Access the full text.
Sign up today, get DeepDyve free for 14 days.
Big Data’s calculative ontology relies on and reproduces a form of hyperindividualism in which the ontological unit of analysis is the discrete data point, the meaning and identity of which inheres in itself, preceding, separate, and inde- pendent from its context or relation to any other data point. The practice of Big Data governed by an ontology of hyperindividualism is also constitutive of that ontology, naturalizing and diffusing it through practices of governance and, from there, throughout myriad dimensions of everyday life. In this paper, I explicate Big Data’s ontology of hyperindi- vidualism by contrasting it to a coconstitutive ontology that prioritizes relationality, context, and interdependence. I then situate the ontology of hyperindividualism in its genealogical context, drawing from Patrick Joyce’s history of liberalism and John Dewey’s pragmatist account of individualism, liberalism, and social action. True to its genealogical provenance, Big Data’s ontological politics of hyperindividualism reduces governance to the management of atomistic behavior, undermines the contribution of urban complexity as a resource for governance, erodes the potential for urban dem- ocracy, and eviscerates the possibility of collective resistance. Keywords Urban governance, ontology, epistemology, pragmatism, individualism, democracy few weeks later reported that ‘‘the number of murders Introduction recorded by the (police) department is almost always Data politics dominated newspaper headlines in lower than those counted as homicides by the city’s New York City at the end of 2015. Controversy erupted medical examiner’’ (Goodman, 2016). The Police when a former Police Commissioner charged that the Commissioner defended such practices, saying that ‘‘I city’s method of collecting crime data underreported stand by my crime statistics because they are factual, actual events. He cited as an example the NYPD’s prac- they are the truth,’’ while a civil liberties advocate tice of recording a ‘‘shooting’’ only if a bullet wounds a countered that ‘‘the controversy highlights just how victim. According to the New York Times account: soft and subjective police statistics can be’’ (Goodman, 2015). a shooting ... is recorded only if someone is hit ....If a Meanwhile, some 100 miles to the south, in the eco- bullet tears a person’s clothing but does not wound the nomically devastated city of Camden, New Jersey, victim, the episode is not included in the Police police oﬃcials reported a large-scale expansion of that Department’s oﬃcial tally of shootings ... Gunﬁre at city’s ‘‘ShotSpotter’’ automated gunﬁre detection a car in which the occupants are wounded by shattered system (Adomaitis, 2015). ShotSpotter is described by glass but not by a bullet is not recorded as a shooting. (Goodman, 2015) Rutgers University, USA As the oﬃcial in charge of the police department’s Corresponding author: CompStat (Computer Statistics) program explained: Robert W Lake, Bloustein School of Planning and Public Policy, Rutgers ‘‘‘We need the bullet to cause the injury ... and we University, 33 Livingston Avenue, New Brunswick, NJ 08901-8554, USA. need blood’’’ (Goodman, 2015). A follow-up article a Email: firstname.lastname@example.org Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 3.0 License (http:// www.creativecommons.org/licenses/by/3.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access- at-sage). 2 Big Data & Society its corporate provider as ‘‘an acoustic surveillance tech- report listed as zero the number of chronically homeless nology that incorporates audio sensors to detect, locate families in New York City not in homeless shelters. and alert police agencies of gunﬁre incidents in real Although the city’s Human Resources Administration time .... The alerts include ... the precise time and loca- (HRA) funds 45 emergency and transitional shelters for tion (latitude and longitude) represented on a map and women and their children forced to ﬂee their homes due other situational intelligence’’ (ShotSpotter Fact Sheet, to domestic violence, HUD also reported as zero the 2016). The expanded ShotSpotter system in Camden number of homeless domestic violence (DV) victims in was part of a larger strategy of augmented video sur- shelters because the DV shelters operated by HRA were veillance and data collection designed to reassert the considered separate from the homeless shelters oper- appearance of police control in a city that routinely ated by the Department of Homeless Services (New tops national rankings in the incidence of violent York City Department of Homeless Services, 2016). crimes (NeighborhoodScout, 2016). Simultaneously, the Mayor’s Oﬃce announced an What counts as a ‘‘gunshot’’ in Camden, in many ‘‘unprecedented expansion’’ in the number of shelter cases, would not register as a ‘‘shooting’’ in New York beds for homeless victims of domestic violence to City. Whereas New York construes a ‘‘shooting’’ in the accommodate ‘‘a 50 percent increase over the current narrowest possible terms requiring the presence of a 8,800 individuals served yearly’’ (New York City Oﬃce shooter, a bullet, and a victim’s blood, Camden’s city- of the Mayor, 2015; Stewart, 2015b). Further con- wide acoustic surveillance system automatically records founding HUD’s data, HUD’s count of 1706 homeless every ‘‘digital alert’’ of an ‘‘actual gun discharge’’ as a youth almost certainly underestimated a signiﬁcant ‘‘gunshot crime in progress’’ pinpointed in time and subgroup of the homeless who, advocates said, might space (ShotSpotter Fact Sheet, 2016). These diﬀerences exceed 10,000 (Gibson, 2011) but ‘‘avoid public places between New York City and Camden cannot be sepa- where they could be counted for fear of referral to rated from their political context. The outcome of may- Child Protective Services and .. . avoid shelters out of oral elections in New York City, as well as the city’s safety concerns’’ (Navarro, 2015; Stewart, 2015a, 2016). attractiveness for residents, tourists, and investors, The selective practices of categorization and meas- depends on the public perception of safety and security, urement illustrated in these examples might easily be exerting downward pressure, in turn, on the practice of dismissed as the intrusion of political agendas in the otherwise objective and politically neutral construction collecting and documenting crime statistics. The NYPD’s CompStat program tracks weekly crime data of data as, in the words of the NYPD Commissioner, by precinct as a tool for managing organizational per- ‘‘factual’’ and ‘‘the truth.’’ If this were the case, a solu- sonnel and resources but it is equally a tool for mana- tion might lie in the rationalization and depoliticization ging public opinion (Eterno and Silverman, 2010). In a of methods of data collection, categorization, and ana- similar manner but conveying a diﬀerent message, lysis, bringing actual practices into closer alignment Camden’s expanded ShotSpotter detection system with normative claims. The ubiquity of Big Data as a deploying sensors and monitors in every neighborhood technique of governance, biopolitics, and bureaucratic also inﬂuences political opinion by establishing a visible control, however, has expanded the scope of the prob- police presence throughout the city. lem and ampliﬁed the challenge of delineating solu- A related controversy over categories, exclusions, tions. My argument in this paper is that the challenge and measurement erupted over data on New York of (and to) Big Data is not conﬁned only to the politi- City’s homeless population at a time when visible cization of its practices but rather is situated in its foun- homelessness, like crime, had become a political liabil- dational ontological premises, involving the ity for the city’s mayor. The annual homelessness count evisceration of context through an ontology of hyper- reported by the U.S. Department of Housing and individualism. An ontology of atomistic individualism Urban Development (HUD) in late 2015 found underlies the construction of calculative data in general 75,323 homeless individuals in New York City but (Hacking, 1990, 1991, 2006) but the arrival of Big Data, that number was quickly challenged by advocates for involving the algorithmic production, manipulation, the homeless and HUD acknowledged uncertainty in and application of very large datasets, has exacerbated the ‘‘reliability and consistency’’ of the data (Stewart, and expanded the scope of the problem by obscuring 2015a; U.S. Department of Housing and Urban from critical scrutiny its foundational hyperindividual- Development, 2015). The ambiguities in the data were ist ontology. manifold. Individuals and families who became home- This paper aims at a partial corrective by examining less through eviction, ﬁre, landlord harassment or other Big Data’s underlying calculative ontology. By ontol- reasons, and were living doubled-up with friends or ogy I mean ‘‘a set of contentions about the fundamental relatives were not considered homeless by HUD’s def- character of human being and the world’’ (Bennett, inition and were excluded from the count and HUD’s 2001: 160) or simply ‘‘a theory of objects and their ties’’ Lake 3 (Theory and History of Ontology, 2016). Specifying Big the data table rather than from the meaning residing Data’s ‘‘ontological imaginary’’ (Bennett, 2001: 161) in the lived experience of the original units of answers the question starkly posed by Wagner-Paciﬁci observation. et al. (2015: 5) who ask, with respect to Big Data: ‘‘Just Consideration of Big Data’s ontology of hyperindi- what is our basic ‘ontological unit?’’’ or, even more vidualism moves beyond epistemological debates over plainly, ‘‘What is a thing?’’ (see also Beauregard, deﬁnitions, categorizations, data collection methods, 2015, 2016). Big Data’s ‘‘onto-story’’ (Bennett, 2001: and data accuracy. The interrogation of such matters 161) can be brieﬂy summarized in the premise that derives from an internal critique of Big Data’s onto- the world is knowable via calculation and measurement logical framework while adopting and remaining and can be represented as the aggregation of discrete, within its ontological assumptions and focusing on independent, empirically observable units. These units problems of operationalization and implementation, are the ‘‘data points’’ representing, to list only a few that is, on problems of method (Lake, 2014). examples, gunshots, homeless people, sociodemo- Motivating such internal critique is the belief that graphic characteristics, credit card swipes, Internet better (i.e. more accurate, consistent, objective, or com- searches, or geo-tagged locational coordinates captured prehensive) methods of data collection, aggregation, from smartphones (Goldstein, 2016; Kitchin, 2013, and analysis will produce better knowledge. Beyond 2014; Wagner-Paciﬁci et al., 2015; Weber, 1946). This merely addressing internal operational mechanics, how- calculative ontology both relies on and reproduces a ever, internecine conﬂicts over the ‘‘how’’ of Big Data form of atomistic individualism in which the onto- have constitutive eﬀects. By performing and naturaliz- logical unit of analysis is the discrete data point, the ing Big Data’s ontological assumptions, debates over meaning and identity of which inheres in itself, preced- what gets counted, through what methods, via what ing, separate, and independent from its context or its algorithms (Kwan, 2016), and despite what omissions relation to any other data point. and (mis)categorizations reproduce its foundational By the hyperindividualism of Big Data, I refer to the premises while deﬂecting attention away from a critical practice of disaggregation and reaggregation that pro- assessment of those underlying principles (Zaloom, ceeds through a multistep process of interconnected 2003). The practice of Big Data governed by an ontol- and interdependent constructions of the world. Big ogy of hyperindividualism is also constitutive of that ontology, naturalizing and diﬀusing it through prac- Data’s ontological imaginary involves (1) the division and disaggregation of data ﬁelds (‘‘variables’’) into tices of governance and, from there, throughout ever-smaller units measured at ever ﬁner-grained myriad dimensions of everyday life. The challenge for levels of resolution, (2) the practice of counting each governance is that problems inherent in the ontology individual observation as an autonomous unit—a underlying a practice cannot be resolved by altering the thing-in-itself—extracted from and independent of its practice but must be addressed at the level of founda- context, and (3) the reaggregation and recontextualiza- tional ontological assumptions. Changing those onto- tion of the resultant data ‘‘bits’’ through the automated logical assumptions, however, destabilizes the entire algorithmic search for statistical patterns and correl- ediﬁce of practice built up on the prior underlying foun- ations hidden within the dataset. While