To search is to google––to use Google’s search engine to find something on the Web. The search for meaning, love, purpose, or God––search as an existential feature of being human––has, in little more than a decade, been reduced to a secondary meaning.

[I delivered various versions of this talk this past fall and spring at Boston University, Ohio State University, and McGill University. It’s just a talk and hasn’t been published. So it’s rough. Comments welcomed and needed. I’m writing a book on the history of search.]

From its now almost apocryphal beginnings at Stanford in 1998, Google was described by its co-founders Larry Page and Sergy Brin as a technology designed to “organize the world’s information.” In Google’s first press release on June, 7 1999, Brin said that a “perfect search engine will process and understand all the information in the world.” In its first decade, Google focused on the former––processing and organizing information––by trying to map a World Wide Web that was essentially an ever-expanding and highly fragile set of documents connected by hyper-links. Google’s search engine helped people navigate the Web my modeling how webpages linked to one another (more on how exactly later). Google’s search engineers thought of the Web as a medium of documents. And its search engine was, thus, document-centric; keyword based; and highly contextual. Search results were always embedded in particular texts, documents that, once you clicked on one, framed information in a particular way.

But search engineers, at Google and elsewhere, were never been satisfied with this “document-centric” form of search. And, as Brin’s comments in that first Google press release show, they have always wanted to move beyond merely “processing” or “organizing” information. A perfect search engine, as Brin claimed, would not only process all the world’s information, it would “understand” it. And now, a little over a decade after its founding, Google engineers have publicaly started to imagine and even predict a new generation of search engines that would do just that. Google, as one of its engineers put it, is about to be transformed from an “information engine” to a “knowledge engine.”[1] The ultimate goal of these new search technologies, as singularity guru and now Google project manager Ray Kurzweil puts it, is to create search technologies that can “know at a semantically deep level” what users want without users even asking.

This brief account of Google’s first decade highlights two distinct notions of search:

1) search as navigational technology

2) search as an epistemically creative technology, one that can adjudicate, evaluate and ultimately create knowledge.

Most contemporary discussions of “search” and search technologies proceed as though Google invented search technologies designed to help people navigate and orient themselves amidst a surfeit of information––an a-historical tendency typical of our digital moment that anticipates the future as much as it ignores the past.

But our search technologies, as well as our own experiences of digital deluge and information overload, have a history [Grafton, Blair, and Zedelmaier]. And today I want to focus on one chapter in that history: eighteenth-century German search technologies or “books about books”–– books designed to help readers make their way through a perceived surfeit of print. Such history of search will:

1) historicize practices and concepts that techno-utopians, in their breathless anticipation of perfect search engines and thinking machines, might have us forget.

2) help us make sense better sense of what Google is up to in trying to transform itself from an “information engine” to a “knowledge engine.”

We may imagine ourselves to be engulfed by a flood of digital data, but late eighteenth-century German intellectuals saw themselves as having been infested by a plague of books, circulating contagiously among the reading public. To be sure the sheer number of texts in Germany did increase at a strikingly sharper rate between 1770 and 1800 than at any other time in the seventeenth and eighteenth centuries, an increase of about 150 per cent. But these fears of plagues and contagions were not merely quantitative anxieties––how many books exactly?––but more basically epistemological and ethical ones. They were anxieties about what counted as real, authoritative knowledge, when information seemed ubiquitous and easily accessible. And scholars and intellectuals dealt with these cultural anxieties, in part, by designing a range of print-based technologies and practices for accessing, organizing, and navigating a world of increasingly available print.

For much of the eighteenth century, German scholars and intellectuals associated authoritative knowledge with comprehensive or complete knowledge––what they termed Vollständigkeit. [You can hear this aspiration echoed in Google’s desire to organize all the world’s information.] This concept was central to scholarly technologies such as lexica, encyclopedias, and related “finding aids” or “books about books.” In the first half of the century, the quality of such books was judged to be a function of how “comprehensive” they were. But scholars rarely agreed on what counted as “complete” or comprehensive” knowledge and how, if it all, it was to be achieved.

