In the spirit of Emile Zola's Le roman experimental (1880) in which the naturalist novel becomes a kind of laboratory subject to the laws of scientific method and experiment, the modernists set up their own laboratories for the production of literature, film, radio, the performing arts, and the visual and plastic arts. From Hugo Munsterburg's psychology lab at Harvard and Alfred Stieglitz's 291 gallery in New York at the turn of the twentieth century to the European and North American art and design labs of the 1920s and 1930s, the modernist period witnessed the emergence of institutional formations that brought together artists, writers, film makers, architects, sociologists, psychologists, anthropologists, economists, scientists, and engineers in a transatlantic cultural movement that traversed disciplinary boundaries and fostered new modes of collaboration. What these laboratories of art and design consistently demonstrate is the extent to which the institutional structures of the artistic avant garde were informed by their critique of scientific and corporate models of research; their critique of the very models they reference varied from ideological rejection of industrial capitalism and Western science to aestheticist reform of industrial design. With the creation of studio-laboratories in the 1960s and 1970s, the late twentieth century welcomed a new generation of collaboration among artists, scientists, engineers, and industry that modeled itself on avant-garde labs of the early twentieth century. At the opening of MIT's Media Laboratory in 1985, Nicholas Negroponte announced that its experiments with digital media were "as much like the Bauhaus as a research lab" (quoted in Bass 1993), which at once moved toward the formation of digital-humanities and new-media laboratories of the late twentieth and early twentieth-first centuries and, at the same time, returned to the avant-garde labs of the modernist period.

This chapter extends from research on the genealogies of modernist laboratories (aesthetic, scientific, and corporate) and their relationship to the formation of contemporary digital-humanities and new-media laboratories and collaboratories.^[1] The trajectory of this project, broadly conceived, leads from the modular principles of early twentieth-century industrial design and avant-garde aesthetics, to the implementation of modular architecture in the construction of mid-century corporate science labs, to the modularization of markup languages, interoperable digital tools, and collaborative and cross-disciplinary lab environments in the late twentieth and early twenty-first centuries. In doing so, I traverse the conjuncture of multiple modularities and disciplinarities—from architecture, art, and industrial design to markup languages, source code, and digital tools—in the historical contexts of cultural, industrial, and postindustrial modernities. In this chapter, in particular, I bring into focus the intersection between the production of modernist art as aesthetic experiments conducted under laboratory conditions—sometimes literally, other times figuratively—and the positioning of digital humanists in lab environments performing text-mining and visualization experiments with data and metadata derived from digitally remediated materials originally produced in analog formats by the historical avant garde. Among the contemporary practitioners and labs whose experiments probe novel approaches to modernist textual studies are those included in this chapter: the Speculative Computing Laboratory at the University of Virginia, founded by Johanna Drucker, Jerome McGann, and Bethany Nowviskie; the Stanford Literary Lab, founded by Franco Moretti and Matthew Jockers; the Harvard Cultural Observatory, founded by Jean-Baptiste Michel and Erez Lieberman Aiden; and the Modernist Journals Project Lab at Brown University and the University of Tulsa, co-directed by Sean Latham, Robert Scholes, and Jeff Drouin.^[2] Over the past two decades, these labs and others have emerged as institutional locations at which critical theorists, textual scholars, librarians, archivists, programmers, scientists, and engineers have converged in the implementation of collaborative methodologies and technologies that have transformed and tested the increasingly expansive scale and versatile disciplinary mix of modernist textual studies.

New-media modernism

As much as tropes of innovation and experiment were commonplace for the ways in which modernists imagined themselves and their aesthetic practices, they have now become synonymous in the co-emergent formations of "new-modernist" and "new-media" studies. One of the common tropes in new-media studies, at least since the publication of Jay David Bolter and Richard Grusin's Remediation in 1999, constructs analogies between late twentieth and early twenty-first century digital media and early twentieth-century modernism and the avant garde. Bolter and Grusin's comparison of "the rhetoric of cyberspace" to the manifestos of the futurists and the suggestion that "cyberspace enthusiasts have a similar relationship to technologies of representation that Marinetti and the futurists had to technologies of motive power" is one example of their collocation of modernism and digital media (1999, 54). One of their main premises is that digital objects call attention to their remediated materiality in a manner analogous to the aesthetics of modernist self-referentiality with its insistence on foregrounding its own mediations. This premise is not theirs alone either: it's a recurrent claim in the work of new-media theorists (see Goble 2010; Manovich 1999; Manovich 2001; Manovich 2002; Manovich 2003; North 2005; and Pressman 2013). In other words, just as the modernists were audacious enough to proclaim themselves the first to be modern, so theorists and creative practitioners working with digital media claim for their discipline and their art the distinction of being the first to be new.

As influential as the work of new-media theorists has been in shaping my thinking about remediated modernism, I find myself uneasy with the analogical correlation of either modernism or the avant garde and digital media. It's a seductive trope, one that permits us the fantasy of a transhistorical modernism that focuses on what Matthew Kirschenbaum calls the "formal materiality" of interface rather than the "forensic materiality" of storage (2008, 10). New media's always-new modernism is troubled by persistent reminders of the historicity and materiality of digital objects. One of the ways in which we experience this online is when we click on a broken link and get an "Error 404: Not Found" message. If, in the normal functioning of a website, we come to expect that the information we seek will be rendered in what Lisa Gitelman ingeniously calls (in a variation on Gertrude Stein) the "continual, continuous present tense" of the Internet (2006, 145), the error message makes us aware of the historicity of hyperlinks whose brokenness refers us to a temporality to which we no longer have access. They interrupt our navigation of digital space, create ruptures, and remind us of the persistent materiality of remediated objects. Errors also force us to acknowledge the labour necessary to the production of digital media, for it's when the link is broken that we have to confront the fact that remediated materialities get old, require rebuilt code, and reside on machines that need maintenance. What gets effaced in reading modernism in the "continual, continuous present tense" of new media's always renewable interface is the sedimentation of labour time and economic mode of production in the storage and retrieval of data. The fantasy of an always-new modernism dissolves once we come to understand the ways in which digital objects are not perennially renewable resources, but rather material entities that age, degrade, and break down and require continual reinvestments of labour and capital.

