Uncategorized

Knowledge Machines: University to Search Engine

For critics on the political left and right, the contemporary and increasingly data-driven university has become a bureaucratic behemoth, a rational system that expunges the personal and particular in favor of the quantifiable and universal. Consider here in the UK, the Research Excellence Framework and its accounting of over 191,150 “research outputs” in the past several years. Or consider the increasingly common use by top US universities of Academic Analytics, a proprietary database of PhD programs and faculty at 385 universities. “Objective data,” promises the company’s website, supports “the strategic-decision making process” at universities.

Such assessment regimes are largely seen as the delayed consummation of a disenchanted modernity, as described by Max Weber almost a century ago. Like other modern institutions and systems, universities now use technical means to control “all things through calculation,” thereby ensuring, as Weber wrote, that in principle nothing remains “mysterious or incalculable”––and, perhaps most important for our discussions today, the human. The university is not just being automated. It has become an agent of automation––an increasingly efficient calculating machine.

These calculating initiatives, claim critics such as Wendy Brown and Christopher Newfield, are part of a broader trend––the remaking of the university into a neoliberal institution. Neoliberal universities see students as revenue, sources for tuition and future philanthropy dollars; and knowledge and its transmission as capital investment.

Neoliberalism and Digital Humanities

Critics of the so-called neoliberal university, especially those within literature departments, have also begun to argue that the digital humanities in particular are fundamentally complicit with the forces of capital and its administrative minders. David Golumbia, a media culture scholar, argues that neoliberalism and digital technologies comprise a historically unique complex of money, metrics, administrative visibility, and automation. “The university,” he writes, “was for a long time not marketized, but became so, in part through the Bayh-Dole Act of 1980 that allowed commercialization of all kinds of academic research.”

Arguments such as Golumbia’s assume that before 1980 the university was unsullied by the market, commercialization, and what Max Weber described more than a century ago as the processes of modernity: accounting, calculation, rationalization etc. But this isn’t true. Since at least the eighteenth century, in Germany at least, universities and humanistic learning have, like all social institutions, absorbed and adapted processes of modernity.

The story I want to recount today, however, is not simply a historical one. Contemporary conversations about the university tend to devolve into either/or thinking: the university is either the object of vague forces such as neoliberalism or a subject of unconditioned autonomy. This false alternative not only obscures a more complex history, it also foreshortens conceptual possibilities for a future in which the university may be but one technology among an increasingly complex media and knowledge ecology.

University as Technology

I’ve introduced a number of terms: neoliberalism, DH, automation, autonomy and the university. I want to bring these together by tracing a metaphor common to them all: the university as technology. The university has long been not just as a user of technology but itself a technology. For centuries, universities have been both denigrated and celebrated as factories, mercantilist centers, encyclopedias, and machines.

But what analytic purchase does considering the university as itself a technology afford?

“The ‘content’ of any medium,” wrote Marshal McLuhan famously, “is always another medium.” The content of writing is speech; the content of print is writing; the content of the telegraph is print . . . etc. Our new digital technologies have not displaced older technologies, but they have allowed us to understand them anew.

The eighteenth-century Enlightenment, in its various forms and iterations, referred not just to emancipation from religious and secular authorities but also to an array of technologies—encyclopedias, dictionaries, taxonomies, philosophical systems and, I would add, the modern research university. Scholars, editors, publishers, intellectuals, and bureaucrats designed these technologies to manage the centrifugal experience of knowledge.

Over the course of the eighteenth century, print technologies proliferated to such a degree that scholars and intellectuals feared that books and other printed matter had themselves begun to pose a threat to human agency. And, most importantly, a threat to what Immanuel Kant called the autonomy of thought: the capacity of a uniquely human reason to set its own ends without the determination of external influences. Human reason was autonomous. Fear of too much print or too many books was not simply a fear that there was too much to know but more basically a fear about the degradation of human reason and, thus, human agency.

Intellectuals such as Kant recast worries about too much print as anxieties that books and periodicals were living things, capable of growing, infecting, expanding, and, most dangerously, thinking. The great threat to Aufklärung, wrote Kant in 1784, was not the priest, the physician or even the despot but rather “ein Buch das für mich Verstand hat”––that is, books that could think for, in place of human knowers. Before twentieth-century readers began to fear autonomous robots, eighteenth-century readers feared autonomous books.

As intellectuals and scholars began to imagine this self-organizing system of print, another institutional system emerged: the modern research university. And its first coherent instantiation was in Göttingen, where in 1734 King George II of England and elector of Hannover founded a university that would go on to serve as the model for the university of Berlin, founded in 1810.