And by the end of the century, many scholars began to pit “complete” knowledge against “systematic” knowledge. According to their critics, projects oriented toward “comprehensiveness” merely aggregated material stuff. “Systematic” projects, in contrast, created or represented coherent wholes. And these different conceptions of knowledge technologies raised a basic question: should they simply facilitate access to other printed texts by locating them (helping make use of them) or should they aim to constitute knowledge itself. To use the language of the eighteenth century: should they be aggregative or critical?

Today, I want to recount this 18C story of “complete” knowledge––of Vollständigkeit––and the epistemological and ethical questions it raises by focusing on a few particular eighteenth-century print projects. In conclusion, I will return to Google’s own struggles to define the purpose and shape of search.

I’ll start in early eighteenth-century Germany with one image of a “complete” account of knowledge.

Christian Jöcher and Burkhard Mencken’s Compendious Lexicon of the Learned, first published in 1715, was a lexicon of scholars and authors that, in the editors’ words, provided a “complete” and “coherent” account of learning from ancient Greece to eighteenth-century Germany.

The frontispiece (above) offers a striking visual image of a “comprehensive” organization of knowledge. Extending from the bottom of the page to the top right side is a bookcase filled with books of all shapes and sizes. Its immensity is highlighted by a marked contrast with two much smaller men standing opposite it on the left-hand side of the page. One of the two men, cast in bright light, stands with an outstretched arm gesturing toward the towering shelf of books, whose spines are visible in a single glance. A vaulted ceiling extends across the top of the page and imbues the entire image with a sense of light-filled openness in which everything is illuminated and knowable. This image is a visualization of what Jöcher and Mencken claim their lexicon to be: an accessible, complete, and clear view of the entire scholarly world. It is an image of what eighteenth-century scholars referred to as the “empire of erudition” [Reich der Gelehrsamkeit]––a unified, homogenous realm of knowledge, theoretically accessible to all fellow scholars in print. The lexicon reveals, as though it were pulling back a curtain as suggested in the top left hand corner, an already organized and established set of knowledge.

And yet, this particular image of knowledge or learning is one of book bindings, closed books shelved one beside the other. And the sheer immensity of the bookshelf, how it dwarfs the two men opposite it, is ominous. With the unlimited vertical horizon, the bookshelf could expand infinitely. With every newly published book, it could continue to grow upward, but the two men would be left to gesture in vain toward a accumulating mass of print. Furthermore, the shelf may well be visible, but there is no ladder, no way to actually reach the books. Would its shadow not ultimately obscure those two men gazing skyward? And what was to prevent the bookshelf, teetering under the weight of ever-more print, from tipping over and crushing them?

[Aside: Common trope of a bookish scholar dying in his library: Zedler’s Universal Lexicon (1746, 50: 1393) tells the story of Johannes Vossius who “finally died . . . at the age of 72 . . . who after falling from his broken book latter was completely covered with books.” The proliferation of print was an existential matter. Metaphors of contagion and disease don’t appear until the 1770’s and ’80’s.]

But if you were to actually open the Compendious Lexicon of the Learned and read it, what you would find is a mass of bibliographic details about printed texts: titles, publication dates and locations, edition information, all arranged according to ‘learned’ authors in alphabetical order. There is little to no discussion about the content of the books: just meta-data.

Like the frontispiece of the Lexicon in which only the spines of the closed, shelved volumes are visible, Jöcher’s “complete” account of learning only identifies where knowledge, complete and contained, may possibly be found. It provided an order not of knowledge per se but of print. It organized a textual inheritance. It didn’t offer a critical evaluation of the particular value of one book over another. Its principled eclecticism––mass of information jambed together––assumed a certain epistemological disposition. The purpose of search technologies was to help scholars navigate and locate texts so that they, not editors, can create knowledge and evaluate texts.

In addition to his lexicon, Jöcher also edited, from 1719 to 1739, the Deutsche Acta Eruditorum, a journal whose subtitle–“or the history of the learned, who apprehend the current state of literature in Europe”–attested to the purported unity and cohesion of erudition. Jöcher’s journal was an amalgamation of previously distinct textual genres: book catalogues, trade fair catalogues, correspondences among scholars, and review periodicals.

