and what, based on her experience, will not work in practice. The interplay of the clinician and the patient, and the comparative seriousness with which the fitting of a medical device is taken, offers an expanded set of moments in which a discussion about needs, desires and experiences can take place.

But of course, such an ideal isn't always the case, if ever. In the conventional method of producing prostheses, as documented above, there is, at the best of times, a degree of conflict between the desires and needs of wearers and the established best-practices of practitioners. One wearer I interviewed recounted her experience working with a prosthetic technician who was also, himself, a wearer. She argued that the advice of a technician who was a fellow user was, to her, more informed and more reliable than advice from someone who had experience building, but not using, a prosthesis. She suggested that his assessments of the comparative merits of different off-the-shelf parts were reliable because they were based on experience. There is perhaps a case to be made here about aligning the interests of users with the interests of producers, and encouraging empathy and the cultivation of common ground between the two.

Though producers cannot always be users of their products, the cultivation of informed groups of stakeholders is key to understanding the needs of end users. In a best case scenario for custom production, the one-to-one mapping of a user and a product should be able to provide such input. But, as we see with the example of the prosthetists, this is not always the case, thanks to issues like time, expense, differing positions of authority between the producer

141 and the user, and even the capacity of the two stakeholders to have a sufficiently open relationship with one another. In chapter seven, I further explore the relationships between users and producers of custom goods produced through practices of mass-customization and digital fabrication. In that chapter, I argue that, in search of economies of scale, current practices of mass-customization do not actually provide the potential for candid relationships between users and producers, and that the parameterization of such customization instead provides an illusion of choice and fit, while instead offering a diminished version of a custom good.

Intersections between the three high-level themes

In detailing and comparing various examples of the three major themes—the social organization of labour, the material conditions of production, and the role of the user in production—in action, I have, up to this point, broadly treated the three as if they are discrete or capable of existing in isolation. This is not the case. Instead, the three themes are inextricably linked, with the material conditions of production and the social organization of labour being especially tightly coupled. This is really no surprise, given the importance and influence of existing work which emphasizes the interconnectedness of the social and the technical. Indeed, Latour (2005a) expounds at some length about the inability to separate the two. Above, in comparing 3D printers to sewing machines, I begin to get at the interconnectedness of the two themes. The material exigencies of the printer (which can only be fully understood by working with it) are navigated through the social structures of the lab. In the LGRU, the structure of a given organization (for example, WORM, the host of the Rotterdam meetings) enables material conditions which facilitate the work being done (the availability of space and equipment, allowing unstructured work time to happen). This is also the case for the dressmakers, where we might well see an indentured apprentice fetching and carrying, tidying, and bringing coffee. Her low-status work, entrenched in the professional guidelines of the guild, is intertwined with the material conditions of labour in the workshop.

Similarly, the interplay between the material conditions of production and the place of the user cannot be ignored. At CoRSU, we might see this in Rosaline's ability to stay at the

142 hospital, which gives her (or her guardian, as the case may be) an increased ability to participate in the decisions taken about her prosthesis. We might also look at the relatively large number of prosthetists in Toronto as a key factor in the knowledge embedded in Socketmixer. The LGRU provides what is perhaps the most straightforward example of this intersection: getting potential users and developers together to define and develop the software together. Among the dressmakers, the combination of the showroom and the workshop provides a similar circumstance, allowing the mistress to continue to preside over her workers, even while fitting a client, and the workers to act as flies on the wall of their mistress's interactions with clients.

But it is also not enough to think of the three elements as pairs. Instead, I contend that none of the three can be adequately explored without the others. At its base, the connection between the three themes is a necessity in negotiating the factors that go into thinking of end users as non-interchangeable. If the fundamental break from mass production is considering users as specific, rather than general, then producing for that condition requires an understanding of the conditions, both social and material, that allow a rich interplay between the user and the producer (and allow that line to be blurred). This is evident in the connector example provided above, where the needs of individual users, with regard to components like feet, run up against the social and material systems which form the substrate of the workshop and its relations with the outside world. In the case of Rosaline's foot, we see a collision of supply chain issues (how long it takes to ship a component from Switzerland to Uganda), individual expertise (Abdullah's ability to make a functional foot out of the materials available), social convention (the desire of Rosaline's family for her to have a more foot-like foot), and a reciprocal negotiation between the workers in the Orthopaedic Workshop and their client. Unless we include all of those factors, we run the risk of missing out on why Rosaline is apparently happy (for the moment) to have two left feet. We could compare this to the conditions prevailing in the LGRU which allowed software to be built: the work of bringing together the involved parties, the resources available at the meetings, and the structuring of activity, all of which allowed decisions to be made collectively about what kind of software would be produced.

143 In later chapters, I will address how we might look at these interconnections, including through existing theoretical frameworks. I now move on to two chapters which do just that: chapters six and seven pick up on scholarly discussions in which my three themes are situated. In chapter six, I bring more theoretical nuance to the twin concerns of the social organization of labour and the material conditions of production, through the shared intermediary of the body. Following that, chapter seven returns to the role of the user in the production of customized goods.

144 6 The body in digital production

This chapter broadly maps onto two of the meta-codes exposed in chapter four and applied in chapter five, namely the social organization of labour and the material conditions of production. My intention here is to take two themes which are emergent in the case studies detailed earlier in this dissertation, and to run them up against a small portion of the huge body of applicable work that has already been done. I do this by addressing both codes through one related idea: the body in digital production, which encompasses elements of both more traditional conceptions of the body in production, and more recent ideas about digital labour. I make use of the idea of the body in production in addressing the two themes because the management of the body in labour is a milieu in which social and material meet. Indeed, considering the labouring body highlights the impossibility of separating the social from the material. I make use of literature on scientific management, expertise, and political economy, as well as conceptions of the body and embodiment, in conjunction with examples from my case studies. I supplement my case studies and my use of the literature with three other examples which provide further insight into the connections between the management of the body in digital labour and historical precedents. By taking this approach, I hope to both inject a little more theoretical nuance into my observations, and to identify areas where the work done in my case studies might contribute to existing understandings of the way we organize ourselves and marshal our material resources in doing work at the intersection of the custom and the computational.

145 Though this chapter uses the idea of the body in digital labour as a way of approaching the social organization of labour and the material conditions of production, there are further themes that arise from questions of bodies in labour, which manifest themselves across the two high-level themes explored here. The first of these has to do with expertise, and where it resides. One of the organizing principles of both scientific management and fractionalization of labour in general is that it is possible to rationalize a task, formalize or instantiate it in some kind of process or document (often both) and use labourers, who are deprived of specialized expertise, roughly interchangeably. We see this in the changes to the garment industry in the 19th century, with the specialized labour of the dressmakers eventually giving way to the routinized labour of the slop workers. Though expertise can be moved into descriptions of tasks, it can also be embedded into machinery or other apparatus for directly performing those tasks. As Haydu (1988) describes it, a transition at the turn of the 20th century from skilled craftsmen manipulating general-purpose machinery, to un-skilled machine tenders operating single-purpose tools led first to displeasure amongst the skilled workmen, but then ultimately to a re-organization of the factory in favour of the specialized tools and un- specialized workers. Today, such fragmentation of labour has been taken to extremes, with garment workers, for example, performing the same small task repeatedly throughout the working day. However, scholars such as Hayles (1999) argue that knowledge held by a human cannot be fully embedded in a digital or mechanical system, because of the embodied nature of such knowledge. Even in a tiny, fragmented task, the embodiment of the worker performing the task comes into play.

The second major theme I address in this chapter has to do with the ways in which people doing intangible labour, like software development, organize their worlds—both materially and socially. Picking up on some of the themes arising from my analysis of the LGRU, and drawing on literature from areas like digital and immaterial labour, I explore the physical infrastructures implicated in the carrying out of what might be considered digital work. With the help of literature from infrastructure studies and some of the more exciting corners of the digital humanities, I make the case that a concern for the material is not just interesting as an area of contention, but as a practical way of gleaning insight from milieux which we may oversimplify when we see them as purely digital. In particular, I argue that a

146 concern for the material conditions of digital production is crucial to a more complete understanding of the factors, stakes and relationships at play in such modes of production. Though an argument like this is made elsewhere (in for example, Dourish and Mazmanian [2011] and Kirschenbaum [2008]), making the case that the infrastructures of digital work, like the cables through which we access the internet and the magnetic charges on a hard drive, have material traces, I wish to go further. I suggest that, not only is it important to consider the infrastructures supporting work we think of as digital, but to also think about the impacts and origins of some of the more marginal material supports used, and the ways in which different tasks mobilize those supports.

The body in labour and the embodiment of immaterial labour

In her book, In the Age of the Smart Machine, Shoshana Zuboff (1988) tells a story about the employees of a pulp mill—in the bleaching division—who, when a new set of digital monitors and controls is installed, find themselves at loose ends, no longer able to do the kinds of physical work they were originally trained for. In recounting the story of their ennui, Zuboff hits on an incident with an automated door. In order to move in and out of the highly- technologized, climate-controlled control room, the employees are meant to walk through one set of automatic doors, wait for those doors to close, and then press a button, triggering the second set of doors to open—an airlock, essentially. As Zuboff tells it, within three years of the installation of the doors and the control room, the doors were rendered useless, literally at the hands of the control room workers, who, rather than wait for the automatic mechanism to do its job, had taken to prying the doors open manually, rendering the airlock broken and useless, and exposing the computers and workers contained in the control room to the fumes of the bleaching plant. Zuboff is harsh in describing the actions of the workers, highlighting the contradiction represented by their treatment of the doors compared against their concerns about the effects the chemical fumes in the plant will have on their bodies. She critiques what she sees as a kind of irrationality in their behaviour, and seems at once to both romanticize their desire to be physical in their work and to infantilize them for it. At the foundation of the story, though, is a concern about what happens when work changes, and specifically what happens when the dynamics of control and expertise in work are upended.

147 ***

During the implementation in Uganda of the PrintAbility project, I had a moment of disagreement with one of my colleagues. The colleague, whose day job is as a software developer, envisioned a future where all of the messy, dusty, noisy machinery in the Orthopaedic Workshop would be gone, replaced by computers and 3D printers. The hand- work involved in the production of prosthetic and orthotic devices would be a thing of the past. I pushed back, arguing that, even if all of the devices were 3D printed, finishing work would still need to be done, which would require grinding machines, at the very least. My colleague suggested that smaller, quieter, hand-held versions of the sanding and finishing machines would do the job. We went back and forth more, never really resolving the disagreement. In part, I now realize that my distaste at the idea of replacing the manual process entirely is founded in the conviction that the professional expertise of the orthopaedic technicians and technologists is housed not just in their knowledge of clinical issues like anatomy and rectification, but in their use of their tools, in the material constraints and possibilities embodied by those tools, and indeed, in the knowledge they hold in their bodies. This knowledge is exemplified by tools like the grease pencils with which they mark up the residual limb during the process of palpation, and the insight they gain into the functioning of the leg by flexing it.45 By removing the tools of their trade, we would be removing one of the sites of their expertise. Such a suggestion parallels Zuboff's story of the pulp mill workers, boxed into a glass bubble of a control room, for whom an important site of activity and expertise was removed, only to be replaced by a technical system which purported to have their expertise built into it and which was meant to be an improvement on their former way of doing things.

*** In 19th century England, restrictions were placed on the export of industrial machinery and the migration of skilled tradesmen (Musson, 1972). Musson tells us that “Manchester manufacturers—whilst demanding removal of duties on raw cotton imports and abolition of the Corn Laws—at the same time remained stubbornly protectionist in their opposition to the

45 For more on this subject, see coons & Ratto, 2015.

148 removal of restrictions on the export of machinery” (1972, p. 19) for fear that manufacturers in the powerful textile-production centre would lose their international dominance in the cotton fabric market. He goes on to suggest that, prior to the 18th century, the state of British-made machinery and industrial expertise had been such that “foreign machinery and artisans were encouraged into this country” (p. 20), but that the advent of the Industrial Revolution changed the direction of the tide, turning British expertise into something with export value. Parliament responded by placing restrictions on export (whether that export constituted the sale of machinery to foreign industry, or the migration of skilled tradesmen). Knowledge inhered in both machines and in the skilled labourers capable of building them, and that knowledge constituted a competitive advantage for England's major industries. Both humans and machines were sites of expertise and economic advantage. The relationship between bodies and expertise, as well as bodies and labour, is the major area of concern in this chapter.

***

In order to discuss the status of the body in labour, we need to consider how the body became a machine generally, and a machine for labouring in particular. Many arguments offer the model of the Cartesian mind-body divide as a way of thinking about the feats that have been attempted in pursuit of controlled bodies. Hayles (1999) offers anorexia as an example of cognition being emphasized over embodiment in humanism: the mind controlling and dominating the body. Brumberg (1988), in her history of anorexia, suggests similarly that appetite suppression in both anorexia nervosa, and its older sibling, anorexia mirabilis, represented an exercise of power or moral fortitude over the apparently unruly and irrational (or unholy) desires of the body. Hayles also argues for the importance of the distinction between the body and embodiment, stating that "[e]mbodiment differs from the concept of the body in that the body is always normative relative to some set of criteria" (p. 196). She goes on: "Whereas the body is an idealized form that gestures toward a platonic reality, embodiment is the specific instantiation generated from the noise of difference" (p. 196). In this way, Hayles makes a case for the paramount importance of considering the body not as a discrete object, but rather for considering embodiment as a contextual, moment-to-moment interaction between the body and the social constructs and material circumstances around it.

149 She further argues that embodied practices and gestures “cannot be separated from [their] embodied medium” (p. 198), that practices exist in their embodiment, and representations of such practices can only be imperfect abstractions of the embodied movement, practice or gesture.

Foucault, with less apparent emphasis on both the individual governing of one's own body and the importance of embodiment, suggests that a move from absolute rulers maintaining either literal or symbolic power over the life and death of their subjects metamorphosed, in the age of less absolute control and an increase in the power of norms, into a need to control the lives of individuals within a polity. In describing the origins of bio- power, he states that

"In concrete terms, starting in the seventeenth century, this power over life evolved in two basic forms; these forms were not antithetical, however; they constituted rather two poles of development linked together by a whole intermediary cluster of relations. One of these poles—the first to be formed, it seems—centered on the body as a machine: its disciplining, the optimization of its capabilities, the extortion of its forces, the parallel increase of its usefulness and its docility, its integration into systems of efficient and economic controls, all this was ensured by the procedures of power that characterized the disciplines: an anatomo-politics of the human body." (1979, p. 139)

The need to govern and make productive human bodies, to integrate them into machines was, Foucault suggests, important to the development of capitalism. Capitalism, he argues, "would not have been possible without the controlled insertion of bodies into the machinery of production and the adjustment of the phenomena of population to economic processes" (1979, p. 141). The body, in both Foucault's telling and Hayles', is something which takes disciplining, either by the Cartesian bit of aether hovering over it, or by first an absolute ruler, and then a capitalist. In my case studies, there is an attention, as well, to the body as something which not only needs to be managed, but as something requiring accommodation. In the production of a prosthetic leg, in the fitting of a dress, in the feeding of a software developer, we see examples of bodies needing to be attended to, either as the focus of labour, or as a necessary

150 precondition for work. Conversely, the production of a prosthetic leg offers an opportunity for a body to be rendered economically and socially useful, with mobility, social inclusion, and employability being desirable outcomes for the prosthesis wearer. Foucault's conceptions of the body and its governance as key elements in the development of capitalism and normative social structures are highlighted even further in the development of scientific management.

Taking the mind out of manual labour, taking the body out of mental work

"The achievement of Taylorism was to contain and discipline the working subject, separating the consciousness of the labour process and its embodiment in labour. From Taylor's time onwards, the fundamental managerial control strategy was to constitute management as the thinking subject, leaving workers as a bodily residue who would perform the gestures of production without autonomy. Subjectivity and identity would be separated from the working body, as its labour was divided and subdivided by management." (Bahnisch, 2000, p. 64)

The premise described above by Bahnisch is emblematic of the Taylorization of labour. Frederick W. Taylor's efforts around the turn of the 20th century made a mark on the way labour is even now divided and specialized, and on the role of management in industry. Through his efforts to determine what was physically possible and constituted “a fair day's work” for an idealized kind of worker, Taylor helped to instantiate the idea that someone other than the worker carrying out the task should be the one to determine which parts of a task should be performed, how, and when (Bahnisch, 2000). As Bahnisch puts it, Taylor's idea of the division of labour, in his pet case study, a steel plant, featured a manager who should be able to "programme every movement of the pig iron handler on charts in the security of his office, then turn his gaze to the window to observe the embodied object of his managerial subjectivity at work" (2000, p. 63). In order for this to happen, though, labour needed to be understood and codified. Scientific management needed to be preceded by scientific characterization. In Taylor's observations of the actions of pig iron handlers, and in his contemporary Frank Gilbreth's efforts to create taxonomies of physical movement, we see the desire of scientific managers to codify labour, in order to transfer its knowledge and control to

151 managers who, themselves, had little or no embodied knowledge or craft tradition of the actions required to complete particular manual tasks.