From its very beginnings, Göttingen was designed to be a modern university. And by modern I mean it was designed and organized as an extension of the modern, increasingly bureaucratic, and revenue-seeking state. It was also understood to be part of what was coming to be understood as the market: an economic system. The modern research university arose not in opposition to the market or the state but as an institutional feature of both. Not entirely reducible to either one but increasingly bound up with both.

In 1780, a former Göttingen student likened his alma mater to a factory owned by the king: “You, Mr. Curator, are the factory director; the teachers at the university are the workers; the young people studying and their parents . . . are the customers; the sciences taught at the universities are the wares. Your king is the master and owner of his scholarly factory.”

grape_-_gottinger_universitats-_und_bibliotheksgebaude_1815

The rector organized the faculty in the production of “academic wares”: printed publications that circulated both within and outside universities as academic currency. Publications boosted the status of the university as it pursued revenues––student fees, especially those of wealthy, foreign English students (what American universities now term out-of-staters and international students who pay the full sticker price of $50K per year). The industrial language of the factory referred here primarily to the value of the university and knowledge for the state.

Göttingen’s coupling of scholarly production and prestige and academic advancement was a key element of the university’s more general system of academic mercantilism: of using all institutions for the interests of the state coffers. The value of the university was much like that of a mine, a forests, or a factory: all of these were resources to be cultivated and exploited for the state’s financial gain.

The key science of Göttingen’s academic mercantilism was Staatspolicey and what would later become Statistik––sciences that sought to make aggregates and masses of individual data visible and, therefore, as John Durham Peters puts it, “intelligible, [interpretable] and manipulable.” These early forms of state-sponsored data-management produced knowledge without an individual knower. It didn’t matter whether the state counted trees, ore, or printed books––it was all information, Statistik.

The University of Göttingen tied knowledge directly to state economic interests. And in so doing it tied the notion of knowledge to modern ideas of progress, production, value, and utility— it turned knowledge into research [Forschung]. Göttingen helped introduce the idea that knowledge could, like any other commodity, be produced and, thus, endlessly circulated, calculated, and cashed out in state coffers. Knowledge as untethered from the capacities and intentions of any particular human. And the university was its place of production; its factory.

University as Factory 2.0 

In 1798, more than 50 years after the founding of the University of Göttingen, Immanuel Kant wrote that “it was not a bad idea to treat the entire content of learning (actually the minds devoted to it) in a factory-like manner through a division of labor.” If Göttingen’s knowledge factory produced knowledge––as opposed to create, share, reveal, or pass it on––Kant’s university divided it.

Whereas encyclopedias, lexica, and other printed objects had been designed to organize the endless proliferation of the categories of knowledge, the “factory-like” university organized the people who produced knowledge. In this modern university, argued Kant, there would be “as many public teachers, professors, [and] trustees” of knowledge as there were “categories of science.” The modern university would subsume printed books and their myriad categories and maps of knowledge. It would, to recall McLuhan’s dictum about the content of one medium being another medium, not displace print but transform it again by making print the content of the university. Print would be subordinated it to a “learned community called a university.”

The idea that the “entire content of learning” could be treated in factory-like terms was, as Kant put it, “really” an idea about how to handle “the thinkers devoted to it.” The university was a factory that produced not just books and articles but the producers of those printed artifacts––that is, the scholars themselves. The university reproduced itself by educating and forming particular types of specialized scholars.

The university’s “factory-like” division of labor––intellectual specialization––also provided the conceptual and political argument for the university’s autonomy as a social institution. “Only scholars,” wrote Kant, “can judge scholars as such.” The cultivation of scholarly expertise differentiated university-based knowledge from other types of knowledge; it lent it a unique cultural authority.

In Kant’s use, the factory metaphor distinguished the university from other cultural institutions such as the state, the church, as well as other knowledge institutions such as libraries, museums, academies, and natural cabinets. It helped establish the university as an institution apart. But it also helped ensure the emergence of a gap between the norm, practices, virtues internal to the university and a uniquely academic knowledge (and prestige, status, truth, success) and what was valued outside the university.

In the Conflict of the Faculties, Kant also reorganized the hierarchy of the university faculties. He makes the lower faculty (philosophy) the higher faculty (traditionally theology) by reserving the capacity for critical thought (Kritik) to the philosophy faculty. The higher faculties (law, medicine, theology) as well as all those publics external to the university––what he terms the “businessmen” of reason––engage in something less then critical thought. And in this sense, the idea of the university as a factory anticipates key elements of automation: hierarchies of thought, mere computation vs. judgment or critical thinking vs. mere repetition, theoretical knowledge vs. technical knowledge.

