Collecting Lives

Front matter

Critical Data Narrative

as Modernist Aesthetic in , Early Twentieth-Century

U.S. Literatures

Contents

Acknowledgments vii

Introduction: “More nearly a transcript of life”: Collecting Lives and Narrating Selves in Early

‘Twentieth-Century U.S. Literatures I 1. “Such a body of information”: W. E. B. Du Bois, Data, and the Re-assemblage of Race and Self 33

Introduction

“More nearly a transcript of life”

Collecting Lives and Narrating Selves in Early Twentieth-Century U.S. Literatures

Most men and many women prefer to come into closer touch with reality and seek it, often in vain, in the newspapers. Consequently fiction is undergoing a process of fission; the cleft between the realistic and romantic novels is widening. The former are becoming more nearly a transcript of life, and the latter, no longer tethered to earth, are soaring into the ether of the imaginary and impossible.

—Edwin Slosson, The Independent, April 13, 1905

Human beings are interested in two things. They are interested in the reality and interested in telling about it.

—Gertrude Stein, “A Transatlantic Interview,” 1946

In or around March 1905, the literary representation of lives changed. Or at least, that’s when Edwin Slosson, chemist and literary editor of progressive New York weekly newspaper The Independent, noticed it.' His description of and explanation for this change plots a complex interaction between readerly desire and a cultural surround of pervasive empiricism. Readers no longer seek the “generalized types of humanity as expressed by the artist in painting and sculpture,” Slosson argues, because “romances and poems do not interest them so much as do individuals.” Realistic novels are, therefore, “becoming more nearly a transcript of life” (849). The selecting and synthesizing acts of literary narration are seen as less valuable, less real and, perhaps surprisingly, less interesting to readers than the more mechanical approach of unselectively recording everything that happens within the scope of a life. The preference for and epistemological value attributed to the rigorous recording of observation is also on the ascendant in scientific practice, Slosson continues to point out, which seems increasingly focused on collecting observations for future analysis rather than the proposal of overarching theories. Now, the “candidate for the Ph.D. watches a single amoeba under a microscope and writes his thesis on one day’s doings of its somewhat monotonous life” (Slosson, 849). In literature, as in science, Slosson suggests, the representation of reality is taking on a distinct new form, composed of discrete observations of actual lives, exhaustively collected rather than interpretively selected. In other words, lives are coming to be seen as data collections.

In terms of modernist aesthetics, Slosson’s conjecture proves prescient. The image of a PhD candidate making a “thesis” of rigorously recording an amoeba’s activities during a single day offers a tantalizing anticipation of the narrative innovation for which the literature of the early twentieth century would come to be known. Works such as James Joyce’s Ulysses (1922) and Virginia Woolf’s Mrs. Dalloway (1925) make a novel out of one day’s doings in the life of an ordinary person, taking up the same chronologically constrained and transcription-like method to rethink the relationship between literary narrative and the reality of lived experience. Robert Musil begins his own experiment with exhaustive observation aimed at an individual life in 1921, eventually published as a three-volume, unfinished novel, The Man Without Qualities. As Liesl Olson has noted, “plotlessness is modernism’s great revolution” (21), and Slosson’s aesthetic pronouncements offer a provocative early twentieth century argument that the rise of data collection as a form of representing lives is part of the reason why.

While none of the writers just mentioned would explicitly claim to be collecting data, the formal confluence between their narrative modes and the practice of data collection is striking. Both modernist narrative and data collection aspire to truly represent a reality that has been obscured by preexisting projections of meaning and causality, and both develop forms that privilege collection over selection as a means of doing so. To frame this confluence using the terms of Gertrude Stein’s observation, cited above, the data collection seems to have become a predominant form of representing “the reality” that “human beings” obsessively seek to find ways of “telling about.” Differentiating between “the reality” and “telling about it,” Stein reminds us that acts of telling mark a persistent space between “the reality” and how we understand it. Even as we amass data in the hopes of arriving at a representation of “the” reality, there remains an inevitable and vital gap between data and narrative. This book examines how modernist writers who collected data and narrated lives navigate this space, pinpointing an underexplored intersection between the rise of data collection as a method of knowing the human and modernist experimentation in narrative form.