an ontology dation that allowed the politicization of data construc- of atomistic individualism underlies calculative prac- tion to proceed in the ﬁrst place. As Garﬁnkel tices in general, the diﬀusion of Big Data both relies observed, there are often ‘‘‘good’ organizational rea- on and produces a form of hyperindividualism of an sons for ‘bad’ clinical records’’ (Garﬁnkel, 1967: 186). unprecedented scope and scale. The hyperindividualiza- Resistance to change on the part of interests invested in tion of Big Data results, ﬁrst, from the hyperdisaggre- those current practices (e.g. the police or the mayor) all gation of data ﬁelds in what Kitchin (2014: 2) describes but guarantees the preservation of the status quo. as the production of ‘‘massive, dynamic ﬂows of My purpose in this paper, accordingly, is to consider diverse, ﬁne-grained, relational data’’ recording and the implications for governance of Big Data’s ontology counting, for example, Internet transactions, selected of hyperindividualism. Rather than taking Big Data’s words within social media posts, demographic ‘‘vari- ontological assumptions as the starting point of the ables,’’ real-time spatiotemporal registers, and so on, analysis, however, my concern is to sketch a brief where the identity or meaning of each data point is genealogical account of their emergence. A genealogical self-evidently and inherently given as a thing-in-itself narrative understands practices (and their conse- divorced from its context. That hyperindividualization quences) as situated in the conﬂuence of the circum- permits, second, the reaggregation and intercorrelation stances from which they emerged (Foucault, 1984; of data observations to construct new observations and Hacking, 1991; Nietzsche, 1913). ‘‘History matters,’’ ‘‘facts,’’ the meaning of which is based on, imposed by, Trevor Barnes (2013: 298) reminds us, but, unlike his- and imputed from the discursive categorical labels in tory’s search for origins or causes, a genealogical 4 Big Data & Society approach problematizes the given-ness of Big Data’s ways of articulating the elements of the world and their ontological premises by unraveling and exposing their mutual connections. (Farias, 2011: 371) contingent emergence. Focusing on emergence rather than origins helps, as Jane Bennett (2001: 11) observes, Underlying the ‘‘bigness’’ of Big Data’s high volume of to ‘‘counter the teleological tendency of one’s observational units—‘‘terabytes or petabytes of data’’ thoughts.’’ For Colin Koopman: (Kitchin, 2013: 262)—is the assumption that each one of those myriad observational units (i.e. each individual Genealogical problematization ... provokes a question data point) constitutes a discrete thing-in-itself, an by rendering the inevitable contingent ....A genealogy independently observable unit, the ontological identity also shows us how that which we took to be inevitable of which precedes and thus channels its entry into the was contingently composed. A genealogy does not just ‘‘correct’’ category or ﬁeld within the dataset. Identity show us that our practices in the present are contingent in this individuated ontology inheres in the observa- rather than necessary, for it also shows how our prac- tional unit: the data point is the ‘‘Thing.’’ That identity, tices in the present contingently became what they are. furthermore, remains intact as the individuated obser- The history of that which was once presumed inevitable vational unit—the data point—is extracted from its not only makes us forget the inevitability, it also pro- context in the world and transported to the dataset, vides us with the materials we would need to transfor- entered in the spreadsheet or visually displayed on a matively work on that which we had taken to be a map. These are the dots on the map captured by necessity. (Koopman, 2011: 545) Camden’s ShotSpotter gunshot detection system. Each dot represents the imprint of a discrete auditory In the remainder of this paper, therefore, I explicate Big signal that a certain acoustical frequency deﬁnes as a Data’s ontology of hyperindividualism as a radical ‘‘gunshot,’’ each individuated point identiﬁed by its extension of atomistic liberal individualism and I con- unique locational coordinates of latitude and longitude trast it to a coconstitutive ontology that prioritizes rela- that ﬁx its position on the gunshot map of Camden. tionality, context, and interdependence. I then situate the Like the children’s game of connect-the-dots, the pat- ontology of hyperindividualism in the longue duree of its tern on the map, the shape of the data distribution, or genealogical emergence, drawing primarily from Patrick the parameters of the dataset derive from, and are con- Joyce’s (2003) history of 19th-century liberalism and stituted through, the aggregation and categorization of John Dewey’s (1929, 1935) pragmatist account of indi- the discrete, autonomous, individuated data points. vidualism, liberalism, and social action. In the conclud- The identity of the whole is the aggregate of the indi- ing section of the paper, I consider the implications for vidual identities inhering in the autonomous units of governance of Big Data’s ontological politics of hyper- which it is comprised. E pluribus unum: out of many, individualism. While Big Data’s hyperindividualist one. ontology extends throughout its applications in informa- In contrast to Big Data’s individuated ontology of tion technology, I focus here on the ways in which that the data point is a relational ontology in which meaning foundational ontology aﬀects the deﬁnition of urban devolves from the whole to its constituent parts and the problems, the dynamics of urban politics, and the prac- identity of the individual data point emerges from its tice of urban governance in the age of Big Data. relationship to and membership in the whole. Kitchin notes the highly relational character of Big Data’s data in which ‘‘common ﬁelds ... enable the conjoining of Big Data’s ontology of diﬀerent data sets’’ (2014: 2) through the ability to hyperindividualism overlay; juxtapose; or correlate data layers, ﬁelds, or Clarifying Big Data’s ontological imaginary and categories at will. But this relationality is correlative answering the deceptively simple question ‘‘What is a rather than constitutive, a relationality of juxtaposition thing?’’ reﬂects the centrality of what Latour (2005) rather than of ontological cocreation and codepend- calls Dingpolitik or the politics of the thing. As ence. These ‘‘relational databases’’ construct an acci- explained by Ignacio Farias: dental relationality in which the meaning of the data point is ﬁrst abstracted and removed from its constitu- Urban politics is .. . not about subjects, subjectivities or tive contextuality and then recontextualized through relational juxtaposition with other data points that discourses, but about things, complex entangled objects, socio-material interminglings. This is what have been similarly distanced from their contextually Latour (2005a) calls a Dingpolitik: the understanding constructed ontological identity. In Big Data’s correla- that urban politics can no longer be understood as con- tive, juxtapositional and accidental relationality, each ﬂict between human or, better, class interests, but ﬁeld, layer, or variable in the dataset can be associated involves conﬂicts over diﬀerent ‘cosmograms’, that is, with any other, it can be contextualized or Lake 5 decontextualized, or it can be correlated with one data institutions, or anything else .. ..[T]here is nothing to layer today and a diﬀerent one tomorrow—all without be known about (objects) except an initially large, altering the inherent, immutable meaning, value, or and forever expanding, web of relations to other identity of each data point comprising the dataset. objects. Everything that can serve as the term of a rela- In a coconstitutive ontology, in contrast, the meaning tion can be dissolved into another set of relations, and or identity of the individual data point does not preexist so on for ever. There are, so to speak, relations all the its context. If in the individuated ontology of Big Data the way down, all the way up, and all the way out in every whole is the aggregation of its individual parts, in a rela- direction: you never reach something which is not just tional ontology the individual parts derive their identity one more nexus of relations .. ..There is nothing to be from and through their membership in the whole. known about anything save its relations to other things. Because the meaning of the individual data point is estab- (Rorty, 1999: 53–54) lished by its context, it cannot be removed from its con- text without losing, altering, or obscuring its meaning. Big Data’s hyperindividuated ontology extracts and When the ontological meaning of the ‘‘thing’’—the indi- displaces the data point—the ‘‘thing’’—from the cocon- vidual data point—resides in its relationality within a con- stitutive, relational ontology of the assemblage. textual network or assemblage of things, ‘‘everything is Camden’s ShotSpotter system strips the gunshot from already within the individual’’ observational unit or data its relational context, reduces it to a common acoustical point (Law, 2004: 22). Indeed, ‘‘the notion of assemblage signal, and categorizes all gunshots as identical involves no outside, no exteriority’’ (Farias, 2011: 369) ‘‘things.’’ It does this by elevating the single criter- and ‘‘there is no distinction between individual and envir- ion—the register of an electronic signal on a detection onment. There are no natural, pregiven boundaries .. .. device—and excluding all other contextual characteris- Everything is connected and contained within everything tics as deﬁning criteria for the data category labeled else’’ (Law, 2004: 22). ‘‘gunshots.’’ Shots that hit their target, shots that Viewed within a relational ontology, therefore, a dot miss, shots ﬁred with intentional malice and shots on the gunshot map of Camden no longer represents from the accidental discharge of a ﬁrearm, aggressive merely the localized discharge of a ﬁrearm marked as shots and defensive ones, shots from stolen ﬁrearms an individuated ‘‘thing.’’ The dot instantiates the rec- and shots from legally registered ones, shots emanating ording of a certain acoustical signal at a designated from drug-related violence and shots from law-enforce- electronic frequency but also so much more. ment actions: these contextual complexities (and Contained in that dot is a set of political and economic others) make each of these very disparate types of structures and processes producing a population diﬀer- shots a diﬀerent category or type of ‘‘thing’’ but these entiated by indicators of poverty and inequality; the ontological diﬀerences are ignored and obscured in operation of urban, suburban, and regional land-use reducing the disparate meanings of these disparate practices of inclusion and exclusion that situate that events to the singular category of ‘‘gunshot’’ denoted dot here rather than there within a regional landscape; by a particular auditory signature picked up by an a portfolio of legal, illegal, and extra-legal provisions acoustical detection device. and practices governing the availability, distribution, Meanwhile, a diﬀerent but equally narrowly con- and cost of ﬁrearms; the design and implementation strued criterion is constitutive of a ‘‘shooting’’ in of law enforcement and surveillance practices and the New York City. Here a gunshot that misses its target training of personnel in their use; the technological cap- is not a shooting, nor is a gunshot that wounds its acity to design, construct, and operate gunshot detec- target by shattering the glass of a car window. The tion devices in a chaotic urban environment; a political NYPD’s narrowly construed deﬁnition excludes from decision-making process allocating scarce ﬁnancial the category all but one of the multitude of contextual resources in a cash-strapped city to acquire, install, relationalities comprising the multitude of diﬀerent and operate the detection system; and more. All this, ‘‘things’’ called gunshots. Gunshots whose identities as John Law notes, ‘‘is already within the individual’’ correspond to those excluded categories are experi- gunshot pinpointed at a speciﬁc place and time but all enced in the world but do not exist in the dataset con- of these layers of meaning fall away and disappear stituting the world of Big Data. A contextual, relational when the identity of the acoustic signal is reduced to ontology of homelessness fares no better in New York a ‘‘digital alert ... of a gunshot crime in progress.’’ City. Homeless youth are not counted as homeless and As the pragmatist philosopher Richard Rorty thus populate the ontological category of ‘‘not home- concludes: less,’’ not because they are in fact not homeless but because they have learned to evade HUD’s count of it does not pay to be essentialist about tables, stars, the city’s homeless population. Doubled-up families electrons, human beings, academic disciplines, social are deﬁned and categorized as ‘‘not homeless’’ despite 6 Big Data & Society having been rendered homeless by ﬁre, eviction, domes- political theory from Rousseau and Montesquieu to tic violence, or landlord harassment. The category ‘‘not Bentham, Locke, and Mill; development of a psych- homeless’’ contains both homeless and not-homeless ology of the self (Rose, 1989, 1998); and more. In individuals who, nonetheless, are ontologically con- Genealogy of Morals, Nietzsche describes the process structed as identical within the decontextualized, hyper- of ‘‘ﬁrst making man to a certain extent ... uniform, individuated dataset of homelessness. When the dataset like among his like, regular, and consequently calcul- of homelessness is then algorithmically correlated with able.’’ And, Nietzsche continues: similarly constructed datasets to reveal unexpected stat- istical patterns and associations, the apparent clarity the actual work of man on himself during the longest enabled by Big Data’s emergent relationality of juxta- period of the human race, his whole prehistoric work, position conceals the incoherence of the data entered ﬁnds its meaning, its great justiﬁcation .. . in this fact: into the analysis. man, with the help of the morality of customs and of These practices of data collection, categorization, social strait-waistcoats, we made genuinely calculable and correlation correspond to and reproduce an indi- viduated ontology in which identity inheres in the dis- until ﬁnally ‘‘at the end of this colossal process, at the crete, autonomous observational unit irrespective of its point where the tree ﬁnally matures its fruits ... then do constitutive context. Big Data’s characteristics of high we ﬁnd as the ripest fruit on this tree the sovereign volume, high velocity, ﬁne-grained resolution, and individual’’ (Nietzsche, 1913/2003, Essay II, 2, emphasis comprehensive scope (Kitchin, 2013), coupled with its in original). increasing pervasiveness throughout more and more That sovereign individual, Nietzsche emphatically spheres of everyday life, have elevated the individuated insists, was not born but made. As Claire Rasmussen ontology to what may justiﬁably be considered an convincingly demonstrates in her genealogy of the ontology of hyperindividualism. Enabled by techno- autonomous subject, ‘‘the ‘individual’ as a form of sub- logical developments in data acquisition, storage, data- jectivity is the product of a particular social imaginary base management, and analysis, hyperindividualism that institutionalizes the individual through secondary proceeds through the categorization of data ﬁelds at institutions such as the economy, a system of rights, ever more ﬁnely grained levels of resolution and ever and so on’’ (Rasmussen, 2011: 11). Those institutions, more comprehensive levels of coverage. To cite only however, do not do their work in the abstract. For two examples, DNA barcoding that allows the deﬁni- social historian Patrick Joyce, the construction of lib- tive categorization of unique biological species eral individualism proceeded through myriad mundane (www.barcodeoﬂife.org) extends individuation across practices of governance that spurred ‘‘the growth of all living things while obscuring ecological interdepen- privacy and the individuation of the subject’’ through dencies and coconstitutive ontologies. The Open which ‘‘people became available to be identiﬁed as indi- Research and Contributor ID system that ‘‘provides a vidual’’ (Joyce, 2003: 22). The invention of letter writ- persistent digital identiﬁer that distinguishes you from ing in the early 19th century, for example, stimulated every other researcher’’ (www.orcid.org) institutional- the development of a postal system which, in turn, izes the individuation of knowledge while obscuring the required the introduction and proliferation of street dense network of inﬂuences and interdependencies addresses and individual house numbers. These within which any process of knowledge production is assigned a unique location and, therefore, a unique, situated (Wyly, 2014a). enumerated identity to each dwelling, each letter writer, and each recipient. City directories soon fol- lowed, aggregating information about individual resi- The genealogy of hyperindividualism dents in the ﬁrst comprehensive urban databases How, then, did an ontology of hyperindividuation (Joyce, 2003: 197–198). What Joyce calls ‘‘the hygieni- become possible and from where did it emerge? For sation of the city’’ accelerated ‘‘the individuation of the Trevor Barnes: ‘‘what is forgotten in the celebration self’’ by ‘‘creating spaces around and between bodies, of big data is history’’ (Barnes, 2013: 297) and Barnes protecting them from others’ contact and smells’’ and situates Big Data’s assumptions and practices in the the resulting privatization ‘‘brought people into a new quantitative revolution in geography and the social sci- encounter with themselves’’ (Joyce, 2003: 73). The introduction of indoor sanitation literally wrapped the ences in the mid-20th century. Those developments, of course, were themselves inscribed within a process of individual body in a cloak of privacy, physically separ- individuation of much longer duration. The making ating and thus distinguishing each privatized body from of individuation has deep roots extending from and another and from the generalized body public. Joyce through the Enlightenment ideal of scientiﬁc objectivity similarly describes the cultural history of the bed in (Hacking, 1990, 1999; Harding, 2015; Latour, 1993); the 18th century as contributing to the privatization Lake 7 of sleeping linked to emerging ideas of liberal being human that escape the commodiﬁcation of labor individualism: power—the fully human context of labor—simply dis- appear from view (Arendt, 1958). In France the individual bed eventually became integral Dewey attributed the loss of individuality to the forces to notions of the Rights of Man, ﬁnding its way into of ‘‘quantiﬁcation, mechanization and standardization,’’ political reason in this form: a decision of the which he called ‘‘the marks of the Americanization that is Convention of 1793 in France ordained that state insti- conquering the world’’ (1929: 52). In an eerily prophetic tutions such as hospitals and asylums should provide statement, Dewey observed that individual beds as a natural extension of the Rights of Man. (Joyce, 2003: 73) The marks and signs of this ‘impersonalization’ of the human soul are quantiﬁcation of life, with