Beyond codifying labour and putting it in the hands of managers, another tactic of scientific management was the embedding of knowledge in machines (an issue highlighted in the brief vignettes above). Hetrick and Boje (1992) cite the use of specialized machines as one of the hallmarks of Fordist styles of management, the newly-specialized tools representing a removal of craft knowledge from the skilled worker and its instantiation in a process or a machine. Haydu (1988) similarly provides an account of skilled tradesmen fighting back against the replacement of general purpose machines in shops with machines designed to perform fewer, more specific tasks. Bahnisch (2000) characterizes the embedding of tasks and knowledge in machines as an "appropriation of workers' autonomy and control over their work, the construction of a politics and a technology of the disciplined body at work" (p. 55), a key facet of Fredrick W. Taylor's particular conception of scientific management. Bahnisch further argues that such an attempt to control the body of the worker through procedure and technical constraints was an effort by capitalists and a nascent managerial class to discipline what were seen at the time as the dangerous, potentially unruly bodies of labourers. This focus on the danger represented by the body of the worker is one that Bahnisch picks up on in analyzing a speech given by Taylor. Bahnisch makes the suggestion that, in contrasting the education level of the manager against the embodied activity of the manual labourer, Taylor is contrasting the comparatively frail body but well-developed mind of the manager against the un-developed mind and strong body of the labourer. Bahnisch later characterizes this as an effort by scientific management to “reinscribe the subjective bodies of workers as corporeal objects of disembodied managerial minds, a colonizing move typical of binary discourse" (2000, p. 65). Or, in Foucault's sense, the "modern man [,...] an animal whose politics places his existence as a living being in question" (1978, p. 143).

This idea of the objectification of the worker, in service of the subjectivity of the manager, is echoed by Hetrick and Boje (1992) who, along with numerous others (eg: Sewell and Wilkinson, 1992) compare Fordist management and some of its postmodern kin (like Just- in-time and Total Quality Management) to Foucault's characterization of the panopticon, in its

152 capacity to subject those contained within it to not just surveillance, but an internalized self- surveillance. In the information age, this internalized surveillance is supplemented by the new tools of surveillance embodied in what Zuboff refers to as “informating” and what we might today find in such concepts as big data and the Internet of Things, both of which can be implicated in the collection and use of data about the activities and functions of workers. Similarly, version control systems can be used to track the contributions of software developers to projects, and the use of 3D scan data in the production of prostheses creates a trail of modifications which is not possible in the manual method. Though the uses of these systems in my case studies does not come with the overt intention of tracking labour, such systems do provide opportunities to do so. The increasing digitization of work comes with built-in opportunities for tracking, version control, and fine-grained management of time and work. With this in mind, it is worth making a digression into the idea of digital labour, and just what that term might mean in the context of this dissertation.

Defining digital labour

In the introduction to Digital Labor: The Internet as Playground and Factory, Trebor Scholz (2013) points out (as many free labour theorists before him have also done) that much of the labour undertaken on the internet is not thought of by the people doing it as labour, but does produce value for the companies harnessing it. With every status update, you create value for Facebook. With every search, you create value for Google. Thus, in creating value for these companies, you must be working. By contrast, William Morris (per Kinna, 2000) argued in the 19th century that leisure, when it is anything but simply lying in contemplation, can be seen as a productive force, and an extension of work. Morris did not position this as exploitation and instead thought of it as something one could do for themself. Contrast Morris's ideas of voluntary labour against arguments like Scholz's: that the recreation we derive from websites like Facebook is actually a form of unpaid labour, providing benefit to corporations. Or, as a popular internet truism puts it, “If you're not paying for something, you're not the customer, you're the product being sold” (Ross 2013, p. 18). Unpicking the tangle of who gets what kind of value from which activities is not a trivial task. Terranova (2000) famously gives the example of the chat hosts in the early days of AOL. These were volunteer

153 facilitators and builders of AOL chatrooms, with special privileges, granted to them in exchange for their contribution to the development of online communities. They did not get paid for their work. But these are not necessarily arguments which need to hinge on the digitality of the work. They are, instead, arguments about free labour more generally. Though there is certainly an argument to be made for the capability of computers and the internet to harness more free labour from individuals, we should not take free labour as an intrinsic part of our definition of digital labour.

A whole slew of issues are tied up in the concept of digital labour. Definitions of just what digital labour is are often either unclear or in conflict with one another. Many of the influential, foundational arguments about the nature of digital labour are from the late 1990s and early 2000s, when the internet was a very different place, and far more discrete from “real life” than it is today. A foundational essay, published in 1997 by Richard Barbrook, gives this rendition of the “digital economy,” as it was at the time:

By allowing people to acquire some basic knowledge of making hypermedia, the hi-tech gift economy is helping to create a skilled and innovative digital labour force. However, it is very difficult to adapt the traditional factory system to managing these new workers. The rapid spread of personal computing and now the Net are the technological expressions of the desire of many people to escape from the petty controls of the shopfloor and the office. Despite the insecurity of short-term contracts, they want to recover the independence of craft labour which was lost during the process of industrialisation. Because of rapid technological innovation, skilled workers within the hypermedia and computing industries are precisely those best able to assert this desire for autonomy.

Terranova sums up Barbrook's view of digital labour somewhat more succinctly: “the digital economy is characterized by the emergence of new technologies (computer networks) and new types of workers (the digital artisans)” (Terranova, 2013, p. 35). This definition positions computer networks as an infrastructure and, perhaps, an object of concern in digital labour. This idea that what makes digital labour somehow different from other kinds of labour is the integration of networked computers makes every kind of labour increasingly digital. In this

154 regard, I am reminded of an insight from an ethnographic study conducted by Rachel and Woolgar (1995), situated in the Systems team of a large utility company. They note that, though the workers on the Systems team spend their days working in front of computers, they do not think of all of their work as "technical" but instead view their "technical" work as taking place when they do work on the mainframe (as opposed to the times when their work is being structured by the assignment-management system, an activity which also takes place through computers, but is not, in their minds, a technical task). For the Systems team (in the early 1990s, it should be noted), “technical” is about making the computer an area of concern, not simply a tool. I similarly suggest below that digital labour, if it is to have any meaning at all, must be situated in work which we somehow view as uniquely digital, not simply work which makes use of computers and networks.

But this poses a problem: work which takes digitality as part of its identity is a moving target. In making sense of that problem, I might suggest that there are a few different kinds of labour: there’s the kind that we’ve historically talked about, working in factories and making physical objects. There’s the labour done by people in the service industry. It has a clear physical instantiation, and relies on a face-to-face interaction, unless it takes place over the phone, or over video chat, or any number of other digital substrates. Are call centre workers digital labourers? Is a phone sex operator a digital labourer? But this approach seems reductive: it is defining digital labour as something that doesn’t involve either the production of a physical good or a face-to-face interaction with a customer. In that respect, most of the people in office buildings are digital labourers, typing away on computers, sending emails, accessing company intranets. Another element that seems to come into the definition as assumed is that the internet or computers (but not just word processors) are involved as a platform or major subject of digital labour (the “computer networks” that Terranova refers to). Despite the fact that the standard office worker may sit in her cubicle, tapping away at a computer and checking her email, computers may not be enough of a subject to her for her to be a digital labourer.

Perhaps we can take as a definition that digital labour is labour which sees digital technology as an issue, not just a tool. Terranova's AOL chat hosts fit into that definition. They

155 would be digital labourers because, at the time they were volunteering their labour to AOL, the internet was new and exciting, and their labour was about the internet46, its growth and its dissemination. The element of being about rather than just using seems to be an important one in the way the idea of digital labour is used. What that means is that, once a given technology becomes naturalized enough (like the word processor or the email client), the person using that technology ceases to be a digital labourer and becomes simply a labourer. It's an interesting idea, but it also speaks to the growing irrelevance of digital labour as a category. In my case studies, the question of who is working with digital tools, versus who is working on digital goods is open to interpretation. In the LGRU, despite the use of non-digital tools like the paper and pens used to discuss, brainstorm and plan, the ultimate output is a piece of software. By the logic that a digital labourer is someone who takes the digital as an issue, the participants in the LGRU would be carrying out digital labour. By this logic, those of us producing the Socketmixer software for the PrintAbility project would also be undertaking digital labour. But what about the prosthetists who ultimately use the software to produce sockets?

The question of who among the members of the PrintAbility project might be called a digital labourer returns us to the problem of the body in so-called digital work. The issue is what the labour produces. Software development is digital labour that produces digital artefacts. Other kinds of digital labour produce other things. The purity of software development is that the product of it is something that goes back into the digital ecology (making the LGRU a comparatively clear case). That’s different from a graphic designer, whose labour may be almost entirely digital, but who produces, say, a book cover, which is ultimately a profoundly physical object (unless it's for an e-book). On a spectrum of digital labourers, we might see software developers and even web developers as the most digital, because the product of their labour is a contribution to the digital world. An industrial designer might be one of the least, and might even fall on the edge of who we can define as a digital labourer. Does he count as a digital labourer if 2/3 of his work happens in a 3D modelling program, but then he works with clay models and hand-drawn sketches? Has he already ceased to be a digital labour because he works with computers, but is not concerned with them? The

46 Although not the actual internet, of course, as AOL was a walled garden with very little real relationship to the larger internet and emergent World Wide Web.

156 prosthetists using the PrintAbility tools suffer from a similar definitional problem: when their work encompasses 3D scanning, 3D modelling and 3D printing, but produces a plastic socket which might then be modified using hand tools, does their work count as digital labour? In short, does a digital labourer need to contribute back into some kind of digital ecosystem in order to be considered as such?

In the time since Barbrook's provocation on the nature of digital labour, even if the definition of the term has become no clearer (and possibly even less) capitalism has arguably caught up with the hypermedia utopia of the 1990s. Two key examples of the management and control of digital workers (meaning here that they conduct all of their work in computer- mediated online environments) are exemplified in gold farmers and Turkers, a pair of terms which may seem opaque now, but which I will use in the next section to draw comparisons between the conditions of digital labour and those of industries which are not typically considered digital.

Labour structures in digital production practices

In the introduction to this dissertation, I cited a number of popular examples which position digital fabrication technologies as being emancipatory. The argument seems to be that, in investing both the power to design and the power to fabricate all in one person, the conditions which cause the alienation of labour can be abnegated. Digital fabrication—and especially 3D printing—is, by this logic, a way of re-uniting the planning and the labour which were ostensibly torn apart by scientific management and the rationalization of labour. However, as I highlight in chapter five, one of the other selling points of 3D printing appears to be its disembodiment. Advocates and popular descriptions of 3D printing highlight its ease of use, and the lack of material work involved in producing a print. Just download a model, hit the print button, and enjoy your new object. While the entire process might well be in the control of one person, the degree of that control seems as if it could be questioned. When the entire process is automated, can it truly be said to be in the hands of the person operating the printer? Can 3D printing really de-alienate us?

157 Unfortunately, an answer does not (and cannot, given the scope of the problem) arise from my case studies. In the case of the LGRU, all hands involved in the scoping and development of software and other artefacts seem invested in their work, with little sign of alienation or lack of agency. But the LGRU is admittedly a case in which many participants are involved in an entirely voluntary manner, with no financial incentive or coercion. It is more a case of Morris's productive leisure than the extraction of value by a capitalist who wishes to profit. Perhaps the production of goods in the framework of community-organized collaboration is the de-alienating feature, and the goods they are producing have little to do with it. In the development of the PrintAbility project, similar issues apply. Though we produced software and developed a workflow, the structuring of our labour was not analogous to what one would typically think of as the alienated labour of the factory or Taylorized office. Instead, participants in the PrintAbility project were often involved for reasons like the desire to do something beneficial for society, an interest in researching the project, or a desire to advance an understanding of new production practices to be used in the developing world. Our labour, though in some cases waged, and certainly less overtly communal than the work done in the LGRU, was still, in its organization, not constrained by a routinized and alienating process. Even in the Orthopaedic Workshop at CoRSU, which meets the most criteria of any site for being like a normal workplace, the technicians seemed to have a modicum of control over what they did, and had a sense of ownership over the devices they were building. In this sense, none of my sites conform to the strictures of a clearly Taylorized or overtly alienating workplace. It is thus impossible for me to make claims about the emancipatory potential of digital fabrication tools in my particular case studies. It is, however, possible for me to suggest that claims about the emancipatory and de-alienating power of digital production technologies need to be made with sensitivity to contextual differences. One cannot say, uniformly, that such technologies move power into the hands of the individual, when a variety of structures for organizing labour (whether paid or voluntary) already offer the potential for de-alienation.

We see above that the issue of the governing of bodies and minds, and sometimes the attempt to decouple the two, is at the heart of modern mass production and scientific management, and that technologies are not the only determinants of where power and agency reside in work. In Barbrook (1997) we see the hope that individuals working in the digital

158 realm (free of the strictures of offices and factories!) might extricate themselves from the kind of control embodied by the sites of scientific management and its habits of organization. We see further, from scholars like Terranova, the concern that working on the internet might instead cause a different kind of shift, with the checks and balances that have come along with industrial production disappearing, in favour of flexible labour and little security. In this regard, an exploration of the social organization of labour and the material conditions of production, and their mutual inextricability, is necessary. In first recognizing that there are material conditions of labour in digital work (not just social hierarchies), I wish to compare two classes of present-day digital labourers (taken here to mean something similar to Terranova's definition—labourers working on the internet) against a somewhat similar case from the end of the 19th century and the beginning of the 20th.

Gold farmers Everyone who wants to talk about the working conditions of gold farmers seems to go back to an article published in the New York Times Magazine in 2007, which presents something of an exposé about the lives of said gold farmers. A gold farmer is a third-party worker in an online, multiplayer video game. They are not employed by the companies which publish and run the games—like Word of Warcraft—in which they reside. Instead, as we learn from Dibbell (2007), the gold farmer (or at least Dibbell's conception of the gold farmer) works for a small-to-medium company in China, earning either a wage or a piece rate for collecting valuable goods and currency in Massively Multiplayer Online Role-Playing Games (MMORPGs). The gold farmers do this by playing the game for 10-12 hours a day, constantly killing beasts and other obstacles in order to collect the valuables dropped when their foes die. The valuables, given over to the supervisors of the gold farming operations, are then sold in online marketplaces where game-currency and goods are sold for dollars, euros, and other legal tender. Dibbell suggests that, with regard to the material and social conditions of their work,

159 "on the surface, there is little to distinguish gold farming from toy production or textile manufacture or any of the other industries that have mushroomed across China to feed the desires of the Western consumer. The wages, the margins, the worker housing, the long shifts and endless workweeks – all of these are standard practice. Like many workers in China today, most gold farmers are migrants." (Dibbell, 2007)

He goes on to suggest that what differentiates the gold farmers is that their labour is also play, that after their work day ends, they go to internet cafes and play World of Warcraft on their own time. To me, that's a slightly spurious argument, implying that work can never be enjoyable, and that all waged, factory-style work must somehow be outside of the human interests of the worker.47 Instead, what interests me is the suggestion that the work of gold farming has much in common with the way manufacturing work is done in China, that what would have to be thought of as digital labour is in fact materially similar to such staunchly non-digital labour as textile production. It offers the possibility that the “petty controls of the shopfloor,” as Barbrook put it, have managed to permeate the digital economy.

