Microsoft Word - 71547-1-11-ZyskowskiandMilland.docx Zyskowski, K and Milland, K. (2018). A Crowded Future: Working against Abstraction on Turker Nation. Catalyst: Feminism, Theory, Technoscience, 4(2), 1-30. http://www.catalystjournal.org | ISSN: 2380-3312 © Kathryn Zyskowski and Kristy Milland, 2018 | Licensed to the Catalyst Project under a Creative Commons Attribution Non-Commercial No Derivatives license ARTICLE A Crowded Future: Working against Abstraction on Turker Nation Kathryn Zyskowski University of Washington kcz@uw.edu Kristy Milland University of Toronto kristy@kristymilland.com Abstract This paper examines digital labor and community through an ethnography of a discussion board supporting short-term digital contract workers on the Amazon Mechanical Turk (mTurk). First, we give a thorough overview of mTurk, the crowdsourcing marketplace, and Turker Nation, a discussion board for workers on mTurk. We trace the experience of interacting with this infrastructure on mTurk as worker and employer. Following, we look at scholarship on software infrastructure and autonomous Marxist theorizations of contemporary work. We demonstrate how the labor of participating on the discussion board Turker Nation helps to counter the abstraction the infrastructure provides. We show how workers on Turker Nation use the platform to structure time, build socializing spaces at work and initiate collective organizing. In doing so, we argue that workers’ labor belies conventional class classification, such as white-collar and 2 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) blue-collar labor and instead lays the groundwork for how to structure future digital workplaces. We argue that this laboring resists the assumed logic of capitalism for digital labor that subsumes and takes over workers’ lives and conclude by looking at the limitations of the community’s collective organizing in terms of agreeing on points to communicated to the public. Introduction “No one understands what we are doing – that is frustrating. I can say I work for Amazon, but that is not really true. Recently I told someone that I was ‘Turking’. They misunderstood me and thought I said I was twerking – you know, some dance. I had to say: No, I wasn’t twerking, I was Turking!” “In an ideal world, what does a union do? Provide support for workers, stand up for workers rights and values, and support better working conditions. We have that all here at Turker Nation.” This paper began as an inquiry into the value of digital communities and communication spaces for workers on Amazon Mechanical Turk (mTurk). We focus on one digital discussion forum that supports Turkers, Turker Nation (Turkernation.com). We found that workers use Turker Nation in specific ways that augment the tools available to workers on the mTurk platform and that employing these tools transforms the experience of short-term digital work for Turkers. Workers use Turker Nation to structure time, learn how to build a Turker reputation, and socialize while working. In doing so, we argue that workers’ labor belies conventional class classification, such as white-collar and blue-collar labor, and instead lays the groundwork for how to structure future digital workplaces. We begin by giving a thorough overview of both mTurk, the crowdsourcing website, and Turker Nation, the discussion board. Next, we look at scholarship on software infrastructure and autonomous 3 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Marxist theorizations of contemporary work. We build on this literature by analyzing how the labor fits into, or not, traditional descriptors of workplace class signifiers. We demonstrate how the labor of participating on Turker Nation helps to counter the abstraction the infrastructure provides. Finally, we identify the labor on Turker Nation as structuring time, building socializing spaces at work, and informal collective organizing. These counter assumptions about what digital short-term labor looks like. We argue that this laboring resists the assumed logic of capitalism for digital labor that subsumes and takes over workers’ lives. However, we conclude by looking at the limitations of the community’s collective organizing, such as failing to find consensus on what to communicate to the public. This points toward how difference such as gender, race, and class is produced in these online communities. Despite the potential shortcomings, the labor done in building this community and making mTurk work viable for Turkers works to envision a possibility of sustainable crowdsourcing work. What Is mTurk and Who Are Turkers? MTurk is a short-term contract work marketplace. Laid out in a similar way as the widely known online marketplace Craigslist, mTurk is a place where workers (Turkers) and employers (Requesters) can post or find digital contract work. Amazon launched mTurk in 2005 when it needed to hire humans to identify duplicate pages on its website (Irani, 2013; Gehl, 2014). MTurk became famous as the first crowdsourcing site, but today is one of many. Crowdsourcing refers to a business model that distributes work to a crowd (Howe, 2009). In mTurk, short contracts are called Human Intelligence Tasks (HITs). Examples of common HITs on mTurk include transcription, image recognition, duplicate recognition, and surveys (Ipeirotis, 2010). Other types of HITs include editorial tasks for video and writing. Academics also increasingly use mTurk as a sample pool for research surveys (Mason & Suri, 2012; Berinsky, Huber & Lenz, 2012). 4 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) The name is inspired by a famous eighteenth century Mechanical Turk, a machine, that appeared to beat human chess players. The machine, it turned out, was a clever deception — a human hid inside and moved levers connected to chess pieces (Irani, 2013; Gehl, 2014). As both Irani (2013) and Gehl (2014) have argued, Amazon’s move to name its crowdsourcing marketplace after the Mechanical Turk adds to the illusion of the work being mechanized, rather than completed by human workers. At the end of 2014, scholars estimated that the numbers of active workers on mTurk ranged from 3,000 to 500,000 (Fort, Adda & Cohen, 2011; Paolacci & Chandler 2014; “Requester Tour,” 2016). It is difficult to pin down exact numbers and demographics for many reasons. One reason is that many people sign up, try, and quit mTurk within a few days. Regardless of how many workers there are, scholars estimate that only a small number of very active workers completes the majority of HITs (Deneme, 2009; Adda & Mariani, 2010; Fort, Adda & Cohen, 2011). Most workers reside in the United States and India – the two countries in which Amazon pays workers in cash – with about a quarter of workers hailing from 188 other countries (Ipeirotis, 2010; Paolacci & Chandler, 2014). Amazon pays the remaining workers through Amazon credit (“Worker Web Site FAQs,” 2016). Scholars have found that, despite coming from diverse locations, Turkers tend to share some defining characteristics: they are well-educated, young, liberal, and underemployed (Berinsky, Huber & Lenz, 2012; Paolacci & Chandler, 2014). Turkers are split pretty evenly between male and female, according to the website mTurk- tracker, which logs daily reports of mTurk demographics (Ipeirotis, 2010). Mainstream media has been highly critical of mTurk, viewing it as exemplary of new forms of labor that circumvent employee protections and comforts such as vacation pay, medical insurance, and company discounts. Other concerns are lack of minimum wage and Requesters' ability to withhold payment yet keep and use the work submitted, with no recourse available to the worker. An article in the leftist magazine