Collecting Lives examines U.S. modernist life writing forms as sites of critical engagement with the data episteme, a cultural surround in which data and its collection are presumed to offer unprecedented access to reality, truth, and power. Data collection, as fundamental to both empiricism and imperialism,’ is one of modernity’s foundational representational forms. Further, it is a form that has an especially prominent role in U.S. attempts to represent and control the minoritized and economically precarious social groups created by modernity. The application of data collection to human selfhood? in the late nineteenth- and early twentieth-century United States provides an instructive prehistory to the underlying question of the relationship between data, life, and narrative that animates contemporary debate over algorithmic modes of identification. This historical frame centers U.S. modernist life writers W. E. B. Du Bois, Henry Adams, Gertrude Stein, and Ida B. Wells-Barnett. Working in human-oriented empiricist disciplines in the midst of methodological reconfiguration around the work of data collection, these writers experiment with data as a form of representing lives. Using a comparative approach that situates works from a range of modernist canons in the data episteme, I seek to recognize the long conceptual history of data as a technology of selfhood and its intersection with material histories of power—who has the ability to collect data about whom and whose narration of that data is seen as authoritative. I argue that these writers draw from their work in sociology, history, psychology, and journalism to formulate critical data aesthetics as they confront questions of identity around race, gender, and nation both in their research and their life writing.

These writers reflect on our own data surround in unexpected ways, because they share the desire for, suspicion of, and aspiration to total data representation that animates much of contemporary data discourse, yet they negotiate these relationships to data in a pre-digital ecology of affordances for collection, manipulation, and interpretation. This means that their engagement with data is more literally an encounter with data—more embodied and more often on a point-by-point basis rather than parsed through filters and statistical summaries. As Sara Ahmed elaborates the significance of encounter, “The term encounter suggests a meeting, but a meeting which involves surprise and conflict” (6). Ahmed’s theorization of the encounter includes both face-to-face meetings and the more conceptual, but equally powerful “coming together of at least two elements,” such as “a meeting between reader and text” (7). Encounters are experiences that force us to ask, “how does identity become instituted through encounters with others that surprise, that shift the boundaries of the familiar, of what we assume we know?” (6-7). The data collection practices and technologies of the pre-digital era produce encounters that surprise and defamiliarize. When Du Bois, for example, collects data of African American life for The Philadelphia Negro, he is not querying a series of dot-gov databases with historical information on household make-up, income, and property values. He walks door to door to complete the five thousand questionnaires that make up his database.

He personally enacts the encoding of perception and interaction into number and word. He then stores, organizes, and reorganizes this data in paper form. When Stein, as a medical student, works to model the human brain, she is not using imaging software to apply multiple comparison methods to a set of functional magnetic resonance imaging (fMRI) scans. She is drawing, by hand, every facet of every tissue section put in front of her. Because of these writers’ more intimate encounters with the data points they collect, their perspective on data collection as a representational form offers a vantage that has been obscured by contemporary interfaces.* While so many of the tools we use as professional and lay readers of a data-enveloped world seem to flatter our pretensions of being able to see everything and know it all, these writers daily perform the lossy translation of experience into data and confront the impossibility of keeping every point in view. This perspective marks them as forebears of critical data studies, a field of “research and thinking that applies critical social theory to data to explore the ways in which they are never simply neutral, objective, independent, raw representations of the world, but are situated, contingent, relational, contextual, and do active work in the world” (Kitchin and Lauriault, 7). In their linkage of data collection and embodied selfhood, the authors considered in this book situate data in time and space, expose the creative processes that underlie its seeming rawness, consider the political ramifications of its collection and interpretation, and above all, challenge the idea that if we only collect enough data, it will narrate itself.

Through their hands-on engagement with practices of collecting lives, these writers confront data’s inherent challenge to, rather than confirmation of, narrative traditionally conceived. Narrative becomes meaningful through selection, the designation of a beginning and an ending, and the exclusion of all points that do not help explain the path between. Data, on the other hand, proposes to uncover meaning through collection, or the accumulation of equally meaningful points. When we recognize data’s fundamental challenge to narrative, the data point comes into view as a representational form that joins the fragment, montage, and impression in the roster of modernist aesthetics.