Together, Jöcher’s journal and lexicon assembled the elements of what he termed “book history,” an account of all knowledge available in print.[2] Both were to provide a “complete” account of the world of learning,[3] where “completeness” meant material completeness: a full account of all available knowledge on a given topic or indentification of every printed item on that topic.A “complete” account would have been a comprehensive account not of the content of books (ideas, concepts, arguments) but of the inherited textual tradition.

But Jöcher repeatedly acknowledged his failure in practice to achieve even this bibliographic completeness. Already in the third edition of his lexicon (1733), he conceded that his lexicon is “still incomplete”[noch unvollständig]. In an almost pitiful tone, he complains that he just couldn’t keep up. “Everyday,” he wrote, he happened upon new books or colleagues who would inform him of a previously unknown scholar or author. In his preface, he even listed the names of over 100 scholars about whom he had no knowledge beyond their name.

Jöcher was acutely aware of the gaps in his “complete” account. Rather than abandoning his pursuit, however, he promised that each new edition would be “more complete” and contain more and more extensive entries.[4] The future offered the promise of fulfillment––a unity deferred. And yet in the last edition that Jöcher edited in 1750, Jöcher conceded that none of the previous editions had, in fact, been “complete.”[5]

The various editions of Jöcher’s lexicon exemplified an entire culture of German learning and knowledge in the first half of the eighteenth century. The underlying epistemological premises and norms of this culture were:

• an often utopian desire to identify and locate all printed texts
• a cultural belief that print technologies afforded a homogenous, contemporaneous world of knowledge. Printed knowledge existed in the eternal presence of print (compare to more fundamentally historicist conceptions of texts and reading that emerged later in the century)
• completeness didn’t imply coherence. Almost none of these “books about books” from the first half of the 18C made claims about the internal coherence of books, how they related to each other.
• Pursuit of an order of print, not an order of cognition
• Vernacular inheritors of early modern humanists such as Konrad Gesner for whom ad fontes meant quite literarly: human knowledge is textual knowledge, bound up with the inherited literary tradition

But even this early eighteenth century technical notion of “completeness” was impossible to achieve. In a final capitulation to the futility of his pursuit of “completeness,” Jöcher changed the title of his journal from Deutsche Acta Eruditorum to Reliable News on the Current State, Transformation, and Growth of the Sciences. Because the journal’s volumes had become “too numerous and confused,” he admitted, he could no longer report on a homogenous, unified empire of erudition. The new title acknowledged that the sciences, and thus knowledge, were not static. As print expanded and changed, so too did knowledge.

Jöcher’s experience was typical of early eighteenth-century attempts to identify and organize all printed texts. Theoretical claims to material completeness and the imagination of a homogenous, contemporaneous world of print, were continually thwarted by the failure of actual projects to achieve anything like comprehensive accounts of everything in print.

By 1750 other notions of Vollständigkeit and different types of “books about books” and search technologies began to emerge. One exemplary text was Johann Sulzer’s Brief Account of All Sciences [Kurzer Begriff aller Wissenschaften]. Over the course of several editions  some of these changes print knowledge were evident.

In the first edition (1745), published almost a decade before the French Encyclopédie, Sulzer argued that an encyclopedic or complete account of learning should fully and accurately represent a natural order of the sciences. Quoting Pope, he wrote: “Order is Heaven’s great Law.”[6]] The sciences had their proper “Rang”––hierarchical order. And for Sulzer this natural order was grounded in the human faculties, namely reason [Vernunft] and sensibility or sense perception  [Sinnlichkeit]. For Sulzer, “complete” account of knowledge  accorded fully and perfectly with the order of the human mental faculties. [Think Bacon, Chambers, and D’Alembert and their trees of knowledge] But Sulzer’s conception of completeness was not directly tied to print. And you we can observe a shift from order of print to order of knowledge itself, from the fullness of the materiality of print to fullness of the human mental faculties. Sulzer wanted an account of knowledge [Erkenntnis] not, as he put it, just the “Stoff der Gelehrsamkeit.”