Utne described mTurk as a “Digital Sweatshop” (Cushing, 2013). Beyond 5 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) popular media, scholars have also expressed concern about mTurk labor conditions. Multiple academic and non-academic projects have emerged to improve workplace conditions including TurkAlert.com, mTurkList.com, IndiaTurkers.com, WeAreDynamo.org and academic extensions such as Crowd-Workers.com, TurkBench, OpenTurk.com, turkmotion.qu.tu- berlin.de. While the intentions of journalists and academics calling for Turker unionization may be well intentioned, some Turkers feel as though they are being told how to organize without being part of the conversation. Turker Nation Overview and Background Turker Nation is an online forum about working on mTurk. On the bottom of the homepage is Turker Nation’s trademark and motto: “Turker Nation: The largest and most active Amazon mTurk forum online! Developed, managed, maintained, and operated by Mechanical Turk workers for mTurk workers”. On the homepage there are multiple forums and sub- forums. The forum is separated into different themes; some of the most popular threads include “daily HITs,” “mTurk questions” (including questions about etiquette and payment), and “Turker introductions.” Other threads include “mTurk and Turker Nation in the media,” “cheap eBooks,” and “off-topic” (cooking, gardening, and anything else). Turker Nation also includes a thread where workers can share programs and documents they have built to make working on mTurk easier and more efficient. Across the top of the webpage is a banner that offers quick links to the hottest topics: updates on the best HITs of the day, the chat room, and an affiliate link to buy things on Amazon (a small percentage of this money returns to support Turker Nation). The full forum is not available until you create an account as either a Turker or a Requester. This encourages community building among Turkers and supports communication between Requesters and Turkers as groups, rather than individuals. In order to access the forums you must introduce yourself to 6 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) the community. Beyond the main forum, there are multiple sub-forums. Some examples of the most popular sub-forums include having trouble with the mTurk site, getting suspended as a worker, and becoming a Masters worker.1 Methods The primary fieldwork for this paper was carried out over three months during the summer of 2014. Kathryn chose Turker Nation as the site of research because it is the largest and oldest forum supporting workers on mTurk.2 As Kathryn explored the Turker Nation site, she came across the following statement for researchers: “If you are a journalist or researcher looking to interview people, you are not welcome here.” Aware of the forum's stance toward academics, Kathryn did not create an account on Turker Nation but instead emailed Kristy, the manager of the site, to initiate a conversation about doing collaborative research. During these initial conversations, Kristy expressed how the community felt misused and abused by academic researchers and therefore had recently barred academics from the forum after two research studies negatively impacted the workers on the site.3 When Kathryn approached the community with openness regarding methodology, intentions to publish, and willingness to collaborate, Kristy reconsidered her stance against academic involvement in the community. Kathryn explained her positionality and research ethics, which were informed by the American Association of Anthropology’s stance on research, feminist research methodologies, and workers’ inquiry with a goal of bridging inequalities between communities and academics through reflexive dialogue, co-authorship, and participatory research design (AAA 2012; Anthias & Yuval-Davis, 1993; Craven, 2013; Waters & Woodcock, 2017; Writers & Nagar, 2006). This conversation led to a decision to collaborate with an aim of creating dialogue between the community and scholars on how to engage ethically with online research of Turker Nation, a community which had previously had a bad 7 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) experience with researchers. We agreed that ethical research rests on individuals on the site and/or the community at large comprehending the use of their words, thoughts, and social activities in research and not by following the site’s by-laws or by arguing that it is a "public" webpage (Boellstorff, 2008; Markham & Baym, 2008; Papacharissi, 2010; Coleman, 2013; Gray, 2014). At the same time, we acknowledge that adopting certain methodologies does not necessarily fix discrepancies in information access, voice, and power (Visweswaran, 1994; Stacey, 1988). Kathryn and Kristy exchanged articles and thoughts on research ethics, and dialogically landed on an approach to the research project. For example, Kristy established safe places for initial interactions and group interviews by creating a separate chat room where users opted in and chose to use either their official Turker Nation username or create a pseudonym. This method was similar to some of the components of the house "Ethnographia" that Boellstorff (2008) built in the game Second Life. Kathryn created an information sheet with details about her academic affiliation, contact information, ethics of the research, and interviewees’ rights. This information sheet was open for amendments on Turker Nation. At all times on Turker Nation, Kathryn used a username (author_anthropology) that identified her as a researcher. She only participated in chatroom conversations where it was explicitly made aware she was participating as a researcher. Kathryn logged into Turker Nation each day as a participant observer, following conversations in the discussion board threads and joining the chat room. In addition, Kathryn conducted twenty-nine semi- structured interviews via Skype, phone and Google Chat; interviewees chose their preferred mode of communication. We decided that Kathryn should do the interviews and that these should be separate from monitoring by Kristy, who, because she held power in the community, felt her presence might alter the responses provided about participation on the forum. Toward the end of fieldwork, we met in person for three days to code the interviews and co-create the primary explanatory schematics in this paper. 8 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Infrastructure, Autonomous Marxism, and Class Our analysis builds on the emerging literature on the politics of software infrastructure (Star, 1999; Irani, 2013; Gehl, 2014). Critical social scientists of software point out that infrastructures carry social and political assumptions. Gehl’s (2014) book highlights the political and social implications of the built infrastructures of software and social media sites. He argues that the infrastructure of mTurk was engineered in a manner that obfuscates the human worker: “The legendary marketplace, where labor meets capital in a personified negotiation, is replaced by a screen interface, where labor finally becomes completely mechanical and rationalized … this makes human work look mechanical” (Gehl, 2014, p. 58). We contend that part of the work that Turker Nation does is to rebuild the “personified negotiation” that is missing in the mTurk interface (p. 58). Irani (2013) further argues that mTurk reworks human labor to fit into existing computing cultures, such as Artificial Intelligence and large algorithms. Communities such as Turker Nation enable workers to re-insert human interaction to the Turking experience. While they are hiring, managing, and paying workers, Requesters only view Turkers’ approval