Turkers A similar management of digital workers is evidenced by Amazon's Mechanical Turk service, which pairs hyper-casual workers with clients to do micro-tasks. As Bergvall- Kåreborn and Howcroft put it, the origin of Mechanical Turk was in an itch Amazon wanted to scratch:

"Amazon's retail website has large numbers of replica product web pages and rather than enlist temporary workers to resolve the problem, Amazon developed a component within their platform whereby a potentially global workforce could check for product duplicates and receive payment for each item in a manner that resembles piece work" (2014, p. 214)

As they tell it, Amazon then opened the platform for the use of others. Turkers (as workers on Mechanical Turk are known) are paid tiny amounts of money to do very small tasks. 90% of 47 Indeed, in an industry like the garment trade, which I address later in this section, it would have been positively wasteful for workers to not make use of their trade skills for their own benefit outside of work hours.

160 tasks are ascribed a value less than $0.10 (USD) (ibid). A larger job (such as transcribing a piece of video) might be broken up into smaller units (the video being cut into shorter segments) in order to maintain the tiny scale of a standard Turker task. In this way, Mechanical Turk buys into the idea that humans are roughly interchangeable and that, where they aren't, that lack of interchangeability can be circumnavigated by measures like numerical scores representing someone's success rate on previous tasks (a key feature of Mechanical Turk), or information about where the person lives (with regard to issues of cultural or linguistic compatibility). Here, we see the return of the rationalized human and the logical extreme of scientific management. All Turkers within a particular score range are roughly interchangeable, and, as Bergvall-Kåreborn and Howcroft suggest, those Turkers, acting as self-employed workers, must manage their time and their tasks in a way that optimizes the amount of income they can make in a given period of work.

Outworkers All of this hearkens back to the outworkers of the late-19th and early-20th century garment trades in North America and England, who worked in a system which supplied the slop shops described briefly in chapter four. Like today's Turkers, outworkers or homeworkers took in jobs from centralized sources, completed them in their homes (in the case of outworkers, the work was most commonly sewing), and delivered them back to the factory or middle man from which they originated (Hapke, 2001). They were similarly paid on a piece- work rate (ibid), although their pieces were significantly more time-consuming than the type of work Turkers do. In short, the structures of the putting-out system and of Mechanical Turk have some notable similarities. Both involve outwork, done, if not always in the home, then at the very least, outside of the employer's facilities. Both involve piece rates and the distribution of pieces to a large number of decentralized workers. Both involve workers who do not enjoy the fringe benefits one would expect from more regular employment. So, though the Turker works in front of a computer instead of with a sewing needle, the structure of their labour is not so different from the way piece-work and outworking were done a hundred years ago. Crucially, the outworkers, the Turkers and the gold farmers are all examples of, as Marx would have it, the stagnant population of the reserve army of labour, or the individuals un- or under- employed to such a degree that they are both flexible in their labour and command lower

161 prices than regularly-employed labourers. Foster, McChesney and Jonna (2011) argue that the exploitation of labour in the global South includes the development of a “global reserve army of labor” (p. 2) which “constitute[s] the real material basis on which multinational capital was able to internationalize production—creating a continual movement of surplus population into the labor force, and weakening labor globally” (p. 2). If we see Turkers and gold farmers as simply a globalized extension of the same phenomenon that made outworkers a key feature of the 19th century garment manufacturing industry, it becomes apparent that there is little difference in the structures governing their labour, and indicates that the promise of digital labour as something with the ability to make real structural change does not ring true.

The social organization of labour and the material conditions of production

As we see in the case of the gold farmers, Turkers, and outworkers, structures of organization and control can and do persist, even as the medium of work moves from the garment in need of sewing (and its kin) to the gold that needs farming or the tiny task that needs efficient doing. I am certainly not the first to suggest that work done on computers employs modes of social organization and control similar to industrial methods. Indeed, even Dibbell alludes to that idea, in his account of gold farmers. This suggestion that the digital is not so different or divorced from everything that came before it also shows in another kind of argument, one about the need to account for the materiality of the digital. In a review paper on the subject, Dourish and Mazmanian (2011) outline five different streams of enquiry in what they call "information materiality" and note that the five are "united by the recognition that the ineffable 'stuff' of digital abstractions is encountered only ever in material forms, and the nature of those forms has consequences for how information practice develops" (p. 8). In defining one of their five streams, which they term “thematerial conditions of information technology production” (p. 6), they suggest that scholars doing this kind of work concern themselves with the things that make information technology work, like labour, corporate, governmental and extra-governmental structures, among others. On a somewhat more detailed scale, both Dourish and Mazmanian (2011) and Kirschenbaum (2008) recount stories of forensic (in Kirschenbaum's words) examinations of digital storage media, with Dourish watching a technician read the magnetism on a tape, using a small device with metal filings

162 suspended in it, while Kirschenbaum examines the layout of data in the sectors of a disk. Both of these explorations of the material structures of data construct a similar argument: the digital is not aetherial.

It is not unusual (as we see with Dourish and Mazmanian) to consider digital or technical artefacts as a whole constructed of a huge variety of socio-material parts. Kirschenbaum puts it particularly elegantly (if also slightly glibly):

"Software is the product of white papers, engineering specs, marketing reports, conversations and collaborations, intuitive insights, professionalized expertise, venture capital (in other words, money), late nights (in other words, labor), caffeine, and other artificial stimulants. These are material circumstances that leave material (read: forensic) traces—in corporate archives, on whiteboards and legal pads, in countless iterations of alpha versions and beta versions and patches and upgrades, in focus groups and user communities, in expense accounts, in licensing agreements, in stock options and IPOs, in carpal tunnel braces, in the Bay Area and New Delhi real-estate markets, in PowerPoint vaporware and proofs of concept binaries locked in time-stamped limbo on a server where all the user accounts but root have been disabled and the domain name is eighteen months expired" (pp. 14-15)

In Kirschenbaum's lyricism is an underlying theme taking up the same cause as that articulated by Dourish and Mazmanian, that digital artefacts are imbricated in a complex structure of socio-material factors encompassing infrastructure, labour, capital and apparent trivialities like whiteboards and coffee cups.

In considering the infrastructures of digital work, I wish to pick up on two key aspects: the material conditions of production and the social organization of labour. In this framework, Kirschenbaum gets closest to the meaning I wish to take, in articulating the importance of even the trivial things in the work environment that structure the ultimate result of digital production. Those immediate conditions, like access to equipment, chart paper, coffee, spaces with reconfigurable tables, and so on, are the often unique or at least varying conditions of production that can make, for example, two apparently similar teams of software developers

163 suddenly seem a little different. In short, I will now follow the idea that thinking of the previously digital object as material is not the same as thinking about the wider material conditions of production, and argue for the importance of considering such marginalia as a way of rhetorically re-incorporating the body into labour, recognizing that, outside of the realm of rhetoric, the body has never actually been absent.

Something similar goes for the consideration of the social organization of labour, which needs to account for not just the stable structures, but the ad hoc ones. Here, we have some recourse to organizational studies. Orlikowski suggests a move away from grandiose understandings of organizational knowledge, and towards an idea of “organizational knowing as emerging from the ongoing and situated actions of organizational members as they engage the world. It is an explanation grounded in what it is people do every day to get their work done” (2002, p. 249). She further emphasizes the idea that knowledge and practice are mutually constitutive, with “situated knowing constituted by a person acting in a particular setting and engaging aspects of the self, the body, and the physical and social worlds” (2002, p. 252). It is this attention to the small-scale, mutually-constitutive, changing practices and organizations which I feel is crucial in both developing deep understandings of the work taking place in an organization (or a conflagration of interconnected organizations, in the case of projects like the LGRU) and in connecting the social organization of labour meaningfully to the material conditions of production. I further wish to highlight an insight from Star (1999), who positions “infrastructure as part of human organization” (p. 380), “processes and relations braided in with thought and work” (ibid). In this view, all of the invisible layers of infrastructure, the unquestioned things from the electrical outlet to the coffee maker to the flip chart and markers, are the necessary substrate for any organization of labour. This is yet another place where the material conditions of production and the social organization of labour intersect and are inextricable.

Such inextricability is evident in my case studies. It is manifested especially clearly in PrintAbility. Everything from the difference between having 3D printers already on site in the Critical Making Lab (versus renting time on printers owned by others) to the difference in communication brought on by having team members either working remotely or being

164 beholden to other organizations shows how the material conditions of production impact the way labour can be organized, and vice versa. Though many aspects of PrintAbility show the interrelation between the material conditions of production and the social organization of labour, one example is particularly apt. The process for creating soft liners for sockets was something we had thought about to an extent during our development process, but it had been a secondary concern. Canadian prosthetists and orthotists had suggested that we use multiple layers of socks as liners, while the prosthesis-wearers we interviewed found the prospect of socks impractical and uncomfortable. We had dismissed the idea of making the liner as it is made in the ICRC process, because we were given to understand that its production required a plaster positive, which our process would not have. We considered other avenues, such as generating a sewing pattern based on the scanned limb, and sewing a liner out of heavy felt. It was only when we arrived at CoRSU and had printed the first socket that one of the technicians there suggested that we try making a foam liner like the ones in the ICRC process, but molding it to the shape of the inside of the socket, in the absence of the plaster positive that would have been used conventionally. Though we had some concerns about whether having a hot piece of foam pressed into it could potentially melt the surface of the socket, the experiment proved to be a success. Without the combination of the experience of the technician, the materials and tools required to make the liner, and the printed socket, the experiment would not have been possible. Though we had expertise at our disposal during the Canadian portion of the project's development (in the form of a former president of ISPO), the combination of the Orthopaedic Workshop, the printed socket, and the technicians made it possible to solve a problem that had previously remained unsolved for a period of months.

In the case of the technicians in the Orthopaedic Workshop, embodied knowledge is omnipresent. As one Canadian orthotist I interviewed put it, much of what prosthetists and orthotists do is learned with their hands. In thinking of prosthetists and prosthetic technicians as craftspeople, there is a close coupling of the mind and body in labour, as the work of the prosthetist or technician is in the handling of materials and the palpating of residual limbs. In the work of the dressmaker, something similar holds true. Because the custom dressmakers were broadly working before the popularization of sewing patterns, the knowledge of the dressmakers was instantiated in the fabric they pinned onto the client and the modifications

165 they made during fittings. How such knowledge is translated to screen-based goods is an interesting problem. But in looking at the material conditions of production in the LGRU, it becomes evident that at least the development and transfer of knowledge still inheres in the bodies, gestures, drawings and poster paper of designers and developers, just as a prosthetic technician working on a digital 3D model still flexes and palpates her own knee while thinking about how a patient's knee will move when flexed similarly. In this regard, attending to the material conditions of production provides insight into a crucial labour issue: the separation of the body and the mind in the organization of work.

In considering the material conditions of production, the social organization of labour, and their crucial intersections, we need to account for not just the role of the producer in production, but, increasingly, the user, who is arguably less different from the producer than in mass production models. As Poster (1990) argues, “[u]ntil now commodities were difficult to reproduce. A complex combination of materials and skills were required to make almost everything. Producer and consumer were separated by the process of production. Clothing, appliances, furniture – few consumers imagined they could provide these for themselves. Books, music and film were no different” (p. 73). Though Poster goes on to construct an argument about the relationship between the content and the medium on which it was delivered, the argument about the historical difficulty of production is one which can be applied to other goods, and which I explore further in the next chapter. In particular, I look at the current move towards mass customization, its varied characterizations and forms, and the varying ways users are (and have previously been) enlisted into the production process.

166 7 The users of custom goods

This chapter addresses an issue which I have highlighted repeatedly in previous sections. In the previous chapter, I looked at the relationship between labourers, their social structures and their material environments as they relate to customization and digital fabrication. Building on those concerns, this chapter explicitly tackles the role of the user in the production of custom goods. Though I take those three elements—the social organization of labour, the matetial conditions of production, and the role of the user in production—to be necessarily related, I address the role of the user in production in this chapter in order to relate it more effectively to existing literature which treats user involvement as separate from production. I specifically treat digitally-fabricated mass-customized goods in this chapter, addressing and evaluate claims made around mass-customized goods in relation to the benefits they are supposed to bring, or the problems they are supposed to solve for their purchasers48. In this respect, I use this chapter as both a place to elaborate some of the broader issues of mass-customization and the user raised elsewhere in the dissertation, and also to look at a few specific examples of mass-customized goods, including the successful NIKEid running shoe customization platform. I contrast these against examples from my own case studies. In so doing, I also confront the issue of the parameterization of user choice, and ask: when does a mass-customized good become meaningfully custom for its user, and when is it simply a parametric extension of mass production that makes use of a new technology or process to leverage existing production infrastructure to increase profit for its producer? In relation to

48 I am concerned with their benefits to their purchasers because their benefit to their sellers is fairly obvious and already well documented: increased revenue and, as Zwick, Bonsu and Darmody (2008) suggest, improved consumer retention and satisfaction.

167 that question I also highlight a key ambiguity in the popularization of mass customization, namely the different meanings of the word “custom” and how customization, when attempted on a mass scale, is often highly parametricized. I further contend that parametric customization, in basing its customizability on production infrastructure, often leaves little room for user agency.

Mass production versus mass customization versus mass personalization

Von Hippel (2005) offers a useful definition of mass production: "Mass manufacturers tend to follow a strategy of developing products that are designed to meet the needs of a large market segment well enough to induce purchase from and capture significant profits from a large number of customers" (p. 5). This definition admits the possibility that a manufacturer can look at a need, determine that it is not a profitable one to meet, and comfortably ignore it. The process of ignoring it, as determinedly rational as anything else in the decision-making processes around mass production, has to do with heterogeneity of needs. Von Hippel posits that heterogeneity is high when it takes a large number of standard products to fulfill user needs, and low when it takes a small number of standard products to fulfill the needs of the same number of users. In a market with low heterogeneity, mass production works well and is, apparently, capable of accommodating the needs of large numbers of users. In markets with high heterogeneity, on the other hand, von Hippel sees an opportunity for new modes of production, like mass customization. Mass customization through digital fabrication technologies is often positioned as promising to allow user choice, while still offering manufacturers a simple calculus of profit.

According to Marsh (2012), there is a difference between mass customization and mass personalization. Mass customization, he suggests, is a process by which a manufacturing company can extend its product range through the use of automation, changes to supply chain management, and modular parts to drastically increase the range of options it is capable of providing in a given product. He offers Toyota cars as an example of this, with what was the apparently industry-changing ability to more easily incorporate a range of purchaser-specified options into the cars they produced. Marsh contrasts this against mass personalization, which

168 he argues is a process by which a unique product is made for each customer, on the fly, as in the case of a company which manufactures lenses for eyeglasses. All of this is of course contrasted against simple mass production, in which a factory is tooled up to produce a highly limited range of products, but in huge numbers. Marsh offers all of these examples and contrasts in making the argument that a new industrial revolution is under way, and that the revolution is exemplified by the move to mass personalization, as he calls it. Marsh even makes a prediction ideally suited to some of the cases discussed in this dissertation, that “[w]hen 3D printing techniques become an everyday part of manufacturing, mass personalization will truly have come of age” (Marsh, 2012, p. 61).

In chapter four, I noted a distinction, highlighted by Green (1994) that, in the garment trade, the quintessential move in mass production was “the production of garments in the absence of known, measurable customers” (p. 729). What is intriguing about Marsh's definitions of mass customization and mass personalization and parametric customization (a concept introduced in chapter two) is that, in varying degrees, they rely on the mass customer, a legion of someones not known to the people designing the system through which their goods are made. The mass customization of an automotive manufacturer not only does not draw a one-to-one relationship between the good and its purchaser/user, but does seem to assume some notional user, albeit one with slightly different tastes than another notional user. In Marsh's articulation of mass personalization, on the other hand, there is a glimmer of the custom as Green would have it, with an object like a lens relying on the existence of a given purchaser for its existence. It is not made speculatively, but is instead made on demand, based on the specifications of its purchaser. Even so, it suffers from a parametric logic: there are a fixed number of variables which can be tweaked in its production, and possible values of those variables are what determines the range in which customization can take place, curtailing user choice. So, though the end product itself certainly does appear to be custom, like most products and systems articulated in the logic of mass customization and mass personalization (the two seem to be used interchangeably in many places), it exists in a system which relies on many of the same assumptions as mass production, most importantly that there is no meaningful change in the actual process of production, something which is especially helped along by the advent of digital fabrication technologies, which offer much scope for broadly

169 unchanging production processes, while still offering products which are individually tailored to their purchasers.

