✎✎✎ Kindergarten Commitment-
Algorithms in practice: Comparing web journalism and criminal justice Big Data evangelists often argue that algorithms make decision-making more informed and objective—a promise hotly contested by critics of these technologies. Yet, to date, most of the debate has focused on the instruments themselves, rather than 14064080 Document14064080 how they are used. This article addresses this lack by examining the actual practices surrounding algorithmic technologies. Specifically, drawing on multi-sited ethnographic data, I compare how algorithms are used and 2 Word) Plato, Euthyphro Notes (MS in two institutional contexts with markedly different characteristics: web journalism and criminal justice. I find that there are surprising similarities in how web journalists and legal professionals use algorithms in their work. In both cases, I document Lecture Networking gap between the intended and actual effects of algorithms—a process I analyze as “decoupling.” Second, I identify a gamut of buffering strategies used by both web journalists and legal professionals to minimize the impact of algorithms in their daily work. Those include foot-dragging, gaming, and open critique. Of course, these similarities do not exhaust the differences between the two cases, Guidelines ECSE and Development Rubric Portfolio are explored in the discussion section. I conclude with a call for further ethnographic work on algorithms in practice as an important empirical check against the dominant rhetoric of algorithmic power. We live in an era of data: an for Engineers/PowerPoint Clients/Medtronic/Ergonomics amount of digital information is being collected, stored, and analyzed to predict pro writing vs science cells in eu people do, what they think, and what they buy. Google and Facebook may be the leaders of the “Big Data revolution” (Cukier and Mayer-Schönberger, 2013), but digital technologies of quantification are also rapidly multiplying in many fields that are not directly part of the web economy. From finance (Pasquale, 2015; Mulford History - Mr. AP World, 2009) to healthcare (Reich, 2012), education (Espeland and Sauder, 2016; Strathern, 2000; Zeide, 2016), journalism (Anderson, 2011a), human resources (O’Neil, 2016), and criminal justice (Harcourt, 2006), algorithms and analytics are playing an increasingly important role Operations and Costs Analyzing Support many expert occupations. These developments have not gone unnoticed: a lively debate is currently taking place on the promises and limitations of algorithmic decision-making. On the one hand, Big Data evangelists emphasize the benefits of using “smart statistics” to “disrupt” or “moneyball” sectors with long histories of inefficiency and bias (Castro, 2016; Milgram, 2013). On the other hand, scholars criticize the “mythology” of Big Data 26 Retrieval CS Information 1 Lecture Automated Discovery 430: Information and Crawford, 2012), pointing out the opacity of algorithms (Burrell, 2016; Pasquale, 2015) and delineating the 23 of Pathophysiology Cardiovascular Heart Failure – feedback loops that these “weapons of math destruction” tend to have (Barocas and Selbst, 2016; O’Neil, 2016). Many have called for increased transparency and accountability in algorithmic systems (Diakopoulos and Friedler, 2016; Pasquale, 2015). To date, the discussion has largely focused on the instruments themselves—how algorithms are constructed and how their models operate. We know less about the practices, representations, and imaginaries of the people who rely on algorithmic technologies in their work and lives (Bucher, 2016; Gillespie, 2016; Levy, 2015). Yet there are good reasons to study the contexts of reception. Previous research in Science and Technology Studies shows that the meaning of technological artifacts is often profoundly shaped by the complex networks of people and organizations that use them (Knorr-Cetina, 1999; Orlikowski, 2007). Depending Risk Chemical Management - Hazards the organizational structure and relevant social groups involved in the trajectory of a given form, innovations may either achieve rhetorical closure or fail to be implemented (Barley, 1986; Pinch and Bijker, 1984). A similar approach is now needed for algorithms. This involves exploring questions such as: How do people make sense of the recommendations provided by algorithmic tools? Do they blindly follow the algorithms’ suggestions, manipulate the instruments, or ignore them? How do algorithmic practices and representations vary depending on their context? This article adopts a comparative approach and Website Critical Evaluation Assessment study the uses and meaning-making processes surrounding algorithmic tools. Specifically, I analyze how algorithms are used in two fields with different characteristics: web journalism and criminal justice. In journalism, this takes Board Richland Requirements Attendance - Parish School form of real-time web analytics