ratings, usernames, and number of tasks completed. At the time this was written, there was no way for Turkers to reach out directly to Requesters on mTurk. Irani (2013) argues that this produces “intensified labor hierarchies by obscuring workers behind code and spreadsheets” (p. 721). Irani (2013) identifies a key element of computing culture as those in which “programmers are taught to construct and respect ‘walls of abstraction’” (p. 732). Abstraction as a core element of computing cultures resonates with anthropologists’ reading of abstraction as a defining characteristic of contemporary capitalist culture (Ho et al., 2015). This abstraction has also been referred to as platform capitalism, a perspective that views platform operators and designers as holding most of the power and extracting value from users of the platform (Srnicek, 2016). Turkers are abstracted by the 9 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) infrastructure on mTurk, and their participation on Turker Nation is a labor that counters this abstraction. Autonomous Marxism, also called Post-Fordist labor studies, refers to the various ways in which the structure, tools, and organization of labor has changed since the 1970s (Dowling, Nunes & Trott, 2007). Scholars theorizing this shift in the organization of labor have described it as flexible, fragmented, and precarious. The Italian school, one body of influential scholars, theorizes these new forms of labor as immaterial and affective (Lazzarato, 1996; Hardt, 2005; Berardi, 2009). They argue that the consequences of these new forms of labor include the separation of labor and commodity, increased demands for affective labor, and unending workdays. Feminist scholars were quick to critique aspects of immaterial and affective labor, contending that fundamental forms of labor, such as child-rearing and care, have never directly related to market values (Weeks, 2011). While Turkers engage in forms of unpaid labor such as learning how to communicate most effectively with Requesters, we feel that the descriptive word closest to the labor Turkers do is Baym’s (2015) "relational labor." Baym (2015), in her research on musicians connecting digitally with audiences, defines relational labor as the “regular, ongoing communication with audiences over time to build social relationships that foster paid work” (Baym, 2015, p. 16). Though the labor of building relationships is unpaid, it is understood as one of the best ways to boost sales of tickets, merchandise, and music. While artists are being told to “connect” with audiences over social media, Turkers have built supporting sites (such as Turker Nation) precisely because there was no way to connect with other workers or Requesters. In a similar vein, Martin et al. (2014) use Star and Strauss (1999) to argue that much of the work on Turker Nation is to make previously invisible labor visible. We agree with this, yet believe that this analysis misses both how the labor of visibilizing works to resist the complete takeover of life to work (an assumed part of cognitive digital labor) and also how the work of visibilizing itself creates fissures within the community. 10 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Turking in many ways can be seen as blue-collar knowledge work. During the time of the Industrial Revolution, cottage work was deskilled and pumped onto assembly lines and worker surveillance was the norm. Crowd work has done the same, with projects being broken into micro tasks and then sent out to the crowd, piece by piece, until the project is done (with the workers never seeing the end result). The way Taylor spoke about factory workers, comparing them to draft animals, is mirrored in how academics speak of crowd workers today. The former referred to low skilled workers as "stupid" and compared them to draft animals, while some scholars have referred to crowd workers as “the dimmest bulbs” (Taylor ,1914, Zittrain, 2009). But there are critical differences between theorizing mechanical labor during the industrial period and the pay-per-piece cognitive work that Turkers do. We landed on the conclusion that we may not have the right word for characterizing Turkers’ labor with regard to class. The multiple competing characteristics of the work defy the class characterizations of the nineteenth and twentieth centuries. On the one hand, the work can be characterized as knowledge work, including the ability to work remotely on laptops, a common characteristic of white-collar professional work. On the other hand, it is short-term, contract work with no job security or upward mobility, characteristics that make it appear most like blue-collar work. The turn to cognitive labor from mechanical labor has been written about as the ultimate takeover of one’s life. Berardi (2013) distances what he calls "the cognitariat" from previous industrial laborers: they are doing things that are not necessarily exchangeable. While the pieces of work posted on mTurk may be interchangeable, what we find here is that Turkers use forums to make the work more personable. This is one defining characteristic that Berardi (2013) sees as unique with cognitive labor. They do this by building relationships on Turker Nation that help them gain reputations, by spending time learning how to build their reputation and make themselves a desirable worker. Berardi (2013) assumes that this work becomes the most 11 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) important part of a worker’s life, with conviviality and family life pushed to the side. Therefore, Turkers on Turker Nation who invest in socializing with the community and structure their work days can be seen as a form of resistance to the logic of capital in digital life taking over a whole person’s life. This is similar to the argument that Amrute (2016) makes about IT workers’ leisure activities in Berlin. She argues these can be seen as “developing alternative ways of living explicitly pitched against the expansion of work into all areas of life” (p. 5). As one interviewee said: “If Turker Nation closed, I would lose my ‘office space,’ my friendships, and a place to talk shop.” Building Turker Reputation One of the common roadblocks for beginning Turkers is figuring out how to build worker reputation. Turkers’ reputations are displayed on the "dashboard," or screen where they can see the approval-and-reject rating for the work they have personally posted. When hiring a worker for the first time, the only way to be selective is to use one of the preset screenings. The preset default for Requesters posting HITs through the Requester website is to only accept work from Turkers with a 95% or above acceptance rate.4 Since the initial research for this paper, this default has shifted from 95% to a Master's qualification. When one’s reputation for work is something determined by an algorithm, and when, to quote one interviewee, workers are seen in the interface as an “alpha- numeric string,” collaboration and the sharing of information are ways to make visible the humans working in the system. The most valuable information for beginning Turkers on Turker Nation is information on how to avoid rejections, how to reach out to Requesters to understand a rejection, and what types of HITs to do to improve the chances of HIT approval. When someone creates an account on mTurk they automatically start at the lowest reputation. Only once they build reputation do Turkers become eligible for higher paying HITs. Unlike a "traditional" workplace 12 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) where a new worker might submit work and receive feedback on how to improve it from a manager, mTurk offers no line of recourse for