Where I do see the productive correlation of early twentieth-century modernism and its digital remediations is in the turn to the laboratory as an environment for research, experimentation, and innovation. Lev Manovich's correlation of the laboratory research of the 1920s avant garde with the software research lab seeks "to reactivate the concept of laboratory experimentation," which he locates in "the kind of research undertaken by Russian and German avant-garde artists of the 1920s in places like Vkhutemas and Bauhaus, as they explored the new media of their time: photography, film, new print technologies, telephony" (2001, 15). While Manovich is correct, I think, in locating among the historical avant garde specific institutional formations that correspond to the research environment of digital humanities labs, his desire to correlate avant-garde software design and modernist new media exposes a tendency towards the aestheticization of technology that abstracts it from the material conditions that figure so concretely in the practice of laboratory research. The actual laboratory environments in which members of the avant garde generated research and experimentation have been the subject of case studies and histories of cultural movements, but these laboratory modernists extend far beyond the Russian and German examples to which Manovich alludes. What remains to be considered is not only the transatlantic reach of modernist laboratory practice from Europe to North America but also the experimental interpretive strategies that digital-humanities and new-media practices bring to the examination of laboratory modernisms using the tools and methods of lab research.

Laboratories, collaboratories, observatories

Bruno Latour and Steve Woogar's landmark 1979 study Laboratory life records their ethnography of a neuroendocrinology lab in which they document the observations of a "fictional character" who posits, after a period of initial observation, that "the laboratory began to take on the appearance of a system of literary inscription" (1986, 52). This hypothesis didn't sit well with the laboratory's workers. "Indeed," Latour and Woogar report, "our observer incurred the considerable anger of members of the laboratory, who resented their representation in some literary activity" (1986, 53). The corollary of Latour and Woogar's hypothesis is obvious enough: if the lab is a literary institution, then the institutional formation of the lab isn't the exclusive property of the sciences. Or, more troubling, the sciences are themselves a literary institution. What if we were to reverse their hypothesis and examine the system of literary inscription in the humanities as one of laboratory experiments?

Amy E. Earhart's study of collaboration in the digital humanities tentatively takes up this premise in an analysis of the field that underscores the need to differentiate collaborative institutional formations and practices in the humanities from concepts of "interdisciplinarity or intradisciplinarity" (2011, 28). While Earhart isn't prepared to advocate that the digital humanities uncritically adopt the "science laboratory model" (2011, 31) as the basis for collaboration, she does concede that the laboratory is predicated on an institutional economic model different in both scale and kind from the centres and institutes at universities in which digital humanities research more typically finds infrastructural support.^[3] "This model," she concludes, "has not made headway into the digital humanities" (2011, 33). While the establishment of digital-humanities laboratories across North America over the past decade or so indicates that the nomenclature of the sciences has been transposed to the humanities, it is evident after even a cursory survey of the economic models and infrastructure supporting these institutional formations that some are comparable to science labs and others are labs in name only. As a research unit, the digital-humanities lab is no less in need of material infrastructure—space, equipment, personnel, administration—than any science lab. This isn't to say that centres and institutes can exist without such support but that the dedicated infrastructure of a lab is recognized for its function as a research unit—one that under the economic regime of academic institutions is capable of securing resources from government, funding agencies, private donors, and industry partnerships to ensure its sustainability. What makes the digital-humanities lab possible under this institutional regime is the development of funding models designed to capitalize on digital economies. I don't want to suggest, however, that this necessitates the capitulation of the humanities to the science or corporate laboratory models. Rather, as in the case of the historical avant garde, the digital-humanities lab may follow a different cultural logic, one that cannot exist without the digital economies of post-industrial capitalist modernity, and one that may instead serve as a structure of institutional critique. The digital-humanities lab is not merely an interdisciplinary or intradisciplinary expression of post-industrial modernity; it is an institutional formation whose cultural logic and interests often subsist in tension with both the economic regime that drives technological innovations and the industry-oriented research agendas of colleges and universities that make such labs possible.

