--- schema_version: "library-reader-v1" source_type: "original_book" title: "If...Then Algorithmic Power and Politics" author: null edition: null publisher: null year: null isbn: "2017054909" library_document_id: "1d3bd28e-c1ba-4af2-8580-60b32ef26053" source_sha256: "9d6111c11832b55e134d36eb9ec970d190a989ee71c78f6d87e076c351752a78" pdf_physical_page_count: 211 pdf_physical_page_index_base: 0 range_start: 10 range_count: 10 html_url: "https://workshop.closeclaw.top/api/ai/read/1d3bd28e-c1ba-4af2-8580-60b32ef26053?start=10&limit=10" markdown_url: "https://workshop.closeclaw.top/api/ai/read/1d3bd28e-c1ba-4af2-8580-60b32ef26053.md?start=10&limit=10" previous_url: "https://workshop.closeclaw.top/api/ai/read/1d3bd28e-c1ba-4af2-8580-60b32ef26053.md?start=0&limit=10" next_url: "https://workshop.closeclaw.top/api/ai/read/1d3bd28e-c1ba-4af2-8580-60b32ef26053.md?start=20&limit=10" --- # If...Then Algorithmic Power and Politics > 页码规则：PDF_PHYSICAL_PAGE_INDEX 是从 0 开始的权威文件物理页索引；PRINTED_PAGE 是页面中识别出的印刷页码，可能为空，二者不得混用。 ## PDF_PHYSICAL_PAGE_INDEX=10 | PRINTED_PAGE=NOT_DETECTED ## PDF_PHYSICAL_PAGE_INDEX=11 | PRINTED_PAGE=1 Introduction Programmed Sociality Let us for a moment consider the following scenario: Copenhagen on a rainy November day. The semester is coming to a close. Students and professors alike are stressed, but it is nonetheless a sociable time of the year. Conference trips, essays to be graded, dinner plans, and Christmas presents in the pipeline. In other words: buying plane tickets, surfing the Web for restaurant tips, streaming music, coordinating dinner plans with friends, shopping for Christmas gifts, preparing lectures, watching a movie after work. Besides describing a random day in an academic life, all these activities imply using the Internet. I found some cheap tickets using a metasearch engine, discovered a nice restaurant through a collaborative filtering system, listened to a music playlist suggested to me by a music-streaming site, bought a gift for my mother at an online designer store, chated with friends on a social-networking site to find a suitable time for a pre-Christmas dinner, and, finally, watched an episode of my favorite TV series on a movie-streaming site. It is not just a story about the good life of a Scandinavian academic. It suggests a life deeply entwined with media. Following media theorist Mark Deuze (2012), we might say that life is not only lived in and through media but in and through specific types of media. What these activities have in common is a high degree of interaction with algorithmic media, media whose core function depends on algorithmic operations. This book starts from the premise that life is not merely infused with media but increasingly takes place in and through an algorithmic media landscape. Key to this premise is the notion of the co-production of social life, practices, and technology. While people interact with specific media companies and platforms, these platforms interact with people as well. Users do not simply consult websites or talk to their friends online. Social media and other commercial Web companies recommend, suggest, and provide users with what their algorithms have predicted to be the most relevant, hot, or interesting news, books, or movies to watch, buy, and consume. Platforms act as performative intermediaries that participate in shaping the worlds they only purport to represent. Facebook is not simply a social networking site that lets users “connect with friends and the world around you.”1 As media scholar José van Dijck has argued, “Social media are inevitably automated systems that engineer and manipulate connections” (2013: 12). By the same token, Netflix is not a website that lets users “see what’s next and watch anytime, cancel anytime.”2 It can’t be seen as a neutral platform that merely queries its vast database about a user’s request to show the movies they explicitly want to watch. Relying on vast amounts of data, Netflix algorithms are used to analyze paterns in people’s taste, to recommend more of the same. Popularity is not only a quantifiable measure that helps companies such as Facebook and Netflix to determine relevant content. User input and the paterns emerging from it are turned into a means of production. What we see is no longer what we get. What we get is what we did and that is what we see. In the case that Netflix suggests we watch House of Cards, it is largely a mater of consumers getting back their own processed data. When the show was released in 2013 it quickly became a cornerstone for data-driven programming, the idea that successful business decisions are driven by big data analytics. 