Based on the definitions above, and emerging from my case studies, I would like to set some criteria for how non-parametric customization processes (what we might simply call “customization”) differ from parametric customization processes. It is important to articulate the differences between the two, as parametric customization is on the rise, but does not necessarily suit all use cases for custom goods. Though an online interface for parametrically defining the measurements of a garment might well incorporate the necessary fields for customizing a pair of trousers, it might well lack the capability to incorporate features crucial for individuals with less standard use cases, such as an amputee who needs lining in the legs of his trousers, thanks to the extra stress put on the fabric by his prosthesis. The addition of lining to a pair of trousers is something easily accomplished by a tailor working directly with a client, but may be prohibitively difficult to accommodate in a parametric system which has variables defined by an industrial production process. I suggest that, in order to differentiate customization from mass or parametric customization (or even Marsh's mass personalization), there are four criteria that need to be met: There needs to be a known client for whom the good is being produced; the mode by which that client interacts with the producer can't be entirely constrained by a structuring agent like a user interface; there needs to be some flexibility in the range of parameters capable of being modified (once again, the parameters can't be completely constrained by technical or interface choices); and there needs to be some flexibility to alter or change the materials being used in producing the good.

My case studies demonstrate those criteria. All three of them start from having a known client, though both the products of the LGRU and the custom dressmakers might well have users beyond the one who initially is involved in the commissioning of the good. All three cases also meet the need for the mode of interaction between user and producer to be somewhat flexible. In the case of the LGRU, those in the “artist” rather than “developer” position were equally involved in the production of the LGRU outcomes. Interactions were not constrained by a formal set of rules. The custom dressmakers, though they were operating within relatively standard commercial constraints, had the flexibility to interact with their

170 clients in a range of ways, whether in their own establishments or in the homes of their customers. Clients equally had flexibility in their dealings with dressmakers, being able to make use of their arsenal of social tactics to change the way they related to their dressmakers. In the PrintAbility case, the final product explicitly aims to leave the same room for discussion and compromise between the prosthetist and the client as in the manual process. In leaving room for new parameters to be altered, the LGRU is perhaps the most clearly flexible, with almost no constraints at all on what can be produced, provided it is within the capabilities of the developers and artists working on it. The PrintAbility case is somewhat less flexible, but still leaves room for the set of parameters to be expanded or contracted. Though the socket needs to fit the residual limb, both the prosthetist and the patient have some room for discretion in making decisions about which parts of anatomy to accommodate, and how. The dressmakers have similar constraints, although their modifiable parameters expand beyond fit to encompass style and embellishment. In that respect, the dressmakers are the clear front- runners on the fourth criterion: that there should be the opportunity for choice in materials. With the dressmakers, not only was there choice, but the acquisition of materials was often carried out by the client herself. This is somewhat less the case in PrintAbility, in which materials are constrained by the production process being used, although there is still room for expansion in the print media available. It becomes obvious, in itemizing the ways in which my case studies convey the four criteria listed above, that there is meaningful divergence between mass-customized goods and those which are produced through a smaller scale customization process. In the next section, I take one of those differences—the labour contributed to production by the user—and examine it in more depth.

171 The role of the user in production: Bespoke versus DIY

Von Hippel (2005) offers a provocative pair of definitions, delineating what he sees as the fundamental line between the user and the producer. As he tells it, users are "firms or individual consumers that expect to benefit fromusing a product or service" (p. 3) while manufacturers (his term, which I take as equivalent to the “producer” at issue in this dissertation) are those who "expect to benefit fromselling a product or a service" (ibid). I view this as provocative because it makes the desire to sell a product the only criterion by which someone becomes a producer in von Hippel's eyes, as well as conforming slavishly to the use- value/exchange-value dichotomy, positioning the user as whoever hopes to benefit from use- value and the producer as benefiting from exchange-value. He takes this argument to its extreme by positioning developers of F/LOSS as yet another group of users, which simply re- confirms the profit motive as his only way of differentiating a user from a producer. I start with this provocation in order to ground a brief discussion about the range of activities which could be classified as user-production. The heading of this section positions bespoke production, in which a user has something produced for them by an artisan or other skilled tradesperson (or, increasingly, a semi-flexible, digitally-aided custom manufacturing process) against DIY (Do It Yourself), in which the user takes on the role of producer and follows the classic F/LOSS tenet of scratching their own itch. If we stay within von Hippel's boundaries of who may be a user and who may be a producer, the purchaser of a bespoke good and the practitioner of DIY are both equally users, seeking out customization, and doing so by purchasing different goods, services or materials. While someone buying a bespoke suit purchases both materials and a service, merged together in the form of a suit, a home sewer might well be purchasing fabric, notions, and patterns, but contributing the labour of constructing a garment herself. This distinction is important in light of current discussions about free labour and the contribution of user-generated labour to digital systems and goods. In part, I aim in this section to contribute to the development of a broader understanding of user-labour, and to trouble the distinction between the user and the producer in such discussions.

172 While my case studies could be positioned as either bespoke or DIY (the LGRU is DIY because its primary users are its developers, while PrintAbility and custom dressmakers offer bespoke products), I question the logic of making the profit motive the sole determinant of the producerness of someone engaged in a production process. Bespoke production certainly involves one or more people—the artisans or practitioners—receiving money in exchange for their work and goods, and a customer or client paying for receipt of those goods, marking one person or group as the producer and the other as the user, based on the exchange of goods for money. In DIY production, though an army of marketers would like to indicate that the materials of production are the product, and the assembler of those materials is merely a consumer engaged in a hobby, the same labour as in bespoke production takes place. The practitioner of DIY takes materials (as would be done by a professional artisan) and fashions them into a custom good. The major difference is that no exchange needs to take place, at least with regard to the finished product. The need to purchase materials is the only mark of userness, and the lack of desire to sell the resulting product is the only indication that the person who has made it is not a producer in von Hippel's sense of the word. But this logic breaks down in a case like the LGRU, in which many of the inputs being used to produce software have not been bought. The compilers, languages, libraries and assets are available to everyone, at no cost. Some of those inputs are considered to be public goods, with their development and distribution managed by non-profit organizations and foundations (such as the Python Foundation, which oversees the development of the Python programming language). It is possible, in effect, for someone to extract use-value from those tools and even to use them to produce other goods, without any monetary exchange taking place. If the mark of userness or producerness is a financial transaction or some other means of monetary gain on the part of the producer, then groups like the LGRU could be said to be neither users nor producers, buying none of their inputs and selling none of their outputs. Or, at the very least, if they derive use-value from tools, they are users, but they are users making use of tools made by other users, given the lack of monetary exchange-value involved in the distribution of the tools.

173 Von Hippel reckons with the problem of the producing user. He suggests that, as the desire or need for customization activities arises from having a market with heterogeneous user needs, many different products are needed in order to satisfy all or most users. Because mass production is cheaper for manufacturers, those manufacturers might well choose to ignore the needs of their more varied users, and simply target a large group with comparatively homogeneous needs. This means that users may be forced to develop their own options if their needs aren't served by what manufacturers offer. Hence the recourse to bespoke production and DIY. As von Hippel puts it, "users generally have a more accurate and more detailed model of their needs than manufacturers have" (2005, p. 8) leaving them the experts on what products they require in order to fulfill those needs. Von Hippel's model relies on the user identifying needs, seeking out goods which are capable of fulfilling those needs, making judgements about which needs can be sacrificed and which cannot, and acting in response to the constraints presented. For von Hippel, something known as a lead user emerges from this calculus. Lead users are "ahead of the majority of users in their populations with respect to an important market trend, and they expect to gain relatively high benefits from a solution to the needs they have encountered there" (p. 4). For von Hippel, these are often people pushing the edge of current best practices in their fields, or athletes participating in niche sports. These lead users have a high level of agency, and may even be lucky enough one day to have their innovations absorbed back into the manufacturing apparatus that was initially unable to serve them. In the interim, von Hippel suggests that "[i]t is becoming progressively easier for many users to get precisely what they want by designing it for themselves" (p. 2). The classic F/LOSS tenet of scratching one's own itch applies here. In scratching an itch, a user (in von Hippel's terms, someone who will not financially profit from their efforts) identifies a personal need and seeks to meet it through the self-production of a new product. In the case of a project like the LGRU, an organization or group of organizations fills the role of the user, identifying a need (in their case, tools which break out of current digital design paradigms) and finding likeminded people interested in contributing to its fulfillment. But this may be where both the itch scratching and lead user logic break down: the work of groups like the LGRU is not based on the simple solution of a problem. Instead, it seeks to explore a conceptual space, making use of the skills and desires of a diverse group of contributors. Arguably, any F/LOSS project of sufficient complexity falls under a similar

174 rubric. It becomes overly simplistic to suggest that everyone involved is contributing in order to meet the same need or scratch the same itch.

Others writing in fields similar to von Hippel's posit different sets of motivations on the part of users. Dahl and Moreau (2007) lay out an argument for the desire to do creative work, but with limitations. They first argue for the popularity of what they call “'constrained' creative opportunities” (p. 357) like kits and how-to guides, before suggesting that a major factor in consumer decisions to undertake structured or semi-structured creative work was “a motivation for personal accomplishment, which was achieved by satisfying the needs of both autonomy and competence” (p. 367). For Dahl and Moreau, like von Hippel, the people performing these DIY tasks are users still, rather than producers. They are the ones purchasing the kit or watching the cooking show, which means that one of the benefits of helping them fulfill their creative desires is growth in the sectors engaged in selling DIY goods and services. For this reason, offering customers/users the opportunity for more investment and more outlets for their efforts or tastes has become an established business practice. Kits occupy the space for constrained creative activity, operating under a similar logic to parametric customization. Both models demand active input from the user, which is supposed to result in a greater investment by the user in the finished product.

Zwick, Bonsu and Darmody (2008) lay out some tools for thinking critically about this issue. They argue that “the co-creation economy is about experimenting with new possibilities for value creation that are based on the expropriation of free cultural, technological, social, and affective labor of the consumer masses” (p. 166). They somewhat endearingly refer to customers structured this way as “post-Fordist customer workers” (p. 167), increasing the value available for exploitation by the producers, through the addition of their free customer worker labour (very much in line with current arguments about free labour and its extraction of value from users of social media). They suggest that not only does such co-creation offer additional value to producers, but that it might well help to stem the flow of dissent. They “argue that by constructing consumers as partners in mutually beneficial innovation and production processes, companies are not only exploiting consumer labor but are also reducing the risk of consumer behavior evolving in ways other than prescribed by the company” (p.

175 168). This is an interesting concern when placed in the context of the production of a prosthesis. One might argue that the satisfaction resulting from user input in the socket production process is not artificial, as crucial concerns like fit, pressure and weight-bearing are worked out in the dialogue between client and prosthetist. However, as is the case in many medical contexts, the clients I interviewed did suggest that they felt dissatisfied with their experiences when their prosthetists did not listen to their concerns. It would be easy to argue that user contributions matter more in the production of goods—like prosthetic legs—which are not only used on a daily basis, but for which success relies on an accurate fit. Perhaps, in porting the trappings of customization over from high-stakes objects like medical implements, the utility of user input is diminished, replaced by an affective but non-functional preference for a product which incorporates the user's tastes and labour.

In assessing DIY and kit-based projects carried out by users, I turn once again to the four criteria for customness I laid out earlier in this chapter: having a known customer or user for the good being produced; a lack of structural or built-in constraints in the way the user interacts with the producer; flexibility in the range of parameters capable of being modified; and flexibility in the choice of materials available to the user/customer. In all DIY projects in which the person undertaking the work plans to use their own output, the first criterion is quite neatly met. There is a known user, and that user is, in fact, the person carrying out the project. In a kit, the question is slightly more complex. In order to answer the question, we must ask who we consider to be the producer: is the producer the person assembling the kit, or the factory which built or collected the parts? If we deem the marketer of the kit to be the producer, then the kit suddenly ceases to meet the first requirement for customness. With regard to the second criterion, in a DIY project such as the construction of a bookcase or the tatting of a rag rug, there are few structural constraints. Crucially, when the intended user is also the person carrying out the project, there are no constraints which cannot be negotiated. When the user and the producer are the same person, all aspects of the project are negotiable and modifiable. In the case of a kit, there are certainly constraints imposed by the materials and the instructions, but the assembler of the kit has the freedom to choose to ignore those constraints in carrying out the assembly. But the kit falls down once again when we consider whether or not the range of parameters to be modified is constrained. Inside the affordances of

176 the kit, there may be few modifiable parameters. However, that does not prevent the kit's assembler from breaking out of the kit's constraints—in effect, mis-using or appropriating the kit. In a DIY project, once again, the number of parameters to be modified is limited only by the planning and decision-making of the person undertaking the project. This also holds true in the choice of materials. While material choice may be limited by the financial means, skill, or tastes of an individual carrying out a DIY project, none of those are constraints inherent to the project being undertaken and are thus not a barrier to such a project being considered custom. In a kit, the choice of materials is more than likely constrained unless, once again, the assembler of the kit chooses to break out of the authorized process provided by parts and instructions. It becomes clear, from considering DIY projects and kits in the context of my four criteria, that, though a marketer might see the effect of the two as similar (they both involve someone characterized as a user buying materials to build or assemble something, creating exchange-value for the seller of those parts), when considered from the perspective of the options open to the user, they are very different. They are different to such a degree that, if we abandon von Hippel's reliance on the use-value/exchange-value dichotomy, the DIY project comes to look very much like a process of craft production, whereas the kit looks very much like a highly-constrained and prescribed assembly project. This distinction based on user involvement and agency in production and customization, rather than based on the use- value/exchange-value dichotomy leads next to the investment the user-producer feels towards a project in which they have a stake.

User-participation and tolerating idiosyncracies

The doing oneself of prescribed actions in order to build something from a kit has been much explored in recent years under the heading of the “IKEA effect.” The term (per Norton, Mochon and Ariely, 2011) refers to the idea that, when presented with two apparently identical objects, with the only distinction being that one is pre-assembled and the other has to be assembled by its user, the user-assembled one will be perceived as both better and more valuable by the user who assembled it. Norton et al refer to this as the IKEA effect because of what they see as a key characteristic of IKEA furniture, an affinity induced by labour. In positioning their work, Norton et al refer back to the story of adding an egg to boxed cake

177 mix, and the realization by manufacturers that consumers wanted to feel an investment of labour in the production of the cake. Franke, Schreier and Kaiser (2010) tackle a similar phenomenon in mass customization, which they refer to as the “I designed it myself” effect. In both articles, the authors refer to the potential that the goods assembled or designed by users might be objectively worse than those designed by a professional or built industrially. Franke et al refer to the love an amateur painter might have of their own art work, despite the distaste others might have for it. Norton et al suggest that a self-assembled IKEA armoire might be wobblier or less true than it would have been, had it been assembled by a professional. Both of these accounts see the user, participating in an extended production process, as only nebulously competent, apparently committing crimes against good taste or stable furniture through their efforts, but feeling more affection towards their goods because of the effort they've invested. Von Hippel puts it slightly more pleasantly, suggesting that

"Users typically expect a solution they have purchased to work correctly and reliably “right out of the box.” In effect, a sharp line is drawn between product development at the manufacturer’s site and routine, trouble-free usage at the purchaser’s site. When the user builds a product for itself, however, both the development and the use functions are in the same organization and may explicitly be overlapped. Repeated tests and repeated repairs and improvements during early use are then more likely to be understood and tolerated as an acceptable part of the development process." (2005, p. 49)

So, von Hippel is suggesting that users have a higher tolerance for tweaking and modifying when a good or system is something they have produced themself. The amount of effort needed to create that investment and relationship has also been a subject of study. Franke et al (2010) put the problem of how much effort needs to be involved, in addressing what they call “Mass Customization toolkits” (similar to configurators) and specifically, ones that require little skill or effort on the part of the user. They ask “[i]s this limited role on the part of the originator enough to elicit feelings of accomplishment that may translate into enhanced subjective ownership and thus also into an economically relevant effect?” (p. 126). This does seem to be one of the basic questions in mass customization: where is the effort/agency sweet spot that makes the process economically valuable for the company providing it while still being desirable and affectively valuable to the consumer?