Turkers whose HITs have been rejected (Irani, 2013). Instead, workers turn to Google to figure out what to do next. One interlocutor recounts: “I first learned about Turker Nation when searching the name of a Requester that had given me a rejection.” Another, a self-employed man in his twenties, recounted that figuring out how to set up an account and start competing HITs was self-explanatory. However, once he had been Turking for a few weeks he realized he did not understand how to build a reputation and work his way up to higher-earning HITs. This was hurting him as many of the HITs are only available to Turkers with a 95% approval rating. “When I started mTurk – I was stupid. My acceptance rate was 75%.” The experience of finding Turker Nation when trying to navigate Turker reputation was a common theme. Receiving a rejection on a HIT affects not only Turkers’ immediate pay but also their ability to be eligible for future work. “I believe I did a Google search for ‘MTurk Questions’ and came across them [Turker Nation] in that way …. When I joined the forum, I learned. I now have an acceptance rate of 99.5%.” The outcome of this type of mentoring and sharing of working strategies is that Turkers participating in Turker Nation garner better reputations and increase their understanding of how to navigate the system of mTurk; the result is higher confidence in navigating the system and higher pay. Structuring Time Scholars of digital labor have raised concerns that crowdsourcers (and other digital laborers) work around the clock (Berardi, 2009; Berardi, 2013). While some may, we found that many build time structures into their day, and that Turker Nation serves as a tool to help them accomplish this. One woman said: “I like chat much better than the forum because it feels like I am coming into the office when we all join chat. When 8 o’clock rolls around and people start coming in, it feels like we all 13 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) gather in front of the water cooler (or coffee pot) at work to say hi and the day gets started”. Numerous interlocutors stated that the first thing they do in the morning was to open their laptop, look over the previous day's work, and say hi to fellow coworkers. The timing and socialization of signing into chat is not just symbolic, but also increases the opportunities for finding good work and in turn getting paid more. Workers that Kathryn interviewed ranged from the United States, Canada, and India, yet all said the same thing: that the thread picked up around eight-to-nine in the morning Eastern Standard Time. Part of this is driven by demand; Requesters tended to be based in north America and hence the higher volume of Tasks appeared beginning in the morning and continuing through a North American nine-to-five day. Scholars have recently written on how new media affects when workers do work, when they are expected to be reachable by email, and how increased availability in a labor system shifts the workers’ relation to employees (Berardi, 2009; Gregg, 2011). Having a central (albeit virtual) place to gather in the morning helped reinforce a sense of a shared workday despite the physical distances between workers. The following interviewee described a common morning: “I always go to mTurk immediately upon turning on the computer in the morning. In fact I do it before even having the first smoke of the day. I turn on the computer, open up my mTurk dashboard and check for rejections/approvals, go have a smoke and make coffee, and begin Turking.” Berardi (2013) contends that one of the characteristics of digital work is the loss of pleasure in day-to-day interactions, which turns people to an “investment of desire in work” (p. 4). This line of argument both imagines people as being unconsciously consumed by their never-ending workday and also imagines people as individuals in competition with one another rather than finding solidarity with one another, something we will return to at the end of this paper (Berardi, 2013). While this time structure reveals how Turkers resist a twenty-four- seven work week, it also demonstrates how race and class are mapped 14 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) onto flexible labor. For instance, Turkers in India adjust their work time- clocks to the night shift to maintain the American-centric nine-to-five (Aneesh, 2006). One Turker Nation interviewee from India works in the IT industry in India and worked US hours at that job so that, although he was working the night shift, he used mTurk as a complementary income to his "day" job and did not adjust his working hours.5 Digital short-term contract work has been portrayed as a system that adheres to employers’ and businesses’ time clocks, dependent on having workers available around the clock (Howe, 2009). Blue-collar work socialization has historically been seen as outside the workplace and work hours. Given the close supervisor surveillance and discrete tasks, workers often did not spend time in the middle of the work day socializing in a shared space. While the work Turkers are doing can be seen as mental factory work – in terms of short, discrete tasks on a timeclock, they simultaneously socialize and communicate with other workers in spaces such as Turker Nation. The fact the mTurk does not provide a socializing place (let alone any form of peer-to-peer communication) demonstrates that the mTurk founders imagined mTurk's workers to be discrete individuals who did not need to or want to communicate with each other. The labor put in to demarcate the work-day and maintain a socializing space on Turker Nation works against this. Anthropologists and other social scientists have long been interested in how different structures of labor relate to time. During the nineteenth century, workers fought over the hours of a working day, receiving the first legislation with the Factory Act of 1833 which limited the working day to twelve hours for those under eighteen (Marx, 1993). Through the process of industrialization and the advent of the time clock, industrialists at first had a hard time finding wage-workers; pre- industrialist labor-time worked on different rhythms, such as the natural rhythms of the harvest and sea tides (Thompson, 1967). Over the last few hundred years, the industrialists’ push toward clocked labor-time has become naturalized, and people largely think in terms of eight-hour work days, clocking in and clocking out of work spaces (Thompson, 1967). 15 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) However, to take this as the "natural" work day (as evidenced by Marx and Thompson’s work) both elides a long history of class struggle over the work day as well as the transformation of time during industrialization. Scholars have contended that mobile computing technologies blur the line between work and home; work becomes something that is done at an office, at the kitchen table, or in bed (Gregg, 2011). This simplified narrative, however, ignores innumerable forms of work— particularly much of the work done in the non-western world and non-formal work that never entered into such a system and has always depended on mobile technologies that straddle work and home such as looms, sewing machines, and farming tools (Carey, 1983). Mobile applications enable similar work to continue. Waters and Woodcock (2017), in their work on Deliveroo, a food delivery app, show how the application enables presumed isolated workers to meet, socialize, and organize protests in London. The results of new spatial configurations of work are thus not pre-determined. While individuals’ ideas about work have certainly shifted with the ability to work remotely, Turkers structured