By the second edition (1759), however, Sulzer had traded the tree metaphor, the one based on human faculities, for a map metaphor that echoed that of Diderot and D’Alembert. Because “human knowledge constantly grows,” he wrote, the “material [Stoff] of learning is endless.” And, thus, he concluded, any efforts to fix the order of knowledge or provide a comprehensive account of it were epistemologically naïve and ethically dubious (arbitrary hierarchies). His new edition, then, would simply “mak[e] do” with eight classes: philosophy, history, arts, mathematics, physics, philosophy, law, and theology (see below).

Sulzer devoted a chapter to each category and filled it not with bibliographic meta-data but with extensive discursive descriptions of each science and its underlying concepts. And he made no attempt to relate one category to another.

For Sulzer, previous finding aids and search technologies indiscriminately collected as much information as they possibly could because the store of human knowledge had been limited. Contemporary finding aids, in contrast, had to be more selective because they were being written in a “modern age” that produced an endless stream of new knowledge.[7] Given such an unprecedented expansion of new knowledge, so visible in print, modern texts had to filter out the “vain, useless, or outrageous.” They had to forego the eclecticism that characterized early eighteenth-century projects and forge new principles and aims.

By the second edition of Brief Account of All Sciences, Sulzer had abandoned his initial attempts to provide a formal account of how the sciences and the different “parts” of knowledge related to one another. He also eschewed the bibliographic character of previous search technologies. What he was left with was the aggregation of discursive information into distinct categories, what he termed “containers.”

Sulzer’s various editions anticipated a surfeit of encyclopedic projects published in the 1780’s and 1790’s, each of which had a distinct notion of Vollständigkeit. These various projects could be conceived of as formal or material; systematic or comprehensive; critical or archival:

Material projects sought to collect and organize as much information as possible and present it in an accessible way, typically through pre-established categories.

Formal projects sought common methods or shared concepts and assumptions about what made knowledge cohere; they tried to reveal the underlying concepts and ideas that bound knowledge together, to discern the “internal relationship” of all knowledge

Many of the material projects to collapsed under their own accumulating ambitions. For example: German Encyclopedia, or Universal Dictionary of all Arts and Sciences [Deutsche Enzyklopädie, oder Allgemeines Real-Wörterbuch aller Künste und Wissenschaften] begun in 1778, collapsed in 1804 having only reached the letter “K”; Friedrich Martinis’s Universal History of Nature in Alphabetical Order managed over a period from 1774 to 1793 to produce 11 volumes that ended with the letter C. One reviewer noted in 1790 that “since the appearance of the first volume 15 years have passed and two letters have yet to be completed. If this continues, there will be over 100 volumes and the last one will appear sometime in the middle of the twentieth century.”[8]

But if the material projects collapsed beneath their excessive content, the formal ones collapsed under the proliferation of categories. They fell victim, as one eighteenth-century critic put it, to a “classificatory fury.”[9] Many of these formal projects were inspired by Kant’s so-called critical project, which one editor claimed had provided a basic set of categories that would “finally” allow for the complete organization of all knowledge.

The philosopher Wilhelm Krug, for example, began his encyclopedia with a relatively standard classificatory scheme. He continued to divide these basic categories into ever more subcategories, and subdivide and subdivide. Krug’s self-described “systematic” organization of the sciences repeated in its very form the hypertrophy of Enlightenment taxonomy projects that he mocked. Vollständigkeit was not just a question of quantitative comprehensiveness but also, perhaps even more fundamentally, cohesion.

System and Truth

Whereas the material projects (like their bibliographic precedecessors) made few claims about the internal coherence of the information or knowledge that they collected, the more formal projects sought to create, or reveal, as Krug wrote,  a “system of human” (Verusch, iii). The goal of such projects was to discover the underlying principles of knowledge, its “internal coherence”: not merely an aggregation of facts, an arbitrary alphabetical arrangement of information, or a map of institutional categories. [These were all common critiques in 1790’s.]