178 That effort/agency sweet spot is, I'd contend, an artefact of attempting to achieve a feeling of ownership for the user, despite their use of a mass-produced good. Critical theorists writing on the alienation of consumption suggest that the drive to consume and to acquire goods as an end in itself, rather than in the satisfaction of real needs, dampens individuality and real self-fulfillment (Kellner, 1983). In the arms race between consumers and marketers, the addition of an artificially-induced feeling of fulfillment might well make a product appear less as if it is being bought in the service of simple consumption, and more as if it is making a contribution to the wellbeing and self-worth of the individual purchasing it. However, I would contend that this concern is not at issue in any of my case studies. In the case of the LGRU, there is no purchase activity taking place. No marketer is trying to make LGRU participants feel a false sense of fulfillment through effort. The participants themselves may well feel a sense of ownership over their output, because they have made it and have defined its purpose and direction. In the case of PrintAbility, it is difficult to argue that the input of the user does not have a real, functional value to the final product. The outcome may well be concretely better as a result of user input. Perhaps the early-Industrial custom dressmakers come closest, in that the input of the user is not always in the service of a strictly functional goal. The goal of fitting into a given social setting in an effort at complying with the constraints of class could certainly be thought of as a kind of alienation of the consumer. But there is still an argument to be made for the greater use-value of the custom dress, compared against the current emphasis on exchange-value. Regardless, it is possible that all of these kinds of customization do, in fact, provide the tolerance of idiosyncrasies that von Hippel suggests come with having an investment in the production process. But I would argue that such tolerance does come organically with the self-production of goods. Knowledge of the process through which an object was made provides an insight into why it might behave in peculiar ways. Indeed, as the IKEA effect illustrates, an investment in process, even when it is the process of assembling a kit, produces a greater investment in a good.

While discussions of mass customization focus on things like how to foster user investment in the end product in order to strengthen their feelings towards it (for example: a child selecting a plush heart and inserting it into a Build-a-Bear, making them feel that they

179 have a pleasant secret), the projects I study in this dissertation have a less clear-cut delineation of user and producer, with the role of the user often actually being meaningful in the production process. Von Hippel (2005) might suggest that this is a case of users identifying their own needs, rather than manufacturers trying to identify user needs. He suggests that this is a favourable process, as the two groups don't overlap in a significant way in the traditional model of manufacturing, leaving the producer ill-equipped to fully understand what the user needs.

All of this indicates a problem in the way we make judgement about who is a user and who is a producer, which colours decisions about who is ascribed which tasks, and what kind of benefit is supposed to accrue from carrying out those tasks. Despite positions like von Hippel's, what we are instead seeing is an extreme blurring of the line between producer and consumer, and that the premise that when one purchases something, one is a consumer or user, is fundamentally flawed. Following on from that observation, the next section begins to examine and trouble the apparently arbitrary distinction between user and producer, and advances the idea that there is, instead, a constantly-in-flux relationship between userness and producerness, and that, at some point or other, everyone is both. Or, as I put it below, it's users all the way down.

Users all the way down

As I assert above, the qualities of “userness” and “producerness” are ones which are, to varying degrees, in constant flux. In areas like user innovation, generally quite specialized users (like those who do extreme sports) make modifications to goods, which may eventually be picked up by producers (Hyysalo, 2009). This is possible because the people doing this so- called user innovation don't seek to take out intellectual property protections on their work, and instead simply use their own developments, or share their ideas within their community of practice.

180 Von Hippel (2005) suggests that is is possible for an organization to be both a user and a producer, giving the example of Boeing, which is a manufacturer of airplanes, but a user of machine tools, which are manufactured by another company. Von Hippel's framework for defining the role of the user and the developer, though it is predicated on who stands to make money off of a transaction, at least admits the possibility that a producer can also be a user. Grabher, Ibert and Flohr (2008) offer a taxonomy of user involvement types in what they refer to as “codevelopment.” In service of the taxonomy, they suggest two axes of user involvement and expertise. These axes are “thedegree of involvement” and the “locus of knowledge production” (p. 263). On the involvement dimension, they suggest three basic categories:

“Consultation designates a type of limited and producer-driven interaction in which the customer primarily collaborates in the role of a layperson. Participation entails a deeper, although still primarily producer-driven, form of involvement in which the customer holds the status of an expert. Generation similarly involves expert knowledge that is accumulated by using and modifying the product. Generation, however, denotes a shift from producer- to user-driven development.” (p. 263)

They go on to lay out their taxonomy of differently-skilled users, positioning everyone from users with a vague opinion about the products they use, to Linux kernel developers as users. This distinction once again relies on the idea that the only people who get to be identified as producer are those operating in companies. Presumably, in this logic, the Linux kernel developers are simply yet more lead users, producing an innovation which is then taken on and monetized by large companies, like the electronics manufacturers which ship goods relying on embedded Linux systems.

If Linux kernel developers can be users, and if, according to von Hippel, Boeing can also be a user, I suggest that the boundaries of userness are exceptionally porous. So too, I argue, are the bounds of producerness. If we throw out the arbitrary notion that the only people or organizations qualified to be producers are those which have financial gain in mind in the production of goods, we can come to a far more nuanced understanding of what production is and who does it. If, for example, a prosthetist is both a user and a producer, in that they use materials, methods and tools developed by others, but produce prostheses for

181 their clients, then perhaps we can see the wearers of prostheses as producers as well, inasmuch as they develop, for example, new accessories to aid in the putting on and taking off of their prosthesis, as in the case of one of the Canadian users I interviewed. Von Hippel would refer to such work as lead user innovation, drawing a distinction between the value of the activity when it is carried out by someone whose profit motive is not based on selling the product of their development to someone else. I would call it, at the very least, an example of DIY. A key opening is provided by von Hippel in pointing out that producers are capable of being users. If a user is simply someone who buys materials or tools from someone else, then all manufacturers and producers are users to someone else. In effect, all methods of production are then predicated on the idea of users stacked on users, stacked on yet more users. Or, to put it another way: users all the way down.

If the process of use and production is, indeed, users all the way down, it might be instructive to instead think of userness and producerness as dynamic, rather than static attributes. I demonstrate in the above cases that being a user or a producer is something which takes place contextually and on a moment-to-moment basis. It is not a matter of deciding whether one is a user or a producer, but of identifying when one is engaged in the producer role, and when one is in the user role. If the question is not one of is but of when, then it can be parsed through a combination of my criteria for customness and the three high-level themes which structure this dissertation. The criteria for customness serve to elaborate two interconnected themes: the social organization of labour and the role of the user in production, with the third theme, the material conditions of production, serving in this case as a supplement to the social organization of labour. In chapter four, I suggested that, though I consider the social organization of labour and the role of the user in production to be discrete, they are connected by the fact that the role of the user is defined by the organization of labour —in determining how much of a stake, and of what kind, the user will have in the production process. What the four criteria of customness illuminate is exactly how that stake is defined and decided on. The four criteria offer an insight into what kind of labour, and when, the user gets to contribute to the production of a custom good, thereby taking on the role of a producer or co-producer. Returning to an example I provided in chapter four—Rosaline and her foot—I suggest that the ability of her brother to insist on a more footlike foot was assisted by the

182 iterative process of the Orthopaedic Workshop and by his proximity to the practitioners. In his ability to consult with the technicians building the prosthesis, Rosaline's brother was able to alter the materials used and to exercise a significant degree of agency in determining how the final leg would look. In the case of parametric customization, which does not generally meet my criteria of customness, the moments at which a user may become a producer or intervene in production are severely constrained, and are pre-defined, with no space for negotiation. It is to this end that I now provide two examples of parametric customization systems and consider their limitations with regard to user agency and involvement.

The limitations of parametric customization

I now make use of two additional examples which represent divergent ways of doing customization. The first example, NIKEid, is a system for shoe customization in line with the concept of parametric customization. NIKEid is an appropriate example to work from because it is a mainstream parametric customization platform which directly leverages existing mass production infrastructure. NIKEid is a service which allows customers to choose the colours which will appear in particular areas on a pair of running shoes including various parts of the sole, the uppers, laces, and tongue. This is parametric mass customization at its best and truest because, though Nike allows its customers to dictate some shoe features, and then makes a potentially totally unique shoe which meets those desires, the customer is only in control of certain choices, and those choices are offered and prescribed through a web-based configurator. The parameters to be customized are decided before the fact, by the company, and the scope available for customization is constrained by those parameters. It is not possible to push it further. The parameters exist as they do because they map to an industrial manufacturing process which is feasible as a means of mass customization. Nike gets to choose which colours and materials are available for which portions of the shoes, allowing them to dictate which supplies and processes they will need to employ. Using a computerized loom, it is possible to craft a pair of shoes with a unique colour combination integrated into their uppers. It is always possible to throw in a pair of shoelaces of a different colour, and to put on a different colour of sole. This process epitomizes parametric customization, relying on having a constrained set of variables, and a pre-approved collection of options for each of

183 those variables. When looked at through the lens of my criteria for customness, the process fails on a number of counts: the system is designed so that any user will have the same experience, and though the user has herself in mind while designing her custom shoe, her identity or needs are not taken uniquely into account by the designers of the system—the customness of the shoe relies entirely on the user. Further, the mode of interaction between the client and the producer is entirely structured by the configurator. There is no room for the user to negotiate a different role in the production process. There is no flexibility in the number or nature of parameters which can be modified, with the configurator offering—and indeed requiring a choice on—all of the available parameters. The same holds true for the choice of material, which is, once again, constrained by the production process and its configurator interface.

Yu and Park (2014), among others, see toolkits like NIKEid as co-design, a process which they define as “a specific form of mass customization that was developed as a novel marking strategy” (p. 30). They go on: “The co-design process allows individual consumers to customize a product to meet their needs more precisely” (p. 31), allowing “apparel companies to build better relationships with customers and improve brand loyalty, customer retention, and customer satisfaction” (p. 32). Moreau (2011) also positions this process as a kind of design by the user, and links its positive results to the “I designed it myself” effect mentioned above. Most literature on platforms like NIKEid, coming from the business process or marketing disciplines, focuses on the benefit of the platform to the company deploying it, the comparative ease of doing so, and treats exceptionally uncritically the value it offers to the customers using it. Though several studies, like the one conducted by Yu & Park (2014), aim to determine whether or not users of the system believe they will get value from the system, the idea of value or what value constitutes is not something which is ever particularly problematized. So, though the customer/user is referred to as a co-designer (and, indeed, in Moreau's case, this application of the term “designer” to customers is taken to sometimes laughable extremes, with even the process of choosing ingredients for a sandwich treated as design), it seems like nothing more than a platitude intended to extract labour and affect from a user, in an attempt to strengthen their affinity for a brand and a product. Given the continued success of NIKEid (the range of shoes offered on the platform is growing, with 91

184 different models, ready for parametric customization, offered to women alone), it does appear that Nike's customers, at least, are very ready to buy into a kind of parametric customization that allows them to choose the colours and, in some models, materials of their shoes and to feel that those choices make their shoes unique.

Rosen (2004) offers a counter-example. He details the practices of custom frame builders in high-end cycling who take measurements from their clients and use specialized software in order to aid in the calculation of appropriate geometry for a given use (requiring that there be a known client/user), before producing the frame to those specifications. This, I would classify as bespoke production, with the intended user participating in the process by providing a set of criteria (which potentially include decisions about materials), but not taking part in the production of the frame itself. In the same realm, custom frame builders can be contrasted against the mechanic in the chop shop who takes parts of bikes and puts them back together in a different configuration, with the wheels from this one going onto the frame from that one, and the old racing bike losing its drop bars in favour of straight ones. Unlike the custom frame builder, the chop shop mechanic is acting in a way which adheres to the idea of users all the way down: the only thing he produces himself is the finished bike, with all of the parts, all of the tools produced elsewhere, and at other times. He acts, in almost all ways, like the avid cyclist who collects up a frame, a chain, a wheelset, handlebars, pedals and any other necessary pieces to produce a customized bike from parts. The user-customizer considers the relative merits of each of the pieces (the lightness of the wheel, or perhaps just how cool it looks; the geometry of the frame, or simply its colour) in order to make a final product which they believe is somehow appropriate or desirable (or the best possible for the purpose, given the price). The only major differences between this bike assembler and the one in the chop shop are the volume of work done, and the intent (or not) to sell the final product. In this sense, in an entirely non-digital form of customization, one which has been possible for decades, we see a very robust blurring of the user and the producer. The mechanic is a user, the user-assembler is a producer, and vice versa. However, while the user-assembler is engaging is what I would call a practice of customization, because he knows who his user will be (himself), he is not constrained by a structuring system, he can modify the range of parameters available to be customized, and he has choice in his materials, the chop shop

185 mechanic is arguably not a customizer, as he is in the business of producing bikes for indeterminate users. The user-assembler and Rosen's custom frame builder are in stark contrast to a system like NIKEid, in which there is a clear delineation between user and producer, thanks to the industrial scale on which production happens.

Between NIKEid and the builders of customized bicycles, we see that both apparently aggressively digital ways of involving the user in the production process and very analogue, greasy, wrench-infested ones are viable and prevalent. I will not suggest that the experience of the NIKEid buyer is necessarily less rich than that of the bicycle builder. Though the two are currently qualitatively different experiences, I do not wish to judge as to whether or not one or the other is inherently better, more authentic or more meaningful. Were I a researcher in consumer behaviour, I might feel compelled to see whether the affinity for an object is stronger when it is designed by a user through a computer-mediated interface, or when it is built by hand. Instead, I will bring in a third vignette—from my body of case studies—which occupies something of a middle ground between the two vignettes already provided. While NIKEid allows a user to make choices which are then implemented in an industrial production process, assembling a bicycle oneself has no such intermediation. I would suggest that the work of prosthetists is somewhere in between these two extremes, with the prosthetist themself serving as the intermediary between the user and the fabrication of the prosthesis.

In both the traditional and the PrintAbility model of designing and building a prosthesis, there are a number of moments when both the prosthetist or technician and the patient/client49 can intervene and exercise agency. For the prosthetist or technician, these moments are in the marking-up of the client's residuum, the rectification of the positive (whether plaster or digital), the production of the socket, and the assembly and alignment of the prosthesis. For the client, the moments of intervention are, for the most part, in the fitting stages, when they get to first comment on the fit of the socket, and then on the functioning of the whole prosthesis. As in the case of the user-assembler of a bike, there is the potential for an iterative design process. This is something that differentiates the prosthesis production

49 All of the prosthetists, orthotists, technicians and technologists I have interacted with have referred to the people they are building devices for as “clients.” To me, this feels slightly awkward, although I do understand and empathize with the idea of not wanting to pathologize someone who will always need to wear a medical device.

186 process from so-called co-design processes like NIKEid, which has no scope for iteration. Because the user and the producer are invested in a relationship which involves the two being in the same place, working with the same materials, and having discussions about the comfort and utility of the product, they are capable of iterating together, even if there are sometimes conflicting interests at work. In the example of the DIY bike assembler, this iteration process is collapsed, as the producer and the user are the same person. Iteration, in such a case, does not require discussion. Iteration in a system like NIKEid is rendered difficult by the industrial mode of production. The user might well do some iteration of her own, coming back to the configurator multiple times, considering her decisions before sending off her order, but that is the extent to which iteration can take place. In a milieu like the LGRU, on the other hand, iteration looks far more like it does for the bike assembler: because the developers are also users (and perhaps because software is comparatively easy to modify), there is the potential for more iteration in the production process. Even in the case of a dressmaker, the process of having multiple fittings provides room for change along the way. Once the dress is complete, there is still the chance for later modification, as its fabric might be used to form the basis of another dress later on, or a hem might be taken up to fit a shorter relative receiving a hand- me-down. The inability to iterate meaningfully appears to be one of the problems in parametric mass customization, and the ways in which such processes attempt to achieve customization while still being designed for the use of the notional mass customer.

Users in the production of custom goods

In studying the Libre Graphics Research Unit, the PrintAbility project, and pre/early- Industrial dressmakers, I have addressed a number of different models for the user-developer relationship and have addressed the ambiguity present in the word “custom.” Comparing the parametric customization relationship embodied by NIKEid against the relationship between the craftswoman dressmaker and affluent client, we see a stark difference in both power dynamics (for example, the ability to not pay one's dressmaker, versus the requirement to pay on ordering a pair of shoes customized on the NIKEid platform) and in the areas available for intervention on the part of the user (like the ability of a prosthesis-wearer to comment on the fit of an in-production socket). The advent of mass-customization has often been described as

187 having the qualitative traits of custom production, combined with the economies of scale of mass production (in, eg. Marsh [2012]). It has been my contention in this chapter that such an assertion does not hold true. Rather, I suggest that there are qualitative differences between the kind of customization which takes place one-on-one (or even for oneself) and the kind of customization which relies on standardized interfaces, parametric customization options, and fixed industrial apparatus50 and processes. Making clear distinctions between the kind of customization represented by my case studies and the practices encompassed by the terms “mass customization” or “parametric customization” is increasingly important, as parametric customization systems are offered up as solutions to problems which they are ill-suited to solve.