workdays contradict an image of an around-the-clock digital factory worker. Another aspect of time structuring is how Turkers on Turker Nation strategize how to think of work over a longer time period than the immediacy the system works in. On a system where work contracts may last five minutes to one day, Turkers discuss how to manage a workload and think on a longer horizon: “Turker Nation encourages us to think about how our actions today are going to affect the opportunities for work we have later down the line.” This is one example of how workers use Turker Nation in ways that work counter to the fact that in the mTurk interface they are treated as short-term workers. Thinking only in the immediate is not a good way to plan for growing in a job; the Turkers encourage actions that will benefit their opportunities in the long term. As Marx pointed out, industrialists manipulated time to get more work out of their workers by “hounding him hither and thither, in scattered shreds of time” (Marx in Harvey, 2010, p. 403). Harvey (2010) explains that Marx was referring to a practice where employers would purposefully 16 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) string together small work commitments so that workers would “work 15 minutes but count 10” (p. 403). This strategy of eking out more time from workers is built-in to the software of mTurk and workers certainly spend more time working than the clock counts. However, Turkers on Turker Nation share strategies for how to be efficient and spend the least amount of time searching and applying for jobs (non-paid time). Turkers also share recommended Requesters based on perceived fairness of accepting jobs and communication with Turkers. In this way, Turker Nation provides a space for workers to strategize on how to minimize this non-clock time, enabling them to work against the abstraction. Socializing at Work Turker Nation is also the site for many types of socializing: customary daily greetings, bonding over personal lives while completing work, complaining about work, and sharing ideas for off-line life, such as favorite music and recipes. Turker Nation serves as a place-maker during the workday. Workers who used the chat room all said that they simultaneously log into Turker Nation and mTurk, and left the chat room open throughout the day. Similar to the new co-working spaces and "open-office" structure widespread in IT offices now, this provides Turkers the possibility of always being social or reaching out to ask a question, yet most of the time people are doing work (Amrute, 2014). One form of socializing was the daily chitter chatter of narrating life while working. One day in the chat room, a worker entered: “Speaking of off topic, brb [be right back], gotta go toss the cheese into the crock pot.” While discussing the day’s work, Turkers added in commentary about what they are doing, which made them personable to other Turkers and enabled some to build lasting relationships within the community. Participants commonly referred to others on Turker Nation as co-workers and the chat room as the community social space at work, akin to the central coffee pot. Interviewees described these "off topic discussions" as an integral part of the workday: “It’s just like at regular work … it’s not 17 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) all work all the time.” One participant said: “The conversation covers whatever is of interest to folks at the time; we definitively build relationships with other forum members.” While discussions on gardening, diets, knitting, crockpots, and music lists may seem, and even be labeled, “off topic,” it is an integral part of how workers’ place-making practices form a supportive working space. “If I had not found Turker Nation, I would not have made as much money for sure. And the fun we have when things are slow: priceless.” Beyond demonstrating the ways in which Turkers build relationships with each other on Turker Nation, these topics also point toward workplace socialization being gendered. The majority of the people interviewed by Kathryn were women who were also doing the bulk of domestic labor in their homes. However, we know that the demographics on mTurk are split evenly by gender. We do not have answers for why there are more women on the forum, though many of the women Kathryn interviewed mentioned the benefit of being able to work around a schedule with their kids. The availability of short-term digital contract work in conjunction with poor maternity leave or lack of flexible work schedules for working mothers is something that should be further investigated. Providing a space to discuss the daily trials and tribulations of work is an integral part of how Turker Nation supports Turking as a sustainable form of employment. One interviewee says that one of the most valuable aspects of the chat room is being able to “talk to others who understand how frustrating this job is.” Workers at any job share struggles that others in a different profession may not understand or empathize with. One interviewee described this as “the bond of doing the same work with the same Requesters and having the same problems with the HITs.” Another explained that Turker Nation “[is] a community of people who understand what it’s like to be a Turker! It’s hard to explain to people who have never done it (i.e., what a HIT is, how varied the work is, how low the pay is, etc).” Anthropologists have noted the importance of workplace socializing as a mechanism for coping with difficult work. In 18 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Gusterson’s (1998) book on employees of a nuclear lab in California, he argues that workers joked to deal with the potential catastrophic outcomes of their labor (Gusterson, 1998). We are not making an analogy here, but rather pointing out the strategies that workers use to handle different types of labor: Turkers used Turker Nation to find solidarity and share a common work experience that could otherwise be isolating and difficult to explain to those not familiar with crowdsourcing. When Kathryn asked Turkers if they had ever interacted with other Turkers off line, many responded that they had communicated via email, phone, or Facebook, but had not met in person. One said: “We got to see pictures of Sarah’s beautiful baby and I’ve interacted with some thru email IRL [in real life].” Turkers on Turker Nation distinguished social time at work, on Turker Nation, from off-work social time. “I consider a lot of people here as friends. Some of us do have off-forum contact. We are friends on Facebook and share pictures of our offline life”. While Turker Nation may be seen as the coffee pot at work, Facebook and email are the outside-of-work social spaces. Communication scholars have long been interested in thinking about the ways in which people collaborate at work, how to design spaces that foster collaboration and co-operative work, and how working across media technologies affects facetime with workers. Beyond the structuring of a “workday,” many Turkers mentioned they do not enter the chat room on the weekends. This shows a discernable difference between the use of other social media sites, such as Facebook, and the use of Turker Nation. For example, sharing pictures of family happened over email or Facebook, rather than on Turker Nation. For some, not going into Turker Nation over the weekend signaled an active boundary- making that is broader than Turker Nation, namely, that of creating non- work spaces and times in spite of the fact that we have the opportunity and societal pressure to always work (Weeks, 2011). 