Rather than intradisciplinarity or interdisciplinarity, the digital humanities and its institutional formations might be better described in terms of extradisciplinarity and paradisciplinarity. Matthew Weinroth defines extradisciplinarity in reference to research that "may take place outside of the traditional confines of disciplines, but it remains within the context of epistemic cultures in that it refers to the structure of disciplines…. The disciplinary frameworks provide the grounds for the pursuit of solving a problem but take these further in the collaborative element" (2009, 165-166). By paradisciplinarity, he points to "collaboration in-between disciplines through dialogue and exchange, drawing on the differences between collaborators. Thus, paradisciplinarity can be conceptualized as counter-positioning disciplines by comparing formal aspects of disciplines (methods, theories, even tools or instruments) in order to open up and/or widen relations between individual areas of knowledge production" (Weinroth 2009, 166). In either case, collaboration is the co-operative keyword in both formations: where extradisciplinary research retains disciplinarity as an epistemological structuration that pursues problem-solving outside traditional disciplines via collaborative disciplinary processes, paradisciplinary research occupies an interstitial position between disciplines that foregrounds their divergence as a collaborative means of generating knowledge via differential critique. Reframed by the "modularity principle," extradisciplinarity provides a model in which the digital humanities works outside traditional disciplinary regimes to access new modes of modular interoperability, and paradiscplinarity locates the digital humanities at the juncture between disciplines so that otherwise contradictory modularities—for instance, corporate and avant-garde—engage in counterpositional and collaborative critique (Manovich 2009, 50). Where extradisciplinary collaboration operates beyond inherited institutional formations, and may lead at times to the transformation of the laboratory itself, paradisciplinary collaboration situates the modernist lab at historical moments of intermodular disjunction and uneven development.

Although the nomenclature of the lab has become increasingly more common in humanities contexts, most of these institutional formations bear closer resemblance to the collaboratory. Coined by William A. Wulf in a 1989 white paper, the collaboratory is defined as "a center without walls, in which the nation's researchers can perform their research without regard to geographical location" (quoted in National Research Council 1993, vii). "In operational terms," as the authors of a 1993 National Research Council (NRC) report elaborate, "a collaboratory is a distributed computer system with networked laboratory instruments and data-gathering platforms; tools that enable a variety of collaborative activities; financial and human resources for maintaining, evolving, and assisting in the use of computer-based facilities; and digital libraries that include tools for organizing, describing, and managing data, thus enabling the large-scale sharing of data" (1993, 7). Given these operational definitions from the NRC, it is not unreasonable to extrapolate from Latour and Woogar's analysis to describe the collaboratory as "a system of literary inscription." Current research initiatives have already claimed the concept for the digital humanities, including the Humanities, Arts, Science, and Technology Advanced Collaboratory (HASTAC), co-founded in 2008 by Cathy N. Davidson and David Theo Goldberg, and the Canadian Writing Research Collaboratory/Le Collaboratoire scientifique des écrits du Canada (CWRC/CSEC), founded in 2010 and directed by Susan Brown. If at its inception the collaboratory excluded the humanities, these recent institutional formations instead foreground the collaborative and complex disciplinary potential of the concept itself.

Most accounts of the mid-century modernist laboratory and its direct influence on contemporary institutional formations located at the intersection of information technologies and the arts and design begin in the 1960s with the emergence of "studio-laboratories." Among the earliest of these was Experiments in Art and Technology (E.A.T.), founded by artist Robert Rauschenberg and Bell Labs physicist Billy Klüver in New York in 1966. As outlined in the proceedings from its 1969 conference, E.A.T. created "an international network of experimental services and activities designed to catalyze the physical, economic and social conditions necessary for cooperation between [sic] artists, engineers, and scientists" (NRC 2003, 122). E.A.T. rejected the idea of a "single laboratory or information center" and positioned itself as a mediator—a "matching agency," to be precise—between artists and engineers to facilitate collaboration (Klüver and Rauschenberg 1967, 2). As an institutional formation, E.A.T. functioned as a pre-digital collaboratory; rather than digital networking, it implemented at its earliest stages the mimeograph technology of its modernist periodical predecessors, publishing E.A.T. News (1967-68) with detailed reports on its artist-engineer experiments, installations, and collaborations. With Klüver's affiliation with Bell Labs, E.A.T. served as an extension of the mid-century corporate laboratory into the New York art world. (For an account of E.A.T. collaborations with Bell Labs and, specifically, computer art projects facilitated by Bell Labs between 1961 and 1972, see Patterson 2015.)

Also in the late 1960s, the Hungarian-American artist Gyorgy Kepes founded in 1967 the Center for Advanced Visual Studies (CAVS) at the Massachusetts Institute of Technology (MIT). A former instructor at Moholy-Nagy's New Bauhaus (and its variant iterations) in Chicago from 1937-43, Kepes began in 1946 teaching visual design at MIT, where he devoted much of his research to the interconnections between technology and art. At CAVS, he revived a New Bauhaus ethos to bring about the "absorption of the new technology as an artistic medium" and "the interaction of artists, scientists, engineers, and industry" (quoted in Finch 2009). A one-time artist-fellow at CAVS, Muriel Cooper carried on its Bauhaus legacy at MIT in two ways: as art director at the MIT Press and as founder of the Visible Language Workshop (VLW). David Reinfurt's concise history of Cooper's work at MIT places considerable emphasis on her engagement with the Bauhaus, calling attention to her innovative engagement with techniques and technologies of graphic design and mass production. Her most durable and visible mark on the MIT Press was her 1964 design of its logo, a row of seven books, with the fourth pulled up and the fifth pulled down, which not only produces "an abstracted form of the abbreviation 'MITP'" but also "a clear barcode—as the products of mass production sit together in an orderly row, dematerialized into the pure information of a machine-readable graphic" (Reinfurt 2005, 3). Further, Cooper's most famous book design is the MIT Press English translation of Hans Wingler's Bauhaus: Weimar, Dessau, Berlin, Chicago (1969), a two-year project which she later described as serendipitous: "My design approach always emphasized process over product, and what better place to explore this than in a tome on the Bauhaus, the seminal exploration of art and design in an industrial revolution" (quoted in Reinfurt 2005, 10). Combined with its expressed purpose to serve as an extension of the Bauhaus-Archiv in Darmstadt, the monumental proportions of the volume speak to Cooper's work as the architect and builder of a mobile institutional structure, one that bears the signature markings of Bauhaus modular typographic design. Beyond her work at MIT Press, Cooper went on to implement elements of Bauhaus pedagogy with her founding of the Visible Language Workshop in 1974. Vividly recalling Walter Gropius's "laboratory workshops," she characterized the VLW as "a unique interdisciplinary graphics laboratory" (quoted in Reinfurt 2005, 10), one that brought a Bauhaus design ethos and production environment into the age of computers and digital media.