Grosswissenschaft and the Archive of the Past

In his inaugural address to the Prussian Academy of the Sciences in 1858, the German historian of antiquity Theodor Mommsen declared that the purpose of disciplines like philology and history was to organize the “archive of the past.” After 40 years of pursuing idealist narratives about the past, historians, philologists and humanistic scholars of all sorts needed to reassess their epistemic assumptions and, quite frankly, rein in their ambitions. Instead of recounting the development of Geist over time, scholars needed to collect and organize the material of the archive. Knowledge of the past depended not on philosophical or literary speculation but on meticulously organized facts.

But organizing the “archive of the past” required a completely different type of scholarly work. It required less the interpretive genius of Hegel and more the unflagging industriousness of an worker [Arbeiter]: Fleiss. As secretary of the Prussian Academy, Mommsen set out to institutionalize what he termed the “industrial production of theimages.jpeg sciences” [Grossbetrieb der Wissenschaften]. Over more than three decades, he oversaw a series of massive projects from the first Egyptian dictionary, an edition of early Greek Church fathers, an edition of Mediterranean coins etc. Each project involved the collection of individual facts or data (words, coins, inscriptions), the management of dozens of scholars working in extended and highly specialized groups over years even decades, and capital, big investment by the state and eventually private foundations.

The most well-known project was Mommsen’s own Corpus Inscriptionum Latinarum, which sought to locate and then organize all inscriptions from across the entirety of the former Holy Roman Empire. It eventually collected more than 180,000 Latin inscriptions from across the former Holy Roman Empire. Eventually it grew to 17 volumes and 13 supplementary volumes. And it is still being published to this day, over 150 years after Mommsen began the project.

For many of Mommsen’s contemporaries, the Prussian Academy’s Grosswissenschaft represented a capitulation to a disenchanted, industrialized modernity. Late nineteenth-century scholars and intellectuals developed a series of tropes to name the increasingly widespread anxieties about the fate of the scholar and the future of knowledge in the modern age. Melancholy moderns from Nietzsche (Leben and Wissenschaft) and Weber (Beruf and Betrieb) to Helmut Plesner (Universitäten and Großforschungszentren) and Georg Simmel (subjective and objective) juxtaposed the authentic and meaningful with the artificial and mechanical.

And yet it was Mommsen who provided the most salient distinction, at once elegiac and candid. He distinguished between the academy’s “loyal workers” and the university’s “genius scholars” [genialien Forscher]. With their tireless organizing, the former rarely engaged in real “scholarly creation” [wissenschaftliches Schaffen]. They simply prepared material for the possibility of a future genius. Mommsen’s “loyal worker” and genius represented two different scholarly subjects or persona of knowledge based on their relationship to their scholarly objects––that is, what they actually created. Whereas the intellectual object, be it a book or a mental experience, of the “genius scholar” belongs to the individual who created it, that of the academic scholar belonged to a scholarly community and, ultimately, science. The academic scholar distributes, cultivates, and cares for the “seeds” of knowledge, hoping that they will “bear fruit in a foreign garden.”

According to its critics and defenders alike, the “Grossbetrieb der Wissenschaften” and its adoption of industrial techniques to scholarship had clear social and ethical effects: the division of intellectual and scholarly labor, the mechanization of work, overemphasis on collecting and processing material as opposed to spiritually penetrating them, and the “stultification of scholars.” Mommsen’s vision of scholarship as Großbetrieb undermined the long and, among nineteenth-century German scholars of all fields, dearly held assumption that Wissenschaft rightly practiced yielded Bildung. Represented estrangment of Bildung (subjective formation) from Wissenschaft (objective production of knowledge).

I could recount the next chapter in this story: the Morill Act of 1864, which cemented state-university-industry relationship, and the rise of American research university at the end of the nineteenth century; the post WWII multiversity and federal research: federal funding imperatives, grant system, etc. But you get my point.

For at least two centuries, universities have adopted and adapted modern processes of rationalization.