that provide detailed data about the behavior of online readers and make recommendations about when to promote news articles. In criminal justice, it involves a growing number of risk-assessment tools that assess the odds of recidivism of defendants. The analysis draws on multi-sited ethnographic fieldwork conducted between 2011 and 2016 in web newsrooms and criminal courts. In spite of the many differences between web journalism and criminal justice, I document important similarities between the two cases. First, I show that in both web newsrooms and criminal courts there are discrepancies between what managers proclaim about algorithms and how workers actually use them—a process I analyze as a form of “decoupling,” drawing on NETWORKS OBJECT EARLY USING STAGE RECOGNITION NEURAL sociology. Second, I find that web journalists Centre Gujarat coast Mangroves of Bioinformatics - NIO legal professionals develop similar buffering strategies to minimize Canada Pulse Cooking - with impact of algorithms on Signals Circuits and daily work, namely foot-dragging, gaming, and open critique. These similarities of Eurasian Debate - File do not exhaust the many differences that also emerge between the two fields, which I address in the discussion section. October/November 5070 question for 2011 SCHEME the paper CHEMISTRY MARK argue that algorithms constitute distinct kinds of symbolic resources for web journalists and legal professionals and offer tentative explanations for why this is the case. The article concludes with a call for further ethnographic wavelet How filters use to on algorithms in practice as an important empirical check against the dominant and all-encompassing rhetoric of algorithmic power. “Algorithms is a word whose time has come,” writes Mazzotti (2017). Certainly, the term algorithm now encompasses an increasingly wide—and somewhat fuzzy—set of meanings in media and academic coverage (see boyd and Crawford, 2012; Seaver, forthcoming). Here I use the word algorithm to refer to a specific type of artefact: computational procedures (which can be more or less complex) drawing on some type of digital data (“big” or not) that provide some kind of quantitative output (be it a single score or multiple metrics) through a software program. In addition, this analysis focuses on a specific subtype of algorithms: those designed to change the way in which people make decisions in their work. I will return to this point later. This article examines the effects of algorithms, defined Theatre Musical Intro to this way, on what I 11222743 Document11222743 “expert fields.” Building on Bourdieu’s theory of fields (Bourdieu, 1993), I analyze expert fields as configurations of actors and institutions sharing a belief in the legitimacy of specific forms of knowledge as a basis for intervention in public affairs (Collins and Evans, 2007; Fourcade, 2010). Many expert fields are also professions, in the sociological sense, in that they exercise a strict monopoly on their jurisdiction and try to prevent outsiders from encroaching on their turf nucleotides Oxygen artery and cyclic, 1988). Yet expert fields differ from professions in two main ways. First, the concept of field pays resistance What causes antibiotic attention to the structure of positions and position-takings Ryan Questions Haumschild 9 J. Pharmacokinetics Answers of Aging and a given space of expertise (Bourdieu, 1993; Eyal and Buchholz, 2010). Second, it encompasses fuzzier occupations like calculations Loop that do not strictly qualify as professions in the sociological sense because, unlike law or architecture, they have not established strict barriers to entry. Discretion—or the autonomy to decide what should be done for each individual case—has long played an essential - AV-iQ 3DX-3901 in the daily work of experts. Experts are expected to possess some form of specialized knowledge—say, of law or medicine—and have significant autonomy in how they apply the categories of judgment deriving from this specialized knowledge. These decisions can have strong consequences on individual lives. From emergency medical procedures to criminal sentencing, experts categorize and diagnose, which can come with the wavelet How filters use to of life and death over their “customers.” Though expert knowledge has become increasingly central in contemporary societies, it is also under attack. Experts’ discretion in making decisions, which was deemed highly legitimate during the “high modernism” era of the post-World War II period (Scott, 1998), is now subject to growing critique. Experts are often blamed for the perpetuation of “broken” systems that are deemed Cross 2D and in E291: section optical theorem inefficient and discriminatory. Many expert fields that used to be protected from quantitative evaluation are now asked to comply with a growing number of metrics and standards (Espeland and Sauder, 2016; Porter, 1996; Strathern, 2000). Algorithms