19 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Self-organized Collective Action In July 2014, a Requester created a number of false accounts and was soliciting HITs through these accounts. Kristy woke up to overflowing email and message boxes. Turkers were very concerned about a HIT. The major concern was that the Requester had posted the same HIT under a number of different Requester names immediately signaling a spam situation to workers. Based on the type of questions and demographic data asked in the posted survey, it was concluded that the Requester was probably an academic researcher. Turkers were upset not only by a number of false accounts but also by illusive research; academic exploitation and secretive research had been widely discussed topics on Turker Nation. As word spread through the Turker Nation community and other mTurk blogs, discussion boards, and subReddits, Kristy reached out to other site managers and strategized how to handle the situation. On Turker Nation, a thread was created to discuss the situation and Turkers added in information they gathered and weighed in on how to proceed as a community. Within a day, Turkers had figured out who the Requester was and sent him a letter telling him why creating false accounts and not being upfront about his research created a stressful situation for the workers. The letter concluded by asking the researcher to explain his intentions and apologize to the workers. Some Turkers even wrote letters to his university’s institutional review board (IRB) claiming he had violated human research ethics. Following this episode, Turkers – alongside a group of academics – created a protocol for IRBs to use to approve academic research on mTurk.6 It is critical to recognize how Turkers understand collective action. Historically blue-collar workers have relied on the union as the arbiter between worker and employer (Lichtenstein, 2013). Unions have been notably absent from white-collar workplaces, with the notable exception of teachers. In the twenty-first century, many patterns of labor organization, such as teachers at Charter Schools choosing not to join 20 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) teacher unions, have brought into question the union as the right body to stand up for workers’ satisfaction, rights, and productivity. Anthropologists and labor scholars attribute much of this change to new formations of relations between workers, their labor, and employers in the contemporary form of capitalism (Robertson, 2008). However, the ability to informally organize and present alternatives to management is a hallmark of white-collar work. Turkers’ preference to organize informally and open lines of communication to Requesters, rather than organize in a union, is another way Turkers are blurring the lines between being categorized as either blue-collar or white-collar workers. Examples of collective action are multiple and Turkers participate in collective action regularly. We’ll give two more examples. First, a Requester had posted a HIT as a survey; however, the task was not actually a survey. Frustrated by the mislabeling, a Turker started a thread on Turker Nation gathering petitions from other Turkers asking the Requester to change the label of the task.7 Within one day, the thread garnered over fifteen signatures. The Turker wrote the Requester explaining what they wanted changed and why it mattered; when HITs are mislabeled it wastes Turkers' time and therefore effects their pay per hour. Turkers noted that showing this collective support for one another was recognized by Requesters (albeit limited to those who frequented Turker Nation and cared about dialogue with Turkers) who came to the site. One said: “Requesters seem to like that we’re organized and show that we’re hard workers.” The "well-organized" comment was in response to group-Turker comments and responses to Requesters, such as pointing out when they have made a mistake that will later harm Turkers. In another instance, workers completing projects for a large American video streaming service felt the payments for HITs were dropping to unacceptable levels. When they contacted the Requester posting the work, they received a response that the payment was fair. The workers chatted in Turker Nation’s chatroom and decided to strike against the Requester. The work sat, unfinished, until finally the Requester gave in and increased the pay for each HIT, at which time they 21 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) were completed in a matter of hours. This sort of collective action typically happens in private as workers fear making such action obvious since any connection between their Turker Nation account and their mTurk account could lead to retaliation in the form of rejections or suspension from the platform itself. It is important to note that Turkers did not always agree on the next steps to take. Turker Nation’s ability to shape future understandings of collective action for digital labor is full of potential, yet within limits. Similar to how Amrute (2016) shows that Indian coders on short-term visas to Germany make themselves indispensable to companies both by writing complicated code and by playing up their global-ness, the ability for Turkers to call out an academic on what they determine to be a form of unethical research on mTurk displays their ability to put stakes down in the future of mTurk and to show that they have solidarities to be contended with. The limits of fighting for Turkers’ rights are circumscribed by the situation at hand: Turkers fear that if they demand changes from Amazon, mTurk may be shut down completely. As Martin et al. (2014) note: along with more formally organized forms of collective action comes increased surveillance and control. Turkers value being able to do their work and recognize mTurk as one of the few legitimate sources of flexible, work-from-home employment systems. Given this, the benefits of fighting for change are sometimes outweighed for workers by the ability to just keep working. Michelle Murphy (2006) in her work on "sick building syndrome," or the rise in chemicals and environmental pollutants making workers sick in newly built office buildings in the 1980s, shows how, in the process of creating a movement for workers, the movement ended up eliding differences, such as gender, race, and class, within that community. Put another way, once a sense of community has been built upon certain characteristics, it is hard to communicate to the public that the picture is more complicated. Murphy shows how certain characteristics of how the work looked —workers wore business dress and showed up for work at new, flashy office buildings —made it difficult to communicate to the 22 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) public about the state of working conditions, and how it affected certain communities in different ways. A similar argument can be made with regard to the ability of Turkers, as a community, to communicate effectively with the public about their working conditions and their wishes for future working conditions. Turkers can mobilize their working for a flashy employer, Amazon, and the freedom to work remotely to portray an outward appearance of a professional, white-collar job. These characteristics often point toward being part of the privileged knowledge-working class that comprises white-collar jobs and the American middle class. Kristy recalls a conversation with an Uber driver in her north American city about her work; he thought it sounded ideal, and was quite jealous, until she shared with him that at this job she usually earned four dollars an hour. Working from home on a computer causes dissonance, for the public, when compared to manufacturing work; at the same time, the