Search 2.0: Google Facts and Algorithmic Knowledge

In conclusion I want to shift to our present moment and consider how digital search engines fit into this broader story of knowledge factories and machines. In Weaving the Web, Tim Berners-Lee describes a two-part dream for the web. In the first part, he envisions the web as a means “collaboration” and as a “space” to which everyone has “immediate and intuitive access.” Such a web would allow for “people-to-people communication through shared knowledge must be possible for groups of all sizes, interacting electronically with as much ease as they do now in person.” Like Kant or Mommsen before him, Berners-Lee understands the pursuit of knowledge to be a social one. The web, like print, the research university, and the academy is a technology that binds people together BUT like any technology it binds (or sunders) in certain ways [factory metaphor helped trace epistemic as social and ethical relations]

On Berners-Lee’s vision, the web binds without hierarchy, without institutions, without interruption—it is “immediate” and “person-to-person.” Since he wrote Weaving the Web in 2001, Facebook, Twitter, Google, Apple and a few other explicitly non-collaborative, capital seeking companies have, of course, deferred Berner-Lee’s dream and the utopian hopes for a radically democratic medium. It is no longer, if it ever was, “de-centralized and localized.” The web or the internet––just like the university or the realm of print or the archive––isn’t a smooth, fully enclosed system.

In the second part of his dream, Berners-Lee imagined a semantic web in which machines analyzed “all the data on the Web – the content, links, and transactions between people and computers.” Such a system, he claimed, would enable machines to handle “the day-to-day mechanisms of trade, bureaucracy, and our daily lives” by “talking to machines, leaving humans to provide the inspiration and intuition.” The intelligent “agents” people have long anticipated will finally materialize. Such a Web would automate intellection and, more particularly, the processing of information.

This is a radically scaled iteration of previous knowledge machines: not just a factory (epistemic production, division of labor, managerial oversight) but a vision of a fully automated epistemic system.

The utopian hopes and relative failures of such Berners-Lee’s semantic web are well known. But the dream lives in at Google, where search engineers were never fully satisfied with Google Search 1.0. It was too “document-centric.” From the company’s earliest days, Sergi Brin made it clear that he wanted Google to design a search engine that wouldn’t just process and organize the world’s information but understand it. [First iterations of PageRank based on hyperlink]

In recent years, Google engineers have been working to transform Google Search from an an “information engine” to a “knowledge engine.” In a research paper published last spring, a crack team of Google engineers proposed and modeled a new search method that relies not on “exogenous signals” (links) but “endogenous” ones (facts). The team wants to evaluate websites based on the “correctness of [their] factual information” by designing an algorithmic method of extracting facts and evaluating the accuracy of websites’ facts. (One example they give is the mountain of gossip that obscured facts in the trumped-up controversy over Barack Obama’s nationality.) This algorithm will yield a trustworthiness score (Knowledge-Based-Trust, in Google-talk), in which trustworthiness is the probability that a web source contains the correct “value for a fact.”

The key point relative to the history of knowledge machines is Google’s plan to design a search engine based solely on endogenous signals. [They do this by creating massive data bases of facts. And the new KBT algorithms compare a website’s content to these databases and produce a probability score] Google Search 1.0 simply organized and collated information about information. It never purported to identify facts, adjudicate truth claims or tell us what we really want. It was profoundly uncritical, but it was based on a legitimate premise: search technologies facilitate but do not replace the messy, context-bound, all-too-human creation of knowledge. In their efforts to make search less about navigating the seas of information and more about understanding human desires and adjudicating facts, Google Search 2.0 could well render us more reliant, more passive, and more dependent on search technologies than we already are. Google’s increasing automation of search, that is, could sideline what made Google 1.0 so radical: the messy, oftentimes contingent interplay of human and machine. In its search for the perfect search engine, Google 2.0 wants to excise the human.

Automation and Knowledge Machines: We could define automation as “a self-operating machine, or a machine or control mechanism designed to follow automatically a predetermined sequence of operations, or respond to predetermined instructions”: process without a self, without a subject, without a human

 Conclusion

The problem with automation or the perfect factory or search engine is not the supposed delegation of thought and decisions to inhuman machines. It is, rather, the reification of human decisions within closed processes and predetermined sequences: be it in the mercantilist policy decisions of Enlightenment universities; the intellectual hierarchies of the modern, disciplinary research university; the management patterns and social alienation of big humanities or big science; or the racial prejudices baked into contemporary digital algorithms. The dangers of automation are all-too human.

Returning to the question of education and the future of the university, we should remember that education is not simply a process of information transfer. It is also about the formation of human beings. [Gradual impoverishment of information as concept. In its initial formulations in English from Latin infomare meaning to instruct, to infuse with form, active shaping of the world. Think Aristotlean or Thomistic notions of forms and cosmological order] Radically automated processes can erode this formative aim by displacing the end of education. The humans who design these systems are inevitably more concerned with maintaining the efficiencies of their processes than they are with the messy and contingent education of humans.

[I delivered this talk at Durham University’s Centre for Humanities Inovation, July 12.]

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s