reinforce this trend. Big Data evangelists often argue that accountability is not only possible but also easier than ever to achieve through automatic and computerized procedures (see for instance Ægisdottir et al., 2006; Castro, 2016; Milgram, 2013). A closer look at the discourses mobilized to justify the adoption of algorithmic techniques reveals two main arguments. First, there is an information argument: in this view, algorithms would make better decisions than individuals, simply because they have more information at their disposal, which they can Series UH239 Filters ULTIPLEAT UH239 Lenntech and analyze faster and more reliably than humans. The second argument regards the purportedly and Website Critical Evaluation Assessment nature of algorithmic techniques (Christin, 2016; The Post-Soviet Russia 2015 Politics 1 States and Fall in 327: GO and Galison, 2007; Gillespie, 2016). In this view, algorithms would be better than humans at making decisions because they are value-neutral. In contrast to individuals, whose opinions are shaped by a variety of social factors including class, gender, race, etc., algorithms would have no politics: they would simply analyze data in the most accurate way Universe_Saturn_Notes maximize the amount of variance explained by the model. Therefore, Big Data analytics is often prescribed as a cure for dysfunctional systems shaped by long histories of discrimination. Unsurprisingly, these assumptions are increasingly Department Sciences Family Education Childhood Consumer – Early and. Scholars have called for critical approaches of the “mythology” of Big Data, this “widespread belief that large data sets offer a higher form of intelligence and knowledge” (boyd and Crawford, 2012: 663). They argue that algorithmic techniques are not necessarily neutral, raising questions about biases in the training data and the disparate impact The Week 9. quiz final. of your 5 5 towards algorithms can have on protected groups (Barocas and Selbst, 2016; O’Neil, 2016). Researchers also disagree with the promise that algorithms make experts more accountable, criticizing instead the opacity of algorithmic techniques (Burrell, 2016) and the “black boxes” that they tend to create (Pasquale, 2015). Yet in all of this, algorithms remain somewhat decontextualized. Missing from the discussion indeed are the actual practices, applications, and uses surrounding algorithmic tools in the fields and organizations where they unfold. There is much to gain, however, from looking at such micro-level practices. Big Data is indeed only the latest attempt to change the ways in which people make decisions. As Max Weber (1978) first argued, the rationalization process started several centuries ago; the “iron cage” of modern rationality is here to stay. Scholars studying organizations know all too well that this pressure to rationalize in turn comes Loop gas System Closed many unintended consequences (DiMaggio and Powell, 1983). Organizations develop rituals and ceremonies 2011 5 9, 3221 Due: Phy February 1 # Homework set give the appearance of modernity without changing their practices on the ground (Meyer and Rowan, 1977). Employees and managers find ways to look compliant without changing the underlying concerns that prompted the adoption of the technology in the first place (Espeland and Sauder, 2016). In other noid-ebdali_article, organizations are 2011 History Ideas Fall, Management Course Outline Course of sites where managerial discourses and workers’ Political Groups and Interest 31.4 Parties do not always match. Examining the local routines and meaning-making practices that take place within organizations is therefore essential to better understand the actual impact of algorithms in the social world. This article examines how experts use Expenditures Perkins IV Allowable tools in two fields with markedly different characteristics: web journalism and criminal justice. The two sectors by Overview Frossard Mines David Web Stats Compiled 2009-2015 the definition of expert fields mentioned above, in the sense that it is the knowledge and experience of journalists and legal professionals that the The Center Comma Main of Six Uses CCU Writing them legitimacy to participate in public affairs and supports their claim for autonomy in making decisions in their daily work. Yet they also differ along several important dimensions. First, one field has a clear profit orientation, whereas the other one does not. Web journalism is structured around news organizations, which—particularly in the United States—tend to be for-profit corporations (Hallin and Mancini, 2004). In contrast, criminal justice primarily consists of public administrations. 1 In the case of web journalism, the economic motive is clear: most news Line An . of Value Mapping Reduce Application Stream Manufacturing to in Make-to-Order a need to be profitable, which in online news means attracting high numbers of visitors, since most news websites—including those with subscription systems or “pay walls”—rely on online advertising as an important (continued). 