These projects, claimed their defenders, were concerned with knowledge itself, unbound from print––systematic, true, and pure. Their authors typically distinguished their projects from the material and the earlier bibliographic ones by appealing to some criterian of truth. Knowledge was not simply something that scholars needed to organize and represent in exhaustive acts of erudition; it was something that scholars had to evaluate and, ultimately, authorize as knowledge.

Knowledge was an honorific, a status bestowed upon information or, what German scholars called, facts. A “system,” as Krug and other post-Kantian scholars understood it, stood in for a critical standard of truth or trustworthiness. For these scholars, “system” was the opposite of Vollständigkeit.[10] Whereas Vollständigkeit stood in for quantity and material completeness, system stood in for internal coherence and “truth.” The aim of more formal projects was not simply to organize information about print and make print navigable; it was to constitute what counted as real knowledge.

The Self-Organizing Index of Print

But as the editors and authors of material and formal projects debated the proper aims of search and the purposes of knowledge technologies, the more bibliographic-oriented compendia common in the first half of the eighteenth century returned (or perhaps they had never disappeared).

As print proliferated and scholars grew ever more anxious about print plagues and floods in the last two decades of the eighteenth century, scholars and intellectuals began to suggest, sometimes implicitly sometimes explicitly, that print had begun to assume an order of its own. Even some of the highly formal, Kantian inspired projects were full of citations, discussions of editions, what they termed Bücherkunde—knowledge of books.

And it is here, in the common turn to citational networks and the indexical forms of books and periodicals, that we can discern the emergence of a different approach to the problem of information overload: the idea that print was not only something to be managed by inventive humans or ever more complex encyclopedias but that print had begun to manage itself. Scholars increasingly referred to a “world of print” endowed with self organizing capacities. [This is the crucial analogue for Google: inherent order of the Web.]

I’m using index and indexical in a much more capacious sense than might be common. [9] Indices are typically understood to be alphabetically arranged lists of topics, oftentimes located at the back of a book. These existed well before the eighteenth century and even before printed books [already in the sixteenth century in individual books or even the Index liborum prohibitorum]. An index was a navigational tool to help locate a topic within a particular book.

Over the course of the eighteenth century, however, the idea of an index took on a much broader function. Indices helped navigate not just particular books but, as one scholar put it, the entire “world of books” [Bücherwelt]. They opened up or made accessible not only individual books collected in one place (library); instead, they identified and located texts that were scattered in many different places. They offered a virtual unity. [Konrad Gesner’s Bibliotheca Universalis]. They were systems of meta-data and they represented not knowledge itself but knowledge about knowledge.

In 1811, Johann Adelung’s dictionary defined an index [Register], as a “Verzeichniß” [register] that locates the particular “page or place” where a certain name or term can be found. An index is a “Blattweiser, Blattzeiger”––a page indicator or a page pointer. An index helps readers navigate not just individual books by organizing topics––commonplaces––but by identifying particular pages within a print realm that was increasingly regarded as distinct and internally coherent.

 

Consider, for example, the development of Jöcher’s Lexicon of the Learned.

The “Aristotle” entry from the 1726 edition (above) describes Aristotle’s relationship to Plato and his role as tutor of Alexander the Great. It mentions just one print edition of Aristotle’s printed works, du Vallio’s seventeenth-century edition.

This (above) is the “Aristotle” entry from 1784 edition. Because Aristotle, as the entry reads, “ruled over human reason for almost 2,000 years …. numbers texts [zahlreichen Schriften]” have been printed, commented upon, and translated.” The remainder of the entry (two images below) lists various editions in their full bibliographic detail.

 

Knowledge was tied not to a person so much as to various forms of print.

In their Universal Subject Index (1790), Johann Heinrich Christoph Beutler and J. C. F. Gutsmuth attempted to publish an index of all articles published in German between 1700 and 1798. Like many similar publications in the 1790’s, Beutler and Gutsmuth’s text embraced what they considered to be the enlightening potential of print–its capacity to “purify and transmit scholarly knowledge broadly”–but they also acknowledged how difficult this had become.[11] “In the countless hoards of periodicals,” they wrote, “the best fruit of scientific thought” lay hidden.[12] Their solution was to produce a “repository of human reason” which sought to make the products of the human mind accessible and ready to hand.[13] Organized according to subject or topical category, each entry gave a terse account of the scholarly knowledge “stored” in printed texts.