I further argue that, at times, it is difficult to clearly see the boundaries between users and developers, and the differing roles played by the two, without taking into account both the material conditions of production and the social organization of labour. Indeed, as I suggest in chapter four and elaborate on in this chapter, one of the key aspects of the social organization of labour is the role the user is asked to play in the production of their custom good. The role of the user could be framed in the context of the free labour debate currently raging. I suggest that the user is not just contributing labour for the benefit of the producer, but is gaining both a good and an affectual connection to the customized object. In some cases, like NIKEid, the difference between custom and parametric production might be quite clear, with a home shoemaker (were such a thing commonplace) quite different from the NIKEid system. In the comparison of chop shop mechanics against individual user-assemblers of bikes, the difference is less obvious. But I should note that the difference is not simply something that exists along analog/digital lines. Rather, a user-assembler with some ambition might choose to use the frame-design software mentioned by Rosen in order to generate a frame with an optimal geometry for a particular kind of cycling activity. The issue here is that, in many conceptions of the difference between a user and a developer, the line is one drawn solely with regard to revenue. Instead of such an arbitrary and market-focused distinction, I suggest in this chapter that, in spaces where the work of production is being taken on by both the traditional producer and the traditional user (as delineated by who hopes to gain use-value and who hopes to gain exchange-value), we should try to fully-admit the efforts and contributions of

50 Or, to invoke Marx, dead labour.

188 the user-developer, rather than isolate them in something like the ghetto of von Hippel's lead users, recognizing that production and use cannot be so clearly demarcated and are instead delineated by time and context.

In the following and final chapter, I summarize the work and findings of this dissertation. First, I revisit my case studies in light of the social organization of labour/material conditions of production/role of the user in production framework my analysis of them has spawned. Further, I outline how the research questions posed at the beginning of the dissertation have been addressed and how they have changed. In particular, I highlight the importance of troubling the idea of both the digital and the custom and reiterate one of the key points from this chapter, namely that the current mode of parametric mass customization is qualitatively different from the more individualized forms of customization by which it is influenced.

189 8 Conclusion

Introduction

On a cold, snowy Thursday morning at the end of February, I'm in my kitchen, biking on a trainer while listening to the radio. I listen to the radio because there is nothing so tedious as doing an endurance bike ride without actually going anywhere. The radio helps a bit. The business columnist for the morning show comes on. He's talking about McDonald's and its apparently falling fortunes, represented by slowed expansion and shrinking year-over- year same-store sales. After going through a number of theories, he gets to what's clearly the big one: consumers want something else from their fast food. I fully expect him to say that they want it healthier or more exciting, or less burger, or any number of things. Instead, he makes the case that what consumers really want is the chance to—wait for it—customize. He brings up the example of Subway, a sandwich shop that's been gaining huge ground in recent years, he argues, because they allow people to customize their sandwiches. He even has a little mantra, “customization, not standardization.” McDonald's is failing because their value proposition has always been based on consistency across both time and space: the same burger every time, whether you're in Toronto or Tasmania. Subway and its kin are succeeding because they give customers something else, something apparently antithetical to that standardization. I laugh out loud when I hear this. To me, in the context of this dissertation, Subway is just as standard as McDonald's, using standardized procedures, supply chains, ways of controlling staff, even ways of cutting a bread roll in half. It is, in my terms, parametric customization, it's an extension of product range, but with very little actual meaningful engagement or agency on the part of the customer. Framed by the three high-level themes that

190 have emerged from my case studies, it becomes apparent that what Subway does is a kind of surface-level customization. Though the person purchasing the sandwich might well get something more in line with their tastes, the underlying system through which it is produced does not change, and does not allow the purchaser to bend rules or venture outside of the prescribed form of interaction. The material conditions of production and the social organization of labour are much the same as they would be if there was no choice in the sandwich: the Subway employees are the same fast food workers in the same kind of working environment, carrying out the same kinds of tasks as they would at a McDonald's. The role of the customer is to choose from a set of pre-defined variables. This is one of the major points and contributions of this dissertation: to understand the stakes involved in customization activities, and to make judgements about customization, we need to take into account the intertwined factors represented by the material conditions of production, the social organization of labour, and the role of the consumer (or user) in production. In considering those three issues, there is the opportunity to generate insight about where the user may intervene, how, and to what extent.

Revisiting the case studies

My contention that the social organization of labour, the material conditions of production, and the role of the user in production need to be thought of as interconnected comes from my case studies. Those case studies—the historical study of custom dressmakers, the analysis of documents from the Libre Graphics Research Unit, and my participation in the PrintAbility project—provided me with three different examples of custom production work, all using apparently different methods, and with apparently radically different outputs. It is from finding similarities between the three cases that I derive my understanding that three factors which are often treated in isolation from each other should be taken together.

The custom dressmakers of the 18th to early-20th centuries provided me with a historical grounding from which to start. Gaining a strong understanding of comparatively recent and sophisticated practices of craft-style customization is, an essential step in building further arguments about present and future customization activities. Because the historical

191 perspective is essential scaffolding for my further arguments and for the contextualization of my present-day case studies, I embarked on a literature review covering existing texts which use historical sources to tell the story of custom dressmakers from the 18th to early 20th centuries. The monographs and articles I consulted cover France, England, Scotland and the eastern United States and focused on the activities of custom dressmakers serving women of a variety of different classes. Learning about the ways dressmakers organized their labour, dealt with clients (and with the challenges presented by their bodies), and operated their workshops provided me with a concrete example of craft-style customization which persisted into the early-Industrial period. That concrete example in turn allowed me to build a framework for understanding and analyzing other customization activities, which is concerned with the existence of a known client or user, the constraints and constraining systems placed on the client's interaction with the production process or producer, the flexibility available in the range of parameters being modified, and the ability for the client to contribute to decisions about the materials being integrated into the custom good. All of these traits are evident in the example of the custom dressmakers, and in their more current counterpoints. In addition to sources addressing the working methods and conditions of custom dressmakers, I also looked at sources which cover the introduction of sweatshops, graded sizing, and fractionalized labour to the clothing industry. It is through the comparison of those new-at-the-time methods to the older craft methods that I begin to compare custom work to mass production. That comparison, combined with my understanding of the craft of the custom dressmakers, forms the basis for my later analysis of projects which are situated in the current mode of understanding customization through its relationship with the methods and systems of mass production.

Within the framework provided by the dressmakers, I then looked at the Libre Graphcis Research Unit. The LGRU, which was active from 2011 to 2013, made use of methods, tools, and formats common in Free/Libre and Open Source Software development to engage designers, organizers, and software developers in a co-design process, with the goal of producing new kinds of software for graphics work. By appearances, the LGRU presented a very different case from the dressmakers. Participants in the LGRU were, for the most part, geographically distributed, travelling to the various locations in which LGRU events were held

192 to meet up and work with each other. The good they were producing was an intangible one: software. They made use of the tools and methods of software development—and some of the participants do indeed engage, at other times, in the development of other F/LOSS graphics tools. These are the superficial differences between the LGRU and the dressmakers. More interesting were the similarities between the two cases. Both the dressmakers and the LGRU incorporate users into the production process, although with varying levels of involvement. The dressmaker takes guidance on style from a woman commissioning a dress, while the LGRU makes artists and designers equal participants in the software development process. Similarly, both the dressmakers and the LGRU show evidence of divisions of labour organized around differences in skill, though the for dressmakers, such differences also informed a clear hierarchy, which is not the case in the LGRU. What I find most exciting about these similarities (among others) is that two working environments, separated by over a century, a huge rift in working conditions, and by a significant difference in the kinds of goods they produce, can still show organizational similarities, assumptions about skill, and attitudes about user involvement. I find the comparison instructive because it shows a fairly clear parallel between two different groups of skilled workers engaged in custom production processes.

Having compared a pre-digital process (dressmaking) to an at-first-glance all digital process (the LGRU), my final case study is situated at the intersection of many of the issues presented by the two previous cases. The PrintAbility project employs 3D modelling software to turn 3D scans of residual limbs into 3D-printed prosthetic sockets, leveraging the existing skills of traditionally-trained prosthetists and prosthetic technicians. The project, over the course of the time I spent working on and studying it (between June 2013 and January 2015), encompassed activities ranging from software development and user testing to the production of prostheses for children in Uganda. It took into account some issues present in the case of the dressmakers, like the use of craft methods and the need to account for the particularities of human bodies. It also concerned issues present in the LGRU, like organizational structures which cross institutional boundaries, and the inclusion of user-collaborators in the development of software. Having already confirmed through the dressmakers and LGRU cases that practices of customization can be effectively studied through their social organization of labour, their material conditions of production, and their inclusion of the user in the

193 production process, I felt confident in applying those lenses to PrintAbility. In so doing, I found that they were not only applicable, but found evidence of their necessary interconnectedness.

The three case studies, all chosen because they offer a window onto a practice of custom production, ended up providing me with a framework through which to study such practices. The development of such a framework was not the initial goal of my dissertation. Rather, I started with a set of research questions intended to trouble the ability of digitally- aided customization to help individuals relate to standardized systems and institutions. Below, I describe how my cases and the framework arising from them have answered those initial questions.

Revisiting the research questions

At the beginning of this dissertation, I posed one major research question, followed by four narrower, related ones. The major question was: How does the increasing popularity and viability of digital methods impact the production (and by necessity, the consumption) of physical goods, as well as the interfaces we build between our idiosyncratic selves and our standardized production systems and institutions? The four minor questions, intended as ways of further exposing to enquiry the major question, and of highlighting some key concerns embodied in it were:

• What kinds of constraints do digital methods impose on the production of physical goods? • What ideas and ideals from digital production methods are kept in the application of those methods to physical goods? • How formulaic (or not) are our engagements with digitally-produced physical goods? Are we developing deeper, more engaged relationships? • How do we negotiate the relationships between our bodies and the standardized world? Are digital production methods changing these relationships?

194 In developing insights into these questions, through my case studies, both current and historical, I found three major themes at work: the social organization of labour; the material conditions of production; and the relationship between the user and the producer. Building on from these three themes, I sought out existing work which provides ways of understanding their implications. Because my three themes are not, individually, particularly unique, there is a plethora of relevant literature, from areas as diverse as the political economy of labour to feminist and postcolonial science and technology studies to literature from marketing and consumer behaviour. In traversing this literature through the lens of my three themes, I aimed to both trouble and find connections to it. In so doing, my aim has been to highlight the crucial importance of understanding both the issues of agency present in custom work, and in the ways user needs and wishes are enacted, both through their direct labour, and through their role as a directing force in the outcome of a custom product. In the following paragraphs, I elaborate further on how these themes address my initial research questions by summarizing the findings detailed in the rest of this dissertation, structured by the original questions.

What kinds of constraints do digital methods impose on the production of physical goods? In addressing the sub-questions posed in my introduction, I have found that some have clearer answers than others, and that some turn out to have been the wrong question to pose in the first place. For one, within the cadre of the three themes, I have found that there is perhaps less differentiation than I initially thought between the production of so-called digital goods and the production of physical goods. I suspect that, as the line between a physical and a digital good blurs even more, the differences will diminish further. A key similarity is the impact of the material conditions of production. Though the software developer may work in a very different environment than the 19th century dressmaker, so-too does the 21st century garment worker. In terms of working conditions and the ability to self-determine, the garment worker has much in common with the Mechanical Turker (the highly fragmented nature of her labour, with large projects broken down into tiny steps to be carried out by an army of workers) or the gold farmer (the environment in which they work, with row upon row of gold farmers or garment workers seated at their own stations for the duration of their work day). By contrast, as we see in the two case study chapters of this dissertation, the participants in the LGRU carry out many of the same micro-level tasks as the CoRSU employees in the

195 Orthopaedic Workshop: they manage and negotiate the allocation of their workspace, they collectively triage tasks, and they seek out input from those in the group who have particular expertise, when such input is needed. Though one group is producing software and the other is producing prostheses, their similarity lies in the way they structure their work and resources, a similarity which goes beyond the facile (and increasingly imaginary) boundary between the physical and the digital. The similarity between practices which exist in different media embodies one of the key points of this dissertation: the digital and the physical are, with regard to the practices surrounding their production, not discrete, fundamentally different things. Instead, it is imperative to consider the conditions under which a given good —whether a piece of software or a prosthetic leg—is produced.

In considering the changing role of the user in production, my key finding has been that this blurring is highly context-dependent, but does not appear to be dependent on any distinction between the analog and the digital. This context-dependency includes the need to consider the organizational and social commitments of the people involved in the user- producer relationship. Whether that's the difference between a patient-clinician relationship and an artist-software developer relationship, or simply between a relationship with a greater expected degree of equality and a relationship where there is a significant discrepancy in technical or procedural knowledge, the skills, hierarchies, and expectations involved on both sides must be accounted for in considering how a user and a producer go about interacting with each other. This is one of the problems in current popular approaches to mass customization: they take their understanding of the user more from the mass perspective than the custom. Services which offer highly parameterized customization use the same conceit as mass production: all users are interchangeable. In the three cases presented in this dissertation, the uniqueness of the user is taken as a given. Though working conditions and social hierarchies might be the same or similar from client to client, there is an expectation that some change will be brought to the process by the individual needs of the particular client. In the case of the dressmaker, even some of the material conditions of labour might change from client to client, if the dressmaker is willing to do house calls. In the case of the LGRU, though some structures (like the roles of the four major partners) might stay the same, smaller structures are organized ad hoc, as needs and specifications are solidified. In the

196 development of the PrintAbility process, we see an even more extreme example, where, in the case of Socketmixer, one-to-one observation of users resulted in changes, such as the removal of a button or the re-structuring of a step to provide more room for practitioners to exercise agency and expertise. Equally crucial has been the realization that the role of user and producer is ever-changing and highly reciprocal. In the PrintAbility case, the software developer and I were users of Meshmixer, but developers of Socketmixer. We were users of other software, like git, web browsers, operating systems and so on. We developed a piece of software for use by yet another set of user-producers, the technicians at CoRSU, who used the software, but were producers of prostheses. As I argue in chapter seven, one of the interesting facets of the blurring role of the user in production is how it assists in noticing that it is, in fact, users all the way down.

The relationship between the user and the producer is also inextricably tied to the social organization of labour. In part, this is because the user is complicit in doing some of the labour of customization. Though it is certainly not unusual for users to be engaged in value creation for the benefit of the companies whose platforms they are using (as in the free labour argument), what is typically positioned as less usual is for the user to be engaged in work that has a direct personal benefit. Where a traditional free labour or attention for entertainment bargain would involve a user or viewer gaining access to entertainment or a communication platform in exchange for giving the platform's owners the ability to sell the user's attention to advertisers (or in the social media model, monetize the content they produce on the platform, or sell their personal information), user customization of tangible goods brings the labour of the user into design or production, ultimately resulting in the user taking ownership of the good they have contributed to, different from the ownership model in social media. This applies especially in parametric customization activities, with platforms designed to make the experience of customization the same for each user, leveraging a standard production process, but with an output that the user feels is actually made for them and which does, pragmatically and legally, ultimately belong to them. This mode of parametricization seems to be dominating current discussions and implementations of mass customization, possibly because it is perceived to offer the benefits of customization (or the appearance of those benefits) while leveraging production infrastructure and fulfillment processes that stay consistent.

197 In situations where the profit motive is somewhat less clear, the line blurs further, with the structure of participation varying to a greater degree. In the LGRU, participants might have different kinds of expertise which they actively bring to the process, and which is acknowledged by other participants. By contrast, at CoRSU, there is some assumption of specialized knowledge possessed by the workers in the Orthopaedic Workshop, which is not possessed by the patients. The patients have their own embodied experience of wearing a prosthesis, and can make suggestions based on that experience, but their knowledge does not generally stretch to the process of production. The LGRU, the PrintAbility project, and the pre/early-Industrial dressmakers each have a different mode of organizing their labour. Their assumptions about the organization of labour—and about the optimal relationship between user and producer—colour the objects they produce, the way their material surroundings are organized, and the amount and quality of user input that they admit. In short, the structuring and division of labour has a bearing on the role of the user in production, as well as a reciprocal bearing on the material conditions of production.