divergent groups within the online Turker community have made it difficult to arrive on one clear wish for an outcome of collective action. While the online Turker communities draw workers together, there are multiple social dynamics at play that reduce their power and make it difficult to work together. Kristy reflects on how she saw this happen as a participant in the discussion boards from the beginning. In 2005, there were only two forums available, and many workers at the time were on both. Soon one closed and there was only one (for a while) where the community could meet and discuss mTurk. Internal fissures began and soon other forums began to pop up, often started by disenfranchised members of the forums which came before. For example, CrowdMeBaby.com grew out of Turker Nation, mTurkGrind.com grew out of mTurkForum.com, and mTurkCrowd.com and TurkerHub.com grew out of mTurkGrind. Many of these splits were acrimonious, resulting in members of different discussion forums refusing to converse with each other, which became a major barrier to organizing a large percentage of the community. For example, while a worker from one community might speak to a journalist about their experience, workers in other forums 23 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) posted online about how inaccurate that workers account was of the “real worker experience” (“Virtual Sweatshops Paint a Bleak Picture of the Future of Work” 2017). Conclusion Driscoll and Schwarz (2014) call for more scholarship to look at the discussion forum as an important part of the decentralized social web. Attention to the politics of infrastructure, particularly in light of quickly shifting new forms of economies, will add to our understandings of new forms of labor, class, and digital infrastructures. Anthropologists and social theorists have debated the effect of the post-Fordist era of labor as precarious and have written extensively on the ways in which new media shifts an individual’s relationship with labor, time, and value (Hardt, 2005; Virno & Hardt, 1996; Berardi, 2009; Muehlebach, 2011). Necessary to understanding and theorizing new forms of labor is examining supportive social spaces. Social spaces give insight into how workers are communicating, supporting each other in their endeavor of working, and understanding their role in the larger economy. Certain forms of visibility and community support allow workers to navigate an online workspace to maximize their profits, and at times, negotiate for better terms of work with employers. Engaging in this labor is working to redefine what the future of digital labor might look like. Acknowledgements Thank you to Mary L. Gray, Nancy Baym, Tarleton Gillespie, Jessa Lingel, Katrin Tiidenberg, Kevin Driscoll, Sareeta Amrute, Lily Shapiro, Darren Byler, Samantha Simon, Jiwoon Yu-Lee, Caleb Knapp, and the journal's anonymous reviewers. 24 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Notes 1 A Master's worker is a worker who attained the Master's qualification, an elusive badge doled out by Amazon. Gaining a Master's qualification is elusive and Turkers often speculate about what type of algorithm is used to determine the qualification. On the mTurk homepage, Amazon states that Master workers are determined “by consistently completing HITs of a certain type with a high degree of accuracy. Masters must continue to pass our statistical monitoring to remain Mechanical Turk Masters” (“Worker Web Site FAQ,” 2016). Masters have exclusive access to work as well as access to a private Masters-only forum. In terms of tips for receiving Master's status, Amazon makes clear that Workers cannot apply for status; rather they should aim to complete a high number of tasks accurately and in a timely manner with a wide range of Requesters. 2 It is largest by number of active users on the site, not by number of total users. 3 In the first case an academic Requester posted HITs from a business account and was not upfront about the fact that it was an academic study. The HITs began by showing positive images to be tagged to the worker, but quickly devolved to include images of gore, torture, and death. Then the HITs varied in pay, and the researchers' intent turned out to be to determine how much compensation would be necessary to get workers to do horrible tasks (Portner, C. C., Toomim, M., & Hassairi, N., 2013 [Testing the Theory of Equalizing Differences Using Online Labor Market Experiments, work in progress]). The researchers never debriefed participants, but the Turkers figured out what was going on within a few days and exposed the researchers on various forums. The second was also posted by an academic Requester but appeared to be multiple business accounts. In this case it was not the HITs which caused the uproar. Academics had created a website where the workers could rate the Requesters, a feature not built into the mTurk platform, called Turkopticon (Irani & Silberman, 2013). The researcher was interested in how these ratings affected the results Requesters would get on mTurk, so he created both fake Requester accounts on mTurk and fake worker accounts on Turkopticon to rate those Requester accounts to test the theory that bad ratings would lead to slower or lower quality work being submitted on mTurk (Benson, Sojourner & Umyarov, 2015). Again, the Turkers quickly figured out what was going on, and went so far as to contact the researcher's institution's ethics board to complain. 25 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) These two incidents, so recent and so deeply disturbing to Kristy, had created a great deal of resentment and distrust toward academics. 4 At the time of research, in August 2014, Amazon was defaulting to only accept Turkers with a 95% or above acceptance rate. We know this by logging in to set up a HIT. You had two options for presets for accepting Turkers, either 95% acceptance rate and above or accept all. As of June, 2015, this preset has changed and now it defaults to Turkers who hold Master's qualification. 5 There are few Turkers from India on Turker Nation. Part of the reason, we surmise, is that there is some resentment towards these workers as they are reputed to accept lower-paying work that the North American workers claim brings down everyone’s wages. Due to this, Indian workers have to work extra hard to make themselves legible to other "professional" Turkers on Turking discussion boards. This topic is not within the purview of this paper, but should be explored further. 6 wiki.wearedynamo.org 7 http://Turker Nation.com/showthread.php?22543-Petition-P9R-please remove-the-keyword-quot-survey-quot-from-your-mTurk- TRANSCRIPTION-HITs References AAA Web Admin, 2012. Ethics Statement. http://ethics.americananthro.org /category/statement/ Amazon Mechanical Turk. (2016, August 23). Retrieved August 23, 2016, from https://www.mturk.com/mturk/help?helpPage=worker#what_is_master_ worker Amazon. “Worker Website FAQs”. Retrieved from: https://www.mturk.com/ mturk/help?helpPage=worker Amazon. “Requester Tour” Retrieved from: https://requester.mturk.com/tour Amrute, S. (2014). Proprietary freedoms in an IT office: How Indian IT 26 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) workers negotiate code and cultural branding. SOCA Social Anthropology, 22(1), 101-117. Amrute, S. (2016). Encoding race, encoding class: Indian IT workers in Berlin. Durham, NC: Duke University Press. Aneesh, A. (2006). Virtual migration: The programming of globalization. Durham, NC: Duke University Press Books. Anthias, F., & Yuval-Davis, N. (1993). Racialized boundaries: Race, nation, gender, colour and class and the anti-racist struggle. London, GB: Routledge. Arthur, C. (2014, June 