its attendant Even death and burial became individualized and pri- disregard of quality; its mechanization and the almost vatized when communal burial within the walls of the universal habit of esteeming technique as an end, not as church was banned in England by an Act of Parliament a means, so that organic and intellectual life is also of 1842. When churchyards were replaced by ceme- ‘rationalized’; and, ﬁnally, standardization. Diﬀerences teries, the communal identity of the churchyard was and distinctions are ignored and overridden; agreement, replaced by a purchased cemetery plot now delimited similarity, is the ideal .. ..Homogeneity of thought and as individual property. Markers identifying individual emotion has become an ideal. (1929: 52) gravesites, Joyce observes, extended the memory of the individual into perpetuity, ‘‘another instance of the The individuating practices and conceptual transform- individuation of the human subject, in death now as ations traced by Joyce and Dewey served over time to in life ..., a universalism of the individual subject’’ naturalize an individuated ontology in which meaning (Joyce, 2003: 91). inheres in the unit of empirical observation. As Svend Writing in 1929 at a moment of global economic Brinkmann reminds us, the etymology of the word data crisis, John Dewey observed that ‘‘the problem of con- is ‘‘‘the given’ (the root Latin form is dare, which means structing a new individuality consonant with the object- ‘to give’)’’ (2014: 721). True to its genealogical proven- ive conditions under which we live is the deepest ance, Big Data’s individuated ontology encompasses problem of our times’’ (Dewey, 1929: 56). In a series those ‘‘terabytes or petabytes’’ of empirically observ- of essays titled ‘‘Individualism, Old and New,’’ Dewey able givens whose inherent meanings make them avail- described the ‘‘perversion of the whole idea of individu- able for calculation and categorization. While alism’’ (1929: 49) wrought by the reformulation of liber- categorization constitutes data, as has been widely alism from the (old) political individualism to the (new) recognized (e.g. Hacking, 2006; Porter, 1995; economic individuation of industrial society. The Schneider and Ingram, 1993; Wilson, 2011), the prac- Lockean, political liberalism of the early industrial revo- tice of calculability relies on the availability of data for lution, Dewey wrote, liberated the individual from the categorization according to their inherently given strictures of religious and monarchical rule: ‘‘liqueﬁed meaning. the static property concepts of feudalism,’ and ‘gave a The radical expansion, intensiﬁcation, and mystiﬁca- secular and worldly turn to the career of the individual’’ tion of this process through the practices of Big Data (1929: 78). The subsequent expansion of the market reproduce an ontology of hyperindividualism. For the economy, however, ‘‘subordinate(d) political to eco- 75,323 individuals comprising HUD’s dataset of home- nomic activity’’ (1935: 18) and replaced political indi- less individuals in New York City in 2015 (and the tens vidualism with economic individuation ‘‘to such an of thousands similarly categorized in other jurisdic- extent that individuality is suppressed’’ (1929: 66). If tions), the essentialized identity of homelessness derives the old liberalism liberated the political individual from from their inclusion in the dataset rather than from the subservience to despotic rule, the transformation from contextual dynamics that produce the condition and political to economic liberalism reduced the individual to experience of homelessness. In the circular logic of so many units of labor power subservient to the rule of Big Data, the homeless are those who are counted as the market. Commodiﬁcation and marketization pro- homeless. The ontological identity of homelessness that duced ‘‘a conception of individuality as something inheres to those individualized and decontextualized ready-made, already possessed’’ (1935: 46) and thus bodies when the dataset of which they are constituents available to be bought and sold in the labor market. is correlated and recontextualized with other datasets, Under the new circumstances of mass production and producing new relationalities, may have little if any- consumption: ‘‘liberty becomes a well-nigh obsolete thing in common with the original context-dependent term; we start, go, and stop at the signal of a vast indus- meaning of homelessness experience by the individuals trial machine’’ (1929: 46). In the process, those aspects of comprising the dataset. 8 Big Data & Society enrich governance by bringing multiple perspectives Hyperindividualism and urban to bear on a problem. governance Third, Big Data undermines democracy in the prac- Big Data’s ontology of hyperindividualism exerts a tice of urban governance. Nikolas Rose describes at debilitating inﬂuence on the conceptualization and length the ‘‘constitutive interrelationship between practice of urban governance. Understanding the quantiﬁcation and democratic government’’ (1991: world as an aggregation of individuated data points 675) in which democratic participation requires a reduces governance to the management of atomistic public conversant with numbers and able to compre- behavior, undermines the contribution of urban com- hend the world in the statistical form through which it plexity as a resource for governance, erodes the poten- is presented for public deliberation. The ascendancy of tial for urban democracy, and eviscerates the possibility Big Data, however, renders obsolete the public’s polit- of collective resistance. ical numeracy as the chart and the map are replaced by First, as Patrick Joyce observes: ‘‘before populations the black box of data management software and the can be governed they must be known or identiﬁed’’ complexity of the visible world is replaced by the (2003: 13) and measurement, quantiﬁcation, statistical hidden complexity of the algorithm (Wyly, 2014b). compilation, and mapping have long been used and The ontology of hyperindividualism, furthermore, understood as techniques of governance (e.g. reduces active political agents to the status of gener- Hacking, 1990, 1999, 2006; Mitchell, 2002; Rose, ators of data, whether through volunteering of data 1991). An optimistic version of this claim is that when (Elwood, 2008; Elwood et al., 2012) or as passive tar- quantiﬁcation and measurement make problems vis- gets of data-scraping technology. ible, governments are provoked to produce a solution