1 1 ## PDF_PHYSICAL_PAGE_INDEX=12 | PRINTED_PAGE=2 Of course, there is nothing inherently wrong with this form of data-driven media production. After all, it seems, many people enjoy watching the show. The interesting and potentially troubling question is how reliance on data and predictive analytics might funnel cultural production in particular directions, how individual social media platforms code and brand specific niches of everyday life (van Dijck, 2013: 22). Starting from the basic question of how software is shaping the conditions of everyday life, this book sets out to explore the contours and implications of the question itself. In what ways can we say that software, and more specifically algorithms, shape everyday life and networked communication? What indeed, are algorithms and why should we care about their possible shaping efects to begin with? Let us quickly return to my rainy November day. While chating with friends on Facebook about the pre-Christmas dinner, two rather specific ads appeared on the right-hand side of my Facebook news feed. One was for a hotel in Lisbon, where I was going to travel for the conference, and the other was for a party dress. How did Facebook know about my upcoming trip, or that I had just bought my mother a Christmas gift from that shop? My musings were only briefly interrupted by one of my friends asking me about a concert I had recently been to. She had seen a picture I posted on Facebook from the concert a few days earlier. My other friend wondered, why hadn’t she seen the picture? After all, as she remarked, she checks her Facebook feed all the time. While these connections might be coincidental, their efects are not incidental. They mater because they afect our encounters with the world and how we relate to each other. While seeing ads for party dresses in a festive season might not appear strange, nor missing a picture posted from a concert for that mater, such programmed forms of sociality are not inconsequential. These moments are mediated, augmented, produced, and governed by networked systems powered by software and algorithms. Understood as the coded instructions that a computer needs to follow to perform a given task, algorithms are deployed to make decisions, to sort and make meaningfully visible the vast amount of data produced and available on the Web. Viewed together, these moments tell the story of how our  lives are networked and connected. They hint at the fundamental question of who or what has power to set the conditions for what can be seen and known with whatever possible efects. To address this important question, this book proposes to consider the power and politics of software and algorithms that condense and construct the conditions for the intelligible and sensible in our current media environment. 2 I F . . . T H E N ## PDF_PHYSICAL_PAGE_INDEX=13 | PRINTED_PAGE=3 The ideas of power and politics I have in mind are both very broad, yet quite specific. For one, this book is not going to argue that algorithms have power. Sure, algorithms are powerful, but the ways in which this statement holds true cannot simply be understood by looking at the coded instructions telling the machine what to do. Drawing on the French philosopher Michel Foucault’s (1982; 1977) understanding of power as exercised, relational and productive, I intend to show how the notion of “algorithmic power” implies much more than the specific algorithm ranking e.g. a news feed. What I am going to argue is that the notion of algorithmic power may not even be about the algorithm, in the more technical sense of the term. Power always takes on many forms, including not only the ways in which it is exercised through computable instructions, but also through the claims made over algorithms. As such, we might say that algorithmic systems embody an ensemble of strategies, where power is immanent to the field of action and situation in question. Furthermore, following Foucault, power helps to produce certain forms of acting and knowing, ultimately pointing to the need for examining power through the kinds of encounters and orientations algorithmic systems seem to be generative of. Neither are the “politics” of this book about politics with a capital P. I will not be discussing parliamentary politics, elections, campaigns, or political communication in the strictest sense. Rather, politics is understood in more general terms, as ways of world-making—the practices and capacities entailed in ordering and arranging different ways of being in the world. Drawing on insights from Science and Technology Studies (STS), politics here is more about the making of certain realities than taking reality for granted (Mol, 2002; Moser, 2008; Law, 2002). In chapter 2 I will describe this form of politics of the real, of what gets to be in the world in terms of an “ontological politics” (Mol, 2002). In ranking, classifying, sorting, predicting, and processing data, algorithms are political in the sense that they help to make the world appear in certain ways rather than others. Speaking of algorithmic politics in this sense, then, refers to the idea that realities are never given but brought into being and actualized in and through algorithmic systems. In analyzing power and politics, we need to be atentive of the