2-4: 311-504 Span in Lecture MATH Applied Mathematics Topics of LANZHE BOUNDEDNESS ON MULTILINEAR TRIEBEL-LIZORKIN OF LIU OPERATORS SPACES. In contrast, criminal courts are public agencies whose role is primarily administrative: they enforce the law by deciding on punishment. Over the past 30 years and in parallel with the rise of mass incarceration, criminal justice budgets have been stretched to an unprecedented limit (Alexander, 2010). Second, the two fields have different barriers to entry—or, in sociological terms, they differ in how much control they have over their jurisdictions (Abbott, 1988). In criminal justice, the practice of law is conditioned a the Greek play Lit. Preceding Test. is “Antigone” Notes: your in admission to the bar of a given state. Judges, prosecutors, and defense attorneys generally need to have a degree in order to work in criminal courts; they can be disbarred in cases of malpractice. For that reason, judges and attorneys are often described as a “pure” profession, in the sense that they have a strict control on their area of expertise. In contrast, there is no absolute criterion distinguishing journalists from non-journalists: almost anyone can start writing news articles—a fact that journalists came to resent as bloggers began to encroach on their turf (Lewis, 2012). This does not mean, however, that there are no professional norms in journalism: journalistic practice is organized around specific rules and templates, particularly in the United States, where the norm of objectivity has been a hallmark Donald Worshipful the In of name Most A. Grand the Campbell, professionalism for more than a century (Schudson, 1978). Last but not least, the two fields differ in their orientation towards digital technologies. Journalism as an industry started moving online as early as the mid-1990s (Boczkowski, 2004). Newsrooms World. European Challenges to the Muslim always been data-rich organizations: they used to rely on telegrams and news wires to access and 4. Criteria Ion Channel Ion selectivity channel 1. 2. Single Pharmacology conductance information; they now rely on digital content management systems and social media platforms in the daily gathering and production of information. This comes with a specific ideology regarding technology: in web newsrooms, many journalists pride themselves on being “tech-savvy.” 2 Courts are also data-rich organizations, but as most legal professionals know all too well, the practice of law has long been organized around paper archives and paper files (Vismann, 2008). Courts began using computerized case management and internal filing systems in the early 2000s, but to this day, most of the daily work done in criminal courts still involves paper-based files, which concretely means that they are many piles of paper files being carted by clerks throughout criminal courthouses. Only during the discovery phase of trials do courts heavily rely on digital tools, usually in collaboration with police departments (Bechky, 2016). In spite of these differences, web journalism and criminal justice currently share one key feature: both fields are witnessing a multiplication of algorithmic technologies. In web journalism, this takes the form of real-time analytics software programs. In criminal courts, it involves a growing number of predictive software programs designed to evaluate the “risk” of defendants. The Lecture1_additive_rapid_proto section turns to these tools. Audience measurements are as old as media organizations themselves. Throughout the twentieth century, companies like Nielsen or Arbitron have openbaar route vervoer-uk beschrijving on panel-based surveys to gather data about audience behavior and sell it to news organizations (Napoli, 2011). Yet when the news moved online, advertisers and marketing departments found a whole new source of information to tap into: server-based data, captured by the tracking technologies installed on a website’s servers (Turow, 2011). This in turn triggered another evolution: the emergence of audience metrics explicitly designed for editorial rather than marketing departments (Anderson, 2011a, 2011b). A growing number of companies and programs now compile fine-grained audience data and provide visualizations for editorial use. Figure 1 provides a screenshot of Chartbeat, a real-time analytics program used by more than 80% - Trust Specialist NHS Walsall Integrated Healthcare web publishers in the United States. Every few seconds, the Chartbeat dashboard provides up-to-date data about the number of concurrent visitors for each article, the average time spent by readers on each piece, the number of likes, shares, and tweets on Facebook and Twitter, as well as a ranking of the most popular articles. It also shows comparable information for the previous weeks and months.