Finally, Big Data’s hyperindividuated ontology evis- or risk a crisis of governmental legitimacy (Habermas, cerates the possibility of collective resistance. The dis- 1975). While the visibility of a problem may prompt aggregating, objectifying, and decontextualizing government action, however, the form of its represen- practices informed by these ontological presuppositions tation critically inﬂuences the form and substance of undermine collective action by inculcating a worldview the governmental response. By constituting urban comprised of atomistic individuals. These individualis- problems as the aggregation of individual empirical tic foundational premises correspond to and reproduce hegemonic commitments to a prevailing ideology of observations, Big Data’s ontology of hyperindividual- ism reduces governance to the management of atomistic individual responsibility and personal culpability. behavior or characteristics. A problem represented as a What Dewey observed in 1923 applies with equal or pattern of dots on a map—a concentration of subprime greater force today: loans, poor test scores, low-income households, home- less individuals, or gunshots, for example—prompts a When the self is regarded as something complete within response aimed at altering the number or distribution itself, then it is readily argued that only internal mor- of dots. A governance strategy that focuses on visible alistic changes are of importance in general reform .... symptoms deﬂects attention from underlying causes, at The result is to throw the burden for social improve- worst blaming the victim and at best addressing acute ment upon free-will in its most impossible form. needs of individuals at risk without preventing the con- Moreover, social and economic passivity are encour- tinuing (re)emergence of problems at their source. aged. Individuals are led to concentrate in moral intro- Second, Big Data’s relationship to urban complexity spection upon their own vices and virtues, and to presents a paradoxical challenge to urban governance. neglect the character of the environment ....And while While technological developments in data production, saints are engaged in introspection, burly sinners run collection, and management have yielded an exponen- the world. (Dewey, 1923: 113) tial increase in the number and variety of data cate- gories available for analysis, the decontextualization, homogenization, and standardization of data within Conclusion categories reduce complexity that might otherwise serve as a resource for governance. The hyperindividu- With Big Data a ubiquitous presence in modern life, alism of Big Data echoes the ascendancy of objectiﬁca- critical reﬂection urges caution in adopting its tenets tion in GIS more than two decades ago, in which and practices to inform policy-making and urban gov- ‘‘access to massive databases causes the analyst to ernance. Most discussion of Big Data and governance transform those to whom the data refer from subject- proceeds through internal critique; ﬁnding inconsisten- ively diﬀerentiated individuals to an objectiﬁed ‘other’’’ cies, ambiguities, and lacunae in the processes of data (Lake, 1993: 408; see also Curry, 1993). The loss of the collection, aggregation, and analysis comprising Big subjective viewpoint ﬂattens complexity that could Data’s operating manual. I have argued in this paper Lake 9 Elwood S (2008) Volunteered geographic information: Future that Big Data’s challenge to urban governance cannot research directions motivated by critical, participatory, suﬃciently be addressed at the level of internal critique and feminist GIS. GeoJournal 72: 173–183. and that its ontological presuppositions provide an Elwood S, Goodchild M and Sui D (2012) Researching vol- unreliable foundation for the practice of urban govern- unteered geographic information: Spatial data, geographic ance in a democratic society. research, and new social practices. Annals of the Association of American Geographers 102: 571–590. Acknowledgements Eterno J and Silverman E (2010) The NYPD’s CompStat: Thanks to Rachel Weber and Phil Ashton for organizing the Compare statistics or compose statistics? International symposium on ‘‘The Crowd, the Cloud, and Urban Journal of Police Science and Management 12: 426–449. Governance’’ at the University of Illinois-Chicago in April Farias I (2011) The politics of urban assemblages. City 15: 2015, where an earlier version of this article was presented, 365–374. and thanks to Rachel, Phil, and Matt Zook for organizing Foucault M (1984) Nietzsche, genealogy, history. this special issue on urban governance. Kathe Newman, Juan In: Rabinow P (ed.) The Foucault Reader. New York, Rivero, Elvin Wyly and two anonymous reviewers provided NY: Pantheon, pp. 76–100. extremely helpful comments on earlier drafts. Garfinkel H (1967) Studies in Ethnomethodology. Los Angeles: University of California Press. Gibson K (2011) Street Kids: Homeless Youth, Outreach, and Declaration of conflicting interests Policing New York’s Streets. New York: New York The author(s) declared no potential conﬂicts of interest with University Press. respect to the research, authorship, and/or publication of this Goldstein J (2016) New York police are using covert cell- article. phone trackers, civil liberties group says. New York Times, 11 February. Funding Goodman J (2015) Police leaders’ competing claims focus on The author(s) received no ﬁnancial support for the research, how New York counts crimes. New York Times, 30 authorship, and/or publication of this article. December. Goodman J (2016) Justifiable homicides, taken off the books, alter a murder tally. New York Times, 18 January. References Habermas J (1975) Legitimation Crisis. New York, NY: Adomaitis G (2015) Body cameras, ‘ShotSpotter’ expansion Beacon Press. for Camden County Police Department. NJ.Com, 19 June. Hacking I (1990) The Taming of Chance. Cambridge: Available at: http://www.nj.com/camden/index.ssf/2015/ Cambridge University Press. 06/body_cameras_shotspotter_expansion_for_camden_ Hacking I (1991) How should we do the history of statistics? coun.html (accessed 6 January 2016). In: Burchell G, Gordon C and Miller P (eds) The Foucault Arendt H (1958) The Human Condition. Chicago, IL: Effect: Studies in Governmentality. Chicago, IL: University University of Chicago Press. of Chicago Press, pp. 181–195. Barnes T (2013) Big Data, little history. Dialogues in Human Hacking I (1999) The Social Construction of What? Geography 3: 297–302. Cambridge, MA: Harvard University Press. Beauregard R (2015) Planning Matter: Acting with Things. Hacking I (2006) Making people up. London Review of Books Chicago, IL: University of Chicago Press. 