The real innovation, however, lay in Beutler and Gutsmuth radical reduction of print to something akin to metadata. Consider the following pages and their layout and information:

[Aufklärung—

Gedanken und Fragen sie betreffend.

79. Mk. J. 79. B. 2. S. 94

“Was heißt aufklären” v. Moses Mendelssohn. B.M. J. 84. Sept. S. 193 und v. Kant. B.M. J. 84. Dez. S. 481. (Refer to Kant’s famous 1784 essay and explain the data listed in the above images.)]

Beutler and Gutsmuth identified particular topics and articles across the realm of print. And, in doing so, they treated print as a coherent, self-referential system. For them, the index achieved “Vollständigkeit” by modeling an order that was internal to the world of print itself. The citational logic is fundamentally recursive – that is, it grounds itself by an iterative process that by definition always refers back to itself. In these indexical projects, texts are not relics of their time, documents a historical past. They are “places” [Orte] of knowledge––places or spaces where knowledge can be found. Their temporal orientation was to the eternal present of print, in which knowledge didn’t developed but was fixed and ready to be engaged.

These eighteenth-century debates about comprehensive knowledge and the bibliographic, material and formal ways to achieve it (or not) give us a sense of the rich conceptual history of search. But they can also help us better understand contemporary attempts, as Google’s motto goes, “to organize the world’s information.” Like late Enlightenment indexical projects, Google’s first generation search technology didn’t “organize” or arrange the Web’s information. It modeled an order inherent to the Web itself. [What Google does would be akin to trying to follow all those citations and indexical pointings of the 18C index projects.] In their original paper outlining the “anatomy of a large scale hyper-textual Web search engine,” Page and Brin wrote that they started from the insight that the web “was loosely based on the premise of citation and annotation – after, all what is a link but a citation and what was the text describing that link but annotation.” [15] The original aim of Google, then, was to trace or model all of these links among pages, but not only the outgoing links but also their backward paths. They wanted a more complete model of the citational structure of the Web. And they did this by developing PageRank—a proprietary algorithm that models the links that constitute the Web.

But what made Google Google, as opposed to the early Yahoo or the WWW Worm, was that it did not simply count or collate citations. PageRank furthers the citational logic by not counting links from all pages equally, but by “normalizing by the number of links on a page” (109). Given the proliferation of Web pages and with them hyperlinks, Page and Brin were interested not just in the quantity but in the “quality of results” that their search engines could return. And PageRank defines the quality of a page in terms of its position within the network. What Page and Brin realized was that what made a piece of information valuable was not the “the overarching class or category that it belong[s] to, but rather the connections it ha[s] to other data”: more links from more linked pages increases a site’s PageRank score. [16]

But the radically imminent structure of the Web and PageRank’s modeling of it also means that there are gaps in Google’s “knowledge.” What the Web values is determined simply by the history of what the Web has valued. If my son’s Web page on the construction of his tree house has no incoming links, then his page, functionally speaking, does not exist. “An unindexed Internet site,” as James Gleick puts it, “is in the same limbo as a misshelved book” or, to recall our eighteenth-century examples, an un-indexed book or article. [Constant worry of eighteenth century: what if the citation is wrong? If the page or even book isn’t there? Broken links, 404 messages––assumption of a smooth, complete system]

Like its late eighteenth-century indexical predecessors, Google’s PageRank technology made no claims about the internal content of the pages it modeled. No interest in truth. Its methods were thoroughly immanent to the system it modeled. The value, authority even, of a webpage rests on its position within the network, its relationship to other pages [although it oftentimes appeals to such transcendent language to bolster its claims to authority.] It measures how well connected the Times website is, not how accurate its information is. Some gossip websites have extremely high PageRank scores, while some more accurate sites on whatever topic have low scores. Search results are, in a sense, a measure of popularity within the network and not an adjudication of truth.