With the 1985 opening of the MIT Media Laboratory, under the directorship of Nicholas Negroponte, Cooper merged the VLW with the new lab. Reiterating the VLW's homage to the Bauhaus—in a 1989 special issue of Design Quarterly on "Computers and design," edited by Cooper—she positioned the Media Lab in the same genealogy: "The Media Laboratory is a pioneering interdisciplinary center that is a response to the information revolution, much as the Bauhaus was a response to the industrial revolution" (1989, 18). In her representation of collaborations between artists and scientists at the Media Lab, Cooper returns to the Bauhaus analogy:

The Media Lab's greatest strength may prove to be the collision of the disparate disciplines and values represented there. The valuation models of a scientific community do not easily mesh with those of the art community although they avowedly seek the same grail. In much the same way, the meaning of the Bauhaus was in the conflict between painters like Klee and Feininger, and technocrats like Moholy-Nagy. (1989, 20)

For a 1993 interview with Wired magazine, Negroponte confirmed Cooper's suggested lineage, saying, "You also have to realize how different we are from Bell Labs, Xerox PARC, or any other lab that's ever existed. We are as much like the Bauhaus as a research lab. No photographers, filmmakers, or typographers go to work at Bell Labs in the same way that they do at the Media Lab" (quoted in Bass 1993). In his 1995 book Being digital, a reworking of material originally published in his monthly column in Wired, Negroponte offers a different historical precedent: "As in 1863, when the Paris art establishment declined to let the Impressionists into its official show, the founding faculty members of the Media Lab became a Salon des Refusés and had one of their own, in some cases too radical for their academic department, in some cases too extraneous to their department, and in one case with no department at all" (1995, 225). From its inception, when Negroponte first convinced MIT to build the Media Lab in 1979, it aimed to be an extradisciplinary formation, a principle that he endorses in his claim that "the common bond" among its founding faculty "was not a discipline, but a belief that computers would dramatically alter and affect the quality of life through their ubiquity, not just in science, but in every aspect of living" (1995, 225). Here he anticipates the way in which the mass production of computers would lead to social and institutional transformation during the 1990s in which the avant-garde was no longer Media Lab but a globally dispersed network of hackers (1995, 226). Already apparent in Cooper's 1989 portrait of the Media Lab, Negroponte's originary vision of its founders' avant-garde extradisciplinarity migrates toward institutional paradisciplinarity. As early as the 1985 opening of the fifty million dollar Weisner Building designed by architect I.M. Pei (who had recently completed his design work on the Louvre's steel and glass pyramid in Paris), the Media Lab could no longer lay claim to its avant-garde status as a salon des refusés. Rather, appropriate to the aesthetic heritage of MIT's studio-laboratories, Pei's style paid tribute to the late-twentieth century's institutionalization of visual and architectural modernism (Cannell 1995; Wiseman 2001).

While the studio-laboratories of the 1960s, 1970s, and 1980s were primarily oriented toward collaborations among engineers, scientists, and practitioners in the visual arts and architectural design, the digital-humanities laboratories from the decades at the millennial turn brought experimentation with literary texts and critical theory into lab environments. Under the auspices of the Speculative Computing Laboratory (SpecLab) at the University of Virginia, the earliest literature-based digital laboratory was founded in the 1990s by Johanna Drucker, Jerome McGann, Bethany Nowviskie and others. SpecLab takes its inspiration directly from the methods employed by the early twentieth-century avant garde. SpecLab's "working unit," which they call Applied Research in 'Patacriticism, pays homage to the work of Alfred Jarry, who in 1893 coined the term 'pataphysics—itself a paronym that plays on metaphysics—to refer to "the science of imaginary solutions" (Jarry 1972, 7). While SpecLab has been ostensibly responsible for the development of tools under the institutional rubric of NINES, Drucker differentiates between the work of making "digital tools in humanities contexts," which she attributes to the digital humanities, and the work of making "humanities tools in digital contexts," which she aligns with the 'patacritical project of "speculative computing" (2009, 25; original emphasis). To say that Drucker makes a fine distinction between the digital humanities and speculative computing is perhaps beside the point, since she is principally concerned with preserving a space for experimental digital research in the humanities that aligns itself with the historical avant garde. And, in doing so, she positions speculative computing squarely beside the avant-gardism of other new-media modernisms. There are legitimate reasons why she wants to correlate speculative computing with the cultural logic of the avant garde, not least of which is to formulate "a set of principles through which to push back on the cultural authority by which computational methods instrumentalize their effects across many disciplines" (Drucker 2009, 5). For Drucker, the digital humanities represents the instrumentalization of the humanities, its reification by the very tools that become both the means and ends of research. Although she does not cite any of the prominent and influential twentieth-century critiques of instrumental reason by Horkheimer and Adorno, Heidegger, Gadamer, and Habermas, it's clear enough that these thinkers stand somewhere behind her conception of speculative computing as a 'patacritical project. That is, she positions speculative computing in relation to the historical avant garde's self-distantiation from an instrumentalized bourgeois society by disarticulating the "working unit" of SpecLab from an instrumentalized digital humanities. What Drucker and her SpecLab colleagues neglect to interrogate, however, is the institution of the modernist laboratory itself—a space of scientific and aesthetic experiment inhabited by practitioners of physics and 'pataphysics alike.

