Remerciez-le!

Remerciez @Admin pour avoir partagé cet document gratuitement, de la manière la plus simple, en partageant sur les réseaux sociaux.

https://doi.org/10.1177/0539018419851045 https://doi.org/10.1177/05390184198510

https://doi.org/10.1177/0539018419851045 https://doi.org/10.1177/0539018419851045 Social Science Information 2019, Vol. 58(2) 238 –260 © The Author(s) 2019 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/0539018419851045 journals.sagepub.com/home/ssi Ethical dimensions of quantification Wendy Espeland Northwestern University, USA Vincent Yung Northwestern University, USA Abstract The ethical dimensions of quantification are seldom analysed. We examine three ethical features that are characteristic of quantification – its capacity to express or mediate power, focus attention, and shape opportunity structures. We do so in the context of three recent examples of new types of quantification : university rankings, the racial classification of Asians in the US, and facial recognition algorithms. Our examples highlight the importance of understanding the varied and complex ways that quantification creates and organizes social relations, and the effect of this on multiple forms of inequality. Keywords algorithms, census, ethics, quantification, rankigs Résumé Les dimensions éthiques de la quantification sont rarement étudiées. Nous analysons ici trois particularités éthiques qui sont caractéristiques de la quantification – sa capacité à donner du pouvoir ou à servir d’intermédiaire à ce dernier, sa capacité à attirer notre attention et enfin sa capacité à façonner des structures d’opportunité. Notre étude s’appuie sur trois exemples récents de nouvelles formes de quantification : les classements des universités, la classification raciale des personnes asiatiques aux États- Unis, et les algorithmes de reconnaissance faciale. Nos exemples soulignent à quel point il est important de comprendre les manières diverses et complexes avec lesquelles la quantification crée et organise les relations sociales, et leur effet sur les formes multiples d’inégalité. Corresponding author: Vincent Yung, Northwestern University – Sociology, 1810 Chicago Avenue, Evanston, IL 60208, USA. Email: vincentyung@u.northwestern.edu 851045 SSI0010.1177/0539018419851045Social Science InformationEspeland and Yung research-article2019 Special Issue Espeland and Yung 239 Mots-clés algorithmes, classements, éthique, quantification, recensement Introduction Ethical questions engage with what people ought to do, what actions benefit society, the rights and obligations we owe one another, and questions of fairness. Because we use quantification to assess merit, allocate scarce resources, and represent ourselves to each other, quantification demands an ethics. But it is ethical consequences that are often neglected (Espeland, 1998; Espeland & Stevens, 1998; Sauder & Espeland, 2009; Espeland & Sauder, 2016). In this article, we describe some broad features of numbers that call for an analysis of their ethical implications and focus on three ethical dimen- sions of quantification: power, attention, and opportunity. We illustrate these ethical dimensions with three contemporary examples, including university rankings, the racial classification of Asians in the US census, and facial recognition algorithms. These exam- ples show how power is expressed and mediated through quantification and how it is justified; who can participate in designing and implementing new quantitative technolo- gies; how discretion is curtailed; and the terms under which groups advocate, defend, or resist the imposition of quantification. With the dizzying pace at which new sources of data and new methods for analysing and exploiting data are being developed, in what has been called audit culture (Strathern, 2000), an audit explosion (Power, 1994, 1997), and the new digital economy, it is imperative that we look behind the numbers to see the assumptions and biases they contain, examine the uneven development of data sources, and critically examine their uses and whose interests they serve (Merry, 2016). Quantification is ethical There is a large literature – in the history of science, anthropology, sociology, accounting, law, public policy, and many other fields – that analyses how numbers intervene in the world as well as describe it. While these studies make explicit quantification’s relation- ships to governance (Miller & O’Leary, 1987; Anderson, 1988; Miller & Rose, 1990; Power, 1994, 1997, 2007; Lam, 2011; Davis et al., 2012; Mennicken, 2013), epistemology (Hacking, 1990, 2001; Porter, 1996, 2018; Bowker & Star, 1999; Carson, 2007), and modernity (Hopwood, 1992; Appadurai, 1996; Merry, 2016), much less is stated about the ethics of quantification. Ethics offers a broad approach for examining moral life, offering guidance for the development of foundational principles in such fields as business, medi- cine, public health, computation, and the environment. In our use of the term, we offer a practical consideration of the ethics of quantification. Quantification is moral insofar as it touches upon concerns about justice, fairness, and harm, or how individuals or groups understand which beliefs, behaviours, or goals are appropriate, better, or worthy (Hitlin & Vaisey, 2013). Quantification is