way in which some realities are always strengthened while others are weakened, and to recognize the vital role of nonhumans in co-creating these ways of being in the world. If . . . Then argues that algorithmic power and politics is neither about algorithms determining how the social world is fabricated nor about what algorithms do per se. Rather it is about how and when diferent aspects of algorithms and the algorithmic become available to specific actors, under what circumstance, and who or what gets to be part of how algorithms are defined. Programmed Sociality 3 ## PDF_PHYSICAL_PAGE_INDEX=14 | PRINTED_PAGE=4 Programmed Sociality Increasingly, we have come to rely on algorithms as programmable decision-makers to manage, curate, and organize the massive amounts of information and data available on the Web and to do so in a meaningful way. Yet, the nature and implications of such arrangements are far from clear. What exactly is it that algorithms “do” and what are the constitutive conditions necessary for them to do what they do? How are algorithms enlisted as part of situated practices, and how do they operate in different setings? How can we develop a productive and critical inquiry of algorithms without reducing it to a question of humans versus the machine? Let’s begin a tentative answer with a conceptual understanding of how software induces, augments, supports, and produces sociality. Here, I suggest the concept of programmed sociality as a helpful heuristic device. Through this we might study algorithmic power and politics as emerging through the specific programmed arrangements of social media platforms, and the activities that are allowed to take place within those arrangements. Facebook and other software systems support and shape sociality in ways that are specific to the architecture and material substrate of the medium in question. To do justice to the concept of programmed sociality, it is important to highlight that it does not lead us down a pathway of technological determinism. In using the term “programmed,” I draw on computer scientist John von Neumann’s notion of “program,” for which the term “to program” means to “assemble” and to “organize” (Grier, 1996: 52). This is crucial, as it frames software and algorithms as dynamic and performative rather than as fixed and static entities. Regarding “sociality,” I refer to the concept of how diferent actors belong together and relate to each other. That is, sociality implies the ways in which entities (both human and non-human) are associated and gathered together, enabling interaction between the entities concerned (Latour, 2005). To be concerned with programmed sociality is to be interested in how actors are articulated in and through computational means of assembling and organizing, which always already embody certain norms and values about the social world. To exemplify how algorithmic media prescribe certain norms, values, and practices, let me describe how programmed sociality plays out in the specific context of Facebook, by focusing on friendships as a particularly pertinent form of being together online. As Facebook has become an integral part of everyday life, providing a venue for friendships to unfold and be maintained, it is easy to forget just how involved Facebook is in what we often just take to be interpersonal relationships. Everything from seting up a profile and connecting with other users to maintaining a network of friends entails an intimate relation with the software underlying the platform itself. As van Dijck has pointed out, “what is important to understand about social network sites is how they activate relational impulses” (2012: 161). It is important to understand that relationships are activated online, but also how and when they are activated: by whom, for what purpose, and according to which mechanisms. 4 I F . . . T H E N ## PDF_PHYSICAL_PAGE_INDEX=15 | PRINTED_PAGE=5 With nearly two billion users, many of whom have been members of the platform for many years, most people have long forgoten what it felt like to become a member, how they became the friend that Facebook wanted them to be. Upon first registering with the site, the user is instantly faced with the imperative to add friends. Once a user chooses to set up an account, he is immediately prompted to start filling in the personal profile template. Users’ identities need to be defined within a fixed set of standards to be compatible with the algorithmic logic driving social software systems. If users could freely choose what they wish to say about themselves, there would be no real comparable or compatible data for the algorithms to process and work with. Without this orderly existence as part of the databases, our connections would not make much sense. After all, “data structures and algorithms are two halves of the ontology of the world according to a computer” (Manovich,  2001: 84). Being part of databases means more than simply belonging to a