A collaboratory by any other name, one of the most news-making digital-humanities institutions of the past decade has been the Stanford Literary Lab, founded in 2010 and directed by Franco Moretti (together with Matthew Jockers from 2010 to 2012). Anyone familiar with contemporary scholarship in the digital humanities will be aware of Moretti's concept of "distant reading," coined in his 2000 essay "Conjectures on world literature," which spawned extensive debates in the New Left Review and elsewhere. When Moretti posited his original definition of distant reading, he placed it in opposition to "close reading (in all of its incarnations, from new criticism to deconstruction)" (2000, 57). In doing so, he aligned his method with "scientific work" (2000, 54). Following Max Weber, Moretti claimed for distant reading the status of a "new 'science'" in which "a new problem is pursued by a new method" (2000, 55). Yet this critique of close reading is premised on a misconception of its object, since I.A. Richards not only conceived of his "practical criticism" as an "experiment" analogous to scientific method (1952, vii) but also recognized the classroom as a "philosophic laboratory" (1955, 104). "A book is a machine to think with" (2001, vii), Richards announced in the opening sentence of his Principles of literary criticism; yet few today would consider him a precursor to the algorithmic "reading machines" (Ramsay 2011) imagined by digital humanists. Rather than a methodological break, then, distant reading might have been better conceptualized as a radical return to an experimental science of literary investigation. Even so, it is symptomatic that digital humanists should have taken up Moretti's claim for a new science and a new method so readily: this is the recurrent trope of modernist innovation that pervades the discipline.

Even as digital humanists adopted his method of distant reading, Moretti separated himself from the term as he moved into his own phase of digital research at the Stanford Lab. Notably, there is no mention of digital technologies or methodologies in his original formulation of distant reading, nor does it appear in the "Graphs, maps, trees" essay series (2003-04) or its revision collected under the same title in his influential 2005 book. Not until 2009 does Moretti raise the possibility that "in a few years, we will have a digital archive with the full texts of (almost) all novels ever published" (2009, 134). More recently, in his 2011 pamphlet Network theory, plot analysis, he speculates that "the rise of quantitative evidence" in literary studies over the "last few years" promises new results "because this time we have digital databases, and automated data retrieval" (2011, 2). At once utopian and determinist, Moretti's digital turn accentuates its investment in the technological progress of modernity. In particular, he points to the collaboration between the Cultural Observatory at Harvard and Google Labs on the culturomics project and their December 2010 launch of the Ngram Viewer as an exhibition of the ways in which "the width of the corpus and the speed of the search have increased beyond all expectations" and have allowed researchers to perform "in a few minutes investigations" what took earlier generations "months and years of work" (Moretti 2011, 2). Echoing Moretti's assertions about distant reading a decade earlier, the lead culturomics researchers Jean-Baptiste Michel and Erez Lieberman Aiden lay claim in their report to "a new type of evidence in the humanities" that they compare to the interpretation of a "great cache" of fossilized bones "from which to construct the skeleton of a new science" (Michel et al. 2011a, 182). That their culturomic analysis of "linguistic changes" as evidence of "cultural change" produces model "trajectories" (Michel et al. 2011a, 182) of linear progress is fitting, since the methodology upon which they base their conclusions is specifically intended to redesign humanities research using the tools of a new science.

What differentiates the Stanford Literary Lab from the Harvard Cultural Observatory and Google Labs? Both are models of institutional and paradisciplinary collaboration engaged in digital text analysis and visualization—one based in the humanities,^[4] the other in the sciences—but the modes of interpretation that each brings to textual data are premised on different analytic concepts. The most prominent of these concepts is culture. For a project that purports to examine the textual records of entire cultures, it's unclear why the Harvard Observatory makes no attempt to define what it means by "culture," let alone unpack its neologism "culturomics."^[5] The Stanford Lab's immediate response after the Ngram Viewer's launch was decidedly skeptical, as evidenced by co-director Matthew Jockers's December 2010 blog post in which he called into question both the conceptualization of culture employed by the Harvard Observatory and its reliance on data derived from Google Books: "To call these charts representations of 'culture' is, I think, a dangerous move. Even at this scale, the corpus is not representative of culture…. More than likely the corpus is something quite other than representative of culture. It probably represents the collection practices of major research libraries…. The leap from corpus to culture is a big one." While Jockers is doubtless correct in his methodological critique that a corpus of books doesn't constitute a culture—especially one that systematically excludes magazines, newspapers, printed ephemera, manuscripts, maps, artwork, and so on—he sidesteps the concept of culturomics itself and its portmanteau terminology. According to Michel et al., culturomics refers at once to the object of analysis ("linguistic and cultural phenomena" [2011a, 176]) and its methodology ("the application of high-throughput data collection to the study of human culture" [2011a, 181]), but the keyword itself is left undefined—an odd omission for researchers working on the relationship between linguistic and cultural change. Linguists have blogged about multiple interpretations, one concluding that the model is most likely the field of biology and its allied sciences in which -omics coinages proliferate (for example, genomics) but conceding that the "temptation to read -omics as connected with economics is a strong one" in the contemporary vernacular (for example, Reaganomics, Freakonomics) (Zimmer 2010). Whether deliberate or not, the portmanteau culturomics is itself a sign of linguistic modernity, an experimental coinage of a new science in a new language.