As I alluded to at the beginning of this talk, however, Google seems intent on upending these notions of search. Frustrated by the “document-centric” character of PageRank, Google now wants to develop search technology that “liberates” data from documents and uses that data to create knowledge. In 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). [And this development is where the comparison to 18C print and epistemology can be helpful: 18C attempts to move from search aids and technologies that merely located information to ones that tried to constitute and create knowledge.] They want to evaluate websites based on the “correctness of [their] factual information.” They are attempting to design an algorithmic method of extracting facts and evaluating the accuracy of websites’ facts. (One example they give is Barack Obama’s nationality.) Such a process yields a trustworthiness score or, in Google talk, a Knowledge-Based-Trust (KBT): trustworthiness is the probability that a web source contains the correct “value for a fact.” There are many epistemological questions that such a project raises [what is a fact, how to you determine facts, what kind of knowledge is such fact-based knowledge etc.]

What is perhaps most telling, however, is Google’s interest in trustworthiness in the first place. As I noted earlier, Google has begun to more sharply distinguish its first generation search technology and practice that locate and model the Web’s inherent link structure, so redolent of eighteenth-century indexical print technologies, and ones that might one day, as Brin put it, “understand,” that might in some sense “know.” KBT is part of a broader shift in Google’s aspirations from mapping and processing hyperlinks to, as Ray Kurzweil puts it, “recreating intelligence,” to understanding and anticipating what humans want. Google wants to read not just the web but us. The perfect search engine wouldn’t model the web; it would model us. Google’s desire to transcend the “document-centric” web is a desire to liberate knowledge from the messy particularity and stubborn pages of texts. At its most ambitious, Google wants to overcome the Web’s relationship to print history––documents, citations, footnotes, texts.

But what kind of search would this be? Search has long had connotations of an open ended exploration that propels readers beyond themselves and their already firm desires to unknown realms that held out the promise of intellectual serendipity. Of finding a book, an idea, a concept that surprised. We don’t always know what we want. [17]

The so-called critical editors and authors of those eighteenth-century formal projects dismissed, as some Google engineers have begun to do, their predecessors as merely aggregating, collecting, and mapping mere information. But these deeply ambitious and self-consciously philosophical projects all failed—on epistemic and on more mundane grounds. By the first part of the nineteenth, they had generally disappeared, a causality of the their classificatory fury. But books like Jöcher’s Compendious Lexicon of the Learned continued to be expanded, revised, and published through the end of the 19C. It may have been pedantic, encyclopedic, un-critical––to use quote its late eighteenth-century critics––but it and similar projects were motivated by a premise that was both epistemological and ethical: search technologies should facilitate not replace the messy, context-bound, all-too-human creation of knowledge. As our own search technologies develop, we would do well to recall these lessons from the history of search.

1.  Googleblog.blogspot.com/2012/05/introducing-knowledge-graph
2. “Vorrede zu der vorigen Auflage,” in Jöcher, Compendiöses Gelehrten-Lexicon, 1:4.
3. Jöcher, Compendiöses Gelehrten-Lexicon, vol. 1.
4. Jöcher, Compendiöses Gelehrten-Lexicon, vol. 1.
5. Jöcher, “Vorrede,” Allgemeines Gelehrten-Lexicon, vol. 1..
6. Versuch einiger moralischen Betrachtungen (1745), 22.
7. Ibid., 6.
8. Allgemeine Literatur-Zeitung May 22, 1790, 409.
9. See Brad Pasanek and Chad Wellmon, “The Enlightenment Index.”
10. Maatsch, Naturgeschichte der Philsopheme, 39.
11. Kant’s definition of system: “the unity of mannigfaltigen Erkenntnisse unter einer Idee” (KrV B 860).
12. Beutler and Gutsmuth, Allgemeines Sachregister ueber die wichstigsten deutschen Zeit- und Wochenschriften, 1:iv.
13. Ibid.
14. Ibid., vii.
15. Battle, 72.
16. Zimmer, 106.
17. See, Nicholas Carr, “The Searchers.”