ethical insofar as these moral concerns share some broad relevance to conduct within and across social institutions. The literature on numbers sug- gests two broad features of quantification that illustrate how quantification may be 240 Social Science Information 58(2) considered both moral and ethical: numbers both facilitate a particularly modern ontology and intervene in the social world they represent. First, numbers facilitate a peculiarly modern ontology (Espeland & Stevens, 2008), one that carries strong assumptions that numbers and quantification are means to achieve a more objective, rational, fair, and therefore moral calculus. Like all symbols, the meanings of numbers vary by type and use. But numbers and calculation have long been linked to such appealing qualities as rationality (Nussbaum & Hursthouse, 1984), rigour, objectivity (Daston, 1992; Daston & Galison, 2007), precision (McCormick, 2009), transparency, and authority (Porter, 1996; Power, 1994, 1997). As Martha Nussbaum shows, Plato associated reason and rationality with commensuration, the practice of turning qualities into quantities (Nussbaum & Hursthouse, 1984). And Cicero invoked numbers to bolster his accusations against a corrupt tax collector, but only after warning his audience his explanation would be tedious compared to his usual oratory (Deringer, 2018). We invest numbers with an authority that we do not grant to other forms of knowl- edge. Numbers help to produce what historian Lorraine Daston might call ‘aperspectival objectivity’, a ‘view from nowhere’ both praised and criticized for attempting to suppress some aspect of the individual or self in the pursuit of truth and certainty (Daston, 1992; Daston & Galison, 2007). One famous early enthusiast for the value of quantification in governance was William Petty, the ‘father’ of political arithmetic, which he defined as the art of reasoning using numbers in order to govern. After England conquered Ireland in 1653, Petty launched a large survey to calculate the resources of Ireland that could be taxed to pay for the war (McCormick, 2009). Accountability is another old idea, but one that has become tantamount to quantifica- tion in the past few decades. While it is exceedingly hard to trace these shifts in associa- tions, the historian Ted Porter and the critical accountancy scholar Mike Power suggest that issues tied to trust and distrust have much to do with the proliferation of quantifica- tion and auditing in many spheres of life. For Porter (1996), quantification is a form of communication that bridges social, geographical, and political distances. Conflict, suspi- cion, cultural or historical differences, and elites without secure legitimacy are some of the conditions that fostered the development of quantitative technologies, such as cost– benefit analysis and actuarial tables, in the late 19th and early 20th centuries. Power (1994, 1997) shows how neo-liberal reforms facilitated the migration of accounting and auditing from business to government. In most cases, this shift was understood as a means to overcome bias, improve efficiency, and mitigate conflicts. Quantitative author- ity derives from standardized methods, rules, classification, and controlled modes of combining and distinguishing that make for comparisons of like to like. Second, numbers are fateful interventions with broad relevance to conduct within and across social institutions. In fields ranging from global governance to philanthropy, more often organizations and their incumbents are subject to evaluations done by numbers. How much numbers influence social life varies, of course, as does their usefulness or accuracy as representations. However, what is undeniable is that quantification intervenes in the social world, enabling the creation of ubiquitous and easily transposable technolo- gies in modern life that have far-reaching and sometimes unimaginable consequences. Espeland and Yung 241 For example, standardized intelligence testing, initially developed and deployed to a wide degree to screen thousands of soldiers for recruitment during the First World War (Carson, 1993), has become a permanent fixture of contemporary American education. Standardized tests – of knowledge, ability, personality – are used to allocate scholarships and inform admissions into elite high schools and universities (Stevens, 2007); hold teachers, schools, and school districts accountable (Darling-Hammond, 2007); diagnose learning disabilities (Eyal et al., 2010); and generate measures for university rankings (Espeland & Sauder, 2016). As another example, proprietary algorithms use digital trace data to determine the content of our social media news feeds, the ads we see, and what we consume (Christin, 2018). Tracking devices measure everything from steps to men- strual cycles. The data generated by these devices is repurposed into any number of com- modities and sold to marketers, credit companies, political operators, insurance companies, the military, and pretty much anyone who can use them (Fourcade & Healy, 2017). As the Cambridge Analytica scandal reveals, there is little oversight in who buys this data and how it is used.1 The undeniable upshot is that tools of surveillance, control, and manipulation are uploads/Litterature/ 2-2-espeland-amp-yung-quantification 1 .pdf