Commonalities between the Harvard and Stanford labs reside in their mutual interests in genomics and economics. Moretti's adaptation of models from evolutionary biology and genomics for his distant readings of literary genre in the "Trees" chapter is examined at length in Alberto Piazza's afterword to Graphs, maps, trees (Piazza 2005). Jockers, too, is the progenitor of a neologistic methodology that he calls "macroanalysis," which is related to but differentiated from Moretti's distant reading, and which "is in general ways akin to the social-science of economics or, more specifically, macroeconomics" (2010; see also Jockers 2013). There are suggestive linkages between macroanalysis and the economic interpretation of culturomics. Similar to macroeconomic principles that focus on the performance and trends of whole economies rather than the microeconomics of individual agents and specific markets, culturomics constructs abstract models that represent short-term fluctuations in cultures and long-term cultural trajectories reflected in changes to linguistic phenomena across decades and centuries. For instance, Michel and Aiden bring this approach to an analysis of the "history of economics" itself that traces the linguistic shift from older concepts such as "banking" during the Great Depression to a "new economic vocabulary" (for example, recession, GDP, the economy) that "entered everyday discourse" at mid-century (2011b, 73). While the cultural trend that they identify here is the emergence into the vernacular of concepts derived from modern macroeconomics, the computational method that they practice is consistent with the principles of "modernist" economics (see McClosky 1985). That is, their text analysis and visualization constitute a model of the cultural trajectory that leads to the formation of culturomics itself.

Modularity

To take the longer historical view of these modernist and new-media modernist institutions—one that spans some version of the long twentieth century—we may begin to see how their recursive formations exhibit what Manovich calls the "modularity principle," whose para- and extra-disciplinary dispersion has informed more than a century of laboratory design and research in the arts, sciences, and industry. The laboratory's progress through this long century has been one loaded with disciplinary obstacles, not least of which are the contradictions between industrial and cultural conceptions of modularity. Manovich's account of the modularity principle attempts to address these contradictions by theorizing it in terms of the uneven development of industrial and cultural modes of production under the economic regime of modernity. This contradiction between corporate and avant-garde modularities speaks to the ways in which the laboratory typifies the asynchrony of industrial and cultural modernities. While the factory division of labour can be seen in the studio production of animated films and video games, this kind of cultural labour is qualitatively and quantitatively different from the "systematic character of industrial standardization" achieved when Ford installed the first moving assembly lines (Manovich 2009, 50). Rather than anticipate the time when cultural modularity catches up to industrial modularity, the uneven development of modularities is such that their respective institutional formations are perpetually out of sync. Cultural modularity required the development of industrial modularity in order to constitute itself (as Adorno argued for modernist art [1997, 34]), but as a mode of negation and critique; both modularities owe their existence to the historical conjuncture of industrial capitalism, and neither can be understood without reference to the other.

Consistent with Manovich's analysis of new-media objects, both the Harvard Observatory's and the Stanford Lab's experiments are predicated upon the modularity principle. Where Manovich extends the "fractal structure of new media" to the modular organization of the Internet as a whole (2001, 31), culturomics and macroanalysis take a subset of that modularized structure—the online database—as the object of their investigation. Both approaches modularize remediated textual objects and organize them into structured collections of data; they redistribute the text's linguistic code into a corpus consisting of statistical tables—which can in turn be ported into different types of databases—that effectively transform the bibliographic object into standardized modular units made interchangeable with other texts. With its transformation into statistical tables, modularization strips the text of its bibliographic code and reifies its linguistic code. The reification of linguistic code renders the bibliographic object as a raw resource, a machine-readable textbase abstracted from its human-readable commodity form and its material history in the social life of exchange and consumption. This creates the conditions necessary for the mass production of textuality: data visualization produced on demand, customizations "on the fly,""just in time" deliveries made possible by the postindustrial logic of digital modularity (Manovich 2001, 36). Aggregated as statistical abstractions, linguistic code is remediated by the culturomic or macroanalytic interface as the graphical inscription of a new modular artefact. Rather than reinscriptions of print media, these are new inscriptions in a new medium. This is the realization of what Fredric Jameson calls the "media paradox" of modularity, "where intensified change is enabled by standardization itself" and where the "module would then constitute the new form of the object (the new result of reification) in an informational universe" (1990, 58). This, too, is an instantiation of new-media modernism: freed from medium specificity, modularized data is released from the bibliographic codes that anchor the text to its prior materiality and cultural modality, not to be recombined in the same repeated form (as with modularity under industrial modernity) but reconfigured in endlessly variable patterns (Manovich 2001, 36; Manovich 2009, 46).

One of the more recent laboratory environments to emerge comes out of a pioneering institution in digital modernist studies. The Modernist Journals Project (MJP), which is based and jointly operated at Brown University and the University of Tulsa, was founded in 1995 as "a website of digital editions of periodicals connected to the rise of modernism in the English-speaking world" (MJP 2015a). It continues to lead the way for digital modernist scholarship and pedagogy with its newest developments, which include its teaching resources pages and instructional wiki as well as the MJP Lab. Where the main MJP site is oriented toward the production of reading and search interfaces for its digital editions, the MJP Lab is organized around the visualization of the data "generated over the course of digitizing magazines from the early 20^th century" (MJP 2015b). If the main site is directed toward the creation of a resource that backs the MJP's motto ("modernism began in the magazines") and its concomitant claims about the institutionalization of an Anglo-American modernist literary-historical tradition, the lab site looks toward the expansion of that institutional formation through the expressive modes of visual abstraction (timelines, tables, charts, trees, sunbursts) and new inscriptions of modularized data in graphical interfaces. The trope of innovation that surfaces on the lab site foregrounds the ways in which the MJP Lab is itself a new institutional formation and its experimentation with visualization tools renders the textual data of modernism as new digital objects. Upon entering the MJP Lab, for instance, we are welcomed into a space that not only announces itself as innovative but also professes to facilitate cutting-edge research: "The site is experimental," we are cautioned, "but it's also dedicated to experimentation―playing with the MJP data, and drawing new patterns and knowledge out of its journal files" (MJP 2015b). This is new media's modernism, one that aligns contemporary digital experimentation with the aesthetic innovations of the modernists themselves.