The lens of modernist aesthetics allows us to register these writers’ innovative engagement with data collection and data representation on the level of both content and form. I bring together the terms “data-driven” and “modernism” to suggest a literary critical framework for apprehending the historical, formal, and epistemological uses of data. In present usage, “data-driven” denotes a process of decision-making that purports to look to data for its direction and justification, implying that decisions can be made more effectively and without the danger of bias or misguided tradition by referring to data.’ I puta different twist on the term to call attention to how imagining reality as a vast data collection drives that imagination toward certain formal parameters. Data-driven modernists, in my definition, commit to the form of data collection first and then see what narratives result. In this way, they highlight how data complicates, rather than clarifies, our narrative representations of reality. I approach modernism from a historical as well as aesthetic perspective, following interpretive pathways opened by the work of scholars such as Andreas Huyssen, Susan Stanford Friedman, Douglas Mao, Rebecca Walkowitz, and Werner Sollors, who have challenged critics to consider how our methodological concatenation of text and context must shift when we loosen the association of modernism with a few representative works and seek the “effects of synergy or friction [that] result when the many, sometimes contradictory criteria of high modernism are tested against less evidently experimental texts” (Mao and Walkowitz, Bad Modernisms, 2).° Considering historical context alongside aesthetic practice is especially important for critical consideration of race and ethnicity as factors shaping literary production, as the work of Joshua Miller, Sarah Wilson, and Leif Sorensen has modeled.’ Focusing on data as a representational strategy with aesthetic as well as epistemological implications brings into a view a friction-generating range of texts and creators, embedded in multiple disciplines and literary communities.

Data-driven modernism entails a critical relationship to the “increasingly dense information environment” (Manovich, 23) of the late nineteenthand early twentieth-century United States. This relationship manifests in particular sensitivity to the role of data in representing, understanding, and shaping human life and is elaborated through aesthetic experimentation, particularly with narrative. Understanding aesthetics, broadly, as practices of provoking perception, I see data as a generative tool for such experimentation. In the readings that follow, I argue that the forms of the data point and the data collection constitute a modernist narrative aesthetic evidenced in three broad ways. Modernist data-driven narratives:

1) are fundamentally constructed through collection rather than selection, producing aesthetics of inclusion, excess, and multiplicity that foreground contingency and often displace the ends of narrative in a temporality of deferral;

2) feature narrators who are self-conscious about the relationship between narrative and reality, often explicitly reflecting on the tension between data and narrative and underlining the selective acts that produce narrative; and

3) through these formal engagements with and thematic reference to data, leverage the epistemological and cultural authority of data to challenge narratives of identity claiming to be objective realities of self.

Du Bois, Adams, Stein, and Wells-Barnett not only tell different life stories with data, they tell life stories differently because of data. We must understand the narrative aesthetic effects of data representation to understand data’s implications for concepts of human identity. Understanding these aesthetic effects, we are better able to harness their potential for intervention in the essentialization of race, gender, and class that the data of lives has been and is being used to enact.

Fundamentally, the writers examined in this study have two things in common: they collected data as part of their work in an empiricist discipline focused on the human and they wrote life narratives. My focus on life narratives derives from the fundamental condition of life writing as being written and read in relationship to a referential concept of human life. Every instance of life narrative implies a relationship to an exhaustive data collection that would represent the whole of a real life. This inherent relationship to an idea of observable reality makes life writing a particularly fruitful site of investigation for a changing cultural and cognitive relationship to data.

I use the analytic of narrative in this project in three senses: as a traditional literary form, the realism of which is being challenged and re-formed during this period and which individual texts may be read to seek, resist, and/or revise; as an intervention in the representation of lived experience through which we can read for ethical arguments and engagements; and as an act of self-representation that not only represents but, in the minds of many critics and theorists, also constitutes self.* Because narrative is a kind of epistemological model for reality, the effects of the formal shifts provoked by data collection as a mode of representing the world are not solely aesthetic. As Michael Elliot notes, “narratives do more than place events into chronological sequence; they arrange those events according to patterns of causation in a way that enables the author and reader to create order out of the chaos of everyday life” (xxiii). Narrative presents an epistemological model for reality, implicitly staking claims about causation and relationship. In this way, the status of narrativity in life writing is a register of concepts of self, agency, and social order. I demonstrate the epistemological effects of conceptualizing lives as data by reading the formal effects data entails for life narrative.