collection of data. It means being part of an ordered space, encoded according to a common scheme (Dourish, 2014). As Tarleton Gillespie (2014) points out, data always need to be readied before an algorithm can process them. Categorization is a powerful mechanism in making data algorithm-ready. “What the categories are, what belongs in a category, and who decides how to implement these categories in practice, are all powerful assertions about how things are and are supposed to be” (Bowker and Star, in Gillespie, 2014: 171). The template provided by Facebook upon signing in constitutes only one of many forms of categorization that help make the data algorithm-ready. The politics of categorization becomes most pertinent in questions concerning inclusion and exclusion. The recurring conflicts over breastfeeding images and Facebook’s nudity-detection systems—comprising both algorithms and human managers—represent a particularly long-lasting debate over censorship and platform policies (Arthur, 2012). The politics of categorization is not just a mater of restricting breastfeeding images but one that fundamentally links database architecture and algorithmic operations to subjectification. To understand how sociality is programmed—that is, how friendships are programmatically organized and shaped, let us consider the ways in which the platform simulates existing notions of friendship. As theorists of friendship have argued, shared activity and history are important aspects of considering someone a friend (Helm, 2010). Simulating and augmenting the notion of a shared history, Facebook provides several tools and techniques dedicated to supporting memory. As poorly connected or unengaged users pose a threat to the platform’s conditions of existence, programming reasons for engagement constitutes a key rationale from the point of view of platforms. On Facebook, connecting users to potential friends provides the first step in ensuring a loyal user base, because friendships commit. Functioning as a memory device, algorithms and software features do not merely help users find friends from their past, they play an important part in maintaining and cultivating friendships, once formed. As such, a variety of features prompt users to take certain relational actions, the most well-known being that of notifying users about a friend’s birthday. While simulating the gesture of phatic communication represented in congratulating someone on his or her birthday, the birthdayreminder feature comes with an added benefit: The birthday feature is the most basic way of making users return to the platform, by providing a concrete suggestion for a communicative action to be performed. As I’ve described elsewhere, platforms like Facebook want users to feel invested in their relationships, so they are continually coming up with new features and functionalities that remind them of their social “obligations” as friends (Bucher, 2013). Programmed Sociality 5 ## PDF_PHYSICAL_PAGE_INDEX=16 | PRINTED_PAGE=6 While the traditional notion of friendship highlights the voluntary and durational aspects of becoming friends and becoming friends anew (Allan, 1989), the software, one may claim, functions as a suggestive force encouraging users to connect and engage with the people in ways that are aforded by and benefit the platform. From the point of view of critical political economy, sociality and connectivity are resources that fuel the development of new business models. Platforms do not activate relational impulses in an efort to be nice. Ultimately, someone benefits financially from users’ online activities. This is, of course, a familiar story and one that many scholars have already told in illuminating and engaging ways (see Andrejevic, 2013; Couldry, 2012; Fuchs, 2012; Gehl, 2014; Mansell, 2012; van Dijck, 2013). From the perspective of companies like Facebook and Google, but also from the perspective of legacy news media organizations (discussed in chapter 6), algorithms are ultimately folded into promises of profit and business models. In this sense, a “good” and well-functioning algorithm is one that creates value, one that makes beter and more eficient predictions, and one that ultimately makes people engage and return to the platform or news site time and again. The question then becomes: What are the ways in which a platform sparks enough curiosity, desire, and interest in users for them to return? The subtle ways of software can, for example, be seen in the manner in which Facebook reminds and (re)introduces users to each other. When browsing through my Facebook news feed it is almost as if the algorithm is saying, “Someone you haven’t spoken to in five years just liked this,” or, “This person whom you haven’t heard from in ages, suddenly seems to be up to something fun.” Somehow, I am nudged into thinking that these updates are important, that I should pay atention to them, that they are newsworthy. Rather than meaning that friendships on