The MJP Lab is a research environment with an emphasis on digital-humanities pedagogy. This is the digital companion to the MJP's teaching pages and instructional wiki, which are mainly oriented toward reading magazines principally as print objects and only supplementally in the remediated format of the digital repository. Informed by instructions from MJP director Robert Scholes and technical advisor Clifford Wulfman in their chapter "How to study a modern magazine" from Modernism in the magazines (2010, 143-67), the website includes its own pedagogical guide titled "How to read a magazine" (MJP 2015c). This online manual advises how to read a magazine in the context of its print culture, how to parse a magazine into its component parts, how to devise potential research projects, and how a magazine can be read as a material, printed object (examples provided include skimming, browsing, surveying, grazing, sampling, rooting around, drifting, and circling back). Readers are briefly instructed to familiarize themselves with the MJP search engine so that they "can integrate searches of the database into [their] reading" (MJP 2015c), but there are no instructions on how to perform customized searches or how to interpret search results. How-to readers are left with the impression that one doesn't actually read magazines by using a search engine, or that database searching isn't itself a mode of reading; it is, rather, presented as a supplement to reading practices oriented around the magazine as a physical object. With the MJP Lab, which "may be viewed as a supplement to the MJP's search pages" (MJP 2015b), the site reorients its pedagogy toward digital reading practices. As a supplement to the search engine and its text-based interface, visualization is figured as an experimental mode of reading the MJP as a collection of digital objects. In order to facilitate its pedagogy, the lab provides open access to a portion of its data and metadata, produces sample visualizations, and proposes ideas for further visualizations. Available for download from the MJP Lab's Sourceforge page (MJP 2011), its datasets include Metadata Object Description Schema (MODS) catalogue records, transcriptions with Text Encoding Initiative (TEI) markup, and plain-text files (derived from the MODS records) for six journals (the Egoist, the Freewoman and the New Freewoman, the Little Review, Others, and Poetry). Complete with instructions on how to use digital tools, perform quantitative analysis, and generate visualizations, these datasets are not merely supplemental to the digital editions of the magazines; they are integral to the development of modes of reading remediated modernist magazines as digital objects and to the inauguration of an experimentally practiced pedagogy that foregrounds the materiality of digital media.

Even prior to the creation of its lab and its experimentation with modularized datasets, the MJP's digitization and markup of modernist magazines remediated their modularity. After all, the modernist magazine is already modular in design: its structural elements were conventionalized and standardized in such a way that allows the MJP to compile "a checklist of categories … for an anatomical study of a magazine" and "to take stock of its many different parts—its anatomy" (MJP 2015c). Given that the historical usage of the term magazine derives from a military storehouse for munitions, it follows that magazines of the avant garde were constructed in their image as mobile architectural units, modular storehouses equipped with the armaments of modern art.^[6] While not every avant-garde magazine should be so narrowly defined along martial lines, it warrants thinking through the ways in which its formation at the nexus of corporate and military-industrial expansion during the early twentieth century anticipates the later emergence of the mid-century laboratory at "the triple intersection of big business, architectural modernism, and war" (Rankin 2010, 771). As repositories of aesthetic experimentation, these magazines were at once extensions of the avant-garde's laboratories and, as such, inscriptions of modular design. Influenced by principles of modular design incubated in modernist laboratory environments, avant-garde periodicals of the 1920s and 1930s were paper prototypes of modular labs, collapsible and portable storehouses of experimental research and art.

Remediation of the modernist magazine is an act of "media translation" (Hayles 2005, 89) in which the MJP's digitization, transcription, and markup workflow transcodes the modularity of the analog source into modular data and metadata of the digital object. With the encoding of its magazines according to the TEI Guidelines and MODS as data and metadata standards to ensure interoperability, the MJP repository implements both modular design through its markup language and metadata. As a modular and extensible markup language, the TEI in the most recent iteration (P5) consists of "a core module with essential common elements, and considers all further tagsets as additional modules which can be combined, modified, and trimmed to suit the user's needs" (TEI 2013). As the Library of Congress cataloging standard, MODS is predicated on the assumption that "[m]etadata modularity is a key organizing principle for environments characterized by vastly diverse sources of content, styles of content management, and approaches to resource description…. In a modular metadata world, data elements from different schemas as well as vocabularies and other building blocks can be combined in a syntactically and semantically interoperable way" (Duval et al. 2002). At once highly modular and strongly hierarchical, XML-based architectures such as the TEI and MODS mount considerable obstacles to reading the non-hierarchical and overlapping organization of the heterogenous parts of a magazine.