In considering the constraints put on the production of physical goods by the use of digital methods, I have noted that such constraints vary based on the task at hand. In, for example, the case of prosthetists working with a digital rectification system, the embodied actions performed while doing a manual rectification clearly show through in early engagements with a screen-based rectification process. Prosthetists take a skeumorphic view of the software, attempting to map their process directly onto a 3D model, while concurrently struggling with the admittedly difficult task of learning to navigate in three-dimensional computational space. Beyond this, however, what has been interesting to note has been that the constraints have moved in the opposite direction. In defining a process for the 3D-printing of prosthetic sockets, we have had far more cause to think about the constraints imposed by the physical portion of the process than the digital portion. In aspects like the need to design sockets which can be printed within the print-height possible on a given machine, or in thinking about the arrangement of support material, we have needed to constrain our digital artefacts within the capabilities of the 3D printers currently available.

198 What ideas and ideals from digital production methods are kept in the application of those methods to physical goods? The issue of the transfer of ideals from digital to physical, one which I will admit springs from a question originally raised (but never answered) in my Masters thesis, is one of the questions which may be wrongly put or, at the very least, somewhat less applicable than other questions. An interesting facet of the cases studied in this dissertation has been that, to put it somewhat colloquially, they've broadly stayed in their boxes. The LGRU, as a software development project, despite its housing in a cultural milieu, has strongly adhered to many of the principles of software development. The PrintAbility project has made great efforts to concern itself with material and institutional issues. This is not to say that processes of software development have not been adopted. Some have, such as git for version control, and such as permissive licensing. But what is interesting about this adoption has been that it is exclusively in the software development portion of the project. At present, unlike in areas such as the Open Hardware movement, the PrintAbility project has not had any clear examples of the transfer of F/LOSS ideals into outputs other than software. Interestingly, I did expect to see the opposite, beginning this dissertation under the assumption that there should be some kind of clear translation of the ideals of F/LOSS into digital fabrication projects, especially those which have a more social and less profit-driven aim.

PrintAbility lacked a clear translation of F/LOSS values into the more material realm broadly because it was actually a process design project, rather than a product design project. In something like an Arduino, it makes perfect sense for schematics of the board to be available, as the board is the platform. In PrintAbility, the actual value of the project is in the process which produces the custom socket. Because the socket is custom and, what's more, is quite sensitive personal information, there is no good reason to distribute its files, and many good reasons to not do so. In this sense, it's not that the question of transfer of values was wrong, but that it did not map well to the projects studied, given that one was situated in a relatively traditional software development process and the other, though more material, did not produce any widely-distributable non-software objects.

199 How formulaic (or not) are our engagements with digitally-produced physical goods? Are we developing deeper, more engaged relationships? I find now that it is misleading to ask how formulaic our relationships with digitally- produced goods are. Digitally-produced goods is a category which could arguably encompass most production today, on some level, given the prevalence of CAD software for the design of stamps and dies, the high uptake of CAM systems, and the computer-control of even apparently non-technical tasks like the production of paper (as in Zuboff, 1988). Digitally- produced is a false category. The metric which matters, and which influences the role of the user in production is the agency they are capable of investing in it, and the amount and kind of input they are allowed to have in construction. The user of the prosthesis should not necessarily need to care that their socket has been designed digitally, but can instead take the speed with which design and testing are made possible by rapid prototyping to get an improved fit, or to have the growth or change of their residuum charted. A digitally-produced custom good does not need to be any more formulaic than the custom dress of the 18th or 19th century. However, there is certainly a danger that the current trend towards parametric customization might indeed produce goods which do not have built into them the same kind of investment or concern. It is not a matter of digital production resulting in a particular kind of relationship between user and good, but rather a question of how many opportunities for user agency, and of what kind, are built into the process of designing and producing the product.

How do we negotiate the relationships between our bodies and the standardized world? Are digital production methods changing these relationships? In questioning how we negotiate the relationships between our bodies and the standardized world, and how digital production methods change these relationships (if they do), I have found the example of the dressmakers and the move to mass production instructive. From Foucault to Bahnish, we see the argument that modernization is about the governing of bodies and activities and their placement in standardized routines. In this framework, the dressmakers represent a kind of pre-modern bastion of attention to the body, contrasted against an attempt to fit the body into a standard. This is not to say that the dressmakers were flouting social norms, but rather that they made use of their craft to

200 emphasize the desirable traits of the particular bodies of their clients. This is the kind of work that a custom process of production for the body can do. As I emphasize in chapter seven, I see little evidence that the introduction of digital fabrication tools to flexible mass production actually creates a fundamental change in the way industrial mass production attends to the needs or exigencies of individuals. So perhaps, the question is not how digital production methods change our negotiation of the relationship between bodies and standards, but how the considerations behind those methods and their implementation might open space for more acts of meaningful individualization. Although I suspect that such ventures will not, for the time being, be coming from the large companies currently selling parametric customization.

On the other hand, referencing Hayles as I do in chapter six, there is the question of body versus embodiment. This is one area where we might see a far greater difference between older methods and newer ones. Were we to develop, say, a NIKEid equivalent for custom bicycles, it would certainly be possible to argue that the embodied experience would vary greatly from the experience of sourcing the parts and assembling them oneself. The grease, the wrenches, the bike shops, the torque, all the parts of the idealized process of production, in the hands of the user-assembler I discussed in chapter seven. Similarly, the difference between the embodied action of measuring oneself and inputting those measurements into a parametric system for custom garments is a world away from the experience of standing in a dressmaker's shop, being measured by her practiced hand. Here, then, is the importance of noting the difference between negotiating bodies and negotiating embodiment. The current importance of screen-based interfaces in parametric customization seems antithetical to the incorporation of embodiment in the production of parametrically- customized goods. Perhaps such screen-based interfaces for parametric customization do still offer an improvement over the entirely alienated activity of buying a pre-made good, but they currently offer few opportunities for the exercise of user agency, or for dialogue and iteration between user and producer.

201 Conclusion

To understand the stakes involved in customization activities, and to make judgements about customization, we need to take into account the intertwined factors represented by the material conditions of production, the social organization of labour, and the role of the user in production. Similar to Star's “call to study boring things” (1999, p. 377), the exhortation embodied in my dissertation is that it is necessary to look at not just the product and its consumption, but to consider the broader context in which it is produced, the roles played by users and producers, and even the circumstances under which different stakeholders take on the role of user or producer. Not only that, but considering the three factors as a whole also offers an opportunity to illuminate the practices in and similarities between mediums which are often thought to be at odds with each other. In short, rather than answering the question “How does the increasing popularity and viability of digital methods impact the production (and by necessity, the consumption) of physical goods, as well as the interfaces we build between our idiosyncratic selves and our standardized production systems and institutions?” I have instead come to a more methodological conclusion. Rather than detailing a list of ways digital production technologies, in the service of customization, change the relationship between non-standard people and standard goods, I have instead found that we first need a way to trouble the question of the custom and the digital more effectively.

Another contention I have advanced in this dissertation is that parametric customization is not an analog for what we has previously been called customization, or custom production in the spirit of older, craft-based methods. This contention is made visible through the combined analysis of production, modes of user involvement, and conditions of workers executing the production. I suggest that customization is best done—and is most effective as an intervention—when production processes are flexible enough to admit meaningful and substantial input from the user. Though all systems circumscribe user choice in some way, a system which entirely dictates what choices a user may make and enforces a range of parameters for those choices is not a system for meaningful customization. If we want to trouble or surmount the violence of categories and our alienation from the goods we use, it will take serious consideration of how users can be involved in the design of goods, as well as a commitment to dialogue between users and producers. The current trend in

202 customization at a large scale (whether digitally-aided or not) is towards highly constrained, paramatric customization, which often admits little input from the user.

Crucially, this dissertation shows that a number of sensitivities are necessary in designing systems for customization and user-production. We must consider the milieux in which systems are to be implemented, the skills and structures which will be used and occupied by individuals implicated in such systems, and how structural, social and material constraints can curtail or empower the involvement of the user. Equally important is the recognition that it is not a question of whether someone is a user or a producer, but when they occupy the role. In line with that blurring of a binary is another such blurring: that the digital and the physical are, in the practice of production, far less binary than they are often expected to be. These considerations are essential in thinking about how practices of digital customization map to historical practices, and how craft methods can be aided by the advent of digital fabrication technologies, if an effort is made to recognize existing practices and to maintain those considered relevant or important by practitioners and users. I hope that some of the work done in this dissertation will highlight the limitations of that approach, and assist in further enquiries into the nature of custom goods and their digitally-aided production, providing a lever to help open up this space for more critical study.

203 References

Aichner, T., & Coletti, P. (2013). Customers' online shopping preferences in mass customization. Journal of Direct, Data and Digital Marketing Practice, 15(1), 20-35. doi: http://dx.doi.org/10.1057/dddmp.2013.34 Andrysek, J. (2010). Lower-limb prosthetic technologies in the developing world: A review of literature from 1994-2010. Prosthetics and Orthotics International, 34, 4, 378-398. Bahnisch, M. (2000). Embodied work, divided labour: Subjectivity and the scientific management of the body in Frederick W. Taylor's 1907 'Lecture on Management'. Body & Society, 6, 1, 51-68. Barbrook, R. (June 17, 1997). The digital economy. Retrieved from http://www.nettime.org/Lists-Archives/nettime-l-9706/msg00143.html Barbrook, R. & Cameron, A. (1996). The Californian ideology. Science as Culture, 6(1), 44-72. Bardzell, J., & Bardzell, S. (2013). What is critical about critical design? CHI 2013 (Paris, France). Beaton, E. (November 22, 2011). Give the people what they want. PROFIT. Retrieved from http://www.profitguide.com/opportunity/give-the-people-what-they-want-30260 Bergvall-Kåreborn, B., & Howcroft, D. (2014). Amazon Mechanical Turk and the commodification of labour. New Technology, Work and Employment, 29, 3, 213-223. Blanchette, J.F. (2011). A material history of bits.Journal of the American Society for Information Science and Technology, 62(6), 1042-1057. Borges, J. L., & Hurley, A. (1998). Collected fictions. New York, N.Y., U.S.A: Viking. Braverman, H. (1974). Labor and monopoly capital: The degradation of work in the twentieth century. New York: Monthly Review Press. Brumberg, J. J. (1988). Fasting girls: The emergence of anorexia nervosa as a modern disease. Cambridge, Mass: Harvard University Press. Bryant, A. (2002). Re-grounding grounded theory. Journal of Information Technology Theory and Application, (4)1, 25-42. Busch, L. (2011). Standards: Recipes for reality. Cambridge, Mass: MIT Press. Castaldo, J. (April 21, 2014). Mass customization: The future of manufacturing.Profit. Retrieved from http://www.profitguide.com/industry-focus/manufacturing/why-mass- customization-is-the-future-of-manufacturing-64553 Castells, M. (1999). Grassrooting the space of flows. Urban Geography, 20, 4, 294- 302. Chalmers, J. (December 10, 2013). 3D printing: Not yet a new industrial revolution, but its impact will be huge. theguardian.com. Retrieved from http://www.theguardian.com/commentisfree/2013/dec/11/3d-printing-not-yet-a-new- industrial-revolution-but-its-impact-will-be-huge Charmaz, K. (2009). Constructing grounded theory: A practical guide through qualitative analysis. Los Angeles: SAGE. Comte, A. (1907). Discours sur l'ensemble du positivisme. Paris: Société positiviste internationale. coons, g. (2014). The replacement of geography by standards and what to do about it. JCITAR Journal of Science and Technology of the Arts, 6(1). doi: http://dx.doi.org/10.7559/citarj.v6i1.126

204 coons, g., & Ratto, M. (2015). Grease pencils and the persistence of individuality in computationally-produced custom object. Design Studies, (forthcoming). Corbin, J., & Strauss, A. (2008). Basics of Qualitative Research (3rd ed.): Techniques and Procedures for Developing Grounded Theory. Thousand Oaks, CA: SAGE Publications, Inc. doi: http://dx.doi.org.myaccess.library.utoronto.ca/10.4135/9781452230153 Cowan, R. S. (1983). More work for mother: The ironies of household technology from the open hearth to the microwave. New York: Basic Books. Cowen, D. (2014). The deadly life of logistics: Mapping the violence of global trade. Minneapolis: University of Minnesota Press. Craig, B. (2001). Petites bourgeoises and penny capitalists: Women in retail in the lille area during the nineteenth century. Enterprise and Society: The International Journal of Business History, 2(2), 198-224. doi:10.1093/es/2.2.198 Crawford, M. B. (2009). Shop class as soulcraft: An inquiry into the value of work. New York: Penguin Press. Crowston, C. H. (2001). Fabricating women: The seamstresses of Old Regime France, 1675-1791. Durham, NC: Duke University Press. D’Mello, M., & Sahay, S. (January 01, 2007). “I am kind of a nomad where I have to go places and places”... Understanding mobility, place and identity in global software work from India. Information and Organization, 17, 3, 162-192. Dahl, D. W., & Moreau, C. P. (2007). Thinking Inside the Box: Why Consumers Enjoy Constrained Creative Experiences. Journal of Marketing Research, 44, 3, 357-369. Department of Justice (2014). Cyberbullying and the non-consensual distribution of intimate images. Retrieved from http://www.justice.gc.ca/eng/rp-pr/other-autre/cndii-cdncii/ Dibbell, J. (June 17, 2007). The life of the Chinese gold farmer. The New York Times Magazine. Dourish, P. and Mazmanian, M. (2011). Media as Material: Information Representations as Material Foundations for Organizational Practice. Third International Symposium on Process Organization Studies (Corfu, Greece). Dyer-Witheford, N. (1999). Cyber-Marx: Cycles and circuits of struggle in high-technology capitalism. Urbana: University of Illinois Press. Eglash, R. (2004). Appropriating technology: Vernacular science and social power. Minneapolis: University of Minnesota Press. Ensmenger, N. (2010). The computer boys take over: Computers, programmers, and the politics of technical expertise. Cambridge, Mass: MIT Press. Escobar, A. (2011). Development and the anthropology of modernity. In S.G. Harding (Ed.), The postcolonial science and technology studies reader (269-289). Durham: Duke University Press. Fernandez, N. P. (1994). Innovations for home dressmaking and the popularization of stylish dress. Journal of American Culture, 17(3), 23. Retrieved from http://search.proquest.com/docview/200616466?accountid=14771 Foucault, M. (1978). The history of sexuality: An introduction. New York: Pantheon. Fouché, T. (2011). From black inventors to OLPC. In P. Chow-White and L. Nakamura (Eds), Race after the internet (61 - 84). New York: Routledge. Foster, J. B., McChesney, R. W., & Jonna, R. J. (2011). The global reserve army of labor and the new imperialism. Monthly Review, 63(6), 1-31. Franke, N., Kaiser, U., & Schreier, M. (January 01, 2010). The "I designed it myself" effect in mass customization. Management Science, 56, 1, 125-140. Gamber, W. (1992). A precarious independence: Milliners and dressmakers in Boston, 1860-