Provoking critical awareness of the gap between data and narrative is a project that has only become more urgent a century or so after these modernist experiments. Data is’ now constantly being used to narrate our lives. Categorizing algorithms draw from amassed personal data to assign narrative destinies to individuals at crucial junctures, simultaneously predicting and shaping the path of their lives’ unfolding. Data is commonly assumed to bring us closer to objective knowledge of reality, perhaps relieving the need for human interpretation altogether if we can only collect enough. In 2008, Chris Anderson, a writer for Wired magazine, went so far as to assert, “With enough data, the numbers speak for themselves” (108). But the narrative paths categorizing algorithms have assigned to human lives in areas ranging from public policy to personal finance seem, more often than not, to replicate biases about who an individual is and could become." It has become increasingly clear that, as John Cheney-Lippold cautions, “Who we are in terms of data depends on how our data is spoken for” (We Are Data, 48). The question of who we are hinges not on what data says, but on who is granted the authority to speak for it and what forms of representation they employ. The data-driven modernists that populate the juncture between life, data, and story model critical data narrative in ways that prepare us to challenge the claims of algorithmic identification and demand alternative modes of narrating our collected lives.

Data: Point, Collection

Data circulates in public discourse as a synonym for reality, the raw foundation of truly objective knowledge. As Lisa Gitelman and Virginia Jackson have observed, this discourse conceives of data simultaneously as material to be explained and as explanation itself. We commonly think of it as “units or morsels of information that in aggregate form the bedrock of modern policy decisions by government and nongovernmental authorities” (Gitelman and Jackson, 1). Yet as inherently crude and insufficient as anything made up of “morsels or units” must be, data is also commonly thought of as having the power not only to represent what exists, but also to narrate what is to come. “Our data isn’t just telling us what’s going on in the world,” IBM advertises; “it’s actually telling us where the world is going” (qtd. in Gitelman and Jackson, 1). In our discourse, data slips, both colloquially and conceptually, from plural to singular, from material requiring explanation to explaining agent. A critical understanding of data begins with awareness of this operational duality and the epistemological desires that underwrite it.

The data point, as we know it, begins its discursive life as the Baconian particular. Francis Bacon, in his 1620 methodological treatise Novum Organum (New Instrument), enshrines the “particular” as the first object of knowledge. “We must bring men to particulars,” he exhorts, “and they must for a while renounce their notions, and begin to form an acquaintance with things” (19). Bacon argues that scientists must first commit themselves to the creation of “a store and collection of particular facts, capable of informing the mind” (78). His formulation encodes the duality of data in its inseparable linkage of “particular” and “collection” as the forms of representation that will bring us closest to knowing and understanding reality.

While I am suggesting that the Baconian particular launches the form and concept of the data point, it is instructive to note that Bacon does not inaugurate the usage of the word “data” as we now know it. Novum Organum uses the word “data” only once and its single usage illustrates the radical shift in reference that occurs, as historian Daniel Rosenberg has identified, in the word’s journey from Latin to English across the later seventeenth and early eighteenth century. From the Latin dare, to give, the most literal meaning of data is “givens” (Rosenberg, 15-16), assumed truths that precede investigation and analysis. Data, in this sense, is exactly what Bacon rails against. Critiquing Galileo’s explanation of tides, Bacon writes, “He has, however, imagined this on data that cannot be granted (namely, the earth’s motion)” (240). “Data” in Bacon’s usage refers to preexisting theories of planetary motion rather than collected particular observations.

Data’s concrete referent has shifted from fully formed theories, principles, or explanatory narratives to massive collections of points. This shift makes sense in the context of pervasive empiricism, the belief that in order to know the world, we must rely upon sensory perception. If seekers of knowledge “restore the senses to their former rank,” Bacon contends, they will “generally reject that operation of the mind which follows close upon the senses, and open and establish a new and certain course for the mind from the first actual perceptions of the senses themselves” (6). These “perceptions” are the “particular facts” (78) that must now be collected. In his instruction to perceive first and reason later, Bacon sets up an ideal order of operations that temporarily suspends theorizing in order to reach a “new and certain course” that will be superior due to its accord with the world as it really is. This process is often referred to as induction, or reasoning from observation, and set in opposition to deduction, or deriving from first principles. Bacon proposes to shift belief from content to method, discarding a priori principles in hope of as-yet unrevealed principles that the collection of particulars will yield, making them the new “data.” Thus, while the literal meaning of data, the “givens” upon which reasoning is founded, has remained the same, the thing to which it refers has become its opposite.