Facebook are less than voluntary, my claim is that the ways in which we relate to each other as “friends” is highly mediated and conditioned by algorithmic systems. People we do not necessarily think about, people we might not remember, or people we might not even consider friends continue to show up on our personalized news feeds, as friend suggestions, in birthday reminders, and so forth. While it is often dificult to recognize how Facebook “actively steers and curates connections” (van Dijck, 2013: 12), moments of everyday connectivity provide a glimpse into the ways in which the efects of software might not necessarily feel as incidental as they might appear to be. 6 I F . . . T H E N ## PDF_PHYSICAL_PAGE_INDEX=17 | PRINTED_PAGE=7 The programmed sociality apparent in Facebook is about more than making people remember friends from their past with the help of friend-finding algorithms. From the very beginning, a friendship is formed and the friendship is “put to test.” From a political economic standpoint, this is very important, as not every friendship is equally valuable. Some relations are more “promising” and “worthwhile” than others. With more friends than the news feed feature allows for, friendships are continuously monitored for afinity and activity, which are important measures for rendering relations visible. Friendships in social media are programmed forms of sociality precisely because they are continuously measured, valued, and examined according to some underlying criteria or logic. As this book will show in more detail, the news feed algorithm plays a powerful role in producing the conditions for the intelligible and sensible, operating to make certain users more visible at the expense of others. The Facebook news feed displays an edited view of what one’s friends are up to, in an order of calculated importance with the most important updates at top of the feed. Every action and interaction connected to Facebook, be it a status update, comment on someone’s photo, or “like buton” clicked, may become a story on someone’s news feed. Not every action, however, is of equal importance, nor is every friend for that mater. Friendships are put to test, because friendships need to be nurtured and maintained to stand the “test of time.” Algorithms decide which stories should show up on users’ news feeds, but also, crucially, which friends. Rachel, a 24-year-old journalist from New York City, whom I interviewed for this book about her perceptions of algorithms, exclaimed that “Facebook ruins friendships.” With more than seven hundred friends, Rachel says that she is constantly taken aback by all the information and people Facebook seems to be hiding from her feed: So, in that sense, it does feel as if there is only a select group of friends I interact with on the social network, while I’ve practically forgoten about the hundreds of others I have on there. An example of this is a friend from high school, who liked one of my posts a few weeks back. I’d totally forgoten she was even on Facebook until she liked it and we started chating. Rachel’s experience is reminiscent of what I mean by programmed sociality, the notion that social formations and connections are algorithmically conditioned and governed by the sociotechnical and political-economic configurations of specific media platforms. Rachel’s worry about Facebook ruining friendship should also remind us that algorithms need to be understood as powerful gatekeepers, playing an important role in deciding who gets to be seen and heard and whose voices are considered less important. Programmed sociality, then, is political in the sense that it is ordered, governed, and shaped in and though software and algorithms. If we want to consider everyday life in the algorithmic media landscape, we need to pay atention to the ways in which many of the things we think of as societal—including friendship—may be expressed, mediated and shaped in technological designs and how these designs, in turn, shape our social values. As we will see throughout the book, such considerations, however, do not stop with values in design, but exceed the purely technical (whatever that is taken to mean) in important ways. Programmed Sociality 7 ## PDF_PHYSICAL_PAGE_INDEX=18 | PRINTED_PAGE=8 A key argument of this book is that the power and politics of algorithms stems from how algorithmic systems shape people’s encounters and orientations in the world. At the same time, I claim that this shaping power cannot be reduced to code. Specifically, I argue for an understanding of algorithmic power that hinges on the principle of relational materialism, the idea that algorithms “are no mere props for performance but parts and parcel of hybrid assemblages endowed with difused personhood and relational agency” (Vannini,  2015: 5). Thus, it is important to acknowledge that while we start with the question of how software and algorithms shape sociality by looking at materiality in the more