According to MJP director Sean Latham, the magazine is an "ergodic" structure, which is to say that every constitutive part is equally representative of the whole.^[7] This poses a problem for remediated magazines marked up in TEI, in so far as any non-sequential reading practice typical of magazines bypasses the hierarchical routes of XML-based encoding schemas (Latham 2016). Dating from the late 1920s, Jan Tschichold's new typography provides a characteristically modernist model of ergodic reading in which he posits that the "principles of asymmetry" release the reader from the linear strictures of "central-axis typography" as "an expression of our own movement and that of modern life":

Every part of a text relates to every other part by a definite, logical relationship of emphasis and value, predetermined by content. It is up to the typographer to express this relationship clearly and visibly, through type sizes and weight, arrangement of lines, use of colour, photography, etc…. It is true that we usually read from top left to bottom right—but this is not law…. It is therefore quite feasible to start reading a text at a different point from the top left. The exact place depends entirely upon the kind of printed matter and the text itself. (1995, 67)

Although Latham's examples aren't limited to avant-garde periodicals—but rather a more generalizable approach to reading modern magazines—the correlation of the new typography's principles of modular design with his mode of ergodic, nonlinear, and spatialized textual navigation produces an historical model of constructivist reading in which the reader becomes a producer of meaning. Ergodics exposes an ideological incompatibility between analog and digital modularities, or what Richards called in the execution of his own reading experiment "the record of a piece of field-work in comparative ideology" (Richards 1952, 6)—that is, if ergodic reading is a constructivist mode of cultural production, nonergodic reading is a consumerist mode of cultural consumption. Historically speaking, the modular XML architecture of the TEI was designed as a standardized markup language to address "a real concern that the entrepreneurial forces which (then as now) drive information technology forward would impede such integration by the proliferation of mutually incompatible standards" (TEI 2014). So if the analog magazine is at once an ergodic structure and expression of modular design, its digital transcoding produces a nonergodic translation, one that reproduces a modular structure that prescribes reading as consumption of a linear and hierarchical program language rather than reading ergodically in a manner that traverses the multidirectionality of a modular affordance. In other words, TEI markup implements its standardized language and modular architecture in the service of an instrumental logic originally intended as a business solution aimed at economic efficiencies rather than an articulation of modernism's cultural modularities. Ergodic reading may therefore provide a mode of critique of the magazine's digitally transcoded modularity, a constructivist reading practice that requires the existence of a programmatic structure—whether tables of contents or XML trees—against which it enacts its "nontrivial" and "nonlinear" traversal of the text (Aarseth 1997, 1-2). To read ergodically may further counteract what Suzanne W. Churchill and Adam McKible call "critical practices of strip-mining" (Churchill and McKible 2008, 4) magazines for textual content—or, translated to digital practice, text-mining magazines whose datasets are scrubbed of typographic codes, graphic design, layout grids, and advertising spreads and either rendered as modularized plain text or abstracted as metadata.

Given the MJP directors' documented critique of periodical studies in which texts and other paratextual materials are stripped away from their original contexts (Latham and Scholes 2006, 520-1) the MJP Lab's practice of "strip-mining" remediated magazines for their data and metadata raises certain questions about the methods used for its sample visualizations. This is not to say that the MJP directors are by any means at odds with the ideological operations of their own lab, but rather that the lab itself and its data-mining methodologies foregrounds the articulation of avant garde, scientific, and corporate modernities in a way coincident with the histories of the early-twentieth century magazine and cultural modernism. To prepare the way toward these kinds of digital laboratory experiments, Scholes and Wulfman's introduction to modernist magazines proposes rethinking the medium in which we study them as cultural objects, a shift from genre to database: "we need to develop a language of magazines" (2010, 70; original emphasis) they implore, which is to say that media translation necessitates not only understanding how to catalogue magazines as bibliographic objects in a database (for example, MODS) but also competency to know how to encode magazines in various markup languages that enable database queries that abstract from their modular design to render transformations into new modularities and reading interfaces.

To seed the databases required to produce this kind of innovative research cannot be the solitary task of the individual scholar; rather, if we are to develop sufficient competencies in reading and encoding the "language of magazines," they must issue from training that takes place in the collaborative research environment of the laboratory. In their prescient and influential 2006 article "The rise of periodical studies," Latham and his co-director Robert Scholes not only anticipate the emergence of the MJP Lab but articulate the para- and extra-disciplinary institutional formation of the modernist laboratory:

To develop such research models, we might look profitably to the sciences, where laboratories are often structured around precisely this kind of intellectual challenge. In such settings, large experiments are broken down into component parts, and particular sets of skills and expertise are brought to bear on them. The final product is then eventually integrated and published either in whole or in logical parts. Articles in the sciences are almost always written not by single individuals but by collaborators who help in varying ways to execute the project. In applying this model to periodical studies, we might therefore consider the creation of humanities labs: similarly collaborative networks of researchers and institutions that lend their collective expertise to textual objects that would otherwise overwhelm single scholars. (2006, 530)

To look at the large number of contributors to the experiments already conducted at the MJP Lab, it is apparent that the networks that Scholes and Latham originally proposed in 2005 have found their realization in their inter-institutional collaboration between Brown and Tulsa. The humanities lab they propose here is highly modularized, and their modes of collaborative research and publication closely resemble the science labs that Latour and Woogar document. All the same, I wonder why they might have overlooked models even closer to hand, specifically the laboratories of the historical avant garde and their mobile extensions—that is, the collapsible labs that the MJP takes as its primary objects of examination and experimentation. Were they to take these institutional and periodical models into consideration, Scholes and Latham might have envisioned the laboratory itself as a site of experimental reading and inscription, an extension of the models of visual and architectural modularity expressed at once in avant-garde periodicals and in data visualizations generated by digital tools.