Bacon’s collection of particulars is intended to disrupt two human cognitive tendencies: the “operation of the mind” (6) that follows from the desire for all-encompassing coherence of explanation and a too-ready equation of individual experience to total reality. Bacon saw the study of the natural world as helplessly stalled due to reliance on syllogistic methods of deduction. Reasoning from prior principles created a sense of internal coherence to every natural process that, while affectively pleasing, ultimately did not account for the world’s diversity of phenomena. Bacon argues, “The human understanding, from its peculiar nature, easily supposes a greater degree of order and equality in things than it really finds; and although many things in nature be sui generis and most irregular, will yet invent parallels and conjugates and relatives, where no such thing is” (22). Bacon envisions data-ascollection as a corrective to the impulse to project underlying order when one should instead be confronted by confusion. Additionally, as individuals, our perceptions are unavoidably constrained not only by our cognitive predispositions but also by our physical limitations. “The foundations of experience (our sole resource),” he asserts, “have hitherto failed completely or have been very weak” (78). A commitment to data collection promises to overcome the limits of individual perception and forestall the habit of projecting order onto reality.

Bacon’s method thus comes with two mandates: data collection must

IO * COLLECTING LIVES be exhaustive and it must be exteriorized. Exhaustivity, as conceived by information science, is a measure of the correspondence of an index to the document(s) indexed, a measure of the distance between the secondary model and the complete world of knowledge, however that world is defined (Van Rijsbergen, 24-25). As Mary Poovey describes, the Baconian ideal insists that the “entire globe and all of its inhabitants ought to be subject to empirical observation” (“Limits,” 193). The aspiration to exhaustive collection is essential to data epistemology for two reasons. First, it is only through exhaustive collection that we can be sure to overcome our selective, interpretive nature. Second, exhaustive data is presumed necessary to maximize the material gain that can be leveraged out of an increased ability to predict and control."

In addition to being exhaustive, a data collection must be exteriorized. The “collection of particulars” (Bacon, 81) requires literal and conceptual externalization from the human subject in order to disrupt our mental inclinations and extend our analytic capacities. Foreshadowing our contemporary interest in data visualization, Bacon instructs that we must “properly and regularly [place] before the eyes” the “collection of particulars” (81) once it has been amassed. For, he cautions, “after having collected and prepared an abundance and store of natural history . . . still the understanding is as incapable of acting on such materials of itself, with the aid of memory alone, as any person would be of retaining and achieving, by memory, the computation of an almanac” (80). To make use of collected data, the human memory must be supplemented by material repositories and visualizations. The desire for data in this way necessitates technological and representational innovation, because the requisite “collection of particulars” promises to overwhelm the capacity of any single human memory just as the vastness of the world surpasses the scope of any single eye. By placing data outside the self, Bacon exteriorizes it both literally and in the conceptual sense, by constructing it as “something we feel we can know, reveal or interpret and which will give us a foundation” (Colebrook, 71). Conceived as a “collection of particulars,” data becomes a term for something much less coherent than a “given” idea while retaining its telling etymological encoding of our desire for a singular bedrock of knowledge.

While the abundance of data visualizations in contemporary life speaks to the ongoing relevance of exteriorization, the desire for exhaustive collection may be the most important continuity between Bacon’s theorizations and the present moment of Big Data. As boyd and Crawford note, “There is little doubt that the quantities of data now available are often quite large, but that is not the defining characteristic of this new data ecosystem. In fact, some of the data encompassed by Big Data (e.g. all Twitter messages about a particular topic) are not nearly as large as earlier data sets that were not considered Big Data (e.g. census data)” (663). A more definitive trait, as Rob Kitchin and Tracey Lauriault argue, is Big Data’s aim to be “exhaustive in scope, striving to capture entire populations or systems (n= all)” (4).