205 1890. Journal of Women's history, 1(4). Gamber, W. (1997). The female economy: The millinery and dressmaking trades, 1860-1930. Urbana: University of Illinois Press. Gandy, O. H. (1993). The panoptic sort: A political economy of personal information. Boulder, Colo.: Westview. Ginsburg, M. (January 01, 1972). The tailoring and dressmaking trades, 1700-1850.Costume, 6. Glaser, B. G. (1978). Theoretical sensitivity: Advances in the methodology of grounded theory. Mill Valley, Calif: Sociology Press. Gonzales, L. (January 16, 2014). PSFK's trending topic: Bridging the physical-digital divide. Retrieved from http://www.psfk.com/2014/01/digital-physical-merging-worlds- trends.html Gordon, S. A. (June 15, 2004). "Boundless Possibilities": Home Sewing and the Meanings of Women's Domestic Work in the United States, 1890-1930. Journal of Women's History, 16, 2, 68-91. Grabher, G., Ibert, O., & Flohr, S. (2008). The neglected king: The customer in the new knowledge ecology of innovation. Economic Geography, 84, 3, 253-280. Green, N. L. (April 01, 1994). Art and Industry: The Language of Modernization in the Production of Fashion. French Historical Studies, 18, 3, 722-748. Hapke, L. (2001). A shop is not a home: Nineteenth‐century American sweatshop discourse. American Nineteenth Century History, 2, 3, 47-66. Harding, S. G. (2011). The postcolonial science and technology studies reader. Durham: Duke University Press. Harvey, D. (1990). The condition of postmodernity. Oxford: Blackwell. Haydu, J. (1988). Between craft and class: Skilled workers and factory politics in the United States and Britain, 1890-1922. Berkeley: University of California Press. Hayles, K. (1999). How we became posthuman: Virtual bodies in cybernetics, literature, and informatics. Chicago, Ill: University of Chicago Press. Hayles, N. (2006). Unfinished work. Theory, Culture & Society, 23(7-8), 159-166. doi:10.1177/0263276406069229 Hetrick, W. P., & Boje, D. M. (January 01, 1992). Organization and the body: Post-Fordist dimensions. Journal of Organizational Change Management, 5, 1, 48-57. Honeyman, K. (2007). Doing business with gender: Service industries and british business history. Business History Review, 81(3), 471-493,419. Retrieved from http://search.proquest.com/docview/274372478?accountid=14771 Huyghebaert, P., de Heij, G., Snelting, F., Marchand, P. & O'Neill, C. (2013). 16 case stories. Retrieved from http://constantvzw.org/verlag/spip.php?article139# Hyysalo, S. (2009). User innovation and everyday practices: micro-innovation in sports industry development. R & D Management, 39, 3, 247-258. Joerges, B. (June 01, 1999). Do Politics Have Artefacts?. Social Studies of Science, 29,3, 411-431. Kalil, T, & Miller, J. (February 3, 2014). Announcing the first White House Maker Faire. Retrieved from http://www.whitehouse.gov/blog/2014/02/03/announcing-first-white- house-maker- faire Kellner, D. (1983). Critical theory, commodities and the consumer society. Theory Culture & Society, 1(3), 66-83. doi:10.1177/026327648300100306 Kidwell, C. B., Christman, M. C. S., National Museum of History and Technology., & Daniel J. Boorstin Collection (Library of Congress). (1974). Suiting everyone: the democratization of clothing in America. Washington: Published for the National Museum of History and

206 Technology by the Smithsonian Institution Press. Kidwell, C. B. (1979). Cutting a fashionable fit: Dressmakers' drafting systems in the United States. Washington: Smithsonian Institution Press. Kinna, R. (2000). William Morris: The art of socialism. Cardiff: University of Wales Press. Kirschenbaum, M. G. (2008). Mechanisms: New media and the forensic imagination. Cambridge, Mass: MIT Press. Langley, N. (May 2002). Write once, run anywhere? Computer Weekly. Retrieved from http://www.computerweekly.com/feature/Write-once-run-anywhere Latour, B. (1996). Aramis, or, The love of technology. Cambridge, Mass: Harvard University Press. Latour, B. (2005a). Reassembling the social: An introduction to actor-network-theory. Oxford: Oxford University Press. Latour, B. (2005b). Science in action: How to follow scientists and engineers through society. Cambridge: Harvard University Press. Law, J. (1992). Notes on the theory of the actor-network: Ordering, strategy, and heterogeneity. Systems Practice, (5)4, 379-393. Law, J. (2009). The materials of STS. Version of 9th April, 2009. Retrieved from http://www.heterogeneities.net/publications/Law2008MaterialsofSTS.pdf Lessig, L. (2004). Free culture: How big media uses technology and the law to lock down culture and control creativity. New York: Penguin Press. LGRU (n.d.). Co-position. Libre Graphics Research Unit. Retrieved from http://lgru.net/archives/category/co-position LGRU Wiki (November 24, 2011). Retrieved February 7, 2015 from the LGRU Wiki: http://ospublish.constantvzw.org/lab/wiki/index.php/Projectdescription LGRU (May 27, 2013). Considering your tools. Libre Graphics Research Unit. Retrieved from http://lgru.net/archives/5349 Lindtner, S. & Guimarin, A. (April 29, 2014). Making! The other story: Robot#10, twins separated at birth, and Hacker Mama. Ethnography Matters. Retrieved from http://ethnographymatters.net/blog/2014/04/29/making-the-other-story/ Lipson, H., & Kurman, M. (2013). Fabricated: The new world of 3D printing. Maddalena, K., & Packer, J. (2015). The digital body: Telegraphy as discourse network. Theory, Culture & Society, 32(1), 93-117. doi:10.1177/0263276413520620 Markillie, P. (April 21, 2012). A third industrial revolution. The Economist. Retrieved from http://www.economist.com/node/21552901 Marsh, P. (2012). The new industrial revolution: Consumers, globalization and the end of mass production. New Haven: Yale University Press. Marx, K. (1844). Estranged labour. In Economic and philosophic manuscripts of 1844. Retrieved from https://www.marxists.org/archive/marx/works/1844/manuscripts/labour.htm Marx, K. (1971). The Grundrisse. New York: Harper & Row. Marx, K. (1861). The Production process of capital. Retrieved from https://marxists.anu.edu.au/archive/marx/works/1861/economic/index.htm Masters, J. (1995) 'The History of Action Research' in I. Hughes (ed) Action Research Electronic Reader, The University of Sydney, retrieved from http://www.behs.cchs.usyd.edu.au/arow/Reader/rmasters.htm Miller, M. R. (December 10, 2006). The Last Mantuamaker: Craft Tradition and Commercial Change in Boston, 1760-1840. Early American Studies: an Interdisciplinary Journal, 4, 2, 372-424.

207 Miller, D., & Merrilees, B. (August 01, 2004). Fashion and commerce: a historical perspective on Australian fashion retailing 1880-1920. International Journal of Retail & Distribution Management, 32, 8, 394-402. Moreau, C. P. (2011). Inviting the amateurs into the studio: Understanding how consumer engagement in product design creates value. Journal of Product Innovation Management, 28, 3, 409-410. Musson, A. E. (1972). The ‘Manchester school’ and exportation of machinery. Business History, 14(1), 17. NASA (April 21, 2008). Moondust and duct tape. NASA Science. Retrieved from http://science.nasa.gov/science-news/science-at-nasa/2008/21apr_ducttape/ NASA (December 22, 2014). Space Station 3-D printer builds ratchet wrench to complete first phase of operations. NASA. Retrieved from http://www.nasa.gov/mission_pages/station/research/news/3Dratchet_wrench/ Negroponte, N. (1995). Being digital. New York: Knopf. Nenadic, S. (April 01, 1998). The Social Shaping of Business Behaviour in the Nineteenth- Century Women's Garment Trades. Journal of Social History, 31, 3, 625-645. Nike.com (2013). NIKEiD Customize shoes and accessories. Retrieved from http://www.nike.com/us/en_us/c/nikeid Nissenbaum, H. (2001). How Computer Systems Embody Values. Computer, 34, 3, 120. Norton, M. I., Mochon, D., & Ariely, D. (2012). The IKEA effect: When labor leads to love. Journal of Consumer Psychology, 22, 3, 453-460. O'Connell, J. (1993). Metrology: The Creation of Universality by the Circulation of Particulars. Social Studies of Science, 23(1), 129. Obama, B. (June 17, 2014). Presidential Proclamation -- National Day of Making, 2014. Retrieved from http://www.whitehouse.gov/the-press-office/2014/06/17/presidential- proclamation- national-day-making-2014 Orlikowski, W. J. (2002). Knowing in Practice: Enacting a Collective Capability in Distributed Organizing. Organization Science, 13, 3, 249-273. Pargman, D. & Palme, J. (2009). ASCII imperialism. In M. Lampland and S.L. Star (Eds.), Standards and their stories (95-114). Ithaca: Cornell University Press. Public Safety Canada (2014). Action plan 2010-2015 for Canada's cyber security strategy. Retrieved from http://www.publicsafety.gc.ca/cnt/rsrcs/pblctns/ctn-pln-cbr-scrt/index- eng.aspx Python Software Foundation (n.d.). Open spaces. PyCon 2015 in . Retrieved from https://us.pycon.org/2015/community/openspaces/ Rachel, J., & Woolgar, S. (1995). The discursive structure of the social-technical divide: the example of information systems development¹. The Sociological Review, 43, 2, 251-273. Radder, H. (Ed.) (2003). The philosophy of scientific experimentation. Pittsburgh, PA: University of Pittsburgh Press. Raymond, E. S. (2001). The cathedral and the bazaar: Musings on Linux and Open Source by an accidental revolutionary. Cambridge, Mass: O'Reilly. Record, I., coons, g., Southwick, D., & Ratto, M. (2015). Regulating the Liberator: Prospects for the regulation of 3D printing. Journal of Peer Production, 6. Rogers, H. (1997). "The good are not always powerful, nor the powerful always good": The politics of women's needlework in mid-victorian london. Victorian Studies, 40(4), 589- 623. Retrieved from http://search.proquest.com/docview/211998377?accountid=14771 Rogers Publishing (n.d.). Profit—Overview. Retrieved from

208 http://www.rogersconnect.com/publication/profit Rosen, P. (2004). Up the velorution: Appropriating the bicycle and the politics of technology. In R. Eglash, J. L. Croissant, G. Di Chiro & R. Fouché (Eds.), Appropriating technology: Vernacular science and social power (365-390). Minneapolis: University of Minnesota Press. Sadowski, J. & Manson, P. (May 17, 2014). 3-D print your way to freedom and prosperity. Al Jazeera America. Retrieved from http://america.aljazeera.com/opinions/2014/5/3d- printing-politics.html Scholz, T. (2013). Digital labor: The Internet as playground and factory. New York: Routledge. Schorman, R. (December 01, 1996). Ready or Not: Custom-Made Ideals and Ready-Made Clothes in Late 19th-Century America. The Journal of American Culture, 19, 4, 111-120. Sewell, G., & Wilkinson, B. (May 01, 1992). `Someone to Watch Over Me': Surveillance, Discipline and the Just-in-Time Labour Process. Sociology, 26, 2, 271-289. Silver, J. (Producer), Wachowski, A., & Wachowski, L. (Directors). (2001). The Matrix [Motion picture]. United States: Warner Bros. Pictures. Skalski, G. (November 20, 2013). How to attract more women to tech conferences. Opensource.com. Retrieved from http://opensource.com/life/13/11/how-attract-more- women-tech-conferences Slop. (n.d.). In Oxford English Dictionary. Sobchack, V. C. (2004). "A leg to stand on" (pp.205-225) in Carnal thoughts: Embodiment and moving image culture. Berkeley: University of California Press. Stallman, R. (2009). Why open source misses the point of free software.Communications of the ACM, 52(6), 31-33. Star, S. L. (1999). The ethnography of infrastructure. American Behavioral Scientist, 43, 3, 377- 391. Star, S.L. (2010a). When shadows become complex. Learning Communities: International Journal of Learning in Social Contexts, 2, 150-158. Star, S.L. (2010b). This is Not a Boundary Object: Reflections on the Origin of a Concept. Science, Technology, & Human Values, 35, 5, 601-617. Stoneley, P. (January 01, 2006). Sewing in Concord. Women's Studies, 35, 4. The Tacoma Times (March 31, 1904). A Simple shirt waist. Page 2. Retrieved from http://chroniclingamerica.loc.gov/lccn/sn88085187/1904-03-31/ed-1/seq-2/ Terranova, T. (2000). Free labor: Producing culture for the digital economy. Social Text, 18(2), 33-58. Terranova, T. (2013). Free Labor in T. Scholz (Ed.), Digital labor: The Internet as playground and factory. New York: Routledge. Trautman, P. (1979). Personal clothiers: A demographic study of dressmakers, seamstresses and tailors 1880–1920. Dress, 5(1), 74-95. doi:10.1179/036121179805298747 van Oost, E. (2003). Materialized gender: How shavers configure the users' femininity and masculinity. In N. Oudshoorn & T.J. Pinch (Eds.), How users matter: The co-construction of users and technologies (193-208). Cambridge, Mass: MIT Press. Von Drehle,. D. (2003). Triangle: The fire that changed America. New York: Atlantic Monthly Press. von Hippel, E. (2005). Democratizing innovation. Cambridge, Mass: The MIT Press. Yu, U.J., & Park, J. (2014). Consumers' Virtual Product Experiences and Risk Perceptions of Product Performance in the Online Co-Design Practice: A Case of NIKEiD. Family and Consumer Sciences Research Journal, 43, 1, 29-46.

209 Walsh, M. (2005). Gendered endeavours: Women and the reshaping of business culture. Women's History Review, 14(2), 181-202. doi:10.1080/09612020500200428 Weber, S. (2004). The success of open source. Cambridge, MA: Harvard University Press. Weinberg, H.B. (n.d.). Mary Stevenson Cassatt (1844-1926).Heilbrunn Timeline of Art History. Retrieved from http://www.metmuseum.org/toah/hd/cast/hd_cast.htm Winner, L. (January 01, 1980). Do Artifacts Have Politics?. Daedalus, 109, 1, 121-136. Winner, L. (1986). The whale and the reactor: A search for limits in an age of technology. Chicago: University of Chicago Press. Zipkin, P. (2001). The limits of mass customization. MIT Sloan Management Review, 42(3), 81- 87. Retrieved from http://search.proquest.com/docview/224963859?accountid=14771 Zuboff, S. (1988). In the age of the smart machine: The future of work and power. New York: Basic Books. Zwick, D., Bonsu, S., & Darmody, A. (2008). Putting Consumers to Work. Journal of Consumer Culture, 8, 2, 163-196.

210 Appendix 1 – Glossary

ABS: Acrylonitrile butadiene styrene, a plastic commonly used for 3D printing. API: Application programming interface, a set of specifications which allow third-party developers to build applications compatible with an existing software platform. CAD: Computer-assisted design, a catch-all term, commonly used to describe software which assists in the production of 3D models or plans. CAM: Computer-aided manufacturing, a term denoting processes which use computer- controlled machinery to produce goods. compiler: A piece of software which turns computer code written in a high-level programming language into the low-level code needed to run on a computer. configurator: A software interface for configuring a parametrically customized good. chop shop: A bicycle shop which takes apart old bikes and uses the component parts to build new ones. distal: As an anatomical location, the portion of a limb farthest from the torso. extruder head/extruder assembly: The portion of a 3D printer which heats and extrudes plastic or other material. extrusion: The process of pushing heated print material through a small print head in order to lay down a layer of a 3D-print. F/LOSS: Free/Libre Open Source Software, see footnote on page 32 for the working definition adopted in this dissertation. git: A popular version control system. Github: A web-based platform which provides an infrastructure for projects and individuals to implement git. medial: In anatomical locations, towards the centre of the torso. PLA: Polylactic acid, a plant-based plastic popular in hobbyist 3D printing. proximal: In anatomical locations, the portion of a limb nearest to the torso. rectification: In the production of prostheses, the process of modifying a positive model to reflect desired areas of indentation and relief on the socket.

211 Appendix 2 – Interview guides

Prosthetics extreme user interview guide

Object biography What have your experiences with your current prosthesis been like? Experiences with previous prostheses? How many prostheses have you had, over the years? When were you first fitted? Have you ever used a non-custom prosthesis? Include some follow-up questions about experiences with the current and past prostheses.

Fitting process Tell me a little about your last fitting. (Try to elicit information here about the process used, tools used, whether or not the residuum was marked up, etc.) What kinds of tools were used? How much experience did the fitter seem to have? Did they seem expert or knowledgeable? What was their manner like? (Eliciting information about things like casting, markup, molding, iteration, etc.) How many visits did it take to get it right?

Institutional infrastructure How did you find your first prosthetist? How did you find your last prosthetist? Have you shopped around? Have you gotten prosthetics from different places? Have you ever had a non-traditional prosthetic? (like: high-performance, non-traditional production method, etc.) Have you ever been fitted for other custom goods? (like: jewelery, clothing, sporting gear, dental appliances, etc.)

212 Appendix 3 – Production colophon

This dissertation was typeset in LibreOffice, using the Linux Libertine family of fonts for body text and OSP-DIN for headings and display text. OSP-DIN is from the OSP type foundry, run by many of the same people involved in the LGRU, and producing typefaces using F/LOSS tools and released under Free Cultural licenses.

213