28: 23–26. Beauregard R (2016) Planning and the politics of resistance. Harding S (2015) Objectivity and Diversity: Another Logic of In: Lieto L and Beauregard R (eds) Planning for a Scientific Research. Chicago, IL: University of Chicago Press. Material World. London: Routledge, pp. 11–25. Joyce P (2003) The Rule of Freedom: Liberalism and the Bennett J (2001) The Enchantment of Modern Life: Modern City. London: Verso. Attachments, Crossings, and Ethics. Princeton, NJ: Kitchin R (2013) Big Data and human geography: Princeton University Press. Opportunities, challenges and risks. Dialogues in Human Brinkmann S (2014) Doing without data. Qualitative Inquiry Geography 3: 262–267. 20: 720–725. Kitchin R (2014) Big Data, new epistemologies and paradigm Curry M (1993) Geographic information systems and the shifts. Big Data and Society April–June, 1–12. inevitability of ethical inconsistency. In: Pickles J (ed.) Koopman C (2011) Genealogical pragmatism: How history Ground Truth: The Social Implications of Geographic matters for Foucault and Dewey. Journal of the Philosophy Information Systems. New York: Guilford Press, of History 5: 533–561. pp. 68–87. Kwan M (2016) Algorithmic geographies: Big Data, algorith- Dewey J (1923/2004) Reconstruction in Philosophy. New mic uncertainty, and the production of geographic know- York, NY: Dover. ledge. Annals of the American Association of Geographers Dewey J (1929/1984) Individualism, Old and New. In: 106: 274–282. Boydston J (ed.) The Later Works 5: 1929–1930. Lake R (1993) Planning and applied geography: Positivism, Carbondale: Southern Illinois University Press, pp.45–123. ethics, and geographic information systems. Progress in Dewey J (1935/2000) Liberalism and Social Action. Amherst, Human Geography 17: 404–413. NY: Prometheus Books. 10 Big Data & Society Lake R (2014) Methods and moral inquiry. Urban Geography Rose N (1998) Inventing Our Selves: Psychology, Power, and 35: 657–668. Personhood. New York, NY: Cambridge University Press. Latour B (1993) We Have Never Been Modern. Cambridge, Schneider A and Ingram H (1993) Social construction of MA: Harvard University Press. target populations: Implications for politics and policy. Latour B (2005) From Realpolitik to Dingpolitik or how to American Political Science Review 87: 334–347. make things public. In: Latour B and Weibel P (eds) ShotSpotter Fact Sheet (2016) Available at: http://www.shot- Making Things Public: Atmospheres of Democracy. spotter.com/system/content-uploads/ShotSpotter_Fact_ Cambridge: MIT Press, pp. 14–41. sheet_-_final_draft_12.13.pdf (accessed 19 October 2016). Law J (2004) And if the global were small and incoherent? Stewart N (2015a) New York’s rise in homelessness went Method, complexity, and the baroque. Environment and against national trend, U.S. report finds. New York Planning D: Society and Space 22: 13–26. Times, 19 November. Mitchell T (2002) Rule of Experts: Egypt, Techno-Politics, Stewart N (2015b) New York City to add housing for domes- Modernity. Berkeley: University of California Press. tic violence victims. New York Times, 20 September. Navarro M (2015) Housing homeless youth poses challenge for Stewart N (2016) Homeless young people of New York, over- Mayor de Blasio. New York Times, 27 March. looked and underserved. New York Times, 5 February. NeighborhoodScout (2016) NeighborhoodScout’s murder cap- Theory and History of Ontology (2016) Ontology: Its role in itals of America – 2016. Available at: www.neighborhoodsc- modern philosophy. Available at: https://www.ontology. out.com/top-lists/highest-murder-rate-cities/ (accessed 6 co/idx00.htm (accessed 6 January 2016). January 2016). U.S. Department of Housing and Urban Development (2015) New York City Department of Homeless Services (2016) HUD 2015 Continuum of Care Homeless Assistance Daily report 1/8/2016. Available at: www1.nyc.gov/assets Programs Homeless Populations and Subpopulations. /dhs/downloads/pdf/dailyreport.pdf (accessed 8 January NY-600 New York City CofC Point-in-Time Date: 2/9/ 2016). 2015. Available at: https://www.hudexchange.info/ New York City Office of the Mayor (2015) DeBlasio admin- resource/reportmanagement/published/CoC_PopSub_Co istration announces unprecedented expansion of shelter C_NY-600-2015_NY_2015.pdf (accessed 11 January for survivors of domestic violence. Available at: http:// 2016). www1.nyc.gov/office-of-the-mayor/news/632-15/de- Wagner-Pacifici R, Mohr J and Breiger R (2015) Ontologies, blasio-administration-unprecedented-expansion-shelter- methodologies, and new uses of Big Data in the social and survivors-domestic (accessed 12 January 2016). cultural sciences. Big Data and Society July–December, 1–11. Nietzsche F (1913/1967) The Genealogy of Morals. New York, Weber M (1946) Science as a vocation. In: Gerth H and NY: Random House. Wright Mills C (eds) From Max Weber: Essays in Porter T (1995) How social numbers are made valid. Sociology. New York, NY: Oxford University Press, In: Porter T (ed.) Trust in Numbers: The Pursuit of pp. 129–156. Objectivity in Science and Public Life. Princeton, NJ: Wilson M (2011) Data matter(s): Legitimacy, coding and Princeton University Press, pp.33–48. qualifications-of-life. Environment and Planning D: Rasmussen C (2011) The Autonomous Animal: Self- Society and Space 29: 857–872. Governance and the Modern Subject. Minneapolis: Wyly E (2014a) Please do not cite this article. Urban University of Minnesota Press. Geography 35: 783–787. Rorty R (1999) A world without substances or essences. Wyly E (2014b) Automated (post)positivism. Urban In: Rorty R (ed.) Philosophy and Social Hope. New Geography 35: 669–690. York, NY: Penguin, pp.47–71. Zaloom C (2003) Ambiguous numbers: Trading technologies Rose N (1989) Governing the Soul: The Shaping of the Private and interpretation in financial markets. American Self. London: Free Association Books. Ethnologist 30: 1–15. Rose N (1991) Governing by numbers: Figuring out democ- racy. Accounting, Organizations and Society 16: 673–692. This article is a part of special theme on Urban Governance. To see a full list of all articles in this special theme, please click here: http://journals.sagepub.com/page/bds/collections/urban-governance.
Big Data & Society – SAGE
Published: May 16, 2017
Keywords: Urban governance; ontology; epistemology; pragmatism; individualism; democracy
Access the full text.
Sign up today, get DeepDyve free for 14 days.