The aspiration toward exhaustivity underwrites the conflation of data and narrative. Writing nearly two centuries further into the empiricist knowledge project that Bacon helped to launch, Pierre Simon Laplace elaborates on the scope of knowledge—and power—that an exhaustive collection of particulars would enable:

Given for one instant an intelligence which could comprehend all the forces by which nature is animated and the respective situation of the beings who compose it—an intelligence sufficiently vast to submit this data to analysis—it would embrace in the same formula the movements of the greatest bodies of the universe and those of the lightest atom; for it, nothing would be uncertain and the future, as the past, would be present in its eyes. (4)

Laplace’s conjecture assumes that there are a set of natural laws that, working together in complex yet predictable ways, are awaiting only our discovery to reveal seeming chance for the order that it really is. This revealed order will explain reality from the micro to the macro, subsuming “all events, even those which on account of their insignificance do not seem to follow the great laws of nature” (Laplace, 3) in its predictive power. It is only “in ignorance of the ties which unite such events to the entire system of the universe” that “they have been made to depend on final causes or on hazard” (3). Both a priori “final causes” and chance will be abolished by exhaustive data collection. Wendy Chun has named this figure of “an intelligence sufficiently vast to submit this data to analysis,” the “Laplacian subject . . . a sovereign subject capable of ‘knowing all’” (109). The imagination of such a subject entails determinist assumptions about the workings of both animate and inanimate life and nearly unimaginable power for an individual or group that could wield such definitive knowledge.”

By the mid-nineteenth century, Auguste Comte’s theorization of sociology has extended the predictive visions of the empiricist project to encompass the human. As Lewis Coser summarizes, “Comte’s aim was to create a naturalistic science of society, which would both explain the past development of mankind and predict its future course,” holding that society “is subject to basic laws just as is the rest of the cosmos, even though it presents added complexities” (3). As Bacon condemns the tendency to parallels and conjugates and Laplace decries explanation via final causes and hazard, Comte rejects the assumption that human social life is “always exposed to disturbance by the accidental intervention of the legislator, human or divine” (Comte vol. 2, 215). Although Comte did not believe that clear-cut laws of social physics would be immediately graspable, he did contend that they ultimately would be. At that time, humanity would find “social phenomena, like all others, [subject] to invariable natural laws, which shall, as a whole, prescribe for each period, with entire certainty, the limits and character of social action” (Comte vol. 2, 240). Thus, by the time the writers of this study are experimenting with data collection as a mode of representation, they are participating in a data discourse that posits the human as object of empiricist inquiry in the same terms as applied to the world of material.

More recent enunciations of the dream of exhaustive data collection echo this desire for and presumption of ultimate predictive certainty. On January 11, 2007, computer scientist Jim Gray addressed the Computer Science and Telecommunications Board with a vision of a “fourth paradigm” (Bell, xi) of scientific research driven by the collection, curation, and analysis of massive sets of data. Experiment, theory, and computation would be supplanted by data collection as the primary method of scientific investigation. Gray’s “dream of establishing a ‘sensors everywhere’ data infrastructure” (Bell, xv) is the Baconian ideal and Laplacian subjectivity technologized and imagined as a realizable goal. This is a direct result of centuries of empiricist theory and desire. As one of Gray’s respondents puts it, “Data is the result of incremental advances in empiricism-serving technology” (Wilbanks, 211). These new technologies have brought us closer to exhaustivity of collection for all domains, natural and human: “Data comes in all scales and shapes, covering large international experiments; crosslaboratory, single-laboratory, and individual observations; and potentially individuals’ lives” (Bell, xiii). The bounds of data’s potential to generate knowledge are, again or still, being promoted as limitless.

The connection that Bacon, Laplace, and others make between exhaustive data collection and ultimate knowledge is the basis of what I am calling the epistemology of data: the belief that reality is most accurately represented as and will ultimately be understood through a collection of data points. Bacon’s “collection of particulars” taken to its Laplacian conclusion would create an exhaustive and externalized representation of the world. Through data we would come to know the world by being able to see it, mediated through time- and place-bound points of information that are then unmoored from time and place to allow us access to a reality that encompasses all times and places. Data’s epistemology is empiricist in that it reifies and relies on the recording of sensory perception. It is realist in that it places primacy on existence rather than consciousness and assumes that there is a reality to be recorded and understood, even if imperfectly. This belief imparts a sense that data collection is imperative if reality is to be known and understood.