30). Facebook emotion study breached ethical guidelines, researchers say. The Guardian. Retrieved from: https://www.theguardian.com /technology/2014/jun/30/facebook- emotion-study, accessed June 21, 2016. Benson, A., Sojourner, A., & Umyarov, A. (2015). The value of employer reputation in the absence of contract enforcement: A randomized experiment). Berardi, F. (2009). The soul at work: from alienation to autonomy. Los Angeles, CA: Semiotext(e). Berinsky, A., Huber, G., & Lenz G. (2012). Evaluating online labor markets for experimental research: Amazon.com’s Mechanical Turk. Political Analysis, 20(3), 351–368. Bernard, H. (2011). Research methods in anthropology (5th ed.). Lanham, MD: AltaMira Press. Boellstorff, T. (2008). Coming of age in second life: An anthropologist explores the virtually human. Princeton, NJ: Princeton University Press. Coleman, G. (2013). Coding freedom: The ethics and aesthetics of hacking. Princeton, NJ: Princeton University Press. Craven, C. (2013). Feminist activist ethnography counterpoints to neoliberalism in North America. Lanham, MD: Lexington Books. Cushing, E. (2013, January/February). Amazon Mechanical Turk: The 27 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Digital Sweatshop. Utne. Retrieved from: http://www.utne.com/science- and-technology/amazon-mechanical-turk-zm0z13jfzlin.aspx Davis, D. (2013). "Border crossings: Intimacy and feminist activist ethnography in the age of neoliberalism." In C. Craven & D.A. Davis (Eds.), Feminist activist ethnography counterpoints to neoliberalism in North America, (pp.23-38). Lanham, MD: Lexington Books. Dowling, E., Nunes R., & Trott B. (2007). Immaterial and affective labour: explored. Ephemera, 7(1), 1–7. Driscoll, K., & Schwartz, L. (2014). “'I hate your politics but I love your diamonds': The Web-based interest-driven messageboard as DIY infrastructure." In M. Ratto & M. Boler (Eds.), DIY citizenship: Critical making and social media. Cambridge, MA: MIT Press. Durrenberger, E. (2007). The anthropology of organized labor in the United States. Annual Review of Anthropology, 36(1), 73–88. Fort, K., Adda G., & Cohen K. (2011). Amazon Mechanical Turk: Gold mine or coal mine? Computational Linguistics, 37(2), 413–420. Gehl, R. (2014). Reverse engineering social media: Software, culture, and political economy in new media capitalism. Philadelphia, PA: Temple University Press. Gray, M. (2014, July 8). When science, customer service, and human subjects research collide. Now what? (Web log comment). Retrieved from: http://marylgray.org/?p=288. Gregg, M. (2011). Work’s Intimacy. Cambridge, GB; Malden, MA: Polity. Gusterson, H. (1998). Nuclear rites: A weapons laboratory at the end of the Cold War. Berkeley, CA: University of California Press. Hardt, M. (2005). Immaterial labor and artistic production. Rethinking Marxism, 17(2), 175–177. Harvey, D. (2010). A companion to Marx’s Capital. Brooklyn, NY: Verso. Ho, K., Bear L., Tsing A., & Yanagisako S. (2015, March 30). "Generating capitalism." Theorizing the contemporary. Cultural Anthropology website. 28 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Retrieved from: http://www.culanth.org/fieldsights/650-generating- capitalism. Howe, J. (2009). Crowdsourcing: Why the power of the crowd is driving the future of business. New York, NY: Three Rivers Press. Ipeirotis, P. (2010). Demographics of Mechanical Turk. Working paper. CeDER Working Papers. New York, NY, USA. Retrieved from: http://archive.nyu.edu/handle/2451/29585. Irani, L. (2013). The Cultural Work of Microwork. New Media & Society: 17 (5), 720-739. Irani, L & Silberman, M. (2013). Turkopticon: Interrupting worker invisibility in Amazon Mechanical Turk. Proceedings of CHI 2013, Apri 28-May 2, 2013. Kramer, A., Guillory J., & Hancock, J. (2014). Experimental evidence of massive-scale emotional contagion through social networks. Proceedings of the National Academy of Sciences of the United States of America, 111(24), 8788–90. Lazzarato, M. (1996). "Immaterial labor." In P. Virno & M. Hardt (Eds.), Radical thought in Italy: A potential politics (pp. 133–150). Minneapolis, MN: University of Minnesota Press. Lichtenstein, N. (2013). State of the Union: A Century of American Labor. Princeton, N J: Princeton University Press. Markham, A. & Baym, N. (Eds.). (2008). Internet inquiry: Conversations about method. Los Angeles, CA: SAGE Publications, Inc. Marx, K. (1993). Capital: A critique of political economy. Reprint edition. London: Penguin Classics. (Original work published 1867) Mason, W. & Suri. S. (2012). Conducting behavioral research on Amazon’s Mechanical Turk. Behavior Research Methods, 44(1), 1-23. Melossi, D. (2008). Controlling crime, controlling society: Thinking about crime in Europe and America. Cambridge, GB: Polity. Muehlebach, A. (2011). On affective labor in post-Fordist Italy. Cultural 29 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Anthropology, 26(1), 59–82. Murphy, M. (2006). Sick building syndrome and the problem of uncertainty: Environmental politics, technoscience, and women workers. Durham, NC: Duke University Press. Paolacci, G. & Chandler, J. (2014). Inside the Turk: Understanding Mechanical Turk as a participant pool. Current Directions in Psychological Science, 23(3), 184–188. Papacharissi, Z. (2010). A private sphere: Democracy in a digital age. Cambridge, GB; Malden, MA: Polity Press. Robertson, S. (2008). “'Remaking the World': Neoliberalism and the transformation of education and teachers’ labor." In L. Weiner & M. Compton (Eds.), The global assault on teaching, teachers and their unions, (pp. 11–30). New York, NY: Palgrave Macmillan. Srnicek, N. (2016). Platform capitalism. Cambridge, GB ; Malden, MA: Polity. Stacey, J. (1988). Can there be a feminist ethnography? WSIF Women’s Studies International Forum,11(1), 21–27. Star, S. (1999). The ethnography of infrastructure. American Behavioral Scientist, 43(3), 377–391. Taylor, F. (1914). The principles of scientific management. New York, NY: Harper. Thompson, E. (1967). Time, work-discipline, and industrial capitalism. Past & Present, 38, 56–97. Virno, P. & Hardt, M. (1996). Radical thought in Italy: A potential politics. Minneapolis, MN: University of Minnesota Press. “Virtual Sweatshops Paint a Bleak Picture of the Future of Work.” (2017). Www.reddit.com/R/Mturk.https://www.reddit.com/r/mturk/comments/63k vrt/virtual_sweatshops_paint_a_bleak_picture_of_the/. Visweswaran, K. (1994). Fictions of feminist ethnography. Minneapolis, MN: University of Minnesota Press. 30 Zyskowski and Milland Catalyst: Feminism, Theory, Technoscience 4(2) Waters, F., & Woodcock, J. (2017, September 20). Far from seamless: A workers’ inquiry at Deliveroo. Viewpoint Magazine. Retrieved from https://www. viewpointmag .com/2017/09/20/far-seamless-workers- inquiry-deliveroo/ Writers, S., & Nagar, R. (2006). Playing with fire: Feminist thought and activism through seven lives in India. Minneapolis, MN: University of Minnesota Press. Weeks, K. (2011). The problem with work: Feminism, marxism, antiwork politics, and postwork imaginaries. Durham, NC: Duke University Press Books. Writers, S. & Nagar, R. (2006). Playing with fire: Feminist thought and activism through aeven lives in India. Minneapolis, MN: University Of Minnesota Press. Zittrain, J. (2009). Jonathan Zittrain at the internet as playground and factory [Video]. Retrieved from https://vimeo.com/8203737) Bios Kathryn Zyskowski is a Ph.D candidate in sociocultural anthropology at the University of Washington. Kristy Milland recently completed her master’s degree in Labour Studies at McMaster University and is starting a J.D. at the University of Toronto; she is also a Turker, community manager of Turkernation.com, and a labour activist.