conventional sense as “properties of a technology,” the answer cannot be found in these properties alone, but rather the ways in which programmed sociality is realized as a function of code, people, and context. Computable Friendships The concept of friendship provides an apt example for the understanding of programmed sociality and algorithmic life, because it shows the discrepancies between our common-sense notions of friendship and the ways in which friendship becomes embedded in and modeled by the algorithmic infrastructures. Friendships are deeply rooted in the human condition as a fundamental aspect of being together with other people, and which is always already contested based on cultural and historical contexts. Traditionally, friendship has been thought of as an exclusive social relation, a private and intimate relation between two persons (Aristotle,  2002; Derrida,  2005; Hays, 1988). For this reason, true friendship has been regarded as  something that one cannot have with many people at the same time, simply because it requires time to build, nurture, and maintain. Compared to Aristotle’s conception of friendship as something rather precious that one cannot have with many people at once (Aristotle, 2002), Facebook seems to promote the completely opposite idea. The way the platform puts friendships at the center of a business model is no coincidence, of course, and is probably one of the core reasons Facebook has evolved into an unprecedented media company during the past decade. In a patent application filed by Facebook concerning the People You May Know (PYMK) feature, no doubt is left as to the value of friendships for Facebook: “Social networking systems value user connections because beter-connected users tend to increase their use of the social networking system, thus increasing user-engagement and corresponding increase in, for example, advertising opportunities” (Schultz et al., 2014). Software intervenes in friendships by suggesting, augmenting, or encouraging certain actions or relational impulses. Furthermore, software is already implicated in the ways in which the platform imagines and performs friendships. Contrary to the notion that “friendship clearly exists as a relation between individuals” (Webb, 2003: 138), friendship on Facebook exists as a relation between multiple actors, between humans and non-humans alike. As Facebook exemplifies in another patent document: 8 I F . . . T H E N ## PDF_PHYSICAL_PAGE_INDEX=19 | PRINTED_PAGE=9 [T]he term friend need not require that members actually be friends in real life (which would generally be the case when one of the members is a business or other entity); it simply implies a connection in the social network. (Kendall and Zhou, 2010: 2) The disconnect between how members usually understand friendship and the ways in which Facebook “understands” friendship becomes obvious in the quote above. According to Facebook, a user can be “friends” with a Facebook page, a song, a movie, a business, and so on. While it might seem strange to consider a movie a friend, this conception of friendship derives from the network model of the Web in which users and movies are considered “nodes” in the network and the relationship that exists between them an “edge” or, indeed, a friend. Indeed, “the terms ‘user’ and ‘friend’ depend on the frame of reference” (Chen et al. 2014). It is exactly the diferent and sometimes conflicting frames of reference that are of interest in this book. A core contention is that media platforms and their underlying software and infrastructures contain an important frame of reference for understanding sociality and connectivity today. If we accept that software can have a frame of reference, a way of seeing and organizing the world, then what does it mean to be a friend on Facebook or, more precisely, what are friends for, if seen from the perspective of the platform? Facebook friendships are, above all, computable. In an age of algorithmic media, the term algorithmic, used as an adjective, suggests that even friendships are now subject to “mechanisms that introduce and privilege quantification, proceduralization, and automation” (Gillespie, 2016a: 27). Measuring the performance of individuals and organizations is nothing new, though. As sociologists Espeland and Sauder (2007) suggest, social measurements and rankings have become a key driver for modern societies during the past couple of decades. According to philosopher Ian Hacking, “society became statistical” through the “enumeration of people and their habits” (1990: 1). Hacking connects the emergence of a statistical society to the idea of “making up people,” meaning that classifications used to describe people influence the forms of experience that are possible for them, but also how the efects on people, in turn, change the classifications: Programmed Sociality 9 下一段：[继续读取](https://workshop.closeclaw.top/api/ai/read/1d3bd28e-c1ba-4af2-8580-60b32ef26053.md?start=20&limit=10)